Energy transfer measurements of fusion between Sendai virus and vesicles corrected for decreased absorption of acceptor probe.

The fusion of Sendai virus at pH 4-7 with artificial lipid vesicles composed of phosphatidylserine or phosphatidylcholine was quantified by measuring fluorescence energy transfer from N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)-phosphatidylethanolamine to N-(lissamine-rhodamine-B-sulfonyl)-phosphatidylethanolamine in the target membranes. About 60% of the phosphatidylserine vesicles and virus appeared to fuse at pH 4 and about 100% at pH 5. Fusion was much less under all other conditions. The apparent fusion at pH 4, however, was due to a decrease in absorption of the acceptor probe, instead of dilution of acceptor as a result of fusion of labeled vesicles with unlabeled virus. After correction for this fusion-independent effect of Sendai virus, the extent of fusion was only 4-20% at pH 4 but still 80-100% at pH 5. These findings paralleled the loss of hemagglutinating and hemolytic activities of the virus induced by incubation at pH 4 but not at pH 5. Vesicle-virus hybrids were observed with the electron microscope after incubation at pH 5 but not at pH 7. The assay of membrane fusion by fluorescence energy transfer can be misleading unless correction is made for changes in energy transfer due to fusion-independent effects.

The measurement of changes in the fluorescence of amphiphilic probes on fusion of the probe-containing membranes is now one of the standard ways of quantifying membrane fusion (Blumenthal, 1987;Duzgunes and Bentz (1987)), including the fusion of membranes in the presence of proteins Chejanovsky and Loyter, 1985;Chejanovsky et al., 1986;Driessen et al., 1985;Eidelman et al., 1984;Hoekstra, et al. 1984;Stegmann et al., 1985;van Meer et al., 1985;Wharton et al., 1986). This widespread use of fluorescence assays attests to their reliability, although a few irregularities have been observed with methods relying on energy transfer (Driessen et al., 1985;Morris et al., 1985;Uster and Deamer, 1981;Wharton et al., 1986). Of relevance to the present work was the recent finding that membrane-depleted influenza hemagglutinin reduced energy transfer between N-NBD-PE' and N-Rh-PE in target vesicles, so that spurious fusion was * This research was supported by National Institutes of Health Grant AI20421. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
As reported here, a similar fusion-independent change in energy transfer was suspected initially when the assay of PS vesicles indicated their fusion with Sendai virus at pH 4 and at 5, since the virus lost its hemagglutinating and hemolytic activities when incubated at pH 4 but not at pH 5. Decreased absorption of acceptor N-Rh-PE in vesicles incubated at pH 4 with Sendai virus showed subsequently that the decrease in energy transfer, by which fusion was quantified, was not entirely due to fusion-dependent dilution of the acceptor probe originally in the target vesicle membrane. After appropriate corrections were made for fusion-independent changes in probe fluorescence, it became clear that Sendai virus does fuse with PS vesicles at pH 5 but not at pH 4. Hence, fusion between Sendai virus and PS vesicles can be assayed by the degree of fluorescence energy transfer between N-NBD-PE and N-Rh-PE despite fusion-independent effects on probe fluorescence, provided energy transfer data are corrected for the latter effects.

MATERIALS AND METHODS
Chemicals-Egg PC, bovine brain PS, egg phosphatidylethanolamine, egg sphingomyelin, N-NBD-PE and N-Rh-PE were from Avanti Polar Lipids, Inc. Py-PC was from Molecular Probes, Inc. These lipids were analyzed for purity by thin layer chromatography and assayed by a modification of the Bartlett procedure (Bartlett, 1959). Only lots of N-Rh-PE free of isomers with pH-dependent quantum yields were used (van Meer et al., 1985). Cholesterol was from Sigma.
Virus-Sendai virus was grown in chicken eggs as described (MacDonald, 1986) and stored at -70 "C. Virus protein was measured (Markwell et al., 1978) and total viral lipid calculated to be 400 nmol/ mg of viral protein (Hoekstra et al., 1984).
Vesicle Preparation and Incubation with Virus-Mixtures of lipids in chloroform-methanol were dried under high vacuum for 1 h, hydrated in 100 mM NaCl + 10 mM MOPS, pH 7.2, +1 mM EDTA + 3 mM NaN3 at 37 "C for 30 min and sonicated in a bath-type apparatus under nitrogen. Vesicles were oligolamellar and ranged in diameter from 40-400 nm. 10-pl aliquots of Sendai virus containing 11 nmol of lipid were incubated for 5 min at 37 "C in 20 pl of 100 mM sodium acetate, pH 4.2 to 6.4, or 100 mM NaCl + 10 mM MOPS, pH 7.2, as well as 1 mM EDTA + 3 mM NaN3, prior to the addition of 2 pl of vesicles containing 12 nmol of lipid. The pH given under "Results" refers to the pH of the virus-buffer-vesicle mixture determined by measuring the pH of a large volume of such a mixture. After 10 min at 37 "C, 370 pl of 100 mM NaCl + 10 mM Tris-C1, pH 8, or 10 mM MOPS, pH 7.2, as well as 1 mM EDTA + 3 mM NaN3, were added to samples at pH 4.2 or at pH 5.1-7.2, respectively, to raise the pH to 7.0-7.2 prior to measurement of fluorescence. To minimize dilution of the samples for electron microscopy, the pH was adjusted with acetic acid and NaOH.
Fluorescence Measurement-Vesicles containing 0.5 mol % Py-PC, 1.0 mol % N-NBD-PE and/or 1.0 mol % N-Rh-PE were incubated with and without Sendai virus at pH 4-7. Fluorescence was proportional to concentration at these probe levels. Samples were excited at 350 nm and scanned from 360 to 500 nm to measure Py-PC fluorescence (peak emission = 380 nm), excited at 475 nm and scanned from 500 to 550 nm to measure N-NBD-PE fluorescence (peak emission = 530 nm), excited at 570 nm and read at 590 nm to measure N-Rh-PE fluorescence in a Farrand spectrofluorometer with excitation and emission band widths of 5 nm. Light scattering due to Sendai virus was subtracted from all samples. Each sample was read in the absence and presence of 1% Triton X-100 which gave a 100% value of unquenched N-NBD-PE fluorescence after allowing for quenching by Triton itself.
Fluorescence Energy Transfer Assay-This assay is based on the quenching of a donor, e.g. N-NBD-PE, by an acceptor, e.g. N-Rh-PE, in the same membrane through energy transfer, since their emission and excitation spectra, respectively, overlap. Because donor quenching or energy transfer depends on the acceptor concentration (Fung and Stryer, 1978), donor fluorescence can reflect the degree of acceptor dilution resulting from mixing of acceptor-containing membranes with acceptor-free membranes (Struck et al., 1981). Energy transfer is related to donor fluorescence according to the equation, E = 1 -FIFO, F being the donor fluorescence in the presence of acceptor and Fo being the donor fluorescence in the absence of acceptor (Fung and Stryer, 1978). According to a standard procedure Struck et al., 1981, Uster andDeamer, 1981), a calibration curve was constructed to assign a percent lipid-mixing value to the FIFO value of an unknown sample by plotting the percent lipid mixing versus F/Fo values of lipid vesicles "mock-fused in varying proportions (ix. PS or PC vesicles labeled with both N-NBD-PE and N-Rh-PE (for F) or with N-NBD-PE alone (for Fo) and equal amounts of unlabeled vesicles of a lipid composition similar to Sendai virus membranes, 18 mol % phosphatidylethanolamine, 8 mol % PS, 16 mol % PC, 14 mol % sphingomyelin, and 44 mol % cholesterol (Quigley et al., 1972). The F/Fo versus percent lipid-mixing standard curves were not affected by pH but differed slightly if PS or PC were the labeled lipid.
Difference Spectra of Absorbance-Samples identical, except for their increased volume, to those prepared for fluorescence measurements were scanned from 700 to 400 nm in a Shimadzu double beam spectrophotometer with opal filters to minimize the effect of light scattering. The absorbance of N-Rh-PE in PS vesicles was obtained by scanning vesicles labeled with N-NBD-PE and N-Rh-PE in the presence or absence of Sendai virus in the sample cuvette and vesicles labeled with N-NBD-PE alone in the presence or absence of Sendai virus in the reference cuvette.
Electron Microscopy-Samples were applied to collodion-coated copper grids, stained with 2% sodium phosphotungstate, pH 7, and viewed in a JEOL electron microscope at 48,000 X magnification.
Hemolytic Activity of Sendai Virus-The hemolytic activity of Sendai virus, which is a measure of its fusion activity, was assayed as described (Kundrot et al., 1983) after preincubation of the virus from pH 4 to 7 in the absence of red cells.

lB, open bars, shows that lipid mixing between Sendai virus
and N-NBD-PE + N-Rh-PE labeled PS vesicles after 10 min of incubation at 37 "C was a maximal 100% at pH 5.2 and 60% at pH 4.1. Lipid mixing did not occur at these pH values when unlabeled vesicles were substituted for Sendai virus (not shown). At pH 6.4 and at pH 7.2, lipid mixing was 25% or less, a relatively low level also found with Sendai virus and PC vesicles from pH 4 to 7 (Fig. lA, open bars). The marked susceptibility of PS vesicles to fusion with Sendai virus at low pH was surprising since Sendai virus fuses over a broad pH range (i.e. pH 5-8) with cell membranes (Lenard and Miller, 1981;White et al., 1983), of which the acidic lipid content is only 10 mol % (van Deenen and de Gier, 1974). Particularly suspect was the apparent fusion with PS vesicles at pH 4, since incubation of Sendai virus at pH 4 for as brief a period as 1 min reduced its hemolytic activity t o 30.8% of the original. The hemolytic activities of Sendai virus incubated for 20 min at 37 "C at pH 4,5, and 7 were 0, 75.8, and loo%, respectively, of the original activity. Hemagglutinating activity was also destroyed at pH 4, as virus incubated for 5 min at pH 5 and at pH 7 contained 30,800 hemagglutination units/ml, whereas virus incubated for 5 min at pH 4 contained 0 hemagglutination units/ml.

Fusion-independent Relief by Sendai Virus of N-NBD-PE
Quenching-To determine whether Sendai virus could affect the fluorescence of either the donor and/or acceptor probes without inducing virus-target fusion, the fluorescence of each probe alone in PS or PC vesicles was measured after incubation with or without Sendai virus. The effect of Sendai virus on the fluorescence of Py-PC-containing vesicles was also measured to determine whether the virus could affect the fluorescence of a probe bearing the fluorescent group in its hydrophobic region as well as a probe bearing the fluorescent group in its hydrophilic region. Table I shows that Sendai virus reduces the fluorescence of N-Rh-PE in PS vesicles, significantly at pH 4 (to 68%) and somewhat at pH 5 (to 80%), and of N-NBD-PE and Py-PC in PS vesicles slightly at pH 4 (to 90 and to 88%, respectively), and enhances the fluorescence of N-NBD-PE in PS vesicles slightly at pH 6 and 7 (to 113 and to 110%, respectively). In contrast with PS vesicles, PC vesicles incubated with Sendai virus displayed a slightly enhanced fluorescence of all three probes as the pH was raised from 4 to 7 . Thus, some or all of the increased N-NBD-PE fluorescence attributed to virus PS vesicle fusion in Fig. 1 B,   N-NBD-PE or Py-PC The ratio of the fluorescence of N-Rh-PE, N-NBD-PE or Py-PC in PS or PC vesicles incubated with Sendai virus at pH 4.1-7.2 divided by the fluorescence of N-Rh-PE, N-NBD-PE or Py-PC in PS or PC vesicles incubated without Sendai virus at the same pH values (X 100) was determined as described under "Experimental Procedures." The protocol was the same as followed for Fig. 1 Table I could be due either to a diminished excitability of some acceptor molecules and/or to a failure of those acceptor molecules to emit light on excitation.

Correction of Energy Transfer Measurements of Virus Fusion
If the fluorescence reduction were due to a reduced excitability, this fusionindependent effect should be correctable by extrapolating the degree of lipid mixing from a percent lipid mixing uersus F/ Fo curve, based on the availability of N-Rh-PE for energy transfer and not on the N-Rh-PE physically present in the membrane. To ascertain diminished excitability of N-Rh-PE, difference absorption spectra of N-Rh-PE in PS vesicles were obtained as a function of pH and incubation with Sendai virus. Samples were identical to those prepared for the assay of fusion except for their larger volumes. The sample cuvette contained PS vesicles labeled with N-NBD-PE + N-Rh-PE and incubated with or without Sendai virus at the appropriate pH, and the reference cuvette contained PS vesicles labeled with N-NBD-PE alone and treated in the same way as those in the sample cuvette. Fig. 2 shows the absorption spectra of N-Rh-PE in PS or PC vesicles incubated with or without Sendai virus at pH 4, 5 , or 7 and the percent change in absorbance due to the presence of Sendai virus. The absorbance of N-Rh-PE in PS vesicles incubated with Sendai virus is clearly lower than that without Sendai virus at p H 4 (average 63.6 f 10.7%) but not at pH 5 (average 91.8 f 8.4%), at pH 7 (average 90.1 f 8.7%) and in PC vesicles at pH 4 (107.3%). Because Sendai virus incubated at pH 4 looked unusually turbid, N-NBD-PE + N-Rh-PE-labeled vesicles incubated alone at pH 4 were mixed after neutralization and dilution of all samples either with buffer or with Sendai virus incubated alone at pH 4. The N-Rh-PE absorbance of these two mixtures were the same (not shown). Hence, the reduced absorbance of N-Rh-PE in PS vesicles incubated with Sendai virus at pH 4 was not due to light scattering by the pH 4-treated Sendai virus.
True Fusion of Sendai Virus and Lipid Vesicles-Since Fig.  2 shows that the absorption of N-Rh-PE is reduced in PS vesicles incubated with Sendai virus at p H 4, it was necessary to re-evaluate fusion which had been determined under the apparently incorrect assumption that the absorption efficiency of 1 mol % N-Rh-PE was the same under all experimental conditions (Fig. 1, A and B, open bars). T o correct for reduced absorption of N-Rh-PE, a curve was constructed in which the N-Rh-PE fluorescence of vesicles containing various concentrations of N-Rh-PE and 1 mol % N-NBD-PE was plotted against the corresponding FIFO of N-NBD-PE. The resulting curve (Fig. 1C) resembles that reported by Struck et al. (1981). Because energy transfer resulting from lipid mixing is a function of the concentration of N-Rh-PE in the membrane (Fung and Stryer, 1978), the curve in Fig.  1C could be used to assign a corrected percent lipid-mixing value to experimentally obtained FIFO values. The FIFO value equal to true 0% lipid mixing on the curve corresponds to the concentration of N-Rh-PE commensurate with its absorption or fluorescence, not its actual concentration of 1 mol %. The effective N-Rh-PE concentration was calculated either as (1 mol % N-Rh-PE x (N-Rh-PE fluorescence in the presence of virus/N-Rh-PE fluorescence in the absence of virus) or as (1 mol % N-Rh-PE X (N-Rh-PE absorbance in the presence of virus/N-Rh-PE absorbance in the absence of virus)). The N-Rh-PE fluorescence values were from Table I and  Electron Microscopy Corroborates Preceding Results-Corroboration of fusion between P S vesicles and Sendai virus at pH 5 but not a t 7 was obtained by electron microscopy. Fig.  3A is a representative electron micrograph of Sendai virus and PS vesicles incubated for 10 min at 37 "C at pH 7 at a magnification of X48,OOO. The arrow points to a nucleoprotein-containing virus particle which is adhering to but not fusing with a liposome. In contrast, Fig. 3B shows virus fused with PS vesicles after incubation at pH 5. The arrow points to a very large product of virus-liposome fusion which contains loosely arranged nucleoprotein and a fringe of virus spikes on its surface. Virus and PS vesicles incubated at pH 4 appeared as large aggregates of obscure structure (not shown), whereas fusion hybrids were not observed following incubation of virus and PC vesicles at pH 5 (not shown).

DISCUSSION
Since vesicles of biological origin, such as Sendai virus, can be leaky and/or are not easily loaded with aqueous space markers, the assay of membrane mixing, rather than compartment mixing, may be the method of choice for the rapid and accurate measurement of the fusion of such membranes. Alternative methods for quantifying membrane fusion present various disadvantages. The assay of virus-vesicle hybrids by sucrose gradient centrifugation (Haywood and Boyer, 1984) and the assay of trapped marker release from target liposomes (Kundrot et al., 1983;Oku et al., 1982;Tsao and Huang, 1985) Fig. 1 at pH 7.2 prior to negative staining and electron microscopy at X48,OOO magnification, as described under "Experimental Procedures." The arrow points to a virus particle adhering to but not fusing with a PS vesicle. B, Sendai virus and PS vesicles were incubated as in Fig. 3A but at pH 5.2 prior to negative staining and electron microscopy at X48,OOO magnification, as described under "Experimental Procedures." The arrow points to a particularly large product of virus-vesicle fusion.
in analogy with Sendai virus-induced hemolysis do not specifically signify the merger of fusing membranes, whereas the counting of virus-vesicle fusion hybrids in electron micrographs (Tsao and Huang, 1986) is quite time consuming. Also relatively complicated is a form of compartment mixing assay in which nucleases or proteases in the target vesicle degrade viral RNA or protein on virus-vesicle fusion (Hsu et al., 1983;White and Helenius, 1980).
On the other hand, the energy transfer assay data in Fig, 1, reflecting fusion-independent effects of Sendai virus, indicate the need for caution in regarding changes in donor fluorescence as due to fusion alone. Similarly, the fluorescence of N -NBD-PE in N-NBD-PE + N-Rh-PE-labeled target liposomes increased in the presence of membrane-free influenza virus hemagglutinin (Wharton et al., 1986). At least part of this increase in N-NBD-PE fluorescence of target liposomes in the presence of hemagglutinin or intact virus was attributed to "quenching" of N-Rh-PE by hemagglutinin and not to N-Rh-PE dilution by the fusion of labeled with unlabeled membranes. Wharton et al. (1986) alternatively labeled their target vesicles with a different donor-acceptor pair, i.e. cholesteryl anthracene-9-carboxylate and N-NBD-PE, which gave no spurious fusion signal. Since cholesteryl anthracene-9-carboxylate behaves anomalously in vesicles composed entirely of PS (Uster and Deamer, 1981) and PS appeared necessary in the present study, however, it was decided to continue using N-NBD-PE + N-Rh-PE-labeled target vesicles but to attempt to correct for fusion-independent changes in probe fluorescence if necessary.
The need to correct for fusion-independent changes in N -NBD-PE fluorescence was apparent from absorption spectra of N-Rh-PE in PS vesicles incubated with Sendai virus at pH 4, 5, or 7. According to the spectra, the absorbance of N-Rh-PE in PS, but not PC, vesicles was reduced on incubation with Sendai virus at pH 4 but not at pH 5 or 7. Because of this decreased absorbance of N-Rh-PE in PS vesicles incubated with Sendai virus at pH 4, the quenching by N-Rh-PE of a certain proportion of N-NBD-PE in those vesicles was relieved. This decrease in absorption of N-Rh-PE in PS vesicles incubated at pH 4 with Sendai virus may be similar to the well known hypochromicity of nucleic acids (Freifelder, 1982). As nucleic acids become more ordered, the extinction coefficient of the constituent nucleotide bases decreases without a change in the wavelength of maximum absorption. Evidence for an "orientation effect" on N-Rh-PE like that causing hypochromicity of nucleic acids is the sizable fluorescence anisotropy of N-Rh-PE in vesicles incubated with influenza virus at low pH (Wharton et al., 1986). The corrections of percent lipid-mixing values obtained in the presence of Sendai virus (cf open bars uersw stippled and solid bars, Fig. 1, A and B ) appear valid for two reasons: 1) corrected (Fig. lB, stippled bars and solid bars), as opposed to uncorrected (Fig. lB, open bars), values for percent lipid mixing indicate negligible PS-virus fusion at pH 4 at which pH Sendai virus hemolysis and hemagglutination were completely inactivated. Although it is not clear why Sendai virus fuses poorly with PS vesicles at pH 6 and 7 but lyses red cells well (Lenard and Miller, 1981) and fuses BHK-21 cells moderately well (White et al., 1983) at those pH values, the correction nevertheless accurately reflects little fusion by a pH 4-inactivated virus. 2) Virus-vesicle fusion products were seen in electron micrographs of Sendai virus and probelabeled PS vesicles incubated at pH 5 ( Fig. 3B) but not at pH 7 (Fig. 3A). The occurrence of fusion at pH 4 seemed unlikely, since large aggregates of obscure structure formed on incubation of virus and PS vesicles at pH 4. Nevertheless, it should be noted that earlier reports involving the use of different assays and/or different fluorescent probes and/or under different conditions have indicated that Sendai virus Klappe et al., 1986) and reconstituted Sendai virus (Chejanovsky et al., 1986) fuse with acidic lipid vesicles at pH values lower than 5.
How the fusion of Sendai virus with PS vesicles at low pH, like the fusion of reconstituted vesicular stomatitis virus G protein (Eidelman et al., 1984) and of influenza virus (Stegmann et al., 1985) with acidic lipid vesicles at low pH, is related to virus fusion with cell membranes remains unanswered and is referred to at greater length in another report.' What is clear is the importance of testing for fusion-independent conditions which may compromise the accuracy of energy transfer measurements. To ensure that accuracy, the influence of fusion-independent effects on energy transfer should be assessed by measuring the fluorescence anisotropy of the probes (Wharton et al., 1986) and/or the absorption of the acceptor probe as done here. It may then be possible, as in this instance, to correct the data for artifacts revealed in these and other ways. Thus, donor-acceptor pairs such as N - NBD-PE and N-Rh-PE should continue to be extremely useful in assaying membrane fusion, provided measures are taken to detect and correct for fusion-independent effects.