Activation of the Sendai virus fusion protein (f) involves a conformational change with exposure of a new hydrophobic region.

The F protein of paramyxoviruses is actively involved in the induction of membrane fusion. This fusion may be between viral and cellular membranes, as in the initiation of infection or in virus-induced lysis of erythrocytes, or between the plasma membranes of different cells. The F protein is activated by proteolytic cleavage to yield two disulfide-linked polypeptides (F1 and F2); however, its mechanism of action is not clear. In the present study, the conformations of the inactive, uncleaved precursor of glycoprotein (F0), and the active, cleaved form (F1,2) have been compared. The UV circular dichroism spectra of the two forms of the F protein indicate that cleavage results in a conformational change. Detergent-binding studies by velocity sedimentation analysis of Triton X-100-protein complexes revealed an increase in exposed hydrophobic surface of the protein on cleavage. The inactive F0 bound an estimated 27 molecules of Triton X-100/F polypeptide; these molecules are presumably bound to the hydrophobic region of the glycoprotein that anchors the spike-like protein in the virus membrane and that is common to both forms of F. The active form, F1,2, bound 67 molecules of Triton X-100. This increase in the number of detergent binding sites upon F protein activation indicates the presence of a hydrophobic region that is peculiar to the active form, and that may be of functional significance in the membrane fusion reaction.

The F protein of paramyxoviruses is actively involved in the induction of membrane fusion. This fusion may be between viral and cellular membranes, as in the initiation of infection or in virus-induced lysis of erythrocytes, or between the plasma membranes of different cells. The F protein is activated by proteolytic cleavage to yield two disulfide-linked polypeptides (F, and F2); however, its mechanism of action is not clear. In the present study, the conformations of the inactive, uncleaved precursor glycoprotein (Fo), and the active, cleaved form (Fl,z) have been compared. The W circular dichroism spectra of the two forms of the F protein indicate that cleavage results in a conformational change. Detergent-binding studies by velocity sedimentation analysis of Triton X-100-protein complexes revealed an increase in exposed hydrophobic surface of the protein on cleavage. The inactive FO bound an estimated 27 molecules of Triton X-lOO/F polypeptide; these molecules are presumably bound to the hydrophobic region of the glycoprotein that anchors the spike-like protein in the virus membrane and that is common to both forms of F. The active form, F1.2, bound 67 molecules of Triton X-100. This increase in the number of detergent binding sites upon F protein activation indicates the presence of a hydrophobic region that is peculiar to the active form, and that may be of functional significance in the membrane fusion reaction.
The membrane of paramyxoviruses contains two glycoproteins, HN and F, which form spike-like projections on the external surface of the virion (1). The HN protein has neuraminidase activity and is responsible for attachment of virions to cell surfaces (2, 3). The F protein is involved in the membrane fusion that joins the viral membrane with target membranes, an event which is reflected in the expression of several biological activities of the virus, i e . virus penetration, virusinduced cell fusion, and hemolysis (4-7). Virions that express these activities contain the F protein in the form of a disulfidelinked complex (F1,2), consisting of two glycopolypeptides (F, and Fs) which are derived by proteolytic cleavage of an inactive precursor (Fa) by a host cell enzyme. F0 is present on inactive virions that are produced by cells which lack a suitable protease for F protein activation (4)(5)(6)8).
Efforts have been made to study the fusion reaction in systems that are chemically better defined than are the nat-ural reactants, i.e. intact virions and biological membranes. Fragments of the viral envelope (9) and glycoprotein extracts reconstituted with lipid are active in the fusion reaction (10)(11)(12)(13). The isolated lipid-free Fl,2 protein cannot induce membrane fusion; however, when the isolated F I~ protein is reconstituted with phosphatidylcholine into a membrane, fusion can be obtained, if a mechanism is provided to attach the F1,2containing vesicles to the target membrane (13). This attachment mechanism can be provided not only by the HN protein, but also by wheat germ agglutinin (13). Evidence has been obtained that the viral membrane can fuse with multilamellar liposomes (14), suggesting that the protein of the natural target cell membranes may not be required.
The mechanism of F protein action is a matter of great interest, and some information has begun to emerge recently. One possible mechanism, i.e. the association of a phospholipase activity, was excluded previously (15), and other enzymic activities, such as a protease activity (16), are difficult to reconcile with the apparent lack of need for protein in the target membrane. Another possibility would involve a direct hydrophobic interaction of the F protein with the target membrane, and this has been discussed in the light of the finding of an unusually hydrophobic polypeptide sequence at the NH2 terminus of FI, which is created by the activating cleavage (17)(18)(19)(20). That this region of the molecule at the cleavage site is important for activity is suggested by the conservation of the NH2-terminal sequence at the cleavage site among different paramyxoviruses and virus mutants (18,19), and by the fact that oligopeptides with a structure resembling this region inhibit the action of the F protein (20).
The present study provides direct evidence for hydrophobic binding sites that are present on the active form of the F protein, but not on the inactive form, and that may represent a functionally important feature of the F protein. These experiments involve a comparison of the active and inactive F proteins with regard to secondary structure and capacity for binding of the nonionic detergent Triton X-100. The active form, F1,2, was isolated from wild type Sendai virus grown in embryonated chicken eggs. The inactive form, Fa, was obtained from a Sendai virus mutant, pa-cl, grown in the same host. This mutant can be cleaved and activated by proteases such as chymotrypsin or elastase, but not by the proteases available in the chorioallantoic membrane of the chick embryo (6). This source of precursor protein was chosen because it yields Fo preparations that contain no F1,Z, in contrast to the presence of some cleaved F protein in preparations of wild type virus grown in cultured cells, and because the use of the same host avoids variations in the host-specified carbohydrate component. The only known difference between the F proteins of the egg-grown wild type and mutant viruses is that the mutant virus F protein is not cleaved in the egg, presumably because of a point mutation at the cleavage site (18).

EXPERIMENTAL PROCEDURES
Materials-Triton X-100 was obtained from New Ehgland Nuclear (Boston, MA); Emulphogene BC-720, from GAF Corp. (New York. NY); and sodium cholate, from Calbiochem (Los Angeles, CA). l'hosphatidylcholine was prepared from egg yolk (21,22 Virus-Wild type Sendai virus and the mutant pa-cl were grown in the allantoic sac of IO-day-old embryonated eggs, purified by repeated pelleting, and stored at -70°C (13). Virus labeled with [:'H]leucine or ["Clamino-acids was grown in vitro in allantoic membranes (13) and diluted with unlabeled virus to approximately 1 pCi/ mg of virus protein.
Protein Preparation-For detergent-binding experiments, viral glycoproteins were selectively solubilized with 27 Triton X-100 in 10 mM Na phosphate, pH 7.5, and insoluble protein removed by centrifugation at 100,OOO X g for 45 min (2, 3). For circular dichroism (CI)) were disrupted in 2$ Triton X-100, 10 mM Na phosphate, 1 M NaCI, measurements. the F glycoprotein was prepared as follows. Virions pH 7.5. The nucleocapsid and M protein were removed by centrifugation and dialysis against low salt buffer, respectively (3.5). The HN and F glycoproteins in the supernatant were separated and purified by column chromatography on Sepharose 4B with covalently linked fetuin (5,23) as described (13).
Circular Dichroism-The F protein obtained from chromatography on fetuin-Sepharose was further purified and transferred from the UV-absorbing Triton X-100 solution by sedimentation into a 10-25$ sucrose gradient containing 0.1' i Emulphogene BC-720 and 10 mM Na phosphate, pH 7.5. Pooled fractions were passed through a Sephadex G-10 column (0.9 X 50 cm) equilibrated with the same buffer to remove sucrose. The final concentration of protein used for CI) measurements was estimated by a modified Lowry method (24) that is not affected by the presence of detergents or lipid in the samples.
Reconstituted particles containing F protein and phosphatidylcholine were prepared as descrihed (13). F protein obtained by chromatography on fetuin-Sepharose was transferred from the Triton X-100 solution into cholate-containing buffer by sedimentation into a 10-25% sucrose gradient with 2ri sodium cholate, 1 M NaCI, and 10 mM phosphate, pH 7.5. F protein-containing fractions were pooled and mixed with phosphatidylcholine (protein/lipid = 2:l. w/w) and cholate was removed by dialysis against 10 mM Na phosphate, 0.15 M NaCI, pH 7.2, at 4OC for 36-48 h.
Circular dichroism spectra were recorded with a Cary 60 spectropolarimeter equipped with a C1) attachment which was generously made available by Dr. M. Sonenberg at the Sloan-Kettering Institute, New York. An added diaphragm between the sample cell and the detector allows changes in the angle of detection by variation of the size and position of the diaphragm, and this enables one to determine effects from light scattering of the sample (25). Cells of 1 cm path length were used. Before each run the spectropolarimeter was calibrated with a solution of d-10-camphorsulfonic acid (Eastman Kodak), twice recrystallized from ethyl acetate. Spectra were taken at room temperature with constant purging of nitrogen. The ellipticity curves obtained were analyzed on the basis of reference spectra derived by Greenfield and Fasman (26) and by Chen et al. (27,28).
Density Gradient Centri~u~ation-Sedimentation coefficients (s:!,,.,,) and partial specific volumes ( i ) of protein-detergent complexes were determined using the procedure described by Smigel and Fleischer (29). Continuous sucrose gradients in cellulose nitrate tubes, '/I+, X 2 % inches, were formed from solutions of 10% and 25% sucrose (w/w) in 10 mM Na phosphate and 0.17. Triton X-100, pH 7.5, in H20 or DzO. Triplicate aliquots of viral glycoprotein solutions, prepared in low salt buffer as described above under "Protein Preparation," or of standard protein solutions (10 mg/ml), 0.2 ml/gradient, were overlaid on the gradients. The standard proteins were ovalbumin (sn,.,, = 3.6), bovine serum albumin (s~,,,,, = 4.6). and human y-globulin (s?,,.,~ = 7.0). After centrifugation a t 50, OOO rpm for 15 h (sucrose-HSO gradient) or 26 h (sucrose-D2O gradient) in a Spinco SW 60 rotor in a Beckman L2-65B centrifuge with the temperature set at 4°C. the run was terminated by deceleration with the brake off. Eight-drop fractions collected from the bottom, and refractive indices of every other fraction were read a t 20°C in a Bausch and Lomb Abbee 3L refrac-tometer immediately after fractionation. Fractions of gradients with standard proteins were analyzed for protein content (24). and viral glycoprotein fractions were analyzed for radioactivity hy mixing aliquots with 0.5 ml of water and 4 ml of toluene/Triton X-lOO/Liquifluor (1660:1000:116) and measurement in a liquid scintillation counter.
T h e measured refractive indices were used to determine the density and the viscosity at each point along the gradient. The numerical integration of the sedimentation equation. using the calculated density and viscosity, resulted in a series of sL",.,, and ; values for the protein-detergent complex in Hz0 and DrO (29). The intersection of the two curves determines the sLll,, and i values.
Because of the difficulty in determining the exact temperature of the gradients during centrifugation. the temperature that gave the correct s~, , , , .
for the standard proteins was used for all calculations (29). The corrected temperature was within :l°C of the setpoint. The computer program used by Smigel and Fleischer (29) was kindly made available by them and adapted to IBM 360 and 1' 111' 11 computers with the generous help of Dr. James S. Murphy (from The Rockefeller University).
Polyacrylamide Gel Electrophoresis-Slab gels of 1.5 mm thickness with 10% polyacrylamide and 0.25% bisacrylamide in the separating gel were prepared as described by Laemmli (30). Gels were run a t 12 mA for 18 h and stained with 0.2c; Coomassie blue in 50' ; methanol and 7% acetic acid.

RESULTS
Circular Dichroism of Cleaved a n d U n c l e a t l e d F Proteins- Fig.   1 shows the polyacrylamide gel analysis of t h e isolated F proteins used for these experiments. The F protein, isolated from egg-grown wild type Sendai virions and purified by column chromatography followed by preparative velocity sedimentation in sucrose gradients (see "Experimental Procedures"), contains the two glycopolypeptides, F, a n d FA which are separated when run after reduction with dithiothreitol (lane B ) , but migrate as a disulfide-linked complex, F1.2, in the absence of reducing reagent (lane C) (8). The F protein isolated from the mutant pa-cl grown in eggs is not cleaved (6), and is seen in the gel as the uncleaved precursor   D and E ) . A slight difference in the migration of reduced and unreduced Fo (lanes D and E ) has been ascribed to interchain disulfide bonds (8).
For measurement of the circular dichroism of the F protein in solution, nonionic detergent must be present to prevent aggregation, and because of the strong UV absorption of the Triton X-100 which was used for F protein isolation, it was necessary to transfer the protein into Emulphogene BC-720, a nonionic detergent without the UV-absorbing benzene ring of Triton X-100. The transfer was accomplished by sedimentation of the isolated F protein out of the Triton X-100 solution into gradients containing Emulphogene BC-720 as described under "Experimental Procedures." The CD spectra of the cleaved and the uncleaved F protein in 0.1% Emulphogene BC-720 solution (Fig. 2) show the negative ellipticity bands around 208 and 222 nm that are characteristic of a predominantly a-helical conformation. The comparison of the spectra indicates a greater a-helical component in the cleaved form of the F protein, as is evident from the greater negative ellipticity and from the position of the minima. A best fit analysis of the ellipticity curves using the reference spectra of Chen et al. (27,28) gave the following distribution of the conformational components: For F,, 60% a-helix, 15% ,&sheet, and 25% random coil; and for Fl;r, 75% a-helix and 25% random coil. The reference spectra used for these calculations take into account the coupling of the transition moments along very long a-helices, and they yielded the curves that were closest to the observed ellipticities. With a carbohydrate content of the F protein of 15% (32), no significant contribution of carbohydrate to the CD spectrum in this wavelength range would be expected (33). The quantitative analysis of the circular dichroism spectra made use of protein concentrations that were determined by a modified Lowry method which eliminates the error from detergent and lipid (24). Without the knowledge of the primary structure of the F protein, a more precise determination of protein concentration is not possible at present, and the above numbers therefore do not indicate the absolute secondary structure of the protein. However, the inherent uncertainty of the Lowry procedure (34) applies equally to the cleaved and the uncleaved forms of the F protein, and would therefore not affect the observed qualitative differences.
T o analyze the ellipticity of the F protein in the absence of detergent and in a state closer to its natural conformation in the virus envelope, the isolated F proteins were reconstituted with phosphatidylcholine (13). Under the conditions used and with a phosphatidylcholine and F protein ratio of 1:2 (w/w), reconstitution yielded vesicles of 40 to 80 nm diameter, with glycoproteins present as spikes that are similar in shape and arrangement to the spikes in the virus envelope (Fig. 3) (13). With particles of this size, light scattering could contribute to the observed ellipticities, however variation of the acceptance angle of the detector between 8°C and 60°C did not influence the observed CD spectra, and this eliminates light scattering as a source of error. There is a small but consistent difference in the ellipticity curves of the cleaved and uncleaved proteins in the reconstituted membranes as is evident from the shape of the curves (Fig. 4). The quantitative analysis of the ellipticity curves (27, 28) resulted in 90% a-helix and 10% random coil for Fo, and 95% a-helix and 5% random coil for FI:2. It has been reported that "bunching" of protein can result in distortion of the CD curve, diminishing the intensity a t 208 nm more than at 222 nm (35), however in virions and in reconstituted membranes the spikes are spaced well apart, and there is no difference in the packing density of cleaved and uncleaved F spikes (13).' It is unlikely that bunching through aggregation of spikes is a problem, because at the low protein concentrations used, i.e. less than 50 pg/ml, the proteins sediment homogeneously in sucrose gradients (see below).
It is apparent that cleaved and uncleaved forms of F show a greater content of a-helical structure in the reconstituted particles than in detergent solution (cf. Fig. 4 and Fig. 2). The reason for this is not yet clear; however, it may be related to the fact that when associated with the membrane, the F protein spikes are oligomers, as is evident from the size of the spike, and by analogy to the oligomeric structure of spike glycoproteins on enveloped viruses, whereas the F protein in detergent solution is present in monomeric form.
Determination of Detergent Binding of F Proteins by Velocity Sedimentation in Sucrose Gradients-In the course of preparative purification of F proteins by velocity sedimentation on Triton X-100-containing sucrose gradients, it became apparent that the cleaved F protein consistently sedimented at a faster rate than the uncleaved form of F. In a co-sedimentation analysis of the glycoproteins isolated from virions containing [3H]F~ and ['4C]F,,z (Fig. 5), the HN proteins cosedimented, whereas the cleaved F protein sedimented distinctly ahead of the uncleaved form. A similar difference in the sedimentation rate was also observed when Emulphogene BC-720 was used instead of Triton X-100 (data not shown). This difference cannot be explained on the basis of size of the glycoproteins because cleavage of the F protein has been shown to entail no significant loss of polypeptide (8), and because the cleaved form sediments faster. Analysis of the glycoproteins by gel filtration on Sepharose 6B in the presence of Triton X-100 showed elution of Fl,z before Fo (data not shown), indicating that the faster sedimentation of the cleaved form of F was due to an increase in the size of the F proteindetergent complex, rather than to a change in the frictional coefficient.
To determine the basis for the change in sedimentation properties of the cleaved F protein, it was important to determine the amount of Triton X-100 in the protein-detergent complexes. Procedures for this using radioactively labeled Triton X-100 in gel filtration (36) and velocity sedimentation analysis (37) were not successful because they require the protein to be present at high concentrations, at which the F protein aggregates even in the presence of detergent. An alternate procedure described by Smigel and Fleischer (29) employs velocity sedimentation of detergent-protein com- plexes in media of different density, i.e. sucrose-H20 and sucrose-D20, to obtain values for both the sedimentation coefficients and the partial specific volumes of the complex, and this permits the calculation of the proportion of protein and detergent in the complex as described below.
Sedimentation of the unfractionated glycoprotein extract amino-acids and [3H]leucine, respectively, were mixed and the glycoproteins solubilized with 2% Triton X-100 in 10 m~ sodium phosphate, pH 7.5, yielding solubilized HN and F at approximately 0.2 m g / d each (2,3,5). Two hundred p1 of the supernatant was overlaid on the gradient and subjected to centrifugation for 18 h at 50,000 rpm. Fractions were analyzed for radioactivity and neuraminidase activity. containing both the HN and the F proteins yielded bands that were as sharp as those of standard proteins, indicating that the glycoprotein-detergent complexes sediment as homogeneous species in the H20and D20-sucrose gradients (Fig. 6). The peak position of the F protein is independent of glycoprotein concentration over a range of 1 0 0 to 400 pg of glycoprotein applied (Fig. 7). In such experiments, the size of the faster sedimenting peak was found to increase with protein concentration. Analysis of the proteins in the two peaks (Fig. 8) indicated that the slower sedimenting peak contained only F protein; however, the faster sedimenting peak contained not only HN, but, at high protein concentrations, also some F protein, presumably in an aggregated form. Fig. 8 also shows that this aggregation occurred only with the cleaved and not with the uncleaved form of the F protein.
As outlined by Smigel and Fleischer (29), and using their computer program, the sedimentation coefficient (s?,,,,,.) as a function of the partial specific volume (V) was calculated from the position of the peaks in the gradient, and the density and viscosity through the gradient as calculated from refractive index measurements (cf. "Experimental Procedures"). Sedimentation through two different density gradient media, sucrose in H 2 0 and sucrose in D20, gives the s~~, .~~. and V simultaneously. Fig. 9 illustrates this for the cleaved F1.S-Triton X-1 0 0 complex. Analysis of the detergent complexes of the cleaved and uncleaved F protein resulted in Sm.rc and V values listed in Table I. The reproducibility of the determinations is shown by the small standard error among different experiments (Table I). Among the triplicate gradients within a given experiment, the standard error was even smaller (not shown).
The foremost practical problem in this procedure is the determination of the exact temperature in the gradients. Following the procedure of Smigel and Fleischer (29), we circumvented this problem by using the gradient temperature which gave the correct S2n.e. value for the standard proteins. The increase in the partial specific volume from 0.750 cm:'/ g for the uncleaved to 0.789 cm:'/g for the cleaved F protein can be explained only by an increase in the amount of Triton X-100 in the glycoprotein-detergent complex. The amount of detergent in the complexes can be calculated because the partial specific volume of the complex is assumed to be determined by the weight fractions and partial specific volumes of the constituents: i, VI., and i l l are the partial specific volumes for the glycoprotein-detergent complex, the glycoprotein, and the detergent, and WI, is the weight fraction of the detergent in the complex. The partial specific volume of Triton X-100 has been reported as 0.908 cm."/g (38). The partial specific volume of the F protein was calculated to be 0.708 cm:'/g from the carbohydrate content of 15% (32). and the partial specific volumes of protein (0.735 cm"/g) and carbohydrate (0.556 cm"/g). We found empirically that this approximation for the calculation of partial specific volumes of glycoproteins in general gives values that are within 3% of the measured partial specific  volumes of glycoproteins with low (~2 0 % ) carbohydrate content (%), i.e. fetuin, plasminogen, transferrin, az-glycoprotein, and al-antitrypsin.
Using Equation 1 and the experimentally determined values for the partial specific volumes of the F protein-detergent complexes (Table I), the weight of Triton X-100 in the Fodetergent complex was calculated to be 21% of the total, and in the Fl,z-detergent complex, 40%. Taking 65,000 as the molecular weight for F, and Fl,z (8) and 644 for Triton X-100 (29), the calculated molar ratios of Triton X-100 to glycoprotein in the complex were 27 for FO and 67 for F1,2.
An alternate approach to the calculation of the partial specific volume of glycoproteins is the procedure described by Gibbons (39), which uses the partial specific volumes of individual carbohydrate moieties for the calculation. This procedure has been shown to give satisfactory values for several glycoproteins with very high carbohydrate content; however, with proteins containing less carbohydrate, we found it to result in discrepancies between calculated and observed partial specific volumes, and we therefore concluded that the values used above are more accurate. However, we did calculate detergent binding using the partial specific volumes reported by Gibbons, and the carbohydrate composition as reported by Kohama et al. (32), and obtained for FO and F1,2 partial specific volumes of 0.722, and molar ratios of detergent to protein of 21 and 59, respectively. Thus, using these values, the differences between the two forms of the F protein are of the same magnitude as found above.
The method of Smigel and Fleischer (29) involves a number of assumptions. In addition to the diffkulty in determining the actual temperature of the gradients during the run, and to calculating the partial specific volumes of the glycoprotein as discussed above, it is assumed that the binding of the detergent is independent of sucrose concentration and that it is the same in H 2 0 and DzO. Also, it is assumed that the partial specific volumes of protein and detergent are the same in the complex state as in the free state. In the present study, we compared two forms of the same protein, and therefore errors in these assumptions would not alter the conclusion drawn, i.e. the approximate doubling of Triton X-100 binding with F protein cleavage. In the course of the calculation, we have also assumed that cleavage of the F protein does not alter the hydration of the protein. If the observed difference in the partial specific volume of the Fo and the F 1 , Z were due to differential hydration, then F 1 ; 2 would have to bind 0.2 g of HzO/g of protein more than that already bound by Fo. Considering the extent of hydration for a typical soluble protein, 0.2 g of HzO/g protein (40), this seems to be unlikely.

DISCUSSION
The present results document changes in the conformation of the F protein that accompany the proteolytic cleavage of this protein. Because they correlate with the active and inactive state of the F protein, these changes may provide clues to the possible mode of its action. The outcome of F protein activity i s membrane fusion between the viral membrane and other membranes; however, the precise mechanism by which this occurs has not been identified. The reaction requires close contact of viral and target membranes, which is provided by the attachment of the receptor-binding protein, HN, to neuraminic acid residues on the target membrane (23), and the F protein then interacts with the target membrane to induce fusion between the bilayers. Conversion of the inactive precursor to the active F protein by proteolytic cleavage does not entail a significant loss of amino acids, suggesting that the activation that results from cleavage involves a change in the conformation of the F protein (8). A conformational change has now been directly demonstrated by the CD spectra of the active and inactive proteins. The hemagglutinating protein of influenza virus is also cleaved to yield two disulfide-linked polypeptides (41), and the cleavage activates the infectivity of the virus (42, 43). A change in the conformation of the hemagglutinating protein on cleavage has been reported previously (44), however the change in the CD spectrum was seen only in the near UV and not in the far UV, as found with the F protein in the present study.
One aspect of the change in the F protein conformation is the increase in detergent binding described here, which presumably represents the exposure on cleavage of additional hydrophobic sites at the surface of the active F protein.
Significant binding of nonionic detergents such as the 27 molecules of Triton X-100/FO polypeptide found in the present study has been observed only with membrane proteins (29, 36, 37, 45, 46). The implication from this general observation is that Triton X-100 binds to sites of the protein that in the native configuration interact with lipid, and this can readily be extended to the F protein, which is inserted in the lipid of the virus membrane by a hydrophobic portion at the base of the spike. The hydrophobic anchor has been shown with other viral glycoproteins to represent only a small portion of the molecule; thus, with the HN protein of paramyxoviruses the hydrophobic base accounts for less than 8% of the protein (la), and for the hemagglutinating protein of influenza A virus less than 6% (47). The maximum area of Fn that would be occupied by 27 molecules of Triton X-100 (0.5 nm2/molecule) (48) would be 10% if the protein-detergent complex were spherical, and with the actual dumbbell shape of the spike this hydrophobic area could be considerably smaller than 10%.
The mode of attachment of the F spike to the viral membrane by hydrophobic interaction with the viral lipid applies to both the cleaved and the uncleaved form of the protein, and even though one cannot be certain that this interaction is quantitatively identical for both forms of F, it is highly unlikely that the observed increase from 27 molecules of Triton X-100 in the.Fo to 67 in the cleaved F protein reflects a change in this portion of the molecule. The likely explanation for the increase in detergent binding is the exposure of additional hydrophobic sites on another region of the cleaved protein.
These sites may represent a portion of the protein that could engage in hydrophobic interaction wit,h other membranes, and such an interaction may be involved in the membrane-fusing activity of the protein.
The possibility that the mechanism of action of the F protein involves a hydrophobic interaction with the target membrane has been suggested previously, based on the structural features of the protein at the cleavage site. The polypeptide sequence at the NH2-terminal portion of the cleavage site is unusually hydrophobic (7, 17-20), e.g. a stretch of at least 26 uncharged amino acids, and conserved among different paramyxoviruses (18-20). Furthermore, the structure of this NH2-terminal region of the active F protein has been shown to be important for activity, because peptides that mimic the NH2-terminal sequence are inhibitors of F protein-induced membrane fusion (20). However, it remains to be established conclusively that, as the results suggest, the hydrophobic polypeptide sequence at the cleavage site is itself part of an active site or of the detergent binding region.
The present evidence for a newly exposed hydrophobic region on the active F protein sheds light on the general question of why the proteins that are involved in virus penetration, ie. the F proteins of paramyxoviruses and the hemagglutinating proteins of myxoviruses are first synthesized as inactive precursors which must be cleaved by a suitable cellular protease to be activated. The present findings suggest that this may provide a means of preventing the F protein from engaging in hydrophobic interactions with intracellular membranes before it is assembled into the viral envelope. Proteolytic cleavage is a late event in the virus assembly, and has been shown to take place at the plasma membrane (41). In keeping with this, the hemagglutinating protein associated with the microsomal fraction of infected cells is present in the uncleaved form in most virus-cell systems (49,50), and in the one system where cleaved protein has been found in the microsomal fraction (51), it is not known whether those cleaved proteins are still suitable for virus assembly.