Cholesterol Depletion of Human Immunodeficiency Virus Type 1 and Simian Immunodeficiency Virus with {beta}-Cyclodextrin Inactivates and Permeabilizes the Virions: Evidence for Virion-Associated Lipid Rafts

Recent evidence suggests that human immunodeficiency virus type1 (HIV-1) particles assemble and bud selectively through areasin the plasma membrane of cells that are highly enriched withglycosylphosphatidylinositol-anchored proteins and cholesterol,called lipid rafts. Since cholesterol is required to maintainlipid raft structure and function, we proposed that virion-associatedcholesterol removal with the compound 2-hydroxy-propyl-ß-cyclodextrin(ß-CD) might be disruptive to HIV-1 and simian immunodeficiencyvirus (SIV). We examined the effect of ß-CD on thestructure and infectivity of cell-free virions. We found thatß-CD inactivated HIV-1 and SIV in a dose-dependentmanner and permeabilized the viral membranes, resulting in theloss of mature Gag proteins (capsid, matrix, nucleocapsid, p1,and p6) without loss of the envelope glycoproteins. SIV alsolost reverse transcriptase (RT), integrase (IN), and viral RNA.IN appeared to be only slightly diminished in HIV-1, and viralRNA, RT, matrix, and nucleocapsid proteins were retained inHIV-1 but to a much lesser degree. Host proteins located internallyin the virus (actin, moesin, and ezrin) and membrane-associatedhost proteins (major histocompatibility complex classes I andII) remained associated with the treated virions. Electron microscopyrevealed that under conditions that permeabilized the viruses,holes were present in the viral membranes and the viral corestructure was perturbed. These data provide evidence that anintact viral membrane is required to maintain mature virioncore integrity. Since the viruses were not fixed before ß-CDtreatment and intact virion particles were recovered, the datasuggest that virions may possess a protein scaffold that canmaintain overall structure despite disruptions in membrane integrity.

INTRODUCTION

Top

Introduction

Lipid rafts are organized membrane microdomains that are enrichedin sphingolipids, cholesterol, SRC family protein kinases, andglycosylphosphatidylinositol (GPI)-anchored proteins (7, 33;reviewed in reference 6). Lipid rafts are implicated as areasof the plasma membrane where human immunodeficiency virus type1 (HIV-1) assembly and budding occurs in infected cells. Immunomicroscopystudies show that the HIV-1 core protein Gag and viral envelopeproteins colocalize with lipid raft markers on the surface ofinfected cells (22, 37). Further, the GPI-anchored lipid raftmarkers CD55 and CD59 are incorporated into budding virions(12, 37), and virions exhibit high amounts of cholesterol, afinding consistent with the lipid composition of rafts (1).The initiation of budding is proposed to be driven by the multimerizationof the myristylated and positively charged Gag polyprotein ofthe virus and its interactions with the inner leaflet of theplasma membrane (31, 37, 40). The association of the assemblingvirus with lipid rafts appears to result from the fact thatmyristylated Gag is found almost exclusively in lipid rafts(37). Viral budding is then dependent on interactions of hostproteins with the PTAP motif in the P6 protein region, or thelate domain, of the Gag polyprotein precursor (reviewed in reference13).

Lipid rafts are also implicated as the area on the cell membranewhere infection is established since, in both primary cellsand cell lines, disruption of lipid raft regions by removingcholesterol with 2-hydroxy-propyl-ß-cyclodextrin (ß-CD)inhibits HIV-1-induced syncytium formation, reduces and interfereswith coreceptor expression and function (27, 39, 47), and renderscells resistant to infection with CXCR4- and CCR5-specific HIV-1strains of virus without losses in cell viability (29). Recentstudies have showed that methyl-ß-cyclodextrin depletionof virion-associated cholesterol reduced the buoyant densityof viral particles (9) and almost completely eliminated virusinfectivity, presumably by blocking virus fusion to target cells(9, 20, 57). The anti-HIV activity of ß-CD has beenevaluated in an animal model in which the compound was foundto block vaginal transmission of cell-associated HIV-1 in aSCID-HuPBL mouse model with high efficacy (24).

In the present study, we have further examined the effect ofß-CD on virus particle infectivity and integrity.Since the lipid composition of HIV-1 is high in cholesterol(1), a circumstance reminiscent of lipid rafts on cells, andsince this lipid appears to be critical in sustaining theirintegrity, we sought to determine whether complete depletionof cholesterol by ß-CD might reveal insights regardingthe presence of lipid rafts on HIV-1 and SIV particles. We thereforeexamined the effect of ß-CD on infectivity and virionintegrity in both HIV-1_MN/H9 cl.4 (HIV-1) and SIV_Mne/Hut-78cl.E11S (SIV), a virus that is well characterized and a candidatevirus for use in vaccine studies in macaques (2, 3, 10, 51).Our results reveal that depletion of cholesterol from SIV andHIV-1 results in a loss of infectivity and the permeabilizationof otherwise intact virions. Biochemical data, along with electronmicroscopy analysis, indicate that the viral membranes may containlipid rafts and may be supported by an organized internal structure.

MATERIALS AND METHODS

Top

Materials and Methods

Cells and reagents. The T-cell lymphoma lines, H9 and Hut78, were obtained fromthe American Type Culture Collection (Rockville, Md.) and maintainedin complete medium (cRPMI), consisting of RPMI 1640 (Gibco-BRL/LifeTechnologies, Gaithersburg, Md.) containing 10% fetal calf serum(HyClone, Logan, Utah) and 10 mM HEPES (pH 7.2). LuSIV cells(50) were maintained in cRPMI with 300 µM hygromycin B(Calbiochem, San Diego, Calif.). Viruses are identified accordingto the virus strain and cell line in which they were propagated(virus strain/cell line; AIDS Vaccine Program [AVP], Frederick,Md.): SIV_Mne/HuT78 cl.E11S and HIV-1_MN/H9 cl.4 are both single-cellclones produced by limiting dilution (3, 45). Pharmaceutical-grade2-hydroxy-propyl-ß-cyclodextrin (Trappsol HPB, lotnumbers 0100122F8 and 0902122HO) was obtained from CTD, Inc.(High Springs, Fla.).

Preparation and purification of ß-CD-treated HIV-1 and SIV. Viruses were purified as described previously (5). Briefly,viruses were isolated from clarified cell culture supernatantsby two successive rounds of ultracentrifugation in 25 to 50%sucrose density gradients. Virus-containing fractions were identifiedby UV absorption at 254 and 280 nm. Peak UV-absorbing fractionswere pooled, diluted to below 20% sucrose with TNE buffer (0.01M Tris-HCl [pH 7.2], 0.1 M NaCl, and 1 mM EDTA in deionizedwater), ultracentrifuged to a pellet, and resuspended in TNEbuffer. Samples were stored at -80°C. For high-pressureliquid chromatography (HPLC) experiments with fresh virus, clarifiedcell culture supernatants were used. ß-CD-treatedvirus was prepared by incubating virus at the indicated concentrationof capsid protein in the presence of the indicated concentrationof ß-CD for 1 h at 37°C with gentle mixing. Viruswas then repurified through a 20% sucrose pad by ultracentrifugationfor 1 h at 100,000 x g at 4°C in either a SW28, a SW41,or a SW55Ti rotor (Beckman, Fullerton, Calif.) depending onthe volume of sample to be processed. The resulting pellet obtainedfor each treatment condition was resuspended to the originalvolume. Treated, 1,000-fold-concentrated virus was divided intoaliquots in small volumes and stored at -70°C.

SDS-PAGE and Western blotting. For silver staining and Western blotting (immunoblotting), equalvolumes of an aliquot of the viral samples were loaded ontoseparate sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) 4 to 20% gradient gels (Novex, San Diego, Calif.)under reducing conditions. One gel was developed with the SilverXpresssilver staining kit (Invitrogen, Carlsbad, Calif.). A matchedgel was used for immunoblotting. Proteins were transferred ontoan Immobilon-P membrane (Millipore, Bedford, Mass.) and brieflydeveloped with Ponceau S to mark unstained molecular weightmarkers (Novex). Rabbit polyclonal antisera was used to detectthe following SIV proteins: gp120 (DJ-36603; AVP) and reversetranscriptase and integrase (RT [no. 366A] and IN [no. 373],respectively; both generous gifts from J. Levin). Monoclonalantibodies were used to detect HIV-1 gp120 (BF5-D5-F8) and SIVp28 (94-01P1F5), and HIV-1 p24 and (96-0020) (all from AVP).Goat anti-mouse or goat anti-rabbit horseradish peroxidase-conjugatedsecondary antibodies (Jackson Laboratories, Bar Harbor, Maine)were revealed by using enhanced chemiluminescence (AmershamLife Sciences, Arlington Heights, Ill.). Densitometry was performedfrom scanned images of immunoblot films with Scan Ace III (PacificImage Electronics) in the TIFF file format. Band intensitieswere quantified by using Scion image analysis software (ScionCorp., Frederick, Md.). The relative intensity ratios of thebands being analyzed were not altered. Appropriate exposureswere used to ensure that the measured intensities were withinthe range of the optical hardware.

Cholesterol measurement. Cholesterol was measured with an Amplex Red cholesterol kit(Molecular Probes, Eugene, Oreg.).

HPLC. Viral samples were disrupted in 8 M guanidine-HCl (Pierce, Rockford,Ill.) with or without 50 mM dithiothreitol (Calbiochem) andthen fractionated by HPLC to isolate viral proteins. HPLC wasperformed at a flow rate of 300 µl/min on a Poros R2/Hnarrow-bore column (2.1 by 100 mm; Boehringer, Mannheim, Germany)by using aqueous acetonitrile-trifluoroacetic acid solventsand a Shimadzu HPLC system equipped with LC-10AD pumps, an SCL-10Asystem controller, a CTO-10AC oven, an FRC-10A fraction collector,and an SPD-M10AV diode array detector. The gradient of bufferB (0.1% trifluoroacetic acid in acetonitrile) was as follows:10 to 36.5% for 12 min, 36.5 to 37% for 4 min, 37 to 41% for7 min, 41 to 70% for 12 min, and 70% for 5 min. A temperatureof 55°C was maintained during HPLC separation. Peaks weredetected by UV absorption at 206 and 280 nm and analyzed bysequencing by using an automated Applied Biosystems 477 proteinsequencer, by SDS-PAGE, and by immunoblot analysis with an enhancedchemiluminescence procedure (Amersham Life Sciences). Quantitationof total viral and purified proteins was performed by aminoacid analysis with a Hitachi L-8800 amino acid analyzer.

Quantitation of viral RNA. Particle-associated SIV Gag RNA was quantitated as previouslydescribed on a Prism 7700 sequence detection system (PE Biosystems,Foster City, Calif.) (30). Specimen preparation were as previouslydescribed (55), with modifications (J. D. Lifson, unpublisheddata). Particle-associated HIV-1 Gag RNA assays were carriedout by using the Roche Amplicor monitor system (detection limit,400 copies/ml; Roche Diagnostic Systems, Nutley, N.J.).

Virus titer determinations. Virus replication and titers were determined by infection ofLuSIV cells as previously described (50). Briefly, 2 x 10⁵ LuSIVcells were inoculated with 10-fold serial dilutions of the indicatedtreated virus in triplicate into a 96-deep-well plate (2.2 ml;Marsh Biomedical Products, Inc., Rochester, N.Y.), followedby incubation at 37°C. Cells were cultured in RPMI 1640with 10% heat-inactivated lipoprotein-depleted fetal bovineserum (Sigma, St. Louis, Mo.). Cells were harvested 60 h afterinfection, and luciferase expression was analyzed with a commercialluciferase assay system (Promega, Madison, Wis.). Wells exhibitingluciferase activity greater that the mean of the negative controlsplus three standard deviations were scored as positive, andthe 50% tissue culture infective dose(s) (TCID₅₀) was calculatedby the method of Reed and Muench (48).

Electron microscopy. Electron microscopy was done by M. Gignac and K. Nagashima (Laboratoryof Cell and Molecular Structure, SAIC-Frederick, NCI-FCRDC,Frederick, Md.). The treated or untreated virus was fixed with1.25% (vol/vol) glutaraldehyde in phosphate-buffered saline.Cell pellets were embedded and processed as previously described(16). Embedded virus pellets were sectioned and examined ina Hitachi H-7000 electron microscope operated at 75 kV. Virusparticle images were captured with a digital camera at 1,024x 1,024-pixel resolution and stored as TIFF files. Images wereprocessed with Adobe Photoshop (Adobe Systems, Inc., MountainView, Calif.).

RESULTS

Top

Results

ß-CD treatment of either SIV or HIV-1 results in permeabilization of virions. To test the hypothesis that cholesterol removal from cell-freevirion particles may be disruptive, purified SIV and HIV-1,fixed at concentrations of 2 and 3.5 µg/ml of capsid proteinequivalents, respectively, were treated with increasing concentrationsof ß-CD for 1 h at 37°C and repurified through20% sucrose by ultracentrifugation. Silver staining and Westernblotting of the samples was done to determine whether virion-associatedproteins were lost (Fig. 1). Permeabilization was defined asa greater decrease in the signal intensity for the capsid proteinrelative to the signal for the envelope glycoprotein (gp120).A slight loss of signal for the internal capsid protein (p28)relative to signal for gp120 occurred at 20 mM ß-CDin SIV, and p28 was clearly lost at 80 mM ß-CD (Fig.1A, C, and E). The amount of SIV membrane-associated gp120 showedonly a modest reduction through the concentration range up to80 mM but was clearly diminished at a ß-CD concentrationof 160 mM (Fig. 1C and E). In HIV-1 the ratio of p24 to gp120was fairly constant up to a ß-CD concentration of80 mM, when p24 was lost (Fig. 1B, D, and F). The HIV-1 gp120signal was relatively unchanged throughout the treatment range(0 to 160 mM ß-CD) (Fig. 1D and F). The decrease insignal intensity for gp120 in the untreated (none) versus 0mM treatment (virus incubated in buffer for 1 h and repurifiedthrough sucrose) represents virus that is lost during the repurificationprocedure (Fig. 1). To ensure that the observed effect was notmerely due to high concentrations of ß-CD but alsorepresented a reproducible effect that correlated with the ratioof cholesterol to ß-CD, we performed an experimentsimilar to the one above, except the ß-CD concentrationwas fixed at 20 mM and the virus concentration was varied (Fig.2). SIV permeabilization, as indicated by the higher loss ofcapsid protein relative to the envelope protein, was evidentby densitometry measurement of bands at 4 µg/ml of p28,but the SIV was clearly depleted of capsid protein at 2 µg/mlof p28 (Fig. 2A, C, and E). gp120 levels for SIV remained stableover the concentration of virus tested. HIV-1 permeabilization,as indicated by the greater loss of core proteins relative toenvelope protein, occurred at 4 µg/ml of p24 (Fig. 2B, D, and F).Fluctuations of signal for gp120 of up to 20% (Fig.1 and 2) were probably attributable to variability in recoveryof the viral pellet after repurification of virus through sucrose.

View larger version (67K):
[in this window]
[in a new window]
FIG. 1. Treatment of fixed concentrations of SIV and HIV-1 with increasing concentrations of ß-CD results in permeabilization of the viral membrane and loss of capsid protein. A fixed amount of either SIV (A, C, and E) or HIV-1 (B, D, and F) was treated with increasing concentrations of ß-CD for 1 h at 37°C. Samples were analyzed by SDS-PAGE and silver staining (A and B). Levels of gp120 and capsid (p28 for SIV; p24 for HIV) were determined by Western blotting (C and D), and band intensity was determined by densitometry analysis (E and F). Bars: , densitometry results for gp120; ${blacksquare}$ , densitometry results for capsid protein. None, concentrated virus stock prior to processing.

View larger version (64K):
[in this window]
[in a new window]
FIG. 2. Treatment of either SIV or HIV-1 with fixed concentrations of ß-CD and decreasing concentrations of virus resulted in permeabilization of the viral membrane and loss of capsid protein. Decreasing concentrations of capsid for either SIV (A, C, and E) or HIV-1 (B, D, and F) were treated with 20 mM ß-CD for 1 h at 37°C. Samples were analyzed by SDS-PAGE and silver staining (A and B). Levels of gp120 and capsid (p28 for SIV and p24 for HIV) were determined by immunoblotting (C and D), and the band intensity was determined by densitometry analysis (E and F). Bars: , densitometry results for gp120; ${blacksquare}$ , the densitometry results for capsid protein. None, concentrated virus stock prior to processing.

Permeabilization of virions correlates with a complete removal of cholesterol from virions, and cholesterol depletion follows a biphasic pattern. To determine the relationship between viral membrane permeabilizationand residual amounts of cholesterol in the virion, virus sampleswere assayed to measure the residual cholesterol content inviral preparations after exposure to a range of ß-CDconcentrations (Fig. 3). When the concentration of ß-CDwas varied, the measurements of residual cholesterol indicatedthat a plateau was reached for HIV-1 at between 1.25 and 5 mMß-CD and a sharp drop occurred at >10 mM ß-CD(Fig. 3A). Similarly, SIV treatment resulted in a plateau ofresidual cholesterol at concentrations between 0.625 and 5 mMß-CD (Fig. 3A), followed by a second phase of moregradual decline of cholesterol at higher concentrations. Theobserved plateau suggests that cholesterol removal from SIVand HIV-1 followed a biphasic pattern, suggesting that virionsmay have distinct pools of virion-associated cholesterol aspreviously described (17), a finding consistent with the presenceof virion-associated lipid rafts. Permeabilization, as indicatedby a loss of capsid for both HIV-1 and SIV, correlated withthe removal of cholesterol to levels below the limits of detectionof the assay used (Fig. 3A). The difference in the amounts ofcholesterol measured in the two viruses may reflect the differencein starting virus concentrations of 2 and 3.5 µg/ml ofcapsid protein for SIV and HIV-1, respectively; however, thecholesterol content varied from virus preparation to preparationand did not correlate strongly with capsid concentration, suggestingthat the cholesterol content of virions is dependent on thecell line used and the conditions of cell growth at the timeof virus harvest. Therefore, to determine whether the molarratios of ß-CD to virion-associated cholesterol determinedwhether permeabilization occurred, the initial amount of cholesterolin each virus sample was measured and the molar ratios of ß-CDto virion-associated cholesterol were calculated by dividingthe moles of ß-CD used under each condition by themoles of cholesterol present in each virus sample. Figure 3Bshows the molar ratios of ß-CD: cholesterol undereach treatment condition.

View larger version (20K):
[in this window]
[in a new window]
FIG. 3. Permeabilization correlated with total depletion of cholesterol from virus and could be predicted based on the ratio of ß-CD to cholesterol present in the sample. Residual cholesterol was measured in both SIV and HIV-1 treated with increasing concentrations of ß-CD (A). The ratio of ß-CD to the initial cholesterol level was calculated for SIV and HIV for each set of treatment conditions (B), and permeabilization levels estimated from Fig. 1 are indicated by intersecting solid lines. Error bars represent the standard deviation of the mean.

Permeabilization of both SIV and HIV-1 occurred at a molar ratioof ß-CD to cholesterol of >80. Since the cholesterolcontent of virus preparations varies, we were able, by utilizingthis method of calculating ratios, to predict when permeabilizationwould occur, and we have now reproducibly prepared multiplepreparations of permeabilized and nonpermeabilized HIV-1 andSIV.

Permeabilization results in the loss of several core proteins. To extend our understanding of the effect of ß-CDto other viral and host proteins present in the virus preparation,virions were prepared under permeabilizing (ß-CD/cholesterolratios of >300) and nonpermeabilizing conditions (ß-CD/cholesterolratios of <60) for HPLC analysis as previously described(10). Virus input was normalized to gp120 content, as determinedby SDS-PAGE, immunoblotting, and densitometry. Under nonpermeabilizingconditions, both SIV and HIV-1 virions showed a reduction inthe cholesterol peak (Fig. 4A and B). The slight reduction inthe signal for p8 in SIV under nonpermeabilizing conditions(Fig. 4A) was recovered under reducing conditions (data notshown). HIV-1 also appeared to have slight reduction in thesignal for p6 after treatment under nonpermeabilizing conditions.SIV that was prepared under permeabilizing conditions showeda marked loss of the internal viral proteins p8, p1, p6, matrix,and capsid (Fig. 4C). Viral envelope proteins gp120 and gp41were unaffected, as were the host proteins HLA class I and classII, which are incorporated into the viral membrane. Host proteinsthat are incorporated inside of the virion (actin, moesin, andezrin) were also unaffected. HIV-1 prepared under permeabilizingconditions also demonstrated a marked loss of p1, p6, and capsidand, to a lesser extent, NC and matrix proteins (Fig. 4D). Aswith SIV, host proteins incorporated into the membrane of thevirus were not affected, nor were the levels of actin. Particle-associatedgp120 levels were also not affected by ß-CD treatmentof HIV-1, as determined by immunoblotting (data not shown).The retention of NC protein in HIV-1 increased if the viruswas oxidized (data not shown) (2). Similar results were obtainedwhen freshly isolated virus was used in place of frozen viralstocks (data not shown).

View larger version (26K):
[in this window]
[in a new window]
FIG. 4. Treatment of SIV or HIV-1 under nonpermeabilizing conditions resulted in loss of cholesterol, whereas treatment of SIV or HIV-1 under permeabilizing conditions resulted in loss of proteins found inside the virion. Concentrated SIV was diluted and prepared under nonpermeabilizing (A) or permeabilizing (C) conditions and was examined by HPLC. Concentrated HIV-1 was diluted and prepared under nonpermeabilizing (B) or permeabilizing (D) conditions and examined by HPLC. The upper trace is untreated virus, and the lower trace is ß-CD-treated virus in all four panels. Loading was corrected for gp120 content. Peaks were identified by amino acid analysis and are indicated on the upper trace.

Differential retention of RT and IN in permeabilized SIV and HIV-1. Since RT and IN are critical enzymes required to establish infectionand are obscured in HPLC analysis by host proteins, Westernblotting was used to detect these proteins in permeabilizedvirus preparations. SIV and HIV-1 (same stocks as in Fig. 1)were normalized for input gp120 (by Western blotting) and probedfor RT and IN (Fig. 5). The signal for RT and IN was significantlyreduced at the 40 mM concentration of ß-CD in SIVand was below the limit of detection at the 80 and 160 mM levelsof ß-CD (Fig. 5A). In contrast, although there wasa significant reduction in the signal for RT in HIV-1 at the80 mM level of ß-CD and above, the weak signal wasretained even at 160 mM ß-CD (Fig. 5B). Integrasewas retained in HIV-1 throughout the concentration range ofß-CD in contrast to SIV (Fig. 5B).

View larger version (76K):
[in this window]
[in a new window]
FIG. 5. Differential retention of RT and IN in permeabilized SIV and HIV-1. Stocks of virus prepared for Fig. 1 were normalized to gp120 content and RT, IN, and capsid protein were examined by immunoblotting. (A) Results obtained for SIV; (B) results obtained for HIV-1.

SIV and HIV-1 lose viral RNA upon permeabilization. To determine whether viral RNA was retained in SIV and HIV-1after permeabilization by ß-CD, virions from Fig.1 were subjected to RNA analysis by RT-PCR with probes for Gag.Virion associated SIV and HIV-1 Gag RNA was measured as describedin Materials and Methods. The results obtained showed a 89 and98% reduction in the amount of viral RNA in SIV at 80 and 160mM ß-CD, respectively, showing that SIV particleslost viral RNA after permeabilization with ß-CD (Fig.6A). Viral RNA was also significantly reduced in HIV-1 uponpermeabilization but less than in SIV (68 and 87% of viral RNAat 80 and 160 mM ß-CD, respectively) (Fig. 6B).

View larger version (26K):
[in this window]
[in a new window]
FIG. 6. SIV and HIV-1 lose viral RNA upon permeabilization. Stocks of virus prepared for Fig. 1 were analyzed for particle-associated viral RNA by real-time RT-PCR. (A) Results obtained for SIV; (B) results obtained for HIV-1. Virus input was normalized to gp120 content.

Treatment of HIV-1 or SIV with ß-CD results in a dose-dependent inactivation of virus. To determine what effect cholesterol depletion and permeabilizationhad on both infectivity and viral replication, samples of virusfrom the same preparations used in the above experiments wereassayed for infectivity on an SIV luciferase tat transcomplementationreporter cell line (LuSIV) as previously described (50). WhenLuSIV cells were infected with virus treated with various concentrationsof ß-CD and assayed for luciferase expression 60 hafter virus exposure, we found a dramatic decrease in the replication(indicated by mean luciferase expression) of both SIV and HIV-1at concentrations of ß-CD of $>=$ 10 mM (Fig.7A and B). The dynamic range of the luciferase assay is suchthat weakly positive signals (>2 standard deviations abovethe negative control) do not show up as positive in graphs displayingthe data. However, TCID₅₀ calculations take into account thewells that are weakly positive, making the TCID₅₀ approach amuch more sensitive method of determining residual infectivityafter ß-CD treatment. We thus calculated the TCID₅₀values for viruses that had been treated with ß-CDand added to the reporter cells. The TCID₅₀ results show thatß-CD treatment reduces HIV and SIV infectivity byas much as 4 and 5 logs, respectively (Fig. 7C and D). The resultsshowed that infectivity is abolished in a dose-dependent mannerand is reduced to levels below the limit of detection once thevirions are permeabilized.

View larger version (27K):
[in this window]
[in a new window]
FIG. 7. Suppression of viral replication and loss of titer correlate with treatment of either SIV or HIV-1 with increasing doses of ß-CD. Stocks of virus prepared for Fig. 1 were analyzed for both replication and infectivity on a cell line (LuSIV) that expresses luciferase upon infection with either SIV or HIV-1. SIV (A and C) or HIV-1 (B and D) was treated with increasing concentrations of ß-CD (see legend in figure). Viral replication as measured by luciferase expression was then determined (A and B), and titers were calculated (C and D), under limiting dilution. Luciferase levels were measured 60 h after infection of the indicator lines. The mean luciferase value of three replicate wells is given for each dilution (A and B). The titer is expressed as the reciprocal dilution of the lowest dilution at which a positive well was scored (C and D). NP, concentrated viral stock not processed.

Permeabilized SIV or HIV-1 virions have holes in their membranes. To determine the morphology of particles recovered after ß-CDtreatment, SIV and HIV-1 were prepared under permeabilizingand nonpermeabilizing conditions as described above and thenexamined by transmission electron microscopy. SIV prepared undernonpermeabilizing conditions (Fig. 8B and E) was very similarin morphology to control virus (Fig. 8A and D). In contrast,SIV prepared under permeabilizing conditions had a markedlydifferent morphology from that of control virus (Fig. 8C and F to I).The most striking difference was that large gaps inthe membrane were evident in the permeabilized virus (see arrowsin Fig. 8C and F to I). Permeabilized virions appeared to havesingle large holes in their membranes (Fig. 8H and I). HIV-1prepared under nonpermeabilizing conditions (Fig. 9B and E)was very similar in morphology to control virus (Fig. 9A and D).In contrast, HIV-1 prepared under permeabilizing conditions(Fig. 9C and F to K), similar to SIV, had a markedly differentstructure from that of the control virus. Again, large singlegaps in the particle membrane were evident in the permeabilizedvirus (see arrows in Fig. 9C and F to K), although, rarely,a particle appeared to have two holes (Fig. 9I). In both viruses,permeabilized virions also appeared to have a less electron-densecore than control virus and the core appeared to be generallyproximal to the hole in the viral membrane (acentric). The membraneof the permeabilized virions also appeared to be of irregularthickness, compared to control virus, especially in proximityto the holes. It is important to note that although large holeswere clearly visible in the membrane of virus prepared underpermeabilizing conditions, the virions, which were not fixedbefore ß-CD treatment, were otherwise intact. Thissuggests that the viral membrane is not solely responsible formaintaining virion size, shape, and integrity as discussed below.The "cores," as seen in permeabilized viruses in Fig. 9, areoften irregularly shaped and positioned. Based on the real-timePCR data shown above, these structures may not contain RNA.Definitive identification of the core composition in permeabilizedvirus will require further study.

View larger version (177K):
[in this window]
[in a new window]
FIG. 8. Electron microscopy of SIV treated under permeabilizing conditions revealed holes in the membrane of the virus. SIV samples that were untreated (A and D), treated under nonpermeabilizing conditions (B and E), or treated under permeabilizing conditions (C and F to I) were examined by transmission electron microscopy. Panels G and I are larger magnifications of the images in panels F and H, respectively. Magnifications: x4,000 (A), x20,000 (B to F and H), and x40,000 (G and I).

View larger version (174K):
[in this window]
[in a new window]
FIG. 9. Electron microscopy of HIV-1 treated under permeabilizing conditions revealed holes in the membrane of the virus. HIV-1 samples that were untreated (A and D), treated under nonpermeabilizing conditions (B and E), or treated under permeabilizing conditions (C and F to K) were examined by transmission electron microscopy. Magnifications: x4,000 (A and F) and x20,000 (B to E and G to K).

DISCUSSION

Top

Discussion

In the present study we examined the effect of cholesterol depletionby ß-CD on the structure and infectivity of HIV-1and SIV. In agreement with other studies, HIV-1 infectivitywas decreased in a dose-dependent manner (20) and SIV was affectedin a similar manner, when the viruses were treated with ß-CD.We also found that HIV-1 and SIV viral membranes were permeabilizedwhen treated with ratios of ß-CD to cholesterol of>80. Permeabilization correlated with the removal of cholesterolfrom the virus to very low levels and caused the loss of internalviral proteins. Permeabilization of virions by nonionic detergentshas also been reported to occur within a narrow window of detergentconcentrations (60). Electron micrographs of HIV-2 preparedby fixation and treatment with 1% NP-40 show a morphology similarto that of the virions reported here, with acentric cores andholes in the membrane (46). Virion permeabilization with ß-CDwas distinct from these previous studies, since permeabilizationresulted in holes large enough passage of fairly large proteinsin addition to small ions or metabolites (38, 61). The resultsof the current study were somewhat surprising since the virionswere not fixed prior to ß-CD treatment and yet showedlarge holes, clearly indicating that membrane integrity hadbeen compromised. Presumably, the edge of the holes were stabilizedby "hairpinning" of the lipid bilayer, resulting in a lipidicpore. We cannot exclude the possibility that some impurity inthe ß-CD is responsible for the permeabilization effect;however, multiple lots were tested to ensure reproducibility(data not shown), and the ß-CD was pharmaceuticalgrade. Permeabilization correlated very well with cholesteroldepletion and virus could be permeabilized by fixing the ß-CDconcentration at 20 mM and varying the virus input. Thus, itis unlikely that permeabilization is attributable to an impurityin the ß-CD. It is possible that at the high concentrationsof ß-CD at which the holes are formed other lipids,in addition to cholesterol, may be lost, critically destabilizingthe viral membrane in local areas.

Measuring the residual cholesterol in SIV and HIV-1 exposedto increasing concentrations of ß-CD showed that thereare two pools of cholesterol in the virions. The bimodal patternof cholesterol depletion of virions is in agreement with thatobserved previously in cells and HIV-1 (17, 21, 25, 59). Thetwo pools likely represent cholesterol that is either tightlyassociated with sphingolipids and proteins (slow pool/lipidraft-associated) or is not tightly associated with other molecules(fast pool/non-raft associated) and is therefore difficult oreasy to remove, respectively (21, 59). This observation suggeststhat the membranes of HIV-1 or SIV particles are organized intolipid raft and nonraft regions similar to the plasma membranesof host cells, an idea in agreement with findings from Gordonet al., who used electron spin resonance spectroscopy (ESR)to demonstrate cholesterol-rich and cholesterol-poor regionsin the membrane of HIV-1 (17).

In the present study the infectivity of HIV-1 and SIV was reducedin dose-dependent fashion at ß-CD concentrations greaterthan 10 and 5 mM, respectively, and the loss of infectivitywas closely associated with reduction in virion cholesterol.Reduction of infectious titers to below threshold levels ofdetection correlated with virus permeabilization, suggestingthat virion infectivity is not completely lost until lipid-raftassociated (slow pool) cholesterol is removed from the viruses.The loss of viral infectivity after cholesterol depletion islikely due to a combination of factors: inihibition of fusion,loss of virion-associated cholesterol, loss of mature Gag andPol proteins, loss of viral RNA, and destruction of some virions.We have shown that infectivity of HIV treated with nonpermeabilizingconcentrations of ß-CD can be restored by replenishingviral cholesterol (unpublished results). However, we were unableto restore virus infectivity by adding back cholesterol to permeabilizedvirions (data not shown). Presumably, the large holes in theviral membrane cannot be repaired by cholesterol alone, andpatching the membrane probably cannot overcome loss of criticalviral proteins and RNA.

Wilk et al. showed that viral RNA is maintained in subviralparticles upon detergent lysis, whereas matrix and capsid moleculesare released (58). In our studies, internal viral proteins werelost in both SIV and HIV-1, with the notable retention of someNC, MA, and RT and virtually complete retention of integrasein HIV-1. The viral RNA, reduced very significantly, also appearedto be more efficiently retained in HIV-1. Since all of the retainedproteins in HIV-1 are thought to be associated with the ribonucleoproteincomplex (19), the partial retention of viral RNA in permeabilizedHIV-1 may be responsible for retention of these proteins. Theloss of RNA from permeabilized viruses raises the question ofwhat is the nature of core-like structures in these particlesrevealed by electron microscopy. Since host cytoskeletal proteinsare retained in permeabilized viruses, these proteins couldcontribute to the residual electron density. Retroviruses maybe quite heterogeneous, and it is therefore possible that permeabilizedvirions with core-like structures represent a subpopulationof treated particles that retain RNA. To minimize heterogeneity,we used molecular clones, and most permeabilized virions appearedto have irregular core-like structures. It is not clear whetherthe presence of permeabilized virions with core-like structurescorrelates with the amount of residual viral RNA retained aftervirus permeabilization. Immunoelectron microscopy and carefulparticle counting will have to be done before we can addressthis question.

The striking morphology of SIV and HIV-1 particles that werepermeabilized by ß-CD treatment in the present studyis reminiscent of p6 mutant virions that are defective in buddingand attached to cells by a thin membranous tether (15, 18).Further, the holes in ß-CD-permeabilized virions appearto be approximately the same size as the region of the virusmembrane tethered to the host cell in the aforementioned buddingmutants, as well as the region of the virus membrane that overliesthe gap in the submembrane Gag polyprotein precursor complexin immature virions (14, 15, 18). These previous observations,combined with those of the present study, raise the intriguingpossibility that HIV-1 and SIV virions carry "scission plaques."These scission plaques may represent stable specialized lipidrafts on the virus membrane that are involved in virus fissionand possibly fusion. A specialized viral membrane region thatorganizes the viral budding machinery has been suggested previously.Craven et al. (11) suggested that Rous sarcoma virus containsan organized microdomain, revealed as a prominent bleb in theenvelope of Rous sarcoma virus when the major homology regionof capsid is mutated. Consistent with the presence of a scissionplaque on HIV-1 and SIV, we generally observed a single largegap in both viral membranes. However, due to the nature of transmissionelectron microscopy, it is impossible to know for certain ifmore than one hole is present. Three-dimensional reconstructionof transmission electron microscopy images or other ultrastructuralapproaches will be necessary to clarify this important issue.Other evidence for lipid rafts on HIV-1 virion particles comesfrom electron microscopy studies showing localization of GPI-anchoredproteins CD55 (DAF) and CD59 (HRF-20) to one pole of the virions(36). A caveat is that the processing of virions in these earlierstudies may have induced capping or clustering of the moleculesas can occur in cells.

The results reported here add to the growing evidence for theexistence of an organized protein scaffold or "viroskeleton"inside of HIV-1 and SIV, associated with the membrane, thatmay be responsible for maintaining virion integrity. This proposedprotein scaffold is distinct from the RNA scaffold upon whichthe core is constructed. The host cell cytoskeleton has beenimplicated in many aspects of the HIV-1 virus life cycle, includingviral assembly (32, 35, 49), budding (49, 52), entry (23), andearly steps in reverse transcription (8). Additionally, cytoskeletalcomponents are required during transport of the preintegrationcomplex to the nucleus (19). Several host molecules, includinghost cytoskeletal and cytoskeletal binding proteins, are incorporatedinto HIV-1 and SIV upon release from the cell (42-44). Actinis a substrate for viral protease activity (35, 53, 54, 56,58) but when complexed with NC it is protected from cleavage(58). Actin or other cellular cytoskeletal components may participatein viral assembly and viral maturation and ultimately form partof the viroskeleton that maintains virion integrity. Since mostHIV-1 and SIV preparations are contaminated with microvesicles(4), which contain actin, the presence of actin in virions maybe overestimated. However, the amount of actin, moesin, andezrin attributed to the virion itself comprises ca. 10 to 15%of the total amounts of these proteins in virus preparations(43), so it is clearly possible that they could contribute tothe formation of an intravirion protein complex. The possibilitythat these molecules may be organized and contiguous with themembrane of the virion is further indicated by recent reportsthat there are interactions between the cytoplasmic tail ofgp41 and cytoskeleton proteins and cytoskeletal regulatory proteins(26, 62). This is a particularly interesting observation giventhe difference in protein and viral RNA retention between SIVand HIV-1 reported here, since the gp41 of SIV used in the presentstudy has a truncated cytoplasmic tail, whereas the cytoplasimictail of the HIV-1 gp41 is full length. We have demonstratedthat integrins incorporated by HIV-1 (41) are constitutivelyactive (unpublished results), and this observation also supportsthe presence of an organized protein complex associated withthe viral membrane. Given the involvement of lipid rafts andcytoskeleton components in regulating integrin function in cells(28, 34), integrins may be perpetually activated on HIV-1 becauseof their presence in virion lipid rafts and association withcomplexes containing host cytoskeletal components inside ofthe virions.

The observation that ß-CD-treated virus is inactivatedand permeabilized provides valuable insights for the developmentof ß-CD as a topical microbicide for AIDS prophylaxis.ß-CD has already been shown to be efficacious in aSCID-HuPBL model of vaginal HIV transmission at nonpermeabilizingconcentrations (24). Permeabilization and loss of viral coreproteins greatly enhance the microbicide potential by causingirreversible virus inactivation.

ACKNOWLEDGMENTS

We thank Michelle Gignac for excellent technical assistancewith the electron micrographs; Nicky Hoffman, Michael Piatak,and Jeff Lifson for SIV viral RNA analysis; Charla Henson forHIV viral RNA analysis; Julian Bess for providing virus forthese studies; and Mike Grimes and Bill Bohn for the productionof virus. Finally, we thank Lou Henderson, Zhaohao Liao, andDavid Ott for insightful suggestions on the manuscript.

This work was supported by grants to J.E.K.H. from the PublicHealth Service (AI31806, AI46291, and HD39613). This projectwas funded in part with funds from the NCI, under contract N01CO-12400.

FOOTNOTES

* Corresponding author. Mailing address: Johns Hopkins School of Medicine, Department of Pharmacology and Molecular Sciences, 725 N. Wolfe St., 320A Physiology Bldg., Baltimore, MD 21205. Phone: (410) 955-3017. Fax: (410) 955-1894. E-mail: jhildret{at}jhmi.edu.

REFERENCES

Top