J. gen. Virol. (1982), 63, 457-467.

Printed in Great Britain 457

Key words: rotavirus/immunoperoxidase/uhrastructure/glycoprotein

Localization of Rotavirus Antigens in Infected Cells by Ultrastructural

Immunocytochemistry
By B E T T Y L. P E T R I E , I * t D A V I D Y. G R A H A M , 1'2 H E N R Y H A N S S E N 1
AND M A R Y K. E S T E S 1'2
i Department o f Virology and Epidemiology and 2Department of Medicine, Baylor College o f
Medicine, Houston, Texas 77030, U.S.A.

(Accepted 1 July 1982)

SUMMARY
Virus structural antigens were localized within a line of monkey kidney (MA104)
cells infected with the simian rotavirus SA11 using electron microscopic immunoper-
oxidase techniques. When hyperimmune guinea-pig anti-SA11 serum was used, virus
particles, membranes of virus-associated endoplasmic reticulum, and viroplasmic
inclusions were most heavily labelled. A general cytoplasmic reaction (ribosomes,
intracytoplasmic membranes, etc.) with anti-SA 11 serum was also observed, but nuclei
were unstained. In addition, several other virus-induced structures were found to
contain rotavirus proteins, including convoluted smooth membrane within the
endoplasmic reticulum, aberrant virus-like particles, and 15 to 20 nm diam.
cytoplasmic tubules. Monospecific antiserum to VP7 (outer capsid glycoprotein, mol.
wt. 38 000) reacted strongly with virus particles and the virus-associated endoplasmic
reticulum, but reacted poorly with viroplasmic inclusions. The nucleus and general
cytoplasm were unstained with anti-VP7. In contrast, monospecific antisera to VP2
and VP6 (inner capsid proteins, tool. wt. 94000 and 41000 respectively) reacted very
strongly with viroplasmic inclusions. Virus particles, endoplasmic reticulum and
cytoplasmic ribosomes were also labelled with these sera. These results indicate that
rotavirus inner capsid proteins are synthesized throughout the cytoplasm and become
concentrated in viroplasmic inclusions, while the outer capsid glycoprotein is
synthesized primarily on ribosomes of the rough endoplasmic reticulum. Thus, the
outer capsid layer appears to be acquired during virus budding into cisternae of the
endoplasmic reticulum.
INTRODUCTION
Rotaviruses, originally identified as veterinary pathogens (Mebus et al., 1969), are now
recognized as a major cause of gastroenteritis in young children and a variety of mammalian and
avian species (Flewett & Woode, 1978; McNulty et al., 1979). A comparison of the morpho-
genesis of the simian rotavirus SAIl (Lecatsas, 1972; Altenburg et al., 1980) with electron
microscopic observations of rotavirus replication in tissues from infected animals and in
cultured cells (Adams & Kraft, 1967; Banfield et al., 1968; Lecatsas, 1972; Stair et al., 1973;
Holmes et al., 1975; Hall et al., 1976; McNulty et al., 1976; Chasey, 1977; Saif et al., 1978;
Pearson & McNulty, 1979) demonstrated that SA11 is a useful representative model for the
study of rotavirus infections. Our previous studies (Altenburg et al., 1980) show that areas of
dense granular material (viroplasm) begin to accumulate in the perinuclear region of the
cytoplasm 8 h after infection. Virus particles assemble near the periphery of these inclusions and
subsequently enter into cisternae of the rough endoplasmic reticulum by a budding process.
Many particles become enveloped during budding, but the envelope appears to be easily lost
since viruses located near the centres of large cisternae usually lack it. The envelope is not
required fo.r infectivity (Elias, 1977; Estes et al., 1979b; Petrie et al., 1981) and its function
t Previous publications written under the name Betty C. Altenburg.

0022-1317/82/0000-5093$02.00©1982 SGM
458 B. L. P E T R I E AND OTHERS

remains unknown. Rotavirus particles have never been noted to bud through the plasma
membrane, but rather enter the culture medium following lysis of infected cells.
SA 11 morphogenesis also exhibits several features not yet reported for other rotavirus strains.
In 1 to 5 ~ of SA1 l-infected cells, bundles of 15 to 20 nm diam. tubular structures (usually
surrounded by an electron lucent space) can be found within the cytoplasm and/or the nucleus
(Lecatsas, 1972; Altenburg et al., 1980). In addition, virus particles are sometimes found within
the intercristal space of host cell mitochondria (Altenburg et al., 1980). The larger nuclear
tubules observed in cells infected with murine (EDIM, epizootic diarrhoea of infant mice) and
porcine rotaviruses (Banfield et al., 1968; McNulty, 1978) have not been reported in SA11
infections.
Rotavirus antigens have been localized within infected cells by light microscopy using both
immunofluorescence (McNulty et al., 1979) and immunoperoxidase techniques (Graham &
Estes, 1979; Altenburg et al., 1980). Although virus antigens can be detected as discrete
cytoplasmic inclusions by these techniques, it is not possible to distinguish viroplasms from
virus-filled cisternae. Furthermore, the smaller and less frequently observed structures cannot
be resolved by light microscopy. Although the major features of rotavirus morphogenesis have
been well-established, little is known at the molecular level about the assembly of rotavirus
proteins into virions. Preliminary studies on the ultrastructural localization of calf (Chasey,
1980) and SAI 1 (Altenburg et al., 1979) rotavirus antigens, using peroxidase as the immuno-
logical marker, have shown a generalized cytoplasmic distribution of virus proteins in cells
undergoing lysis with heavy labelling of virus particles, viroplasmic inclusions and rough
endoplasmic reticulum. However, these studies used broadly reacting antisera and examined
only a few cells. In the present study, we confirm and extend these results using hyperimmune
antiserum to SA11 rotavirus and monospecific antisera against (i) the major glycoprotein VP7
(mol. wt. 38 000) from the outer shell, (ii) the major inner capsid protein VP6 (mol. wt. 41000),
and (iii) the second most abundant inner capsid protein VP2 (mol. wt. 94000).
METHODS
Cells and virus. Stocks of simian rotavirus SAI 1 were grown at low multiplicity in cultures of foetal rhesus
monkey kidney cells (MA104; Microbiological Associates, Bethesda, Md., U.S.A.) in Eagle's minimal essential
medium (MEM) without foetal bovine serum (FBS) as previously described (Estes et al., 1979b). Stocks were titred
by plaque assay in MA 104 cells under an agar overlay supplemented with pancreatin and DEAE-dextran (Smith et
al., 1979).
Antisera. Hyperimmune antiserum to SAI 1 was prepared in guinea-pigs using purified double- and single-
shelled SAI 1 particles (Mason et al., 1980) as previously described (Estes et al., 1979b). This antiserum has been
shown to immunoprecipitate most of the SA 11 structural proteins (i.e. VP2 to VP9 but not VP1) from infected cell
lysates, whereas preimmune serum from guinea-pigs did not (Mason et al., 1980; Ericson et al., 1982). Immuno-
logical specificity was confirmed in the immunocytochemical experiments by absorbing the antiserum with
double-shelled SA 11 virus purified on CsCI gradients (Mason et al., 1980) as follows : a 40 gtl amount of anti-SA 11
serum diluted 1 : 100 was mixed with 360 gtl virus (260 ~tg/ml) or 0.01 M-phosphate-buffered saline (PBS), then
incubated 1 h at 37 °C and overnight at 4 °C. The immune complexes were removed by centrifugation in a
Beckman airfuge at 20 lbf/in 2 (138 kPa) for 15 min. No peroxidase staining occurred in infected cells incubated
with absorbed serum, while mock-absorbed serum retained this activity.
Monospecific antisera were prepared against VPT, the 38 000 mol. wt. (38K) protein, the major glycoprotein in
the outer capsid layer: VP6, the 41000 mol. wt. (41 K) protein, the major inner capsid protein; VP2, the 94000 mol.
wt. (94K) protein, the second most abundant inner capsid protein by injecting guinea-pigs with purified peptides
eluted from SDS-polyacrylamide gels. The nomenclature of the SA 11 peptides used here is that described by Estes
eta/. (1981 ). Specificity of the monospecific antisera was confirmed by immunoprecipitation of only the indicated
proteins from infected cell lysates (Ericson et al., 1982; H. Hanssen and M. K. Estes, unpublished results). The
anti-VP7 serum has been shown to precipitate with equal efficiency either the glycosylated protein or the
unglycosylated precursor protein detected only after treatment of infected cells with tunicamycin (Ericson et al.,
1982). Preimmune sera from all of the immunized guinea-pigs failed to react with SA 1l-infected cells in immuno-
peroxidase procedures and did not precipitate any proteins from SAil-infected cells.
Ultrastructural immunocytochemistrv. Confluent monolayers of MA 104 cells in 35 mm plastic culture dishes were
washed three times in Tris-buffered saline (TBS) and inoculated with 5 to 10 p.f.u./cell SA 11. After 1 h adsorption
at 37 °C, cells were washed three times with TBS to remove unadsorbed virus and incubated at 37 °C. At various
times after adsorption, cells were washed twice with PBS pH 7.4, and fixed for 7 min in 1 ~ 1-ethyl-3-(3-dimethyl-
 Rotavirus antigens in infected cells 459
aminopropyl)carbodiimide (Sigma) and 0-25 % glutaraldehyde (Electron Microscopy Sciences, Fort Washington,
Pa., U.S.A.) as described by Willingham et al. (1978). After fixation, the cells were washed in PBS and made
permeable to immunological reagents by incubation (30 rain at ambient temperature) in PBS containing 0.02%
saponin and 10% FBS (shown to be free from antibodies to rotavirus by immunofluorescence tests). This solution
was used for all antibody dilutions and for all washes between incubations. Permeabilized cells were incubated for
1 h with guinea-pig antiserum to SA 11 diluted 1:500, with antiserum to VP7, VP6 or VP2 diluted 1 : 100, with
virus-absorbed antiserum or with preimmune guinea-pig serum diluted 1 : 100. The cells were then washed for 30
rain to remove unabsorbed antiserum, reincubated with a 1 : 10 dilution of goat anti-guinea-pig IgG (Cappel
Laboratories, Cochranville, Pa., U.S.A.), followed by a second 30 min wash and finally incubated with a 1 : 50
dilution of guinea-pig peroxidase]antiperoxidase (PAP) complex (Cappel Laboratories) (Sternberger et al., 1970).
In later experiments, including all of the experiments using monospecific primary antisera, peroxidase-conjugated
goat anti-guinea-pig serum (Cappel Laboratories) diluted I : 100 was substituted for the goat anti-guinea-pig serum
and PAP reagents. This substitution (which did not alter the results) was made because the PAP reagent was no
longer available from Cappel Laboratories and because it shortened the procedure by 1-5 h. Secondary
glutaraldehyde fixation, enzymic reaction with diaminobenzidine, post-fixation in osmium, dehydration, and
Epon 812 embedding in situ were performed as previously described (Willingham et al., 1978). The embedded cell
monolayers were screened for the presence of peroxidase reaction product by light microscopy. Antigen-positive
cells, when present, were excised from the embedded monolayer, mounted on preformed Epon blocks, and thin-
sectioned for electron microscopy. When no visible peroxidase reaction was present, cells showing typical
rotavirus cytopathic effect (Estes et al., 1979a) were selected. Sections were examined without further staining
using an Hitachi HUI 1B electron microscope operated at 75 kV.
Isolation of subeellular fractions. MA104 cells were grown to confluency in 850 cm 2 Falcon roller bottles and
infected with 5 p.f.u./ceU SAI 1 or mock-infected with TBS. After a 1 h adsorption period, cells were washed to
remove unadsorbed virus and the cultures were incubated at 37 °C in Eagle's MEM without serum containing 10
~tg trypsin/ml (TRL; Worthington). At 18 h post-infection, the cells were washed three times in buffer containing
65 mM-NaCI, 75 mM-KCI, 0.25 mM-MgC12, 10 mM-HEPES, pH 7.4, and resuspended at a concentration of 5 × 106
cells/ml in the same buffer. Cell membranes were disrupted by N 2 cavitation, and a 'large granule' fraction was
obtained by differential centrifugation as described by Schmidt-Ullrich et al. (1976). The granules, containing
mitochondria, endoplasmic retieulum and various membrane-bound vesicles, were washed three times in PBS by
centrifugation in an Eppendorf microcentrifuge and incubated with the appropriate antisera without prior
fixation as described above, except that saponin was omitted from all reagents. The granules were fixed and
processed for ultrastructural immunocytochemistry using the methods described for cell monolayers, except that
the granules were embedded as pellets.

RESULTS

Immunocytochemistry using hyperimmune a n t i - S A i l serum

SA 11-infected cell cultures were processed for ultrastructural localization o f virus a n t i g e n s
b e t w e e n 18 and 24 h post-infection, a t i m e w h e n virus-induced cytological alterations were
widespread. A t low magnifications ( < x 10000) a definite, overall c y t o p l a s m i c staining could be
seen (Fig. 1 a), while rotavirus antigens a p p e a r e d to be absent f r o m the nuclei o f i n f e c t e d cells.
T h e structures most heavily labelled by anti-SA11 serum w e r e v i r o p l a s m i c inclusions, virus
particles and the m e m b r a n e s o f virus-associated cisternae. M a n y o f the infected cells had
reduced cytoplasmic density, as o b s e r v e d previously in r o t a v i r u s - i n f e c t e d cells ( M c N u l t y et al.,
1976; Pearson & M c N u l t y , 1979), and s o m e cells were p r o b a b l y lysing at the t i m e o f fixation.
T h e i m m u n o l o g i c a l reagents p e n e t r a t e d lysing cells readily and reacted w i t h a v a r i e t y o f host cell
m e m b r a n e s , including the nuclear and outer m i t o c h o n d r i a l m e m b r a n e s (Fig. 1 b), as well as w i t h
virus-induced structures. T h e m o r e intense staining seen at the p e r i p h e r y o f the v i r o p l a s m s in
Fig. 1 (b) was c o m m o n l y o b s e r v e d with the anti-SA11 serum. A t h i g h e r magnifications, the
outlines of d e v e l o p i n g virus particles could occasionally be seen w i t h i n v i r o p l a s m i c inclusions
(see Fig. 2a). N o peroxidase reaction product was o b s e r v e d in S A 1 1 - i n f e c t e d cells i n c u b a t e d
w i t h p r e i m m u n e guinea-pig serum (Fig. I c), or in m o c k - i n f e c t e d cells i n c u b a t e d w i t h the hyper-
i m m u n e anti-SA11 serum (Fig. 1 d) or any of the o t h e r i m m u n e sera.
By scanning a large n u m b e r o f infected cells, we were able to o b s e r v e that several o f the virus-
induced structures w h i c h occurred relatively rarely (e.g. in 1 to 1 0 ~ o f the cells in a p a r t i c u l a r
e x p e r i m e n t ) were specifically labelled with anti-SA11 sera. T h e s e i n c l u d e d bundles o f cyto-
plasmic tubules or filaments p r o b a b l y c o r r e s p o n d i n g to the 15 to 20 n m d i a m . tubules previously
460 B. L. P E T R I E AND OTHERS

Fig. 1. SA1 l-infected MA104 cells processed for immunocytochemistry. (a) Intact cell labelled with
anti-SAil serum by the PAP technique; (b) lysing cell labelled with anti-SAil serum by indirect
immunoperoxidase (IP); (c) cell labelled with preimmune guinea-pig serum by the IP technique; (d)
mock-infected MA 104 cell reacted with anti-SA 11 serum. Bar markers represent 1 lain. Abbreviations
in this and subsequent figures are: Nu, nucleus; er, endoplasmic reticulum; m, mitochondria; vi,
viroplasmic inclusion; vp, virus particles.
 Rotavirus antigens in infected cells 461

Fig. 2. Virus-induced structures in SA1l-infected ceils labelled with anti-SA11 serum using the PAP
procedure. (a) Viroplasmicinclusionand virus particles within the endoplasmicreticulum, arrowheads
indicate developing particles; (b) convoluted smooth membrane (sm) within the endoplasmic
reticulum; (c) aberrant, elongatedvirus-like particles in the infected cell cytoplasm;(d) bundle of 15 to
20 nm diam. tubules in the cytoplasm, arrowheads denote individual tubules. Bar markers represent
0.5 ~tm.

reported in SAil-infected cells (Lecatsas, 1972; Altenburg et al., 1980) (see Fig. 2d), larger
tubules (70 nm diam.) with a dense core which resembled elongated virus particles (Fig. 2 c) and
convoluted smooth membrane material within some of the virus-filled cisternae (Fig. 2b).
Because the fixation procedures used in these experiments did not preserve the plasma
membrane of infected cells well enough to evaluate their antigenicity, we attempted to label the
outer membranes of unfixed (living) and formaldehyde-fixed cells with the anti-SA11 serum
using PAP and with anti-SAil immunoglobulins bound to colloidal gold (Geoghegan &
Ackerman, 1977). No external cell surface antigens could be detected by these methods,
although extracellular virus particles were specifically labelled (data not shown), suggesting that
SAIl antigens are not inserted into the plasma membrane although virus proteins may be
associated with its internal surface.

Immunocytochemistry with monospecific antisera to isolated structural polypeptides

Results obtained using antiserum to VP7, the 38K outer shell glycoprotein (Fig. 3a, b), were
different in several respects from those described for the multivalent anti-SA11 serum. The viro-
plasmic inclusions reacted poorly with anti-VP7, and the general cytoplasmic staining
(ribosomes, various membranes) was tess intense in relation to the virus particles and
endoplasmic reticulum enclosing them, which were heavily labelled (compare Fig. 3a, b with
Fig. 1a). The nuclear membrane, which is often associated with virus particles, also reacted with
the anti-VP7 serum in many cells (Fig. 3 b). Close examination of the virus particles in Fig. 3 (b)
reveals that they are stained primarily on the external surface, in keeping with the location of
glycoprotein VP7 in the outer capsid.
In contrast to the reaction pattern observed with anti-VP7, monospecific antiserum to VP6,
the most abundant inner capsid protein, stained viroplasms intensely and uniformly (Fig. 4a).
The ribosomes and cytoplasmic membranes appeared similar to those in cells stained with anti-
SAIl serum. Virus particles which reacted with the anti-VP6 serum (Fig. 4b, c) appeared
462 B. L. PETRIE AND OTHERS

Fig. 3. SAil-infected cells labelled with anti-VP7 (outer capsid glycoprotein) serum by the IP
procedure. (a) Viroplasmic inclusions and general cytoplasm are weakly stained in comparison to
endoplasmic retieulum and virus particles; (b) virus particles are stained primarily at their periphery,
the nuclear membrane is also peroxidase-labelled. Bar markers represent 1 Itm.

Fig. 4. SA11-infected cells labelled with anti-VP6 (major inner capsid protein) serum. (a) Viroplasmic
inclusions and ribosomes are labelled, but virus particles within the endoplasmic reticulum are not; (b)
virus particles within the endoplasmic reticulum of a different, but identically processed cell from (a);
(c) inset of virus particles shown in (b) at higher magnification. Bar markers in (a) and (b) represent
1 ttm, and in (c) 0-25 p.m.
 Rotavirus antigens in infected cells 463

Fig. 5. SA11-infectedcell labelled with anti-VP2 (inner capsid protein) serum. Viroplasmic inclusions
are heavily labelled. Endoplasmic reticulum and virus particles are labelled but are difficult to
distinguish from the cytoplasmic background (ribosomesetc.) which also reacted with anti-VP2. Bar
marker represents 1 ~tm.

smaller than those stained with anti-SA11 (Fig. 2a), which is consistent with the inner capsid
location of this protein.
Antiserum to VP2, the 94K inner capsid protein, gave a cytoplasmic staining reaction similar
to that of anti-VP6 (Fig. 5). Viroplasmic inclusions also reacted well with this serum. However,
virus particles did not stain as intensely with anti-VP2 serum as they did with any of the other
antisera (Fig. 5) and were difficult to distinguish.

Technical considerations in localizing virus antigens by ultrastructural immunocytochemistry

Antibody penetration was quite variable in SAil-infected cells that retained their cyto-
plasmic density, a problem also noted by other groups of workers using similar techniques
(Bohn, 1980; Chasey, 1980; Stanislawski et al., 1980). Staining of large viroplasms, especially
with the anti-SA11 serum, was usually heavier at the periphery than in the centre, suggesting
that penetration was not always complete. However, these experiments do not exclude the
possibility of differential distribution of antigens. SA 11 antigens were not detected in the nuclei
of infected cells, even though nuclear inclusions with tubules similar to those in Fig. 2 (b) are
occasionally observed in SA 11-infected cells (Lecatsas, 1972; Altenburg et al., 1980). However,
experiments with cells expressing the simian virus 40 nuclear T antigen have indicated that the
nuclear membranes are often not permeabilized by the procedures used here (B. Petrie,
unpublished results). A similar situation seems to occur occasionally with the rough endoplasmic
reticulum membrane (see Fig. 4a), where virus particles within the cisternae are not stained.
Demonstration that the immunocytochemical reactions were not technical artefacts
Glutaraldehyde, a bifunctional cross-linking reagent, was used in the primary fixation prior to
the addition of antiserum in the experiments described above. Therefore, it was possible that
rotavirus proteins within the cytosol were artificially and non-specifically bound to cellular
membranes and to other structures by this procedure. To check this possibility, a subcellular
'granule fraction' consisting of mitochondria, endoplasmic reticulum and various membrane-
bound vesicles was isolated from SAil-infected cells, incubated with anti-SAil serum and
subsequently with immunological reagents before fixation and further processing. PAP
labelling of the granule fraction prior to glutaraldehyde fixation detected the presence of
rotavirus antigens in the mitochondria and intraceUular membranes (Fig. 6a). The lack of
464 13. L. PETRIE AND OTHERS

(d)
iii!~ ';~i~; ::i~
!
~!!i ~:~i

Fig. 6. Large granule fraction isolated from SA 1!-infected (a, c) and mock-infected (b, d) cells. Washed
granules were labelled using the PAP procedure with antiserum to SA11 (a, b, d) or preimmune guinea-
pig serum (c) prior to fixation and embedding for electron microscopy. Granules in (d) were exposed to
soluble fraction from infected cells before labelling. Bar marker represents 0-5/~m.

staining in granules from mock-infected cells (Fig. 6b) and in infected cell granules incubated
with preimmune serum (Fig. 6c) demonstrated immunological specificity. As an additional
control, a sample of the granules from mock-infected cells was incubated for 2 h at 0 °C with the
post-granule supernatant from infected cells and then washed twice in PBS before incubation
with anti-SA11. These granules also appeared negative for rotavirus antigens (Fig. 6d).

DISCUSSION
The present work was undertaken to obtain a more detailed understanding of rotavirus
morphogenesis than can be achieved by conventional electron microscopy or by immunocyto o
chemical techniques used with light microscopy. Successful ultrastructural immunocytochemi-
cal labelling of rotavirus antigens was accomplished before embedding by using a mild
detergent, saponin, to permeabilize the cells after a brief prefixation procedure. This procedure
did minimal damage to intracellular membranes and left most of the cellular organelles easily
recognizable. However, many of the cells which were heavily labelled had reduced cytoplasmic
density, a common feature of rotavirus-infected cells, which may have been enhanced by
leaching during the permeabilization, antibody incubations and washing steps. Since it was
primarily in these lysing cells that virus antigens were detected on mitochondrial and other host
cell membranes, the possibility of redistribution of virus proteins must be considered. The
finding of S A l l antigens associated with partially purified and washed mitochondria from
infected cells indicates that their presence was not caused by reagents used during pre-fixation
or saponin treatment. It therefore seems likely that the presence of rotavirus antigens in or on
cellular structures which are not directly involved in virus replication results from association
with soluble proteins or proteins released from the breakdown of viroplasms and virus-
containing cisternae during lysis of the host cell.
The present study confirmed that several cytoplasmic structures specifically induced by SA 11
infection of cultured cells are composed of virus structural antigens. As expected, the dense
granular inclusions, believed to be the site of virion assembly, reacted strongly with anti-SA 11
serum. In areas where the penetration of the immunoperoxidase reagents was extensive, outlines
of partially assembled virus particles were visible within these viroplasmic inclusions. The 15 to
20 nm diam. cytoplasmic tubules were also noted to contain virus antigens, suggesting that they
 Rotavirus antigens in infected cells 465

Table 1. Use of monospecific antisera to detect differential distribution o f rotavirus antigens

within infected cells
Sera
A
f
Structures Anti-SA11 Anti-VP2 Anti-VP6 Anti-VP7 Interpretation
Viroplasmic inclusions + ++ ++ ± Assembly of single-shelled
particles
Virus particles ++ + ++ ++ Staining reflects location of
proteins in particles
Endoplasmic reticulum ++ + + ++ Synthesis of VP7, assembly of
membrane outer shell
Other virus-induced + ND* ND ND Composed of structural
structures protein(s)
*ND, Not detected.

may result from unbalanced synthesis of one or more S A l l proteins. We were unable to
determine whether the 38K glycoprotein, the 41K protein or the 94K protein were components
of these tubules, as they were not seen in any of the cells reacted with antibody to these
polypeptides. Although rare in rotavirus-infected cells, similar appearing tubules or fibres are
commonly seen in cell cultures infected with members of the related orbivirus group. However,
in contrast to our data with SA11, they are thought to be composed o f non-structural virus-
specified proteins (Oshiro & Emmons, 1968).
A second type of tubule similar to one reported in SA11 infection of primary rhesus monkey
kidney cells (Altenburg et al., 1980) and to those seen in cells infected with E D I M virus
(Banfield et al., 1968) was observed in close proximity to the virus particles in some of the SA11-
infected cells. As these tubules were the same diameter as the particles, contained an electron
dense core structure and were heavily coated with the immune reaction products, they probably
represent aberrant virus assembly. Accumulations of smooth membrane-like material within
virus-containing cisternae also reacted with anti-SA 11 serum, suggesting that these membranes
may be similar in composition to virus envelopes.
The use of monospecific antisera to three major polypeptides of the capsid allowed us to detect
a differential distribution of rotavirus antigens within infected cells. A summary of these
findings is presented in Table 1. The two inner capsid proteins, VP6 and VP2, were
concentrated in the viroplasmic inclusions, where assembly of rotavirus particles is believed to
occur. However, these inclusions were only weakly stained with antiserum to VP7, the outer
capsid glycoprotein, suggesting that this protein is present in smaller quantities than the other
proteins. It is also of interest that antiserum to VP2, the 94K protein, stained virus particles
rather weakly (Fig. 5). Recent evidence from work with the calf rotavirus (Bican et al., 1982) has
shown that VP2 protein is associated with an inner 'core' structure and therefore might be less
accessible to antibody than either VP6 on the inner shell or VP7 on the outer shell.
The data from these experiments suggest that rotavirus proteins are synthesized both on the
free and membrane-bound ribosomes. The antisera to VP6 and VP2, as well as the multivalent
anti-SA 11 serum, reacted with ribosomes distributed throughout the cytoplasm of infected cells.
However, the glycoprotein was associated primarily with ribosomes of the rough endoplasmic
reticulum. These data are consistent with other experiments (McCrae & Faulkner-Valle, 1981 ;
Ericson et al., 1982), suggesting that glycosylation of VP7 occurs co-translationally in the lumen
of the endoplasmic reticulum. This pattern of glycoprotein synthesis is known to occur in a wide
variety of enveloped R N A viruses (Klenk & Rott, 1980) in which the glycoproteins are major
constituents of the virus envelope. Rotaviruses are unique in that the infectious virions are not
enveloped, although they initially acquire an envelope derived from the endoplasmic reticulum
(Petrie et al., 1981). It seems likely that the 38K glycoprotein is added to the outer capsid layer
during the enveloped stage of virus assembly.
Our results demonstrate the usefulness of combining electron microscopic cytochemical
techniques with highly specific immunological probes to study virus replication and assembly at
466 B. L. P E T R I E AND OTHERS

t h e m o l e c u l a r level. F u r t h e r e x p e r i m e n t s w i t h t h e a n t i s e r a d e s c r i b e d h e r e , a s w e l l a s w i t h
a n t i s e r a b e i n g p r e p a r e d t o o t h e r S A 11 s t r u c t u r a l a n d n o n - s t r u c t u r a l p r o t e i n s , s h o u l d a l l o w u s t o
determine the polypeptide composition of the many rotavirus-induced structures and help
identify the function of each protein in morphogenesis.

The authors thank Generina Hanssen for her help with cell culture and Edward Calomeni for his expertise in
thin sectioning. This work was supported by grant A M 30144 from the National Institute of Arthritis, Metabolism
and Digestive Diseases and by a grant from Vicks Health Care Division of Richardson-Vicks, Inc. (Mount
Vernon, N.Y., U.S.A.).
REFERENCES
XDAMS,W. S. & KRAFT,L. (1967). Electron microscopic study of the intestinal epithelium of mice infected with an
agent of epizootic diarrhea of infant mice. American Journal of Pathology 51, 39-60.
ALTENBURG, B., GRAHAM,D. Y. & ESTES,M. K. (1979). Ultrastructural immunocytochemistry of rotavirus-infected
cells. Proceedings of the 37th Annual Meeting of the Electron Microscopy Society of America, pp. 40-41.
ALTENBURG, B. C., GRAHAM,D. Y. & ESTES,M. K. (1980). Ultrastructural study o f rotavirus replication in cultured
cells. Journal of General Virology 46, 75-85.
BAnFmLO, W. <3, I~SNIC, <3. & BLACKWELL,J. , . (1968). Further observations on the epizootic diarrhea of infant
mice. An electron microscopic study. Virology 36, 411-421.
BICAN,P., COHEN,J., CHARPILIENNE,A. & SCHERRER,R. (1982). Purification and characterization of bovine rotavirns
cores. Journal of Virology 43, 1113-1117.
BOHN, W. (1980). Electron microscopic immunoperoxidase studies on the accumulation of virus antigen in cells
infected with Shope fibroma virus. Journal of General Virology 46, 439-447.
CHOOSEY,D. (1977). Different particle types in tissue culture and intestinal epithelium infected with rotavirus.
Journal of General Virology 37, 443-451.
¢I-IASEY, D. (1980). Investigation of immunoperoxidase-labelled rotavirus in tissue culture by light and electron
microscopy. Journal of General Virology 50, 195-200.
ELIAS,M. M. (1977). Separation and infectivity of two particle types of human rotavirus. Journalof General Virology
37, 191-194.
ERICSON, B. L., GR~d-t~M,D. Y., MASON,B. B. & ESTES, M. K. (1982). Identification, synthesis and modification of
simian rotavirus SA11 polypeptides in infected cells. Journal of Virology 42, 825--839.
ESTES, M. K., <3R~tA.M,D. Y., <3ERBA,¢. P. & SMITH, E. M. (1979a). Simian rotavirus SA11 replication in cell cultures.
Journal of Virology 31, 810-815.
ESTES, M. K., <3R~d~,M, D. Y., SMITH, E. M. & <3ERBA,C. P. (1979b). Rotavirus stability and inactivation. Journal of
General. Virology 43, 403-409.
ESTES, M. K., GRAHAM, D. Y. & M~ON, B. B. (1981). Proteolytic enhancement of rotavirus infectivity: molecular
mechanisms. Journal of Virology 39, 879-888.
FLEWETT, T. H. & WOODE, <3. N. (1978). The rotaviruses. Archives of Virology 57, 1-23.
GEOGHEGAN, W. D. & ACKERMAN,<3. A. (1977). Adsorption of horseradish peroxidase, ovomucoid, and anti-
immunoglobulin to colloidal gold for the indirect detection o f concanavalin A, wheat germ agglutinin, and
goat anti-human immunoglobulin G on cell surfaces at the electron microscope level : a new method, theory
and application. Journal of Histochemistry and Cytochemistry 25, 1187-1200.
GRAHAM, D. Y. & ESTES, M. K. (1979). Comparison of methods for immunocytochemical detection of rotavirus
infections. Infection and Immunity 26, 686-689.
HALL, <3. A., BRIDGER, J. C., CHANDLER,R. L. & WOODE, G. N. (1976). Gnotobiotic piglets experimentally infected
with neonatal calf diarrhoea reovirus-like agent (rotavirus). Veterinary Pathology 13, 197-210.
HOLMES, I. H., RUCK, B. J., BISHOP, R. F. & DAVIDSON,G. P. (1975). Infantile enteritis viruses: morphogenesis and
morphology. Journal of Virology 16, 937-943.
KLENK, H.-D. & ROT'I', R. (1980). Cotranslational and posttranslational processing o f viral glycoproteins. Current
Topics in Microbiology and Immunology 90, 19-48.
LECATS~S,G. (1972). Electron microscopic and serological studies on simian viruses SA 11 and the 'related' O agent.
Onderstepoort Journal of Veterinary Research 39, 133-138.
McCRAE, M. A. & FAULKNER-VALLE,G. P. (1981). Molecular biology of rotaviruses. I. Characterization of basic
growth parameters and patterns of macromotecular synthesis. Journal of Virology 39, 490-496.
McNULTY, M. S. (1978). Rotaviruses. Journal of General Virology 40, 1-18.
McNULTY, M. S., CURRAN,W. L. & McFERRAN,J. B. (1976). The morphogenesis of a cytopathic bovine rotavirus in
Madin-Darby bovine kidney cells. Journal of General Virology 33, 503 508.
McNULTY, M. S., ALLAN, G~ M., TODD, D. & McFERRAN, J. B. (1979). Isolation and cell culture propagation of
rotaviruses from turkeys and chickens. Archives of Virology 61, 13 21.
M~SON, B. B., GRAHAM,D. Y. & ESTES,M. K. (1980). In vitro transcription and translation of simian rotavirus SAI 1
gene products. Journal of Virology 33, 1111-1121.
MEBUS,C. A., UNDERDAHL,N. R., RHODES,M. B. & TWIEHAUS,H. J. (1969). Calf diarrhoea (scours) reproduced with a
virus from a field outbreak. Bulletin of the Nebraska Agricultural Experimental Station 233, 1-16.
OSHIRO,L. S. & EMMONS,R. W. (! 968). Electron microscopic observations of Colorado tick fever virus in BHK21 and
K B cells. Journal of General Virology 3, 279-280.
 Rotavirus antigens in infected cells 467

PEARSON, G. R. & McNULTY, M. S. (1979). Ultrastructural changes in small intestinal epithelium of neonatal pigs
infected with pig rotavirus. Archives of Virology 59, 127-136.
PETRIE, B. L., GRAHAM,D. Y. & ~TES, M. K. (1981). Identification of rotavirus particle types. Intervirology 16, 20-28.
SAIF, L. J., THEIL, K. w. & BOHL, E. H. (1978). Morphogenesis of porcine rotavirus in porcine kidney cell cultures and
intestinal epithelial cells. Journal of General Virology 39, 205-217.
SCHMIDT-ULLRICH,R., WALLACH,D. F. H. & DAVIS, F. D. G., II (1976). M e m b r a n e s of normal hamster lymphocytes
and lymphoid cells neoplastically transformed by simian virus 40. I. High-yield purification of plasma
m e m b r a n e fragments. Journal of the National Cancer Institute 57, 1107-1116.
SMITH, E. M., ESTES,M. K., GRAHAM,D. Y. & GERBA,C. P. (1979). A plaque assay for the simian rotavirus SA11. Journal
of General Virology 43, 513-519.
STAIR, E. L., MEBUS,C. A., TWIEHAUS,M. J. & UNDERDAHL,N. R. (1973). Neonatal calf diarrhea. Electron microscopy
of intestines infected with a reovirus-like agent. Veterinary Pathology 10, 155-170.
STANISLAWSKI,L., MONGIAT,F. & HAGUENAU,F. (1980). Immunoelectron microscopic localization of Rous sarcoma
virus structural protein p27 in infected chicken embryo fibroblasts, lntervirology 14, 261-271.
STERNBERGER, L. A., HARDY, P. H., JR, CUCULIS,J. J. & MEYER, H. G. (1970). The unlabeled antibody enzyme method
of immunocytoehemistry. Journal of Histochemistry and Cytochemistry 18, 315 333.
WlLLINGHAM,M. C., YAMADA,S. S. & PASTAN,I. (1978). Ultrastructural antibody localization of ct2-macroglobulin in
membrane-limited vesicles in cultured cells. Proceedings of the National Academy of Sciences of the United
States of America 75, 4359-4363.

(Received 9 February 1982)