Structure of the Bluetongue Virus Capsid

Transcription

1 JOURNAL OF VIROLOGY, OCt. 1972, p Copyright 1972 American Society for Microbiology Vol. 10, No. 4 Printted in U.S.A. Structure of the Bluetongue Virus Capsid D. W. VERWOERD, H. J. ELS, ETHEL-MICHELE DE VILLIERS, AND H. HUISMANS Molecular Biology Sectionl, Veterinary Research Iitstitute, Oiiderstepoort, Pretoria, Republic of South Africa Received for publication 5 July 1972 Seven polypeptides were found to be present in the capsid of the bluetongue virus (BTV), four of which are major and three are minor components. This number and size distribution is the same as that found in reovirus, which has a similar segmented, double-stranded ribonucleic acid genome. In both viruses an excellent correlation is found between the molecular weights of certain genome segments and those of the polypeptides, suggesting a direct coding relationship between them. Careful comparison, however. indicates that the correlation patterns of the two viruses are different, suggesting that different genome segments may be utilized to code for certain capsid proteins. This phenomenon may be related to the fact that these viruses can be distinguished morphologically, reovirus having a larger diameter and possessing a double-layered capsid against the smaller single capsid of the BTV. Utilizing the degrading effect of CsCl gradients, it was demonstrated, however, that two of the BTV polypeptides are present as a diffuse protein layer surrounding the capsid. This outer layer probably has some of the functions of the reovirus outer capsid, one being the "masking" of a viral transcriptase which could be demonstrated after its removal. Bluetongue virus (BTV) is one of a group of viruses characterized by the possession of a segmented, double-stranded ribonucleic acid (RNA) genome (22). A remarkable fact about this group is that it encompasses viruses with mammalian, insect, and plant hosts. All the members so far identified, with the exception of reovirus, are either transmitted by arthropod vectors or have an insect as the primary host. Reovirus is also distinguished from the other members by the fact that it is not associated with a specific disease and by its ubiquitous distribution among mammals. Morphologically, a distinct division can be made within the group. Reovirus, which can be regarded as the type species of the one subgroup, is 75 nm in diameter and possesses a double-layered capsid. The outer layer was generally thought to consist of 92 capsomeres, but it is likely to have a different type of capsid structure with a trimer or dimer arrangement of structural units (1, 20). BTV, which typifies the other subgroup, has been reported to be 55 nm in diameter and its capsid to consist of a single layer of 32 morphological units or capsomeres (8) Ṫhese differences in the structure of the protein coat are reflected in both the biological and the physical characteristics of the respective viruses. For example, no serological relationship has been found between BTV and reovirus, and hemagglutination, a common characteristic of reoviruses, has never been observed in the case of BTV. Reovirus is also relatively stable to both high temperatures and acidity, whereas BTV is very sensitive to both. It is obvious, therefore, that in spite of a strong resemblance in the structure of their RNA genomes, the two viruses differ considerably in their protein capsids. It was therefore of considerable importance to study in greater detail the polypeptide composition of the BTV capsid and to attempt a clarification of the relationship between the two viruses. Of equal importance was an attempt to find a correlation between the genome segments and polypeptides in BTV as part of an investigation of the biological function of the viral components. The structure of reovirus capsid proteins has been investigated by Loh and Shatkin (13) and by Smith, Zweerink, and Joklik (18). Both groups reported seven polypeptides, three of which occur in the core particle. One of the four residual polypeptides was demonstrated not to be a primary gene product, but to be derived from one of the other polypeptides, leaving six primary products in the capsid (25). Recently, two additional "noncapsid" polypeptides were found in reovirus-infected cells. A reasonably good correlation was found between the molecular weights of the eight primary gene products and 783

2 784 VERWOERD ET AL. J. VIROL. eight of the ten genome segments, indicating a possible "gene" function for the segments (27). A similar investigation of the polypeptide components of BTV capsid protein is reported in this communication. In addition, a method for the selective removal of certain capsid proteins from BTV is described and the activation of the viral transcriptase is demonstrated. MATERIALS AND METHODS Virus and cells. Type 10 BTV and type 3 (Abney strain) reovirus were used throughout this work. Virus was produced mainly on monolayers of BHK-21 cells. Origins of both viruses and cells, and methods used in their culture, have been described previously (9). Virus production and purification. BHK monolayers were washed once with Eagle minimal essential medium (MEM) without serum and inoculated with virus at an input multiplicity of 10 to 20 plaqueforming units (PFU)/cell. After incubation at 37 C for 36 to 48 hr the cells were shaken off, centrifuged into a pellet, and stored at 4 C until used for purification. The purification procedure used for both viruses was essentially the same as that described previously (24). Cells derived from 100 roller bottles were washed once in 72 ml of M tris(hydroxymethyl)aminomethane (Tris)-hydrochloride buffer at ph 8.8, and suspended in 150 ml of the same buffer by means of an Ultra-Turrax homogenizer. One-tenth volume of a 1% Sephadex G200 suspension was added and the mixture was extracted with 50 ml of Freon 113 in a cold room at 4 C. The cell debris and Freon phase was washed once with 70 ml of Tris buffer and the combined water phases were extracted two more times with Freon. To the final extract bovine serum albumin was added to a final concentration of 0.5%, followed by one-tenth volume of a 10% Tween-80 solution. The mixture was then extracted with 0.5 volume of freshly distilled ether by shaking in the cold. After adjusting the salt concentration of the water phase to 0.1 M with sodium chloride, virus particles were pelleted through a layer of 40% sucrose in a Spinco 6OTi rotor at 40,000 rev/min for 75 min. Higher velocities tended to disrupt the BTV particles. Pellets were suspended in M Tris buffer and layered on 10 to 30% sucrose gradients in the same buffer. Gradients were centrifuged for 50 min at 24,000 rev/min in a Spinco SW27 rotor. The clearly defined virus bands were collected and pelleted once more at 40,000 rev/min. For most purposes these pellets were of sufficient purity. Isopycnic density gradient centrifugation in CsCl was used for both final purification and selective degradation of BTV (see Results). Preformed 40 to 55% gradients, buffered with 0.2 M Tris-hydrochloride, were centrifuged at 47,000 rev/min in the Spinco SW50.1 rotor for 16 hr. Virus bands were collected and dialyzed in the cold against M Tris buffer at ph 8.8 to remove CsCl. Preparation and assay of radioactive viral protein. Radioactive BTV protein was prepared by labeling virus with "C-protein hydrolysate using a pulse technique. Monolayers of BHK cells were inoculated with BTV as usual in Eagle MEM. After 8 hr at 37 C, the medium was changed to Earle salt solution containing "4C-protein hydrolysate (Radiochemical Centre, Amersham, England) at a final concentration of 0. 1,Ci/ml. After a further 2 hr at 37 C, the medium was replaced with Eagle MEM again, and the cultures returned to 37 C for 36 hr. Virus was purified as described above, dissociated, and electrophoretically separated as described below. After staining, the virus bands were cut out of the polyacrylamide gel, dissolved in 0.2 ml of a 10% piperidine solution and counted in 8 ml of Kinard's scintillator in a Packard Tri-Carb scintillation spectrometer. Infectivity assays. Infectivity of virus preparations were determined by means of the plaque test previously described (9). Dissociation of virus. Virus samples at a concentration of 10.0 mg/ml in Tris buffer were made 0.4% with regard to sodium dodecyl sulfate (SDS) and 1% to 2-mercaptoethanol. SDS was obtained from Mercks, Darmstadt, W. Germany, and recrystallized from ethanol. The samples were then heated for 15 min at 37 C, followed by 1 min at 70 C for BTV and 1 min at 96 C for reovirus just prior to electrophoresis. From 50 to 100,g of the dissociated protein was used for electrophoresis. Polyacrylamide gel electrophoresis. Conditions for electrophoresis of polypeptides were based on those used in the fractionation of poliovirus capsids (19). For comparative purposes and for the determination of the molecular weights of the polypeptides, best results were obtained using a polyacrylamide gel slab and the Ortec 4200 high-resolution electrophoresis system (Ortec Inc., Oak Ridge, Tenn.). Slabs were prepared containing 8% acrylamide in both the separating and well-forming gels. Both also contained 0.24%c N,N-methylenebisacrylamide 0.2% N,N, N',N',- tetramethylenethylenediamine, 0.2% SDS, 6 M urea, 0.02 M ethylenediaminetetraacetic acid (EDTA), 0.05 M phosphate buffer (ph 7.2), and 0.08%/c ammonium persulfate as catalyst. Acrylamide and bisacrylamide were recrystallized following the method of Loening (12). Urea was purified by passing through a mixed-bed deionizer column just prior to use. For urea-containing gels, it was found unnecessary to subject them to pre-electrophoresis, but when urea is omitted, distortion of the bands, probably caused by persulfate in the well-forming gel, can be avoided by overnight pre-electrophoresis. Ethylene diacrylate was used as cross-linking agent at one-fifteenth the acrylamide concentration, in cases where the gel was used for radioactivity determinations. Samples containing about 10 mg of dissociated virus per ml were diluted with an equal volume of a 40% sucrose solution containing some bromophenol red as marker, and 10 to 20,liters applied per well. Electrophoresis was carried out at a potential of 60 v, a capacitance of 1.0 mf, and a rate of 1,000 pulses/sec for 7.5 hr. For optimal resolution in urea-containing gels, a discontinuity in buffer concentration between gel and reservoir was found to be essential. The reservoir buffer therefore contained 0.01 M phosphate, 0.2% SDS, and 0.02 EDTA, adjusted to ph 4.2. Gels were fixed and stained overnight in a solution

3 VOL. 10, 1972 STRUCIURE OF BTV CAPSID 785 of 0.25% Coomassie brilliant blue in a mixture of methanol, acetic acid, and water (5:1:5) and electrophoretically destained in distilled water containing 10% methanol and 10% acetic acid. RNA was fractionated essentially according to the method of Loening (12). Ethylene diacrylate (0.2%) was used as cross-linking agent to prepare 3% gels in Loening's buffer. The time allowed for fractionation of double-stranded RNA components was 4 hr at 100 v for 10-cm gels at 20 C. After electrophoresis, the gels were fixed in 1.0 M acetic acid and stained for 60 min in 0.2% methylene blue solution in 0.2 M sodium acetate buffer of ph 4.0. Destaining of gels was done overnight in distilled water. Scanning of electropherograms. Stained gels were scanned in a Vitatron model TLD100 flying spot densitometer equipped with an integrating recorder. Molecular weight determinations. Molecular weights were determined as described by Weber and Osborn (25) and Dunker and Rueckert (6). The following polypeptides of known molecular weights were used as reference substances to obtain a linear plot of mobility against molecular weight (the latter on a logarithmic scale): bovine serum albumin dimer (132,000) and monomer (66,000), ovalbumin (86,000) and ovalbumin H-chain (43,000) (both from Nutritional Biochemicals Corp., Cleveland, Ohio), glucose oxidase (185,000), alcohol dehydrogenase (84,000), and lactic acid dehydrogenase (33,000) (Seravac, Cape Town, Republic of South Africa). Samples containing 15 to 30 jg of protein were treated with 1% SDS before applying to a slab gel for co-electrophoresis together with dissociated BTV and reovirus samples. Electron microscopy. Negative staining of purified virus was done with 3% phosphotungstate as previotusly described (8). Virus particle counts were obtained using a loop drop method and a Dow latex particle suspension as standard reference. Assay for polymerase activity. The standard assay mixture contained per 0.5 ml: 0.5/,Mmole each of the four ribonucleotide-triphosphates; Tris-hydrochloride, 30 Mmoles; MgCl2, 3 Mmoles; phosphoenol-pyruvate (PEP), 1 Mumole; pyruvate kinase, 25,g; MnCl2, 1 umole; dithiothreitol (DTT), 1 mole; and 200,ug of virus. 3H-labeled uridine triphosphate was added to a final concentration of 2 MtCi/Mimole. Samples were incubated for 3 hr at 28 C unless otherwise indicated, and the reaction was stopped by the addition of carrier yeast RNA and 1 ml of 10% trichloroacetic acid containing 0.05 M sodium pyrophosphate. After 1 hr at 4 C, the precipitates were collected by filtration on Millipore filters which were then dried and counted in a toluene scintillator solution in a Packard scintillation spectrometer. RESULTS Electrophoretic fractionation of BTV polypeptides. BTV type 10 was purified, dissociated, and analyzed by means of polyacrylamide gel electrophoresis as described under Materials and Methods. The stained gels were scanned in a flying spot densitometer and a typical example of the pattern obtained is shown in Fig. 1A. For A. ITV S. REOVI RUS FIG. 1. Densitometer tracings of the fractiontationi pattern2s obtained for BTV (A) and reovirus (B) usinig polyacrylamide gel electroplhoresis. Migration is from left to riglht. comparison, a fractionation pattern of reovirus polypeptides, prepared in the same way, is shown in Fig. lb. In both viruses a distinction can be made between major components, present in relatively large amounts, and minor components which are represented by either very small or intermediate sized peaks. Four major components are found in both viruses and although their molecular weights, reflected by the rates of migration, are different, their distribution into three size classes, of which the largest contains two, are obviously similar. An equal number of three minor components is also found in the two viruses, but in this case the mobilities of the components, relative to those of the major components, are quite different. These minor components were present in all of more than 50 different virus preparations analyzed. Their concentrations relative to those of the major components, as judged by the size of the relevant peaks in densitometer tracing, remained fairly constant. These observations indicated that they probably represent viral polypeptides. In addition to the seven components constantly present, a number of very faint bands could sometimes be seen in BTV purified on sucrose gradients only. Both their occurrence and their relative amounts, when present, were variable, however, suggesting that they represent cellular contaminants. To determine their identity, BTV was isolated from a mixture of uninfected,

4 786 VERWOERD ET AL. J. VIROL. labeled cells and infected, unlabeled cells. A detectable number of counts was only found in the variable fraction with a mobility higher than that of the fastest major component. The absence of label in the other minor components cannot be regarded as proof of their viral origin, however, because their concentration is so low that the level of labeling attained in the experiment could conceivably have been insufficient to allow detection. Virus was also isolated from a different host cell, i.e., from L-strain fibroblasts instead of BHK-21. Again the result was not conclusive. Most of the faint bands were also seen in preparations from L cells, but again their presence and concentrations were variable from one preparation to another. In a final attempt to solve the problem, other methods of purification of the virus was investigated. Isopycnic banding in CsCl density gradients proved to be the most successful. Purification and degradation of BTV by CsCl density gradient centrifugation. It has been known from previous work (21) that BTV is sensitive to high salt concentrations, and that 90% or more of the infectivity of a preparation is lost even after only 2 hr of centrifugation through a CsCl gradient. These experiments were all done in neutral, unbuffered solutions. While re-investigating the possibility of purifying BTV by isopycnic banding for the purpose outlined above, the effect of varying the ph in the gradient was investigated because the virus is also known to be acid-sensitive in contrast to reovirus which is stable at low ph values. The ph was found to have a profound influence on the equilibrium pattern obtained, as shown in Fig. 2. Three identical runs of the same BTV preparation are shown in Tris-buffered CsCl gradients of ph 8.0, 7.0, and 6.0, respectively. The set of gradients used for the photographs were heavily loaded to show the faint bands of low density. Identical results were obtained when one-tenth the amount of virus or less was used. Such gradients were used to determine the buoyant density of the bands by collecting drops from the bottom of the tube and measuring its refractive index. At ph 8.0 the main band formed has a density of 1.38, which is the same as that previously found for infective virus. When analyzed by means of polyacrylamide gel electrophoresis (Fig. 3), this band was shown to possess all seven polypeptides constantly present in sucrose gradient-purified virus, but to have lost the additional variable components. This result seems to indicate that these components are indeed of cellular origin and not part of the viral capsid. ph 8-.0 p H 6.0 ph 7.0 FiG. 2. Isopycnic density gradienit sedimentation patterns of BTV in 40 to SS% CsCI gradients buffered wit/i 0.2 m Tris-HCI to the ph values inzdicated. Sedimentation conditions: 16 hr at 200,000 X g at 4 C. In addition to the main 1.38 band, the ph 8.0 gradient also has a band of higher density (1.39) and a faint band of lower density (1.29). The 1.39 band coincides in position and density with the main band formed in the ph 6.0 gradient. In Fig. 3 this band represents viral particles which have lost the major polypeptide with the lowest mobility, thereby increasing its relative RNA content and therefore its buoyant density. In addition, the ph 6.0 gradient also has a virus band in the 1.38 density region, representing complete virions, as well as the 1.29 band. The ph 7.0 gradient in Fig. 2 gives the best illustration of the degradative process, because it clearly shows all four density bands obtained. In addition to the 1.38 and 1.39 bands already discussed, a main band of still higher density, i.e., 1.42, is found. Gel electrophoresis (Fig. 3) demonstrated that it consists of virions that have lost two polypeptides: the one that is lost at ph 6.0 plus the major polypeptide component of medium mobility. All three gradients have bands of the same low density of about This band is sometimes resolved into two bands very close to each other. Electrophoretic analysis shows that this band represents the polypeptides lost from the virions with higher densities. The apparent discrepancy in the effect of rising ph on the degradation of BTV in CsCl gradients can possibly be explained by another interesting observation. It was found that if the CsCl solutions were buffered with m Tris instead of 0.2 m, the pattern of degradation at ph 8.0 was the same as at ph 7.0. The presence of the much higher CsCdconcentration seems to exclude the ionic strength of the Tris-ion per se as an

5 VOL. 10, 1972 STRUCTURE OF BTV CAPSID 787 explanation and to indicate a spe cific stabilizing fectivity of BTV after short exposure to the salt. effect of the Tris-ion at ph 8.' 0. At ph 5.0 We therefore attempted to correlate the loss of essentially the same result was ( obtained as at certain polypeptides, described in the previous ph 6.0, and a further increase in acidity led to section, with the loss of infectivity. The various complete disruption of the virions3. BTV fractions from CsCl gradients were col- Infectivity of partly degraded BTV particles. lected, dialyzed for 3 hr against M Tris- CsCl gradi- hydrochloride buffer at ph 8.0, and titrated by Previous experiments in unbuffered means of the standard plaque assay. Electron ents indicated an almost complewte loss of in- microscope particle counts were done on the same samples as described under Materials and -._4 Methods. The results are summarized in Table 1. It is obvious that the loss of either one or both of the polypeptides by the virion results in an almost complete loss of infectivity as reflected by its ability to form plaques. That this is not due to instability of the resulting particles is shown by the 10,000-fold increase of the particle number to plaque-forming unit ratio. Infectivity could be partly restored by mixing equal amounts of the 1.42 and 1.29 fractions D _ derived from a ph 7.0 gradient, and dialyzing overnight at 4 C against 0.2 M Tris-hydrochloride buffer (ph 8.0) to allow recombination (Table 1). _ " Conceivably, the function of the two polypeptides in the 1.29 density fraction might have up, Aml something to do with adsorbtion to or penetration of the host cell by the virus. To test for this ls di d possibility, i.e., that the high-density BTV particles have essentially lost their ability to adsorb RED REO BSTV TY STY B T V T V to or penetrate cells, the effect of adding reagents SUCROSE that is known to facilitate infection with viral RNA was investigated. Cell monolayers were ctrophoresis of treated for one minute with Earle salt solution FIG. 3. Polyacrylamide gel ele polypeptides *from reovirus cores reovirus (1.36), BTV purified on sul BTV densities 1.38, 1.42, and 1.39 c CsCl gradients with ph values of 8.1 respectively (Fig. 2) and the 1.29 de) the ph 7.0 gradienit. Virus fractions antd electrophoretic conditions were unider Materials and Methods. Migra. to bottom. (density 1.44), containing 10%0 2crose dimethyl sulfoxide and 160 gradients, g of diethylaminoethyl-dextran per ml just is isolated from prior to infection. Only a slight enhancement of si,y7b.and fr6om the infectivity of fraction 1.42 was obtained were dissociated (Table 1). those described Morphology of partly degraded BTV particles. tion is from top In previous electron microscopy studies on the morphology of BTV (8), virus preparations TABLE 1. Infectivity of bluetonigue virus (BTV) fractions from density gradients of BTV band (g/cm3) Infectivity (PFU/A260 Particle count Gradient Density of BTV band (g/cm3) unit) per PFU Sucrose 1.1 X CsCl PH X CsCl ph X X 105 CsCl PH X X 105 CsCl PH X 104 CsCl PH a 1.4 X 107 CsCl ph DEAE-D + DMSOb 1.0 X 105 a One A260 unit of each band were combined and dialyzed overnight against 0.2 M Tris-hydrochloride buffer, ph 8.0, at 4 C to allow recombination. I I Cell monolayers were treated with diethylaminoethyl-dextran (DEAE-D) and dimethyl sulfoxide (DMSO), as described in the text, before infecting with virus for the plaque titration.

6 788 VERWOERD ET AL. J. VIROL. purified by means of sucrose density gradients were mainly used. These preparations contained only relatively few particles in which the capsid structure could be seen completely unobstructed. Preparations purified on CsCl gradients contained many more of these particles, which were considered to represent highly purified virions. In the light of the discovery that CsCl partly degrades the BTV particles, this assumption had to be reinvestigated. In Fig. 4 the morphology of the different types of undegraded and selectively degraded particles are shown. Virions from sucrose gradients (Fig. 4a) were generally amorphous in appearance, showing little or no symmetrical arrangement of morphological units. They seem to be covered by an outer layer or appear to be "elaborated by surface projections" as reported by Bowne and Ritchie (3). Particles often had an irregular appearance when the outer layer becomes detached or permeable to the negative stain. The diameter of this type of BTV particle averaged 68 nm. The virions with a density of 1.38 obtained from a CsCl gradient of ph 8.0 (Fig. 4b) were indistinguishable from those described above. The 1.42 density fraction obtained from ph 7.0 gradients (Fig. 4d) corresponded to the nucleocapsids previously described by us (8), showing 32 clear morphological units arranged in a symmetrical particle with a diameter of 55 nm. The particles with an intermediate density (1.39) also had an intermediate appearance showing some symmetrical arrangement (Fig. 4c). Fig. 4e represents aggregates of polypeptides dissociated from the low-density virions by the action of the CsCl and constituting the 1.29 density band. Finally, Fig. 4f illustrates the reassociation obtained when nucleocapsids (Fig. 4d) were dialyzed with the dissociated polypeptides (Fig. 4e) against buffer with a high salt concentration. Clearly the resultant particles have reverted to the original structure represented by Fig. 4a-c. Many particles in all the BTV fractions examined displayed lesions or holes filled with phosphotungstic acid, or were lying on positions of fivefold symmetry. It is not yet clear what the explanation for this phenomenon is, but it could mean that pentamers are being lost more easily than hexamers, or perhaps it could be explained by a dimer or trimer arrangement of the structural units. Molecular weights of the polypeptide components. The electrophoretic mobilities of polypeptides in SDS-containing polyacrylamide gels have been shown to reflect a linear relationship to the logarithms of their molecular weights by several workers (6, 25). Marker proteins enumerated under Materials and Methods were therefore subjected to co-electrophoresis on the same gel slabs as dissociated BTV and reovirus, and their mobilities were used to construct a plot from which the molecular weights of the viral polypeptides were determined. Reovirus was included in these experiments both as a positive control and to allow a direct comparison of the polypeptide composition of the two viruses under identical conditions. Molecular weights obtained are shown in Table 2. In order to investigate a possible relationship between the size of the polypeptides and the size of the genome segments as described previously for reovirus (26), the molecular weight of the polypeptide that would theoretically be coded for by each genome segment was calculated. It was assumed that each segment is fully transcribed and translated, and a conversion factor of 18 was used, based on the ratio of the average molecular weight of double-stranded RNA to that of a polypeptide of equivalent size (26). During the course of this work, it was found that some of the previously reported molecular weights for the genome segments were inaccurate, probably due to difficulties in comparing mobilities using separate gels. The values therefore had to be redetermined using either gel slabs or co-electrophoresis of BTV RNA and reovirus RNA on the same gel. Such an experiment is illustrated in Fig. 5. Molecular weight values were obtained by plotting our reovirus mobilities against the molecular weights published by Shatkin, Sipe, and Loh (16) and by reading off the final values for both viruses from the line of best fit. These values were used for the calculation of the theoretical molecular weights of the polypeptides in-table 2. The seven polypeptides of BTV were assigned a position of best fit in Table 2. Their molecular weights represent the average results from five experiments. In the case of reovirus, the positions of the polypeptides relative to the ten genome segments are those proposed by Zweerink et al. (27). Major components have been italicized for convenience. In both viruses a remarkably good correspondence is seen between the calculated and the experimentally determined values for most of the polypeptides, providing strong evidence that the presumption of a direct coding relationship between the genome segments and the polypeptides is indeed valid. It further indicates that most of the genome segments are fully expressed. One polypeptide in the case of reovirus, with a molecular weight of 81,000, would correspond best to the size of segment 6, but it has been

7 VOL. 10, 1972 shown (26) to be derived from the primary gene product 4 by the splitting off of a small peptide. In the BTV series, two of the minor components, shown in parentheses opposite segments 4 and 7, show rather large deviations from the calculated STRUCTURE OF BTV CAPSID 789 size. These polypeptides could conceivably also be derived from primary products by hydrolysis as in reovirus, they could represent incomplete expressions of their genome segments, or each could be coded for by more than one segment. I I e ] ~~~~~~t FIG. 4. BTV particles negatively stained with phosphotungstate. Final magnification in all figures is X 160,000. a, BTV particles from sucrose gradient. Note diffuse appearance. b, BTV particles from CsCl gradient, ph 8; their appearances are very similar to that of (a). c, BTV particles from CsCl gradient, ph 6; more surface detail can be discerned. d, BTV particles from CsCl gradient, ph 7; note the appearance of a regular arrangement of morphological units that can be discerned compared with (c), (b), and (a). In addition, a reduction in size ofparticles in (d) compared to (a) is obvious. e, Aggregates ofprotein components lost from virus particles shown in (d) and recovered in the 1.29 density band. f, Virus particles obtained as the result of a recombination of the fractions represented by (d) and (e). An increase in size and a more diffuse type ofparticle compared with (d) is again obtained.

8 790 VERWOERD ET AL. J. VIROL. BFV REO REO BTV xl x 1t E i m I i 250 S x i9 X t 1zo 4. l- '60 X l X10 fl x 1o6 _ 9. 08C4x x lop -X--,1.." :'5-2M ^^^*. ~4. -5 _ ~~~~ X X X 106.,,... -_10 o030 x 10 FIG. 5. Polyacrylamide gel electrophoresis of the RNA segmentts of BTV antd reoviruts geliomes. Molecular weightts of the segmenits as ilidicated were determined from the middle gel where the two viral RNA preparationis were subjected to co-electrophoresis, as described ini the text. Migrationt is from top to bottom. An attempt was made to determine whether all the BTV polypeptides are primary gene products by means of the pulse-chase technique used for reovirus (27). The results were inconclusive, mainly because the amount of BTV protein synthesized in an infective cell is only a fraction of that found in reovirus-infected cells. In spite of the fact that BTV causes an inhibition of host cell protein synthesis, sufficient BTVspecific protein synthesis could not be obtained to distinguish it from cellular products. The assignment of certain other components to specific positions in Table 2 are somewhat arbitrary. It is, for example, impossible to determine whether the major component in reovirus with a molecular weight of 138,000 is derived from genome segment 1 or 2, or perhaps from both, because the molecular sizes of the RNA segments are so similar that they are not sepa- TABLE 2. Molecuilar- weights of polypeptides in bluetongue viruts (BTV) an1d reoviruts Equivalent genome segment Molecular weights, BTV Molecular weights, reovirus Calculated" Fioundb Iaclte ae Fud Calcl :ud 138, , ,100 72,700 64,400 60,000 33,300 30,000 27,700 16, , , ,000 (82,000) 61,000 (42,000) 29, , , ,300 88,800 86,000 82,700 48,300 42,200 35,500 33, , ,000 89,000 (81,000) 47,500 40,500 34,500 a For calculation see text. Major components have been italicized. Parentheses indicate uncertain assignments (see text) S x 10- rated by gel electrophoresis under our conditions s 106 Similarly the polypeptide of molecular weight -? O054x 10 29,500 in BTV, shown opposite segment 8, could X 10 also possibly be derived from segment 7 or even 9, or perhaps from more than one of them. Attempts to obtain a resolution of component 8 by using higher polyacrylamide concentrations and smaller sample volumes were unsuccessful. An interesting result that emerges from Table 2 is the fact that the major polypeptide components of the two viruses, with the exception of the polypeptide coded for by segment 3, do not seem to be derived from equivalent genome segments. Thus polypeptide 1, which in reovirus is a major component, is only present in minor amounts in BTV. Conversely, polypeptides 6 and 2, both major components in BTV, are completely absent in reovirus, whereas polypeptide 10 is present in reovirus as a major component but is absent from BTV. Molar ratio of BTV polypeptides. To determine the relative amounts of the different polypeptides in the bluetongue virion, virus protein was labeled with 14C-protein hydrolysate, dissociated, and subjected to gel electrophoresis as described above. A series of virus concentrations was applied to the same gel slab and, after staining, the bands were cut out and the radioactivity was determined. The values obtained were averaged by plotting it against the relative concentration (Fig. 6) and using the linear relationship to calculate for each component the fraction of the total radioactivity it represents. These values were divided by the molecular weights of the respective polypeptides to obtain their molar ratios (Table 3). An estimate of the number of molecules of each of the polypeptides per virus

9 VOL. 101) 1972 STRUCTURE OF BTV CAPSID 791 pl BTV FIG. 6. Lin2ear relationiship betweeni the volume ofbtv samples labeled with '4C-amino acids atdfractionzated by meanis ofpolyacrylamide gel electrophoresis and the amount of radioactivity recovered for each band. Tlhese curves were used to obtain average values Jor the calculationz of molar ratios as showii in Table 3. Curves are ntumbered accordinig to the system followed in Table 2. TABLE 3. Molar ratios of bluetongue virus polypeptides Fraction of Molecular M\ola rtio Estimated Moecultar total ar ra-6 number per weighta Fraciountof 6c10 viriond 140, , , , , , , a Major components have been italicized. b Calculated from values obtained from Fig. 6. c Molar fractions were obtained by dividing the fraction of total counts for each polypeptide by its molecular weight. d Obtained by multiplying each molar fraction (column 3) by the estimated total molecular weight of the viral protein. particle was made by multiplying the estimated total molecular weight (MWt) of the viral protein by the molar ratios obtained. MWt was derived from the total molecular weight of the BTV- RNA (1.1 x 107, see Fig. 5) and the previous demonstration that BTV contains close to 20% nucleic acid (21). The resultant values are shown in Table 3. RNA polymerase activity in degraded bluetongue virions. It has been shown by various workers that the removal of the outer capsid of reovirus by means of chymotrypsin or heat treatment induces a RNA polymerase activity (2, 15, 17). The outer layer which is removed is composed of polypeptide components 5 (plus 4), 7, and 10 (, 1,,u 2, 6 1, and 6 3 in Joklik's notation). The inner core particle therefore consists of polypeptides 1, 2, and 8 (X 1, X 2, and 6 2), and the transcriptase activity is probably associated with either one or all three of them.

10 792 VERWOERD ET AL. J. VIROL. TABLE 4. Polymerase activity of bluetonigue virus (BTV) fractions from density gradients Density of Polymerase Gradient BTV band activity (counts Gradient (g/cm3) (cm) per unit/2 min/a2s60 hr) Sucrose 0 CsCI ph CsCI ph CsCl ph CsCl ph Chymotrypsin and heat have been shown to have no equivalent effect on BTV nor do any other proteolytic enzyme, detergents, or a variety of other chemical and physical treatments tested (23). The discovery during this work that BTV can be selectively degraded on buffered CsCI gradients immediately suggested the possibility that an equivalent enzyme might be "activated" by this treatment. Virus bands of the different densities shown in Fig. 2 were collected for their respective gradients, dialyzed, concentrated, and incubated with a standard incubation mixture as described under Materials and Methods. RNA-polymerase activity was clearly demonstrated in virions which have lost polypeptides 2 and 6, or 2 only, as shown in Table 4. A complete characterization of the enzyme will be published elsewhere (Verwoerd and Huismans, in press). It is significant for the present discussion, however, that it can be distinguished from the reovirus enzyme, both by its dependence on manganese ions and in its optimal temperature of only 28 C. DISCUSSION Electrophoretic analysis on SDS- and ureacontaining polyacrylamide gels demonstrated the presence of seven polypeptides in bluetongue virions. Four of these represent major and three represent minor components with regard to their relative concentrations. This result is identical with that reported for reovirus (18), probably reflecting the similarity of the genomes of these two viruses. As far as the mobilities of the various components are concerned, the electrophoretic patterns obtained for the two viruses look quite different, however, and can easily be distinguished from one another (Fig. 1). Since it was known that the 10 genome segments of BTV are not identical in size to those of reovirus, although of similar size distribution (24), it was conceivable that the dissimilarities found in the mobilities of the polypeptides could be explained by these differences, provided that a direct coding relationship exists between genome segments and polypeptides. To test for this possibility, the molecular weights of the polypeptides were determined by means of a comparison of their mobilities with those of known marker proteins. These values were then compared with the calculated molecular weights of theoretical polypeptides derived by a process of full transcription and translation from the respective genome segments. Two main results were obtained. First, an excellent agreement was found between the calculated and the experimental values for the majority of polypeptides in both viruses (see Table 2). Although it is no proof, this result can be regarded as strong evidence that a direct coding relationship indeed exists between the genome segments and polypeptides. Again, this confirms a similar deduction previously reported for reovirus. In the second place, the differences between the molecular weights of the equivalent RNA segments in the two viruses could not explain the dissimilarities in the polypeptide fractionation patterns. Careful matching of the polypeptides with the genome segments indicates that these dissimilarities are due to the fact that the main polypeptides of the two viral capsids do not seem to be derived from the same genome segments. Thus, the four major components of BTV correlate best with genome segments 2, 3, 6, and 8, whereas the major reovirus polypeptides are considered to be derived from genome segments 1, 3, 4, and 9. Genome products apparently present in one virus but absent in the other are 2 and 6 in BTV, and 10 in reovirus. A certain degree of uncertainty exists about the precise assignment of some of the polypeptides. For example, component 3 in reovirus could be coded for by either or both of genome segments 2 and 3, and component 8 in BTV could possibly be derived from segment 7, although in size it is closer to segment 8. Component 5 in reovirus is shown in parentheses in Table 2 because it is not a primary gene product but derived from polypeptide 4. Attempts to detect a similar conversion of primary gene products into secondary products in the case of BTV failed. Although the possibility cannot be excluded completely at this stage, this does not invalidate the deductions made, as it is clearly impossible, even after taking the uncertainties into account, to arrange the two sets of molecular weight values (Table 2) in such a way that identical correlation patterns are obtained for the two viruses. The discovery that BTV can be degraded selectively in CsCl gradients at different ph values opened up further possibilities of com-

11 VOL. 10, 1972 STRUCTURE OF BTV CAPSID parison. In reovirus the outer capsid, which can be removed by treatment with chymotrypsin, consists of polypeptides 4, 5, 7, and 10. In the case of BTV, which does not possess an outer capsid, polypeptides 2 and 6 can be selectively removed in CsCl without disrupting the nucleocapsid. It is probably significant that they are equivalent to the only two reovirus polypeptides (X 2 and,u 3) which, so far, have not been found in either the virion or the infected cell. Electron microscopy revealed that these two polypeptides form a diffuse protein layer around the capsid proper, obscuring the arrangement of structural units in the latter but not possessing a definite structure itself. This observation solves a discrepancy that has existed for some time between reports on the structure and size of the bluetongue virion. On the one hand, particles were described with a diameter of 68 nm and reoviruslike structure not showing typical capsomeres, but possessing surface projections (3, 14). On the other hand, we described a particle 55 nm in diameter, consisting of 32 clearly discernible capsomeres, and regarded the larger particles as contaminated with cellular material (8). Obviously, the former particle represents the virion possessing the outer layer whereas the latter particles are the nucleocapsids which have lost this layer. Evidence that the outer layer is indeed virus specific derives from the fact that the molecular weights of the constituent polypeptides correlate closely with those of two of the viral genome segments, as well as from infectivity studies. Infectivity to particle count ratios clearly demonstrates that infectivity is lost on removal of the outer layer and regained partially after recombination (Table 1). The structure of the BTV capsid that evolves from this work is closer to that of reovirus than we previously thought, both in the size of the complete particle and in the possession of a double layer. However, the way in which these layers are assembled still seems to be quite different for the two viruses. In reovirus the outer layer consists of structural units arranged in a regular, though still undetermined, way and the inner capsid is the lesser structured part. In BTV the opposite is true; the outer layer is diffuse and nonstructured, whereas the inner layer consists of structural units clustered as conventional pentamer-hexamer morphological units or capsomeres. These differences in capsid structure form the basis of important differences in the characteristics of the two viruses. For example, the greater stability of reovirus is probably due to its outer capsid. Likewise, the absence of hemagglutination 793 in BTV, as well as other biological and serological differences, are mainly dependent on differences in the outer protein layer. The greater stability of reovirus might also be related to the differential effect of the two viruses on cellular protein synthesis. Huismans has shown that, in contrast to reovirus, BTV has a definite inhibiting effect on cellular protein synthesis following infection (10). In enteroviruses an equivalent effect is caused by double-stranded RNA (5). In reovirus the inner particle remains intact throughout infection (4). Although it has not yet been determined whether the BTV nucleocapsid remains intact during the infectious cycle, it is quite reasonable to expect that the much less stable capsid of this virus could result in liberation or leakage of its double-stranded RNA genome which could be responsible for the inhibition of protein synthesis. A common function of the outer layers of the two viruses appears to be the masking of the viral transcriptase. The fact that no eauivalent polypeptides, i.e., polypeptides derived from equivalent genome sections, are found in these layers suggests that the masking is probably a nonspecific effect. The possession of similar transcriptase activities can certainly be regarded as an indication of relationship, but again the characteristics of the two enzymes are sufficiently distinct to warrant differentiation between BTV and reovirus. Of special interest is the observation that in BTV the transcriptase has an optimal temperature of 28 C in comparison with the 47 to 52 C of reovirus (11). This is probably related to the fact the BTV is transmitted by, and has been shown to replicate, in an insect intermediate host. Both the low optimal temperature of the transcriptase and the fact that the yield of infective BTV in animal cells at 37 C is only about 20% of that of reovirus might serve as indication that BTV is basically an insect virus, the mammalian host being only of secondary importance. In conclusion, the results reported in this paper have some bearing on the problem of classification of the double-stranded RNAcontaining viruses. It has been proposed that all the viruses possessing a segmented doublestranded RNA genome be classified together in one genus, the diplornaviruses (22). Mainly based on morphological and biological grounds, it was proposed that a subdivision was necessary with reovirus and BTV as the two type species. Although it remains obvious that there is a basic relationship between the two groups, the differences in the protein coat structure reported here strengthens the arguments for a distinction between them. At this stage, however, it still seems desirable to retain them in one genus,

12 794 VERWOERD ET AL. J. VIROL. with a subdivision into subgenera reflecting the capsid structure. Further work is necessary to see if the capsid structure can be correlated with the host specificity of these and other related viruses. Too little is known at this stage about the capsid structure of the considerable number of viruses provisionally included in the BTV-group (22, 14) to decide if they should be classified together. The possibility cannot be excluded that other forms of capsid arrangement could exist. A monkeyderived virus, S.A. 11, for example, has been shown to possess an outer layer staining much more distinctly than that of BTV with phosphotungstate (7). This could possibly represent an intermediary form. To clarify the relationship between these viruses, further comparative studies are in progress. ACKNOWLEDGMENTS The continuing support of the Director of the Onderstepoort Veterinary Research Institute of the South African Department of Agricultural Technical Services is gratefully acknowledged. The nvestigation was only possible with the excellent technical assistance of J. Broekman, P. A. L. Eksteen, and P. A. M. Wege. LITERATURE CITED 1. Amano, Y., S. Katagiri, N. Ishida, and Y. Watanabe Spontaneous degradation of reovirus capsid into subunits. J. Virol. 8: Borsa, J., and A. F. Graham Reovirus: RNA polymerase activity in purified virions. Biochem. Biophys. Res. Commun. 33: Bowne, J. G., and A. E. Ritchie Some morphological features of bluetongue virus. Virology 40: Chang, C. T., and H. J. Zweerink Fate of parental reovirus in infected cells. Virology 46: Cordell-Stuart, B., and M. W. Taylor Effect of doublestranded viral RNA on mammalian cells in culture. Proc. Nat. Acad. Sci. U.S.A. 68: Dunker, A. K., and R. R. Rueckert Observations on molecular weight determinationis on polyacrylamide gel. J. Biol. Chem. 244: Els, H. J., and G. Lecatsas Morphological studies on simian virus S.A. 11 and the related "O-agent". J. Gen. Virol. 17: in press. 8. Els, H. J., and D. W. Verwoerd Morphology of bluetongue virus. Virology 38: Howell, P. G., D. W. Verwoerd, and R. A. Oellermann Plaque formation by bluetongue virus. Onderstepoort J. Vet. Res. 34: Huismans, H Host cell protein synthesis after infection with bluetongue virus and reovirus. Virology 46: Kapuler, A. M An extraordinary temperature dependance of the reovirus transcriptase. Biochemistry 9: Loening, U. E The fractionation of high-molecularweight ribonucleic acid by polyacrylamide-gel-electrophoresis. Biochem. J. 102: Loh, P. C., and A. J. Shatkin Structural proteins of reovirus. J. Virol. 2: Murphy, F. A., E. C. Borden, R. E. Shope, and A. Harrison Physicochemical and morphological relationships of some arthropod-borne viruses to bluetongue virus-a new taxonomic group. J. Gen. Virol. 13: Shatkin, A. J., and J. D. Sipe RNA polymerase activity in purified reoviruses. Proc. Nat. Acad. Sci. U.S.A. 61: Shatkin, A. J., J. D. Sipe, and P. Loh Separation of the reovirus genome segments by polyacrylamide gel electrophoresis. J. Virol. 2: Skehel, J. J., and W. K. Joklik Studies on the il vitro transcription of reovirus RNA catalyzed by reovirus cores. Virology 39: Smith, R. E., H. J. Zweerink, and W. K. Joklik Polypeptide components of virions, top component and cores of reovirus type 3. Virology 39: Summers, D. F., J. V. Maizel, and J. E. Darnell Evidence for noncapsid proteins in poliovirus-infected HeLa cells. Proc. Nat. Acad. Sci. U.S.A. 54: Vasquez, C., and P. Tournier New interpretation of the reovirus structure. Virology 24: Verwoerd, D. W Purification and characterization of bluetongue virus. Virology 38: Verwoerd, D. W Diplornaviruses; a newly recognised group of double-stranded RNA viruses. Progr. Med. Virol. 12: Verwoerd, D. W Failure to demonstrate in vitro as opposed to in vivo transcription of the bluetongue virus genome. Onderstepoort J. Vet. Res. 37: Verwoerd, D. W., H. Louw, and R. A. Oellermann Characterization of bluetongue virus ribonucleic acid. J. Virol. 5: Weber, K., and M. Osborn The reliability of molecular weight determinations by dodecyl sulphate-polyacrylamide gel electrophoresis. J. Biol. Chem. 244: Zweerink, H. J., and W. K. Joklik Studies on the intracellular synthesis of reovirus-specified proteins. Virology 41: Zweerink, H. J., M. J. McDowell, and W. K. Joklik Essential and nonessential noncapsid reovirus proteins. Virology 45: