Fusogenic Virosomes Prepared by Partitioning of Vesicular Stomatitis

Fusogenic Virosomes Prepared by Partitioning of Vesicular Stomatitis

THEJOURNAL OF B l o m ~ c u C . ~ S T R Y 0 1994 by The American Society for Bioebemistry and Molecular Biology, Inc. VOl. 269, No. 6, Issue of Febr...

2MB Sizes 0 Downloads 11 Views

Recommend Documents

Vesicular stomatitis virus - Swine Health Information Center
swine vesicular disease (SVD), and Seneca Valley virus (SVV). Two genetically distinct ... identified in white-tailed de

Recombinant Vesicular Stomatitis Viruses Gag and Env - CiteSeerX
Karl Haglund,1,2 Ingrid Leiner,3,4† Kristen Kerksiek,4‡ Linda Buonocore,1 Eric ...... Mergener, K., M. Facke, R. Wel

Prepared By Supervised By
2,105,000. 88,396,000. [30][31]. Fergana Valley Uzbekistan March 2, 1992. 285,000. 2,090,000. 87,780,000. [9]. Nowruz Fi

Benton County, Arkansas Prepared by:
Jul 14, 2012 - Bill Edwards. Mayor, City of Centerton. Delton Bush. Fire Chief, City of Centerton. Greg Neely. Fire Chie

Dynamic Partitioning of Synaptic Vesicle Pools by the SNARE-Binding
Nov 2, 2016 - Ca2 channels and SNARE protein-mediated synaptic vesicle ...... Carruthers CJ, Sutton MA (2010) Local pres

Prepared by Kimberly Vachal and Mark Berwick Prepared for: Illinois
Kimberly Vachal and. Mark Berwick ...... ports with inland river terminals on the Mississippi, Illinois, and Ohio waterw

Prepared for: Prepared by: Arcadis US, Inc. - Hollywood, FL
HIGH SERVICE PUMP STATION UPGRADES. PROJECT NO: 14-4233. Prepared by: ARCADIS. 8201 PETERS ROAD, SUITE 2400. PLANTATION,

System​​Partitioning
Use​ ​ibus​ ​first,​ ​if​ ​not,​ ​try​ ​sogou. Sogou. Download​​deb,​​install,​​andâ

SURVEY FIELD PROCEDURES MANUAL Prepared By THE
location, design, construction, maintenance and operation of engineering projects. 4. Construction Surveys - Surveys whi

'. HISTORICAL INFORMATION on REDWOOD CREEK Prepared by
Mar 1, 1994 - facility on Redwood was at Bair's ranch, which is shown on the 1909 ...... time there used to be lots of c

THEJOURNAL OF B l o m ~ c u C . ~ S T R Y 0 1994 by The American Society for Bioebemistry and Molecular Biology, Inc.

VOl.

269, No. 6, Issue of February 1

1 9

PP. 4050-4056, 1994

Printed in U S A .

Fusogenic Virosomes Preparedby Partitioning of Vesicular Stomatitis VirusG Protein Into Preformed Vesicles* (Received for publication,August 9, 1993, and in revised form, October 28, 1993)

Peter Hug and RichardG. Sleight$ From the Department of Molecular Genetics, Biochemistry, and Microbiology, University of Cincinnati College of Medicine, Cincinnati, Ohio 45267-0524

Viiosomes were prepared by the insertion of vesicular site of their intendedaction. To develop these approaches to the stomatitis virus glycoprotein, a pH-sensitive fusion propoint of actual therapeuticuse, it will be necessary to combine tein, into preformed liposomes. The fusogenic activitythem of witha method of protecting the molecules from degradathese virosomes was characterized in cell-free fusion as- tion while promoting their introduction into the cytoplasm of says using liposomal targets. Fusion was monitored by the target cells. concentration-dependent changes in the efficiency of Liposomes enter cells by endocytosis (8, 9) and are subseresonance energy transfer between N-(lissamine rhodaquently degraded in lysosomes (10, 11).It is in avoiding this mine B sulfonyl)-phosphatidylethanolamine and N44- fate thatpH-sensitive viral fusion proteins may become useful. nitrobenzo-2-oxa-l,8-diazol)-phosphatidylethanolamineEncapsulation of a therapeutic macromolecule such as DNA or and by electron microscopy. The fusogenic activity was a transcription factor within the lumenof a virosome protects dependent on the presence of vesicular stomatitis virusit from extracellular degradation by the host organism and glycoprotein, was pH-sensitive, and hadpH a threshold allows it to be released efficiently into the cytoplasm of the of activation similar to that of the native virus.The ex- target cell following virosomal binding, endocytosis, and fusion. tent of fusion was dependent upon the lipid composition We have recently reviewed the use of liposomes and virosomes of the vesicles. This technique will allow vesicles pre- as vehicles for gene delivery to eukaryotic cells (12). pared by any methodto be made fusogenic. To date, theonly methods of preparing virosomes with active fusion protein haveinvolved detergent dialysis (3, 13, 14). This approach is limited as detergent dialysis-mediated encapsulaEnveloped viruses are surrounded by a lipid bilayer and tion of macromolecules is inefficient (15).Moreover, reconstituenter cells by fusing their membrane with that of the target tion of the viralproteins with endogenous lipids does not result cell, followed by the releaseof the viral contents into the cyto- in a membrane composition that is well suited to long persistplasm. This fusion process is caused by specific proteins in the ence of the vesicle i n uiuo. However, methods of liposome prepaviral membrane. One such proteinis the 66-kDa vesicular sto- ration exist that provide efficient entrapment of macromolchoice of membrane matitis virus G (or glyco-) protein (VSV-G).l VSV enters cells ecules as well as flexibility inthe constituents (12). through theendocytic pathway. The VSV-G protein is not active A solution to this incompatibility between fusogenicity and at neutral pH but is activated in the lower pH of the endosomal compartment, at about pH 6.1 (1).The protein has been recon- efficient entrapment would be to develop a method of making stituted in active form by detergent dialysis, forming vesicles vesicles fusogenic, which is independent of their method of thataretermed “virosomes” (2, 3). These virosomes have formation. One approach to accomplishing this is to make the proved useful in analyzing the requirements andmechanistic liposome itself fusogenic. Huang and co-workers as well as other groups (16-21) using mixtures of PE and various deteraspects of viral protein-mediated fusion. Another potential use for virosomes is the microinjection of gents havedeveloped successful methods to accomplish this. At substances into cells, both in culture and in vivo. Many mac- the lowered pH of the endosome, the detergent,often oleic acid, romolecules, such as antibodies (4), DNA (51,antisense oligo- becomes more water soluble. When it leaves the membrane, the nucleotides (61, and ribozymes 171, have been proposed for PE remaining in the membranegoes through a phase transitherapeutic use. Use of these therapeutic strategies has been tion into the invertedmicellar (HII) phase and apparentlydeslowed, because these molecules do not normally cross plasma stabilizes adjacent membranes. Liposomes prepared using this membranes and therefore cannot be readily introduced to the approach are fusogenic at low pH in vitroand incell culture but lose their pH sensitivity when exposed to serum proteins, in cell culture or in vivo (20, 22). * This work was supportedby a grant from the Cystic Fibrosis Foun- In this reportwe show that VSV-G protein, whenpartitioned dation and National Institutes of Health Grant GM-39035. The costs of publication of this article were defrayed inpart by the payment of page into preformed liposomes, is fusogenic in virosome-liposome charges. This article must thereforebe hereby marked “advertisement” fusion assays. Thefusion activityof these virosomes was found in accordance with 18 U.S.C. Section 1734solely to indicate this fact. to exhibit a pH sensitivity similar to that of native virus. We $ To whom correspondence should be addressed: Dept. of Molecular of Cincinnati Col- have characterized the lipid and size requirements of the viroGenetics, Biochemistry, and Microbiology, University lege of Medicine, 231 Bethesda Ave., Cincinnati, OH 45267-0524. “el.: somes for fusion. In addition,we demonstrate thatthis method 513-558-5537; Fax: 513-558-8474. can be used to confer fusogenicity upon liposomes made by a The abbreviations used are: Chol, cholesterol; DOPA, dioleoylphos- variety of methods and lipid compositions. phatidic acid; DOPC, dioleoylphosphatidylcholine;DOPE, dioleoylphosdioleoylphosphatidylglycerol;DOPS, phatidylethanolamine; DOPG, EXPERIMENTALPROCEDURES dioleoylphosphatidylserine;N-NBD-PE,N-(4-nitrobenzo-2-oxa-l, 3-diaMaterials-Dioleoylphosphatidylcholine (DOPC), dioleoylphosphatizo1)-phosphatidylethanolamine;N-Rh-PE, N-(lissamine rhodamine B dylserine(DOPS), dioleoylphosphatidylethanolamine (DOPE), diosulfonyl)-phosphatidylethanolamine; PC, phosphatidylcholine; PE, phosphatidylethanolamine; SM, sphingomyelin; VSV, vesicular stoma- leoylphosphatidic acid (DOPA), dioleoylphosphatidylglycerol (DOPG), N-(4-nitrobenzo-2-oxa-1,3-diazol)-phosphatidyleth~olamine (N-NBDtitis virus; VSV-G, vesicular stomatitis virus G (glyco-) protein.

4050

Fusogenic Virosomes Prepared by Partition PE), and N-(lissamine rhodamine B sulfonyl)-phosphatidylethanolamine (N-Rh-PE) were purchased from Avanti Biochemicals(Pelham, AL). Cholesterol(fromporcine liver), sphingomyelin(frombovine brain), poly-L-lysine (approximate molecular weight, 87,000), sodium dithionite, octylglucoside, and Triton X-100were purchased from Sigma. 14C-Cholesterololeate was purchased from DuPont NEN. BioBeads SM-2 werepurchased from Bio-Rad.VSV (Indiana strain, ATCC VR-158) was obtained from the American Type Culture Collection (Rockville,MD). Preparation and Purification of VSV-VSV was obtained by a modification of the method of Bruns and Lehmann-Grube (23). HeLa cells in 1585 cm2 roller bottles were infected at high multiplicity of infection. The infected cells wereincubated for 24 h. Cellular debris was removed from the medium by centrifugation at 3000 x g (4,000 rpm) for 30 min in a Sorvall HS-4 rotor. The virus was pelleted by centrifugation at 30,000 x g,, (12,500 rpm) for 2 h in aBeckman SW28 rotor. The pellet was resuspended in 10 m~ HEPES, 0.9% (w/v) NaCI, pH 7.4, and purified on a 1040%(w/v)continuous sucrose gradient (buffered by 10 rn HEPES, 0.9% (w/v) NaCI, pH 7.4) by centrifugation for 19 h at 150,000 x ga, (38,000 rpm) in a Beckman SW55Ti rotor. The visible virus band was removed from the gradient, pelleted to remove the sucrose, and resuspended in 10 rn HEPES, 0.9% (w/v) NaC1, pH 7.4. After purification, the virus was stored at 4 “C. Purification of VSV-&Lipid- and detergent-free VSV-G was obtained according to the method of Petri and Wagner (24). Briefly,pure VSV was diluted with 10 m~ HEPES, 0.9% (w/v)NaCl, pH 7.4,to a total protein concentration of 0.5 mg/ml. The solution was made 30 n m in octyl glucosideby the addition of solid detergent and incubated without agitation at room temperature for 30 min. Nucleocapsids were pelleted by centrifugation at 150,000 x g . , (38,000 rpm) for 2 h in a Beckman SW55Ti rotor. The supernatant, containing detergent, lipid, and VSV-G, was layered onto a 15-30% sucrose (w/v) gradient, 60 rn octyl glucoside, 10 rn HEPES, 0.9% (w/v) NaCI, pH 7.4,and centrifuged for 24 h at 250,000 xg,, (43,000 rpm) in aBeckman SW55Ti rotor. The gradient was fractionated into 0.6-ml aliquots. Each fraction was assayed for phospholipid (25,26) and for protein by Coomassie staining of samples subjected to gel electrophoresis. Lipid-free fractions containing VSV-G and no other protein were pooled and dialyzed against threechanges of 10 rn HEPES, 0.9% (w/v) NaC1, pH 7.4, buffer in 1000-fold volume excess. The concentration of the isolated VSV-G was measured (27),and the protein was stored at 4 “C for up to 1month. Preparation of Liposomes-The lipid in all liposome preparations was dried down under N2, resuspended in ethyl acetate, redried, and vacuum dessicated for 1h. All liposomes were usedwithin 24 h of their preparation. Sonicated liposomes were prepared by resuspending the dried lipid in buffer and thensonicating to clarity with a Heat SystemsUltrasonics W385 sonicator (Farmingdale, NY). Octyl glucoside detergent dialysis liposomes were prepared by a modification of the method of Mimms et al. (28). A dried lipid film was resuspended in buffer containing 30 rn octyl glucoside.This was then dialyzed for8 hagainst 1000 volumes of buffer. Triton X-100 liposomes were prepared by the method of Patemostre et al. (31, except that no protein was present. The dried lipid film was resuspended in buffer containing 1%(w/v) Triton X-100. A 20-fold excess by weight of Bio-Beads SM-2 with respect to detergent was added, and the solution was stirred for 2 ha t 4 “C and 2 h a t room temperature. The liposomes were usedafter being separated from the Bio-Beads. Reverse-phase evaporation vesicles wereprepared by the method of Straubinger and Papahadjopoulos (29). Vesicles prepared by extrusion were made as described previously (30). Ethanolinjected liposomes were prepared as described by Kremer et al. (31). Unless otherwise specified, the liposome preparations were 1 m~ with respect to lipidphosphate and were prepared from an ethanolic solution containing 30 pmolof lipid/ml of ethanol. These conditions produce liposomes with a diameter of 100 nm (31).All lipid ratios given in this paper are on a molar basis. Preparation of Virosoms-Virosomes containing VSV-G were prepared according to the method of Petri and Wagner (32). Lipid- and detergent-free VSV-G was incubated with preformed, ethanol-injected liposomes for30 min a t 37 “C. The resulting solution of virosomes was used immediately in fusion assays. Unless otherwise specified, the VSV-G concentration was 0.1 mol % with respect to the lipid present in the virosome. The pH-sensitive nature of the fusion seen in these virosomes was extremely fragile in this system. Fusion became constitutive if 10 m~ phosphate was used as a buffer instead of HEPES. pH sensitivity, but not fusion activity, was also lost if the final concentration of the VSV-G stock after purification was above 0.015 mg/ml and was gradually lost during storage of purified VSV-G at 4 “C over a period of about 1month.

405 1

Asymmetric Labeling of Virosomes--Virosomes were labeled exclusively on the inner leaflet by the method of McIntyre and Sleight (33). This method uses dithionite to destroy the NBD label on the outer leaflet of a previously symmetrically labeled vesicle. Dithionite was added after integration of the VSV-G. Virosomes were labeled exclusively on the outer leaflet by adding 1 mol % (outer leaflet only) NNBD-PE in ethanol to virosomes already having 1mol % N-Rh-PE in both leaflets. Fusion Assays-The extent of liposome fusionwas determined by the resonance energy transfer assay originally described by Struck et al. (34). By comparing the energy transfer efficiency between N-NBD-PE and N-Rh-PE with that of a standardcurve, the absolute concentration of N-Rh-PE was determined. The extent of fusion was calculated by comparing the initial and final concentration of N-Rh-PE in the membrane. One round of fusion corresponds to the fusion expected if every virosome were to fuse with one target liposome. A 10-fold excess of target vesicles with respect to virosomes was used. Therefore, the maximum possible extent of fusion is 10 rounds. Preparation of Electronmicrographs-Samples for electronmicroscopy were fixedand stainedby a modificationof the method of Williams et al. (35).2Briefly, liposomes, virosomes,or fusion products were futed in solution for 60 min in 2.5% (w/v) glutaraldehyde, 1%(w/v) tannic acid, 0.1 M sodium phosphate, pH 7.3. They were then pelleted in an Eppendorf centrifuge for 30 min, washed, postfixed in 1%(w/v) OsO,, 0.1 M sodium phosphate, pH 7.3, for 60 min a t 4 “C, washed, and embedded in a plug of1% (w/v) ultra-low temperature-melting agarose (Sigma) to keep the pellet together. The pellet was then gradually dehydrated and embedded in IX112 for sectioning. Thin sections were stained with lead citrate and examined in a Phillips 300 electron microscope. Polyacrylamide Gel Electrophoresis-Polyacrylamidegels ofVSV and the VSV-G protein examined. in the course of purification were made according to the method of Laemmli (36). Polylysine Aggregationand Precipitation of Vesicles-Polylysine aggregation of liposomes, virosomes,and fusion products was performed by a modification of the method of Rosieret al. (37). Polylysine (approximate molecular weight,87,000) was added to a solution of vesicles at a ratio of 10 pg of polylysine/pmol of lipid. Samples were incubated a t room temperature a t pH 7.4 for 30 min and vesicles pelleted by centrifugation in an Eppendorfmicrocentrifugefor 20min. This treatment pelleted more than 99%of vesicles containing 5 mol % DOPA, while leaving more than 95% of vesicles without DOPA in solution. This was true even when both populations of vesicles were present in the same solution at the ratio used in fusion assays. Polylysine irreversibly aggregated less than 5%of vesicles composed of DOPC:cholesterol(70:30) and 0.1 mol % inserted VSV-G. RESULTS

Requirements for Fusion Time Course of Fusion-The extent of fusion between virosomes containing 0.1 mol % VSV-G with a 10-fold excess of liposomal targets is shown in Fig. 1. The membranes of both the virosomes and theliposomes contained 70 mol % phospholipid and 30 mol % cholesterol. The assays were performed at 37 “C, at pH 5.5 and pH 7.4. The timecourse was carried out to 1h to verify that fusion had progressed to its maximal extent. The fusion a t acid pH was both extensive and rapid. After 10 min of incubation, 0.97 rounds of fusion were completed. A low level of fusion also occurred in this systemat neutral pH. This residual fusogenic activitymay be dueto nonincorporated VSV-G that displayed a nonspecific membrane fusion activity or to VSV-G that partitioned into thevirosomal membrane ina nonnative conformation (see “Discussion”). The neutral pH activity was consistently present. Based on the time course of fusion in Fig. 1,endpoint assays with 30-min incubations were used to further characterize the requirements of fusion. Effect of VSV-G Concentration in Virosomal Membrane on Extent of Fusion-When VSV-G is present a t 0.1 mol %, we calculate that there are about 1000VSV-G moleculesAiposome. This is comparable with the estimated 1200 copies ofVSV-G present per native virion (38). To see if changing the ratio of S. Wert, J. Breslin, and G. Hug, personal communications.

Fusogenic Virosomes Prepared by Partition

4052

l .5 n

m

‘p

5 1.0 e .-m0

U

C

2 0.5

I

0

10

20

40 Time (minutes)

30

50

60

FIG.1. Time course of fusion. Virosomes composed of D0PC:Chol: N-NBD-PE:N-Rh-PE (68:30:1:1) and containing 0.1mol % VSV-G (with a 10-fold molar excess of respect to total lipid) were incubated with DOPC:Chol(70:30) targets at 37 “C at eitherpH 5.5 (open circles) or pH 7.4 (closed circles).Aliquots were removed at the indicated times and extent of fusion measured as described under “Experimental Procedures.” Data points are the average k S.D. of three triplicate experiments.

protein to lipid made a difference in thefusion behavior of the virosomes, the amountof protein was vaned from 0-0.5 mol % and the extent of fusion measured (Fig. 2). Although the extent of fusion after 30 min of incubation continued to increase at levels above 0.1 mol % protein, the increasewas proportionally much less.Virosomes containing 0.1mol % VSV-G with respect to lipid were used in subsequent assays. Effect of Virosomal Cholesterol Concentration on Extent of Fusion-The presence of cholesterol in the membranesaffects the level of fusion at pH 5.5 but not at pH 7.4 (Fig. 3). As the relativeamount of cholesterol inthemembranes was increased, the extent of fusion also increased. The cholesterol content of both the virosomes and the target liposomes was kept the same, so that the cholesterol content of the membranes of the fusion products does not change as the assay progresses. Because the efficiency of resonance energy transfer between N-NBD-PE and N-Rh-PE changes as a function of the cholesterol concentration, it was necessary to use a separate standard curve for each data point. pH Dependence of Fusion-Native VSV has a pH-sensitive membrane fusion activity. The threshold of activation for this fusion is pH 6.1 (1).VSV-G reconstituted intovesicles by Triton X-100 dialysis has a threshold of fusion (defined here as halfmaximal activity) of pH 6.3 when fused with Vero cells in culture (3). To determine the pH sensitivity of our preparations, The initial initial ratesof fusion at various pHs were measured. rate of fusion of virosomes made by partitioning VSV-G into preformed liposomes was measured by taking aliquots of a fusion assay and reading them various at times. Half-maximal activity was at pH 6.25 (Fig. 4). Temperature Dependence of Fusion-The change in the initial rate of fusion as a function of temperature wasexamined. Fig. 5 shows an Arrheniusplot of initial rateof fusion over the range 1 7 4 2 “C. The initial rate of fusion vanes linearly as a function of temperature between 17 and 42 “C. Effect of Lipid Composition on Fusion-The effect of different lipids on the extentof fusion was surveyed using virosomes and targets with 15 mol % of DOPS, DOPA, DOPE, DOPG, or SM (Table I). The composition of virosomes and liposomal target vesicles was kept the same. The presence of either of the negatively charged lipids, DOPS or DOPA, substantially enhanced the degree of fusion. This increase was even more pronounced in the presence of added calcium. The extent of fusion de-

0.0

0.1

0.2

0.4

0.3

VSV-G (mole

0.5

X)

FIG.2. Effect of VSV-Gconcentration in virosomal membrane on extent of fusion. Virosomes composed of D0PC:Chol:N-NBD-PE: N-Rh-PE (68:30:1:1) and containing the indicated amount ofVSV-G were incubated with a 10-fold molar excess of D0PC:Chol (70:30)targets at 37“C for 30 minat either pH5.5 (open circles) or pH 7.4 (closed circles). Extent of fusion was measuredas described under “Experimental Procedures.” Data points are the average* S.D. of three triplicate experiments.

0.0

0

10

30 40 Cholesterol (mole %)

20

50

60

FIG.3. Effect of cholesterol concentration in virosomal membrane onextent of fusion. Virosomes composed of DOPC, cholesterol, N-NBD-PE, and N-Rh-PE were prepared. The cholesterol concentration was varied between 0 and 60 mol %, and N-NBD-PE and N-Rh-PE were each present at 1mol %. The remaining fractionof the membrane was DOPC. Virosomes contained 0.1 mol % VSV-G (with respect to total lipid). Virosomes were incubated witha 10-fold molar excessof DOPC: Chol target liposomes having the same molar fraction of cholesterol as the virosomes at 37 “C for 30 min, at either pH 5.5 (open circles)or pH 7.4 (closed circles).Extent of fusion was measured as described under “Experimental Procedures.” Theefficiency of energy transfer between N-NBD-PE and N-Rh-PE was found to varyas a functionof the cholesterol concentration. Therefore, eachpoint’s extent of fusion was determined using a standard curve made with the proper concentration of cholesterol. Data points are the average S.D. of three triplicate experiments.

creased when EDTA was added instead. This may occur because calcium causes the aggregation of liposomes containing negatively charged phospholipids (391, allowing them more opportunities to fuse with one another. The presence ofEDTA abolishes this interaction.DOPE, a lipid that forms an inverted micellar (HII)phase under physiological conditions in the absence of other lipid, also increased the extentof fusion substantially. Neither calcium nor EDTA had aneffect on the fusion of DOPE-containing vesicles. DOPG increased fusionmoderately. Its level of activity was increased by calcium and decreased by EDTA. Sphingomyelin had no significant effect upon the extent of fusion. Effect of Virosomal Size on Rate of Fusion-Fig. 6 shows that the initial rate of fusion, as well as the pH dependence of the

4053

Fusogenic Virosomes Prepared by Partition

5.0

5.5

6.0

7.0 6.5

7.5

8.0

PH FIG.4. pH dependence of fusion. Virosomes composed of DOPC: Cho1:N-NBD-PE:N-Rh-PE (68:30:1:1) andcontaining 0.1 mol % of VSV-G with respect to total lipid were incubated with a 10-fold molar excess of D0PC:Chol (70:30) targets at 37 "C. Starting at time zero, aliquots were removed, and the extent of fusion measured. The assay buffer was adjustedto the proper pHbefore the virosomes were added by the addition of the appropriate amount of 1 N HCI. The pH of the assay was verified after the assay wascomplete. Extent of fusion was measured a s described under "Experimental Procedures." The initial rate of fusionwascalculated from these data. Data points are the average 2 S.D. of three triplicate experiments. -0.4

1

4

micrographs of liposomes, virosomes, and fusion products were made. These are shown in Fig. 7. Ethanol-injected liposomes of 100 nm nominal diameter (Fig. 7A) were incubated with lipidand detergent-freeVSV-G for 30 min at 37 "C (Fig. 7B). These were then incubated in a fusion assay with a 10-fold excess of target liposomes at pH 7.4 (Fig. 7C) andpH 5.5 (Fig. 70). The presence of the targets at low pH produced a heterologous population of very large, multilamellar fusion products. These vesicles were present at much lower frequency and smallersize after incubation at pH 7.4. Mixing of Inner and Outer Leaflets and Virosomal Contents During Fusion-Table I1 shows the results of a fusion assay using asymmetrically labeled virosomes. The inner and the outer leaflets mixed to the same extent during vesicle fusion. This indicates that essentiallyofall the fusion events occurring involved both leaflets and therefore resulted in the generation of a new, larger vesicle. This result suggests that reversible hemifusion (40) does not take place. Fusogenicity of Liposomes Made by Different Methods-A major benefit of this method of virosome preparation is the ability to cause liposomes made by any method to become fusogenic. Table 111 shows the resultsof fusion assays performed on virosomes prepared from liposomes made by different methods. All methods of liposome preparation so far assayedproduce vesicles that canbe made fusogenic. The differences in fusogenicity may result from differences in size (see Fig. 6) and size distribution within a preparation, aswell as residual effects of the preparation, such as trace detergent, fluorescent or probe degradation by sonication. DISCUSSION

The vesicular stomatitis virus G protein has been functionally reconstituted by several methods, all of which use detergent dialysis (2, 3,41).Although this has permitted aninspection of the fusogenic properties ofVSV-G reconstituted with endogenous lipid, these virosomes do not provide an optimal system of intracellular delivery of macromolecules, as dis-1.2 I I cussed above. Thepartitioning of VSV-G into preformed 3.1 5 3.35 3.25 3.45 vesicles overcomes these difficulties. 1 OOO/T (K) The results presented above show that virosomes made by FIG.5. Temperature dependenceof fusion. Virosomes composed partitioning lipid- and detergent-free VSV-G into preformed of D0PC:Chol:N-NBD-PE:N-Rh-PE (68:30:1:1)and containing 0.1 mol liposomes become fusogenic in a manner similar to thatof the % of VSV-G with respect to total lipid were incubated at pH 5.5 with a native virus. This method will have applications in several 10-fold molar excess of D0PC:Chol (70:30) targets at varying temperatures from 17 t o 42 "C. Starting at time 0, aliquots wereremoved every areas. First, itwill allow the dissection of the VSV fusion process to a degree heretofore impossible. Second, the fusogenic 60 s, and the extentof fusion was measured. The initial rate of fusion was calculated from these data and presented as a n Arrhenius plot. vesicles themselves will have a wide use in theintroduction of Data points are the average 2 S.D. of three independent fusion assays. foreign substancesinto cells in a controlled and innocuous initial rate, is affected by the radiusof curvature of the vesicles manner. These virosomes may be used both experimentally to involved. Vesicle sizes wereestimated by the useof dithionite to microinject soluble and membrane-bound probes into cells in determine the ratio of lipid in the inner and outer leaflets (33). culture, as well as therapeutically to add drugs and DNA to As the virosomes and targetsbecame smaller, the initial rateof cells in uiuo. Several pieces of evidence lead us to conclude that thefusion fusion increased. The decrease in vesicle size also resulted in a of virosomes made by partitioning VSV-G into preformed loss of pH-sensitivity of initial rates of fusion. vesicles is the result o f VSV-G protein activity and that it is Characterization of Virosomes and Fusion identical to theactivity of the native protein. First, the extent Fraction of Virosomes That Are Fusogenic-Fusion assays of fusion is dependent upon the concentration of VSV-Gpresent were performed in triplicateusingvirosomes composed ofDOPC: in the membrane (Fig. 2). This indicates that the virosomeChol:VSV-G:14C-cholesterololeate (70:30:0.l:trace) and lipo- liposome fusion observed here is caused by the protein. Petri somes composed of DOPAD0PC:Chol (5:65:30). After 30 min, and Wagner (32) have shown that the cytoplasmic portion of polylysine was added to aggregate those vesicles containing reconstituted VSV-G is resistant tothermolysin digestion. This DOPA, and the resulting solution centrifuged to pellet the ag- indicates that protein i s inserted into the membrane in the gregates (37). When this was done at pH 7.4, the pellet con- native orientation. tained 8.2 e 2.5% of the virosomes, while a t pH 5.5, it contained Second, the fusion is pH-sensitive. This is the salient char50 2 8.5%,indicating that about half (42 2 11.3%)the virosomes acteristic of native VSV fusion behavior. Although the pH senwere active in a pH-sensitive fashion. sitivity of the reconstituted virosomes is less absoluteand more Electron Microscopy of Liposomes, Virosomes, and Fusion labile than thatof the native virus, under the assay conditions Products-% confirm that fusion was taking place, electron used in this paper itis substantial andreproducible. Incubation

4054

Fusogenic Virosomes Prepared by Partition

TABLE I Effect of lipid composition on fusion Virosomes containing 0.1 mol % ofVSV-G with respect tototal lipid were incubated with a 10-fold mass molar excess oftarget liposomes at 37 "C for 30 min. The assay buffer was 10 l l l ~Hepes, 0.9%(w/v) NaCI, pH 7.4. Either calcium or EDTA was added to determine the effect of divalent cations upon the extent of fusion. Data points are the average t S.D. of two independent triplicate fusion assays. Fusion (rounds) Lipid composition Hepes of virosomes

D0PC:Chol (70:30) D0PS:DOPC:Chol (15:55:30) DOPAD0PC:Chol (15:55:30) D0PE:DOPC:Chol (15:55:30) D0PG:DOPC:ChoI (15:55:30) SM:DOPC:Chol (15:55:30)

+2 m Ca2+

buffer pH 7.4

0.27 0.02

+2 m

EDTA

pH 5.5

pH 7.4

pH 5.5

pH 7.4

pH 5.5

1.34 t 0.04

0.27 t 0.04

1.34 t 0.03

0.29 t 0.02

1.31 t 0.01

0.35 t 0.03 1.93

t

0.04

0.41 t 0.03

2.09 t 0.03 0.33

f

0.04

1.84 t 0.04

0.36 t 0.02

2.03 t 0.03

0.43 t 0.01

2.16 t 0.02

0.38 t 0.03

1.99 f 0.02

0.30 t 0.03

2.06 t 0.06

0.30 t 0.01

2.05 t 0.05

0.29 2 0.03

2.01

0.27 t 0.04

1.71 5 0.03 0.32

t 0.01

1.78 t 0.03

0.26 e 0.02

1.65 f 0.03

0.25 t 0.04

1.38 t 0.02

0.27 t 0.02

1.36 t0.25 0.03

t

0.03

1.27 t 0.02

0.03

that of both native VSV and otherpreviously reported methods of reconstitution. While it is at first surprising that purified VSV-G should spontaneously partition across a membrane bilayer and that these proteins should then be active, this is not the first example of this typeof event. Annexin V and synexin have been reported to undergo spontaneous translocation and partitioningintomembranesunderthe properconditions, where they form cation channels (42). Zakim and Scotto (48) have also described a generally applicable method of reconstituting integral membrane proteins into existing bilayers. To date, four proteins, bacteriorhodopsin,the GTzp form of microsomal UDP glucuronosyltransferase, mitochondrial cyto0.0 I I chrome oxidase, and the human placental insulin receptor, 0 20 40 60 80 120 100 have been reconstituted using thismethod. Vesicle Diameter (nm) There are several possible explanations for the low level of FIG.6. Effect of vesicle size on initial rate of fusion. Virosomes neutral pH fusion seen in these virosomes. First, it could be due and target liposomes were prepared by ethanol injection to the indi- to thepresence of some fraction of the VSV-G as a monomer in cated sizes as described under "Experimental Procedures." Both the of the homotrimeric fusion virosomes and target liposomes wereof the same size in this assay. the membrane rather than as part Relative size of the vesicles was assessed using dithionite to measure unit. Such a protein might be unable to protect its fusogenic the ratio of lipid in the inner leaflet to that in the outer leaflet of the moiety from exposure a t neutral pH. It could also be due to the bilayer 133). Virosomeswerecomposed of D0PC:Chol:N-NBD-PE:N- presence in the fusion system of some membrane-free VSV-G Rh-PE (68:30:1:1)and contained 0.1 mol % VSV-G with respect to total lipid. Targets were composed of D0PC:Chol (7030). Virosomes and a that did not partition intoa membrane at all and that might 10-foldexcess of targets were incubated at 37 "C.At varioustimes up to nonspecifically fuse membranes.Another possibility is that the 10 min, an aliquot was removed, and the extent of fusion was deter- virosomes are undergoing a low degree of hemifusion (40) and mined. From these data an initial rate of fusion was calculated. Data that the apparent fusion seen at neutral pH is rather due to are the mean * S.D. of three independent assays. mixing of the lipid in the outer leaflet only and not truefusion. The fusion of liposomes containing the energy transfer pairon of the virus,as well as virosomes reconstituted by Triton X-100 solubilization and Bio-Beads SM-2 dialysis, at 56 "C for 15min, only one of the two bilayers (Table 11) rules out this explanaleaflets mix in thefusion seen at irreversibly inactivates theVSV-G protein (3). This is also the tion. Both the inner and outer case with virosomes made by partition (data not shown). Fi- neutral, as well as acidic, pH. Hemifusion is also ruled out by nally, the initial rateof fusion is temperature-sensitive (Fig. 5). the presence of a small fraction of vesicles of increased size in The Arrheniusplot shows a single slope between 17 and42 "C, the electronmicrographs of the fusion assay at pH 7.4 (Fig. 7C). indicating that there is no change in .the interaction of the Whatever is causing the neutral pH membrane mixing, it is doing so by actual fusion of the virosomes with their targets. membrane and theprotein within this temperature range. Perhaps the most likely explanation is that the neutral pH Several control experiments allow us to conclude that the VSV-G is acting in its native fashion. First, polylysine-medi- fusion is an artifact of the assay system or the reconstitution ated precipitation of the products of a fusion assay show that itself. The fusion events observed here are the resultof fusion nearly half of the virosomes fused at least once with a target between virosomes and liposomes present in 10-fold molar exvirosome with an excess of targets, allof liposome. Second, fusion assays using liposomes labeled with cess. This presents the the N-Rh-PE on both leaflets, but N-NBD-PE on only one leaflet, which possess a high degree of curvature that is in opposite face of the show that both the inner and outer leaflets mix during viro- direction relative to VSVs natural target, the inner some-liposome fusion (Table 11). Finally, electronmicrographs endosome. The initial rate, and indeed the pH-sensitivity, of (Fig. 7) of liposomes and virosomes before and after fusion fusion in thissystem dependson the curvatureof the virosomes assays show the pH-dependent generation of very large fusion and their targets(Fig. 6). An extrapolation of this trend to the size regime of the endosome (0.3-1 p m ) (43) would result in products, exclusively at low pH. The results reported here indicate thatvirosomes prepared negligible amounts of neutral pH fusion. The lipid environment of the VSV-G appears to have a major by this method are fusogenic and that their activity parallels 0.3

I

Fusogenic Virosomes

Prepared by Partition

4055

%. '.

FIG.7. Electronmicrographs of liposomes, virosomes, and fusion products. Liposomes and virosomes were prepared, fixed, stained, and sectioned as described under "Experimental Procedures." The micrographs are at x 12,000 magnification. Scale bar is 2 p. A, ethanol-injected liposomes; B , virosomes (0.1 mol % VSV-G); C , virosomes and 10-fold excess of target liposomes, pH 7.4,30-min incubation at 37 "C; D,virosomes and 10-fold excess of target liposomes, pH 5.5,30-min incubation a t 37 "C. Bothpanels C and D have three distinct populations of liposomes. These are liposomal targets, unfused virosomes, and fusion products. Note the different relative abundancesof the threepopulations in panels C and D. TABLEI1 Lipid mixing of inner and outer leaflet of virosomes during fision Virosomes were prepared composed of D0PC:Chol:N-NBD-PE: N-Rh-PE (6830:l:l)and contained 0.1 mol % of VSV-G with respect to total lipid. These labeled virosomes, having N-NBD-PE present inboth inner and outer leaflets, were treated with dithionite to destroy the NBD fluorophore present in the outer leaflet (33). The resulting vimsomes have N-NBD-PE present only in the inner leaflet. Virosomes containing N-NBD-PE only in the outerleaflet were produced by first preparing virosomes without N-NBD-PE and then inserting the fluorescent lipid into the outerleaflet by addition in ethanol(33). All vimsomes werelabeled on both leaflets with1 mol % of N-Rh-PE. Virosomes were incubated with a 10-fold molar excess of D0PC:Chol (70:30) targets at 37 "C for 30 min. Data are the average * S.D. of two independent triplicate fusion assays. Location of N-NBD-PE in virosome

Both leaflets Inner leaflet Outer leaflet

Fusion (rounds) pH 7.4

pH 5.5

0.25 2 0.01 0.26 = 0.02 0.29 * 0.02

1.35 0.01 1.37 * 0.01 1.32 0.02

*

TABLEI11 Fusogenicity of liposomes prepared by diffent methods Virosomes composed of D0PC:Chol:N-NBD-PE:N-Rh-PE (68:30:1:1) and containing 0.1 mol % VSV-G (with respect to total lipid) were prepared by the indicated methods. They were incubated with a10-fold molar excess of target liposomes made by the same method having a molar composition of D0PC:Chol (70:30) at 37 "C for 30 min. Data are the average f S.D. of two independent triplicate fusion assays. Method of preparation

Sonication Extrusion Detergent dialysis Octylglucoside Triton X-100 Reverse-phase evaDoration

Fusion (rounds) pH 7.4

pH 5.5

0.76 0.04 0.27 0.03

*

1.91 2 0.03 1.18 2 0.04

0.44 2 0.04 0.30 f 0.03

1.52 f 0.05 1.54 = 0.03

0.26 * 0.02

0.99 0.04

ment infusion seen withincreasing cholesterol concentration is that at higher concentrations cholesterol may exclude VSV-G effect on its activity. The substantialeffect that cholesterol has from a fraction of the membrane,leading to a higher effective upon fusion maybe due toits stabilizing thefluid phase DOPC concentration of VSV-G in the portion of the membrane still bilayer and giving the VSV-G trimer a resistive foundation accessible to it. Table I shows the effect that a variety of lipid compositions upon which to brace itself during the fusion process. Blumenthal andco-workers (441,using both influenza and VSV (45)as have upon the extentof fusion. Both DOPS and DOPA increase models, have suggested that while only one fusion protein tri- the extentof fusion. The addition of 2 m calcium to the assay mer is involved in the membrane fusion event, surrounding increases thelevel of fusion seen stillfurther. When EDTA was VSV-G trimers may act as a scaffold to hold the fusion inter- added instead, the enhancement was substantially lowered. PS mediate inplace. A contributoryrole to this process may also be has been proposed as the targetmolecule of VSV-G (46). Either played by elements of the viral nucleocapsid. If this were so, VSV-G in the virosome is binding to PS and PA in the target cholesterol would modulate fusion only from within the viro- liposomes, or calcium-mediated interactions exclusively besoma1 membrane andwould be unimportant in the target lipo- tween virosomal and targetliposome phospholipid are acting to somes. This possibility cannot be addressed in thissystem, as enhance virosome-liposome binding and therefore allowing the effciency of energy transfer between N-NBD-PE and N- more efficient fusion. The addition of DOPE to the membranesalso enhances the Rh-PE changes at a given N-Rh-PE concentration as a function of the cholesterol concentration. This introduces the require- extent of fusion. This enhancement is not affected by either 2 ment that both membranes have identical cholesterol compo- m calcium or 2 mM EDTA. This suggests thatDOPE is acting sitions so that the extent of fusion maybe accurately measured. intrinsically upon the fusion process. Fusion of Sindbis virus Another possible explanation for at least some of the enhance- with model membranes is enhanced by the addition of PE to the

Fusogenic Prepared Virosomes

4056

by Partition

9. Nandi, P. K., Legrand, A,, and Nicolau, C. (1986) J. Biol. Chem. 261, 16722target liposome (47). I t is possible that the PE is acting inboth 16726 the Sindbis and VSV-G fusion by enhancing the formation of a n 10. Straubinger, R. M., Hong, K., Friend, D. S., and Papahadjopoulos, D. (1983) inverted micellar intermediate at the site of fusion. If thiswere Cell 32, 1069-1079 11. Cudd, A,, and Nicolau, C. (1985)Biochim. Biophys. Acta 645,477491 so, it would be interesting to see if PE enhances fusion when Hug, P., and Sleight, R. G . (1991)Biochim. Biophys. Acta 1097, 1-17 present in either membrane or if it is only necessary in the 12. 13. Gould-Fogerite, S . , and Mannino, R. J. (1985)Anal. Biochem. 148, 15-25 virosomal membrane. 14. Lapidot, M., and Loyter, A. (1989)Biochim. Biophys. Acta 980, 281-290 The pH threshold of activation for native VSV is about 6.1 15. Vainstein, A,, Razin, A,, Graessman, A,, and Loyter, A. (1983) Methods Enzymol. 101,492-512 (11,while reconstitution by dialysis from n-dodecyl octaethyl- 16. Yatvin, M. B., Kreutz, W., Horwitz, B. A,, and Shinitzky, M. (1980)Science 210, 1253-1255 ene monoether gives a threshold of 5.8 (2), and reconstitution . Sci. (I. S. A. from TritonX-100using Bio-Beads SM-2 has a threshold of 6.3 17. Connor, J., Yatvin, M. B., and Huang, L. (1984) hoc.N ~ t lAcad. 81, 1715-1718 (3).The pH at which half-maximal initial rateis observed for 18. Huang, L., Connor, J., and Wang, C. (1983) Methods Enzymol. 149,88-99 reconstitution by partitioning is 6.25(Fig. 4). This agrees well 19. Collins, D., and Huang, L. (1987) Cancer Res. 47, 735-739 Liu, D., and Huang, L. (1990) Biochim. Biophys. Acta 1022,34%354 with the valuesobserved by other groups;however, the transi- 20. 21. Collins, D., Litzinger, D.C., and Huang, L. (1990) Biochim. Biophys. Acta tion from the neutral to the acidic level of activity is more 1026,234-242 gradual withvirosomes formed by partition thanfor the native 22. Liu, D., Zhou, F., and Huang, L. (1989)Biochem. Biophys. Res. Commun. 163, 32G333 virus. This maybe due to the factthat the VSV-G in thevirus 23. Bruns, M., and Lehmann-Grube, F. (1984) virology 1 3 7 , 4 9 5 7 is much more highly organized at the supramolecular level, 24. Petri, W. A,, Jr., and Wagner, R. R. (1979) J. Biol. Chem. 254,43134316 allowing cooperativity between the VSV-G trimers to enhance 25. Bligh, E. G., and Dyer, W. J. (1959) Can. J . Biochem. Physiol. 37,911-917 26. Rouser, B., Siakotos, A,, and Fleischer, S . (1966) Lipids 1, 8 S 8 6 fusogenicity. Alternatively, the VSV-G and viral matrix protein 27. Bradford, M.M. (1976) Anal. Biochem. 72,248-254 may interact. It was not possible to compare absolute levels of 28. Mimms, L. T., Zampighi, G., Nozaki, Y., Tanford, C., and Reynolds, J. A. (1981) Biochemistry 20,833440 fusion with those seenby other groups, as they calculated per29. Straubinger, R. M., and Papahadjopoulos, D. (1983) Methods Enzymol. 101, cent dequenching of a fluorescent probe, rather than actual 512-527 30. Hope, M. J., Bally, M. B., Webb, G . , and Cullis, P. R. (1985)Biochim. Biophys. fusion events. Acta 8 1 2 , 5 5 4 5 The ability to produce fusogenic virosomes from liposomes 31. Kremer, J. M. H., Esker, M. W. J.v.d., Pathmamanoharan, C., and Wiersema, made by any method representsa major advance in the potenP. H. (1977) Biochemistry 16,3932-3935 32. Petri, W. A,, Jr., and Wagner, R. R. (1980) Virology 107,543-547 tial uses of these vesicles. Many potential applications have 33. McIntyre, J. C., and Sleight, R. G. (1991) Biochemistry 30, 11819-11827 been stymied because previously the only way to make a fuso- 34. Struck, D. K., Hoekstra, D., and Pagano, R. E. (1981) Biochemistry 20,4093genic virosome was by detergent dialysis. This places severe 4099 limitations on the size of the virosome, as well as on its encap- 35. Williams, M. C., Hagwood, S., and Hamilton, R. L.(1991)Am. J . Respir Cell. Mol. Biol. 5, 41-50 sulation volume. Allowing the methodof preparation tobe cho- 36. Laemmli, U.K. (1970) Nature 227,680-685 sen independently of the requirementfor fusogenicity permits 37. Rosier, R. N., Gunter,T.E., Tucker, P. A., and Gunter, K. IC (1979) Anal. Biochem. 96,384-390 the useof efficient encapsulation techniques such as extrusion,38. Thomas, D., Newcomb, W.W., Brown, J. C., Wall, J.S., Hainfeld, J. F., Trus, B. reverse phase extrusion, and interdigitation fusion to be used. L., and Steven, A. C. (1985) J . vim1 54,598-607

Acknowledgments-We thank Joan Breslin, Gail Chuck, George Hug, Esther Tombragel, and Susan Wert for help in the preparation of the electronmicrographs. REFERENCES 1. White, J., Matlin, IC, and Helenius, A. (1981) J . Cell Biol. 89, 674479 2. Metsikko, K., van Meer, G., and Simmons, K. (1986)EMBO J. 5, 3429-3435 3. Patemostre, M., Lowy, R. J., and Blumenthal, R. (1989) FEBS Lett. 243, 251-258 4. Morgan, D. O., and Roth, R. A. (1988) Immunol. Today 9, 84-86 5. Friedman, T.(1989) Science 244, 1275-1281 6. Zon, G . (1988)Pharmacol. Res. 5,539-549 7. Haseloff, J.,and Gerlach, W. L. (1988) Nature 334, 585-591 8. Magee, W. E., Goff, C. W., Schoknecht, J., Smith, D. M., and Cherian,K. (1974) J . Cell Bid. 63, 492404

39. Walter, A., and Siegel, D. P. (1993) Biochemistry 32,32714281 40. White, J. M., Bodian, D. L., Kemble, G. W., and Kuntz, I. D. (1992) J. Cell. Biochem. 16C, (suppl.) 111 41. Eidelman, O., Schlegel, R., Tralka, T. S., and Blumenthal, R. (1984) J . Biol. Chem. 259,46224628 42. Pollard, H. B., Guy, H. R., Arispe, N., de la Fuente, M., Lee, G., Rojas, E. M., Pollard, J. R., Srivastava, M., Zhang-Keck, Z., Merezhinskay, N., Caohuy, H., Burns, A. L., and Rojas, E. (1992) Biophys. J. 62, 1 5 1 8 43. Helenius, A,, Mellman, I., Wall, D., and Hubbard, A. (1983) “hpnds Biochem. Sci. 8,245-250 44. Guy, H. R., Durell, S. R., Schoch, C., and Blumenthal,R. (1992)Biophys. J. 62, 113-115 45. Blumenthal, R. (1988) Cell Biophys. 12,l-12 46. Schlegel, R., Tralka, T. S., Willingham, M. C., and Pastan, I. (1983) Cell 32, 639-646 47. Scheule, R. K. (1987) Biochim. Biophys. Acta 899, 185-195 48. Zakim, D., and Scotti, A. W. (1989) Methods Enzymol. 171,253-264