JOURNAL OF VIROLOGY, Oct. 1968, p. 972-978 Copyright 1968 American Society for Microbiology Vol. 2, No. 1 Printed in U.S.A. Fractionation of astern quine ncephalitis Virus by Density Gradient Centrifugation in CsCl H. G. AASLSTAD,1. J. HOFFMAN, AND ARTHUR BROWN Biological Scienices Laboratory, Fort Detrick, Frederick, Marylanld 2171 Received for publication 6 June 1968 When partially purified astern equine encephalitis () virus was centrifuged to equilibrium in CsCl, three virus specific bands were observed. A hemagglutinin was detected at a buoyant density of 1.18 g/cm3. Infectious virus banded in two positions; most of the virus banded at 1.2 g/cm3 and a lesser amount banded at 1.22 to 1.23 g/cm3. Analysis of radioactive profiles of CsCl-fractionated virus labeled with either 32PO4 or 3H-uridine suggested that the hemagglutinin was stripped from the intact virion. The viral origin of the hemagglutinin was verified by inhibition with specific antiserum. Attempts to differentiate between infectious virus of the different buoyant densities showed that the denser particle was neither a virus contaminant nor a density mutant. No evidence was obtained to indicate that the denser particle was an immature form of virus. The two infectious species obtained after CsCl fractionation were indistinguishable antigenically. Furthermore, unfractionated as well as CsCl-fractionated virus sedimented at about 26S in sucrose gradients. These results together with the results of rebanding experiments suggested that the denser species (1.23 g/cm3) results from a salt (CsCl)-induced alteration or breakdown of the virion (1.2 g/cm3), and that it arises as the hemagglutinin is stripped from the surface of the virion. Many laboratories have shown that considerable density heterogeneity exists in purified virus preparations. This is easily understood when the particles are deficient or devoid of nucleic acid or are mixed with soluble viral antigens. However, infectious virus of significantly different densities has also been observed, particularly among some of the enveloped viruses such as the arboviruses and the myxoviruses. Two representative arboviruses, Sindbis (4, 7) and Dengue-2 (1), have been characterized with respect to their buoyant densities in CsCl. In each case, considerable density heterogeneity was apparent, and, in some cases, virus specific hemagglutinins and complement-fixing antigens could be resolved from the major band of infectious virus. Sindbis and Semliki Forest virus hemagglutinins are considered to be identical with the virion, which is located in the envelope of the virus (6, 8). In this report, we have determined that the buoyant density of the group A arbovirus, astern equine encephalitis (), is 1.2 g/cm3 in CsCl. In addition, two other bands of viral material were observed; a light hemagglutinin which had a buoyant density of 1.18 g/cm3 and 1 Present address: Department of Microbiology, University of Georgia, Athens 361. an infectious virus of a higher density (1.22 to 1.23 g/cm3). These two bands appear to be the salt-induced breakdown products of the virion. MATRIALS AND MTHODS Virus anid cell culture. The Louisiana strain of virus, the origin and properties of which were reported by Brown (2), was used in all experiments. The virus was propagated on chick embryo (C) monolayers and was assayed by the plaque technique described by Zebovitz and Brown (12). Virus was harvested 16 to 2 hr postinfection; titers averaged 4 X 18 to 8 X 18 plaque-forming units (PFU)/ml. Density gradient cenitrifugation. Before density gradient centrifugation, virus was partially purified from 4 to 6 ml of tissue culture fluid that had been decanted from infected cells by adsorption onto AlPO4 gel (9). The virus was eluted from the gel in.3 M phosphate buffer (ph 8.) and was concentrated by centrifugation at 65, X g for 6 min. The resulting pellet was allowed to resuspend overnight in borate-saline buffer (ph 9.) containing.1% bovine serum albumin (3). Typically, only 15 to 2% of the total PFU were recovered; however, an overall purification of about 5-fold was obtained (2. X 11 PFU/mg of protein). CsCl was added to 4.5 ml of the partially purified virus to a mean density of 1.21 g/cm3. The virus particles were unstable in the salt when bovine serum 972
VOL. 2, 1968 FRACTIONATION OF ASTRN QUIN NCPHALITIS VIRUS 973 albumin was omitted from the buffer. quilibrium centrifugation was performed in a Spinco model L preparative ultracentrifuge with an SW-39 rotor at 1, X g for 24 hr at 5 C. After centrifugation, five-drop fractions (.5 ml) were collected from the bottom of the tube, immediately diluted 1:1 with cold beef-heart infusion, and assayed. Generally, every tenth fraction was held undiluted so that its refractive index could be determined. Density was calculated from the empirical formula p25' = 1.861 (nd21') - 13.4974. Linear sucrose gradients were constructed with 1 and 4% (w/w) sucrose dissolved in borate-saline buffer (ph 9.). Virus was overlaid onto a 4.-ml sucrose gradient and was centrifuged for 6 to 9 min at 4, X g in the SW-39 rotor. Immunzological methlods. To assay arbovirus hemagglutinating activity, we employed goose red blood cells, the microtiter tray system, and the general procedures of Clarke and Casals (3). sucroseacetone hemagglutinin and immune monkey sera which had been treated for the removal of nonspecific hemagglutinin inhibitors were also prepared (3). Plaque reduction tests utilized high-titer rabbit anti- serum. Prewarmed antiserum and virus were mixed and incubated at 25 C, with occasional shaking for various time intervals. Nonimmune rabbit serum was used for the control. Samples of virusantiserum mixtures were immediately diluted 1:1 in cold beef-heart infusion and were assayed for plaque formation. Plaque titers represent an average of triplicate platings. Radiological methods. Isotopically labeled virus was prepared by including 32PO4 (1 lc/ml) or 3H-uridine (5 MAc/ml) in the virus growth medium. Actinomycin D (2,Ag/ml) was also included when 3H-uridine was employed. All radioactivity measurements represent cold trichloroacetic acid-insoluble virus material trapped on membrane filters. The dried filters were counted in toluene-based BBOT with a liquid-scintillation spectrometer. RSULTS Density gradient centrifugation. xperiments in which a partially purified virus suspension was sedimented in linear 1 to 4% sucrose density gradients revealed only one visible band. After fractionation of the tube contents, we observed that maximal plaque-forming activity, hemagglutinin activity, absorbancy at 26 nm, and 3H-uridine radioactivity were coincident. Visual examination of virus centrifuged to equilibrium in CsCl revealed three lightscattering bands. The uppermost band was diffuse, the middle band was well defined and contained most of the banded material, and the lower band was sometimes composed of two distinguishable, although poorly resolved bands. The latter were considered as one band, because the two minor bands did not always appear and were, in all cases, inseparable during fractionation. Most of the PFU (8 to 9%) were found in the middle-band at a buoyant density of 1.2 g/cm3, and this band is presumed to contain the complete virion (Fig. 1). Significant titratable virus was also detected in a saddle-shaped peak in the density range of 1.22 to 1.23 g/cm3. All plaques produced from these various fractions were of the same size and plaque morphology as parental virus. Although hemagglutinin was found under the peaks of infectious virus, the maximal value of these peaks was only FIG. 1. Distribution ofpartially purified virus infectivity and hemagglutinin in a CsCI density gradient. Symbols:, infectivity; *, density; bar diagram, hemagglutinin (HA) titer.
974 AASLSTAD, HOFFMAN, AND BROWN J. VIROL. FIG. 2. Symbols:, infectivity;, denisity; A, radioactivity. Distributiont ofpartially putrified virus infectivity anld 32PO4 radioactivity in a CsCI dentsity gradielnt. ).25 5 4 1.2 -s m t Z, _ J1.15O I s l1 --- 2 x D. 3 I 2 II_I ~~ ~ ~ ~ ~ ~ ~ -I1 -'5 1 2 3 4 5 6 7 8 FIG. 3. Distributtiont ofpartially purified virus WH-uridine radioactivity antd heemagglutintini in a CsCI delnsily gradientt. Symbols:, radioactivity;, density; bar diagram, hemagglutininl (HA) titer. 1c%G of the hemagglutinin titer located in fractions 48 to 5 (buoyant density, 1.18) in Fig. 1. Although the maximal hemagglutinin titer was found in the least dense band, this band contained only 3%c, of the PFU titer. virus labeled with either 32PO4 or 3H-uridine was used in the next two experiments. Figure 2 shows the data obtained when 32PO4- labeled virus was centrifuged through a CsCl gradient. The infectivity profile shown in this figure is identical to that shown in Fig. 1. Radioactivity was found to correspond with the infectious virus peaks as well as with the band of hemagglutinating material. Next, virus labeled with 3H-uridine was centrifuged through the CsCl gradient. Only particles containing ribonucleic acid (RNA) would be detected in the radioactivity profile.
VOL. 2, 1968 FRACTIONATION OF ASTRN QUIN NCPHALITIS VIRUS 975 TABL 1. Specific inhibition of the CsClfractionated virus hemagglutinin Test system HA titer Control antigena... 1:64 Control antigen plus anti- serumb... <1:2 Control antigen plus normal serumb... 1:64 antigen from CsCl gradient... 1:128 antigen from CsCl gradient plus anti- serum... <1:2 antigen from CsCl gradient plus normal serum... 1:64 a Control antigen was a sucrose-acetone preparation. b Monkey anti- serum and normal serum were treated to remove nonspecific inhibitors. The results (Fig. 3) suggested that, since no tritium was detected, the heamgglutinating material banding at 1.18 g/cm3 may be attributed either to fragments of the viral envelope or to a particle split from the surface of the envelope. Nature of the hemagglutinin. Proof that the hemagglutinating material banding at a density of 1.18 g/cm3 was virus specific was afforded by a hemagglutinin inhibition test (3). In Table 1, the CsCl-derived hemagglutinin is compared with an sucrose-acetone antigen. It is clear that both hemagglutinins were strongly inhibited by an immune serum. The virus-specific nature of the CsCl-fractionated hemagglutinin was also indicated by the requirement for an acidic ph for optimal activity. Nature of the two infectious CsCl bands. The detection of a hemagglutinating particle separate from infectious virions in CsCl gradients was anticipated; however, the nature of the infectious particle banding at a density of 1.22 to 1.23 g/cm3 was not clear. We examined the possibility that this denser species might be a virus contaminant, a density mutant, an immature or subviral form of virus, or a breakdown product of the complete virion. The possibility that the denser species was a contaminant was ruled out by the fact that freshly plaque-purified virus contained infectious virus having a density of 1.23 g/cm3. vidence against the possibility that a density mutant was present in the population was provided by a rebanding experiment. The virus was fractionated in a CsCl gradient, and virus that banded at each of the three densities was plaqued on C monolayers. Virus from the resulting plaques was used to infect C monolayers, and the progeny virus obtained was subjected to density gradient centrifugation. Results from this experiment indicated that virus populations of heterogeneous densities could be obtained from plaques of virus fractionated at each of the different densities. The possibility still existed that this denser particle was a natural form of virus. We thought that it might be similar to the viral form described as immature virus (J. I. Colon and J. B. Idoine, Bacteriol. Proc., p. 159, 1963; 11). It has been suggested that immature virus (i) might have a higher titer by hypertonic than by isotonic methods of plaque assay on C monolayers, (ii) might be susceptible to ribonuclease digestion, and (iii) might yield infectious RNA when extracted with cold 83% phenol whereas the complete virus would not. Table 2 presents the results of an experiment which tested some of these possibilities. Both bands behaved similarly; the denser virus did not assay more efficiently by the hypertonic method as would be characteristic of an immature form. In addition, ribonuclease failed to inactivate either preparation. Comparison of the yields of infectious RNA obtained after hot or cold phenol extraction (11) showed that virus of both buoyant densities behaved similarly. Since an immature form should yield infectious RNA by the cold method whereas the mature virion should not, this test was inconclusive. It may be that mature or intact virus purified to the degree that this material was becomes susceptible to cold-phenol extraction. We think that these data offer no evidence to link virus with a density of 1.23 g/cm3 TABL 2. Assays of isotonic and hypertonic infectivity and ribonuclease sensitivity of CsCI-fractionated virusa Infectivity (PFU/ml) Assay method or treatmentb, density, density 1.23 g/cm3 1.2 g/cm3 Isotonic... 6.5 X 19 44. X 19 Hypertonic....24 X 19.83 X 19 Ribonuclease... 5-7 X 19 31. X 19 Ribonuclease control 4.2 X 19 32. X 19 a virus having densities of 1.2 and 1.23 g/cm3 was removed from a CsCl density gradient tube by side puncture. b Isotonic assay was performed with beef-heart infusion as diluent. Hypertonic assay was performed on C monolayers washed with.5 and 1. M NaCl, with 1. M NaCl as the virus diluent. Ribonuclease treatment consisted of incubating the virus with 2,g of ribonuclease for 3 min. at 37 C. The ribonuclease control was handled in a similar manner, except that beef-heart infusion replaced the enzyme.
976' AASLSTAD, HOFFMAN, AND BROWN J. VIROL. to an immature or subviral form, as these terms are usulally understood. To determine whether the virion was breaking down during centrifugation in CsCl, a rebanding experiment was performed. 32PO4_ labeled virus was centrifuged in a CsCl density gradient, and the virus material banding at each of the three densities was obtained. A second CsCl centrifugation was then performed, with each band in a separate tube. The radioactive profile obtained is shown in Fig. 4. Most of the densest material rebanded at a density of 1.23 g/cm3, while a small amount was detected at 1.18 g/cm3. On the other hand, virus from the middle band appeared to break down, yielding radioactivity at each of the three densities; this breakdown was similar to the behavior of freshly purified virus on an initial CsCl fractionation. The band containing hemagglutinin rebanded quantitatively at a density of 1.18 g/cm3. The radioactivity present where the hemagglutinin banded can be accounted for by the fact that this virus contains phospholipids within its outer membrane or envelope. Thus, it appears that virus with a density of 1.23 g/cm3 was produced by the salt-induced alteration or degradation of the complete virion. We think that the hemagglutinin-rich material banding at a density of 1.18 g/cm3 also resulted from saltinduced breakdown. We next attempted to differentiate between the two infectious bands by serological methods. virus was again centrifuged in CsCl and fractionated; the virus bands having densities of 1.2 and 1.23 g/cm3 were pooled. First, we used a 5% plaque neutralization test, in which a constant amount of virus was incubated with various dilutions of rabbit virus antiserum. This test failed to differentiate between the two virus species. A more sensitive test, based on the kinetics of neutralization, was then employed. The results are presented in Fig. 5. The virus used as a control in this experiment was purified by both AlPO4 gel concentration and high-speed sedimentation. A control consisting of unfractionated virus and CsCl-banded virus incubated with normal rabbit serum maintained its titer. Since the kinetics of plaque neutralization were not significantly different, we could not differentiate between the fractionated virus bands. Finally, when virus was fractionated on CsCl, the sucrose gradients (Fig. 6) indicated that the resulting bands of infectious virus did not contain virus particles of different sedimentation characteristics. The upper third of Fig. 6 shows that unfractionated virus sediments at about 26S as a single band in the sucrose gradient. When 32PO4-labeled virus was fractionated on CsCl and the two infectious bands were isolated, dialyzed, and centrifuged in the sucrose gradient, they could not be resolved. DISCUSSION It is currently accepted that arboviruses consist of a ribonucleoprotein core surrounded by a membrane (envelope) which is derived partly from the host cell (1). In addition, electron C~4 C., 7h 6-5i O 4~- Density (g/cm3) FIG. 4. Distribution of 32PO4-labeled infectious virus (1.2 and 1.22 to 1.23 glcm3) and hemagglutinin (1.18 g/cm3) after recentrifugation in a CsCl density gradient. ach band was fractionated from a CsCI gradient and was rebanded in a separate tube. Symbols:, 1.22 to 1.23 g/cm3 band; A, 1.2 g/cm3 band; rl, 1.18 g/cm3 band.
VOL. 2, 1968 FRACTIONATION OF ASTRN QUIN NCPHALITIS VIRUS 1..5~ *.1 c >.O o.o.1 I * virus X O A band I a B bond A 5 Time, minutes ability to hemagglutinate and therefore are not considered to be altogether stripped of fringe. The experiment in which the three virus specific bands of 32PO4-labeled virus were rebanded supports this explanation (Fig. 4). In addition, when e=1.23 32P4-labeled virus was centrifuged in CsCl p= 1.2 in the absence of bovine serum albumin, we observed that the radioactivity banding at 1.18 and 1.23 g/cm3 increased at the expense of the 1.2 g/cm3 band. Conversely, when bovine serum albumin was added to a final concentration of.75%, the 1.2 g/cm3 band was protected. We think that the CsCl effect is dependent on the total amount of protein in the centrifuge tube; i.e., those virus suspensions relatively free from contaminating protein are more subject to mild salt-induced alterations. Faulkner and Dobos (5) recently reported that three hemagglutinins may be resolved from 1-2'~-'-5 fluorocarbon-purified Sindbis virus by CsCl equilibrium centrifugation. Their studies showed that a light fraction (1.19 g/cm3) contained only FIG. 5. Antiserum neutralization kinetitcs oi CsCl- hemagglutinin material, whereas an intermediate fractionated and wifractionated virusv. Test mix- (1.21 g/cm3) and a heavy fraction (1.24 g/cm3) ture contained.5 ml of virus (abolut S X 16 were composed of particles containing both PFU/ml) and.5 ml of a 1:4 dilution of high-titer rabbit anti- serum. Incubation was at 25 C, and samples were withdrawn at the designecated times. 8S morker Symbols:, unfractionated virus;,,1.23 g/cm3 3 virus virus; A, 1.2 g/cni virus. " 2- micrographs have shown that short fringe-like ~ projections cover the surface of the Nvirion. Ac- l - cording to Osterrieth and Calberg--Bacthe hemagglutinin of Semliki Forest virus may (8), 1 be carried on these projections. The se investi- a pro- gators treated Semliki Forest virus wiith teolytic enzyme and observed a loss of hemag- 2. Bond 1.23 glutinin titer that paralleled the remc val of the projections. They also reported theat Semlikd 1.- Forest virus which had received this treatment was still infectious. The data presented in Fig. 1 to 3 nnay be ex-. plained in a manner analogous to the above e N Band Q=1.2 observations; that is, that the concer.111 ijtratinn 4LlUll nf(ls Ul CsCl used to construct the densit3y gradient (initially 1.55 M) was sufficient to dlisrupt the surface of the virion and free a hemag,glutinating fragment. This is consistent with tthe lighter buoyant density of the hemagglutininl and with the labeling pattern presented in Fig. 2 and 3. top 4 8 12 16 2 24 Infectious virus banding at a densitty of 1.23 g/cm3 may therefore be virus which has lost some of its projections or fringe. S!uch a FIG. 6. Sucrose loss density gradient centrifugation of may result in CsCl-fractionated and a unfractionated virus.. denser particle if tihe particle coli phage MS-2 was used as a sedimentation marker. becomes relatively less rich in lipid or if its prop- Unfractionated virus is shown in the top section, erties of hydration are altered, or if b)oth occur. whereas 1.23 g/cm3 and 1.2 g/cm3 CsCl-fractionated Particles in this density range still p )ossess the virus are shown in the middle and bottom sections. 977
978 AASLSTAD, HOFFMAN, AND BROWN J. VIROL. hemagglutinin and infectivity. These investigators also reported rebanding experiments (light, intermediate, and heavy fractions were separately centrifuged in CsCl) which showed that the intermediate fraction dissociated into heavy and light fractions. These experiments agree with our results for virus (Fig. 1 and 4). Faulkner and Dobos (4) suggested that the intermediate form may consist of a microaggregate of Sindbis virions (heavy fraction) surrounded by a fixed amount of light fraction material. We think that this explanation is unlikely in regard to the results reported in this paper, since infectious virions having densities of 1.2 and 1.23 g/cm3 could not be resolved by sedimentation velocity experiments in sucrose gradients (Fig. 6). The examination of each band by electron microscopy would resolve this question; such experiments are now in progress. The CsCl buoyant density of Sindbis virus has been reported to be.2 to.4 g/cm3 greater than that reported here for virus (4, 7). It was also recently reported that virus, in the form of a 2% mouse brain suspension, banded at 1.21 to 1.23 g/cm3 in CsCl and that no hemagglutinin was detected in the upper regions of the gradient (5). The differences between these data and our results may be related to a number of factors, such as the relative purity of the virus suspension or the number of fractions collected from the density gradient. ACKNOWLDGMNTS We thank Delores Michael for able technical assistance and Orville Brand for providing antisera and goose red blood cells. LITRATUR CrrD 1. Acheson, N. H., and I. Tamm. 1967. Replication of Semliki Forest virus: an electron microscopic study. Virology 32:128-143. 2. Brown, A. 1963. Differences in maximum and minimum plaque-forming temperatures among selected group A arboviruses. Virology 21:362-372. 3. Clarke, D. H., and J. Casals. 1958. Techniques for hemagglutination and hemagglutinationinhibition with arthropod-borne viruses. Am. J. Trop. Med. Hyg. 7:561-573. 4. Faulkner, P., and P. Dobos. 1968. Investigations on the formation and interconversion of Sindbis virus hemagglutinins. Can. J. Microbiol. 14: 45-51. 5. Henderson, J. R., S. I. Levine, N. Karabatsos, and T. B. Stim. 1967. Antigenic variants of arboviruses. II. Variation in antigen expressions of strains during replication in different hosts. J. Immunol. 99:925-934. 6. Mussgay, M. 1964. Growth cycle of arboviruses in vertebrate and arthropod cells. Progr. Med. Virol. 6:193-267. 7. Mussgay, M., and M. Horzinek. 1966. Investigations on complement-fixing subunits of a group A arbovirus (Sindbis). Virology 29:199-24. 8. Osterrieth, P. M., and C. M. Calberg-Bacq. 1966. Changes in morphology, infectivity and hemagglutinating activity of Semliki Forest virus produced by the treatment with caseinase C from Streptomyces albus G. J. Gen. Microbiol. 43:19-3. 9. Pfefferkorn,. R., and H. S. Hunter. 1963. Purification and partial chemical analysis of Sindbis virus. Virology 2:433-445. 1. Stevens, T. M., and R. W. Schlesinger. 1965. Studies on the nature of dengue viruses. I. Correlation of particle density, infectivity, and RNA content of type 2 virus. Virology 27:13-112. 11. Wecker,. 1959. The extraction of infectious virus nucleic acid with hot phenol. Virology 7:241-243. 12. Zebovitz,., and A. Brown. 1967. Temperaturesensitive steps in the biosynthesis of Venezuelan equine encephalitis virus. J. Virol. 1:128-134.