Jpn. J. Infect. Dis., 61, 123-127, 2008
Reduced Incorporation of SARS-CoV Spike Protein into Viral Particles
Due to Amino Acid Substitutions within the Receptor Binding Domain
Shu-Ming Li1,2, Gui-Mei Li2, Shota Nakamura3, Kazuyoshi Ikuta2,3 and Takaaki Nakaya1*
International Research Center for Infectious Diseases and 2Department of Virology,
Research Institute for Microbial Diseases, Osaka University, Osaka 565-0871, Japan, and 3Thailand-Japan
Research Collaboration Center on Emerging and Re-emerging Infections, Nonthaburi, Thailand
(Received December 6, 2007. Accepted January 25, 2008)
SUMMARY: Cell clone #21 is a long-term producer of the infectious SARS-coronavirus, although the incorpora-
tion rate of spike (S) protein into virions is significantly lower. Sequencing analysis of the viral structural proteins
revealed four and one amino acid substitutions in the S and membrane (M) proteins, respectively. We demon-
strated, using a viral-like particle formation system, that the S mutations were involved in the lower incorporation
of the S protein into virions, although the M mutation that disrupts the glycosylation was not present in this
phenotype. Further mutational experiments identified two substitutions, Y442C and L472F, within the receptor
binding domain that could be critical for the reduced S incorporation, as well as reduced binding affinity between
the S protein and ACE2 receptor. Thus, these two amino acid substitutions might lead to a conformational
change in the S protein, resulting in reduced incorporation into viral particles.
as well as Western blotting analyses demonstrated that the
incorporated number of S proteins on the viral particles from
The severe acute respiratory syndrome (SARS)-coronavirus #21 cells was lower than that from acutely infected cells (11).
(CoV) was identified as the etiological agent for an acute Sequencing analysis of the viral genome of #21-derived virus
infectious respiratory disorder (1). SARS-CoV is an envel- revealed several amino acid substitutions in S and M genes
oped, positive-strand RNA virus with a ~30 kb genome that (GenBank accession no. AB257344). In this study, we focused
encodes replicase (1a and 1b), spike (S), envelope (E), mem- on the relation of the amino acid substitutions in the struc-
brane (M), nucleocapsid (N) and several accessory proteins tural proteins with the reduced S incorporation into virions
(2,3). The virus particle consists of four structural compo- in #21.
nents which are S, E, M and N proteins. The generation of
viral-like particles (VLPs) bearing four structural proteins has
MATERIALS AND METHODS
been reported (4,5). It was demonstrated that the S, M and
N proteins were necessary for pseudovirus assembly (4). In Virus and cells: The Vero E6 cell line was used for the propa-
addition, it was established using a VLP system that the N gation of SARS-CoV (Frankfurt-1 strain) (12). A #21 cell clone
protein played an essential role on the packaging of SARS- obtained from Vero E6 cells infected with the Frankfurt-1
CoV RNA (5). Thus, the VLP system is very useful for under- strain was previously prepared (11). Vero E6, #21 and HEp-2
standing virion assembly and morphology at the molecular cells were maintained in MEM (GIBCO BRL, Carlsbad,
level. Calif., USA) supplemented with 10% fetal bovine serum
S protein is a type I integral membrane glycoprotein that (ICN Flow), 100 U/ml penicillin and 100 g/ml streptomycin
makes up the crown-like appearance of the viral particles (6). (GIBCO BRL) and passaged every 3 days.
It was reported that angiotensin-converting enzyme 2 (ACE2) Reverse transcription (RT)-polymerase chain reaction
is the functional receptor for SARS-CoV (7). The interaction (PCR) for SARS-CoV genome RNA: Total RNA was ex-
of the receptor-binding domain (RBD) of S protein with ACE2 tracted from #21 cells with TRIzol (Invitrogen, Carlsbad,
was demonstrated by crystal structure analysis (8). Further Calif., USA) and subjected to RT-PCR. The RNA was re-
studies showed that the RBD contains major determinants verse-transcribed with SuperScript III reverse transcriptase
for viral entry and neutralization (9,10). (Invitrogen) using random primers (Invitrogen). The PCR
We previously reported that a total of four (#13, #18, #21 amplification was carried out as described previously (11).
and #34) of 87 cell clones isolated from persistently infected Direct sequencing of PCR products: To perform genome
cells were shown to be viral RNA-positive (11). However, sequencing, ~3 kb DNA fragments were obtained by RT-PCR.
several passages of subsequent culturing cleared the viral RNA The PCR products were sequenced with a BigDye Termina-
from cell clones #13, #18 and #34, and only #21 was a long- tor v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster
term producer of infectious viral particles at a high rate for City, Calif., USA) according to the manufacturer’s protocol.
more than one year (11). Interestingly, electron-microscopic Western blotting: The cell fractions were subjected to
Western blotting and probed with an anti-V5 monoclonal
*Corresponding author: Mailing address: International Research antibody (Invitrogen) for 1 h at 37°C, then with horseradish
Center for Infectious Diseases, Research Institute for Microbial peroxidase-conjugated anti-mouse secondary antibody for 1
Diseases, Osaka University, Suita, Osaka 565-0871, Japan. Tel: h at 37°C (Jackson ImmunoResearch Laboratories, West
+81-6-6879-4251, Fax: +81-6-6879-4252, E-mail: tnakaya@ Grove, Pa., USA).
biken.osaka-u.ac.jp Generation of VLPs by co-expression with S, E, M and
N-expressing plasmids: The whole ORF region without the body for 30 min at 37°C, then incubated with fluorescein
termination sequences of S, E, M and N were amplified using isothiocyanate-conjugated anti-mouse IgG (Jackson Immuno-
RT-PCR (S: forward primer 5´-GAATTCCTCGAGATGTTTA Research Laboratories) for 45 min at 37°C and observed
TTTTCTTATTATTTC-3´ and reverse primer 5´-TCTAGATG by confocal microscopy (FLUOVIEW FV1000; Olympus,
TGTAATGTAATTTGACACCC-3´; E: forward primer 5´-GA Tokyo, Japan).
ATTCATGTACTCATTCGTTTCGGA-3´ and reverse primer Structural analysis of SWT and S#21 proteins using a com-
5´-TCTAGAGACCAGAAGATCAGGAACTC-3´; M: for- puter program: The RBD of the S#21 protein with the substi-
ward primer 5´-GAATTCATGGCAGACAACGG-3´ and re- tutions Y442C and L472F was modeled by replacing the side-
verse primer 5´-CTCGAGCTGTACTAGCAAAGCAATAT chains of SWT protein (PDB ID: 2AJF) using PyMOL v.0.99rc6
TG-3´; N: forward primer 5´-GAATTCATGTCTGATAATGG (http://www.pymol.org).
ACCCCAATC-3´ and reverse primer 5´-TCTAGATGCCTGA
GTTGAATCAGCAG-3´). The PCR products of S, E and N
were cloned into the EcoRI-XbaI sites of the pcDNA3.1/
V5-His expression plasmid (Invitrogen). The PCR product Sequencing analysis of SARS-CoV S, E, M and N genes
of M was cloned into the EcoRI-XhoI sites of the same ex- in the #21 cell clone: The SARS-CoV genome in #21 was
pression plasmid. These plasmids express C-terminal V5-His independently amplified at the S, E, M and N genes by RT-
fusion proteins of S, E, M or N. The wild-type M (MWT) PCR, and individual PCR products were subjected to direct
and S (SWT) genes were prepared from Vero E6 cells acutely sequencing. In total, six amino acid substitutions were iden-
infected with wild-type virus. Mutant M (M#21: T6A) and tified: Y442C, L472F, V594F, H641Y and P794S in S and
S (S#21: Y442C, L472F, V594F and P794S) genes were T6A in M, as summarized in Fig. 1A. Since H641Y was also
prepared from #21. Four types of chimeric S protein (Smut1: detected in the original wild-type virus (Frankfurt-1 strain)
Y442C and L472F, Smut2: V594F and P794S, Smut3: Y442C from acutely infected cells (data not shown), the other four
and Smut4: L472F) were generated by the GeneTailorTM Site- substitutions in S could be present in the phenotype of the
Directed Mutagenesis System (Invitrogen) according to the #21-derived virus. One amino acid substitution, T6A at the
manufacturer’s protocol. The E and N genes with the termi- N-terminus of the M protein, resulted in disruption of the
nation sequence were also PCR-amplified and cloned into putative N-glycosylation site (4NGT6 to 4NGA6). In contrast,
pEF-BOS expression vector (13), and the M gene with the there were no mutations in the E and N proteins (data not
termination sequence was cloned into pcDNA3.1 vector shown). All of the mutation sites that were found in the #21-
(Invitrogen). A monolayer of Vero E6 cells in a 100-mm dish derived virus are summarized in Table 1.
was infected with recombinant vaccinia virus expressing Formation of VLPs with S, E, M and N proteins: We
T7 RNA polymerase (14) at a multiplicity of infection (moi) previously reported that a smaller amount of S glycoprotein
of 1 for 1 h and were transfected with S, E, M and N expres- was incorporated into #21-derived virions in comparison with
sion plasmids (6 g in each) using Lipofectamine 2000 the wild-type virus, as shown by electron-microscopic and
(Invitrogen). The transfected cells were cultured for 4 days Western blotting analyses (11). Therefore, we examined
and were harvested for Western blotting analysis using an whether the above-described mutations were related to the
anti-V5 monoclonal antibody. SARS-CoV VLPs were par- lower incorporation of S protein into virions using a VLP
tially purified from the culture supernatant of the transfected system co-expressing S, E, M and N proteins. Reconstitution
cells by ultracentrifugation through 20% sucrose in PBS with analysis using VLPs confirmed that a smaller amount of
a Beckmann SW28 rotor at 25,000 rpm for 2 h at 4°C. VLPs S protein was incorporated into the #21 VLPs than into the
collected in the pellet were resuspended in 100 l of PBS wild-type VLPs, although the intracellular expression level
for Western blotting with the anti-V5 monoclonal antibody. of each structural protein was comparable (Figs. 1B and 1C).
To separate various VLPs with different compositions, the In addition, since the amino acid substitution T6A in the M
virus suspension was further loaded on a discontinuous su- protein abolished the N-glycosylation in the M protein as
crose gradient consisting of 20, 30, 50 and 60% sucrose as determined by Western blotting (Fig. 2A), we next investi-
described by Hsiesh et al. (5) and then centrifuged in the gated whether this mutation was also involved in the reduced
SW41.1 rotor at 26,700 rpm for 3.5 h at 4°C. Fractions con- incorporation of the S protein into virions. However, there
taining VLPs with different compositions were analyzed for was no difference in the amount of S protein incorporated
the presence of SARS-CoV structural proteins by Western into VLPs between the M#21 and MWT proteins (Fig. 2B). In
blot analysis. contrast to this finding with the M protein, when S#21 was
Expression of M and S proteins: HEp-2 cells were infected co-expressed with the MWT, E and N proteins, the level of S
with recombinant vaccinia virus expressing T7 RNA poly- incorporation was reduced to a level similar to that of the #21
merase at an moi of 1 for 1 h, and were transfected with MWT VLPs (Fig. 2B). We therefore concluded that a mutation(s)
and M#21 expression plasmids using Lipofectamine 2000. The in the S protein determines the reduced incorporation of S
proteins were detected by Western blotting with an anti-V5 protein into virions. Further mutational analysis of the S gene
antibody. For the expression of S proteins, Vero E6 cells were
infected with recombinant vaccinia virus expressing T7 RNA
Table 1. Summary of amino acid substitutions in #21-derived virus
polymerase at an moi of 1 for 1 h, and were transfected with
SWT, S#21 and Smut1-4 expression plasmids using Lipofectamine ORF1a 4 amino acids L2430F, S3321L, D3503A, M3986V
2000. The reactivity levels of an anti-S monoclonal antibody S protein 5 amino acids Y442C, L472F, V594F, H641Y, P794S
3A2 (11) with these proteins were compared by Western blot- M protein 1 amino acid T6A
ting. ORF7-ORF8 218-nt deletion and insertion1)
Immunofluorescence assay (IFA): Fixed cells with 2% 1)
: Deletion of 218-nt (nt 27,886-28,103) was found in ORF 8a and 8b
paraformaldehyde were permeabilized by treatment with and the deleted nucleotides were found to be inserted into ORF 7a
0.1% Triton X-100. Cells were reacted with an anti-V5 anti- (nt 27,571-27,572).
Fig. 1. Mutations identified in the SARS-CoV structural genes derived from #21. (A) Amino acid changes in the S (upper) and
M (lower) genes are shown. (B) WT and #21 indicated Vero E6 cells transfected with SWT, MWT, E, N expression plasmids,
and S#21, M#21, E, N expression plasmids, respectively. VLPs and intracellular viral proteins were detected with an anti-V5
antibody by Western blotting. (C) For separation of various VLPs with different compositions, the VLP-suspension was
further purified on discontinuous (20 - 60%) sucrose gradient centrifugation. Ten fractions were collected from the top of
the gradient and each fraction was analyzed by Western blotting.
Fig. 2. Effect of amino acid substitutions of M#21 and S#21 on S protein incorporation into virions. (A) HEp-2 cells were
transfected with MWT and M#21 expression plasmids as described in Materials and Methods. The proteins were detected by
Western blotting with an anti-V5 antibody. (B) VLPs produced by transfection with S (SWT or S#21), M (MWT or M#21), E and
N expression plasmids were subjected to Western blotting with the anti-V5 antibody. (C) Four types of chimeric S protein
(Smut1, Smut2, Smut3 and Smut4) were generated by site-directed mutagenesis. The amino acid substitutions (shown by bold) in
these proteins were summarized in the low panel. The resulting VLPs were purified and compared by Western blotting with
the anti-V5 antibody.
showed that the first two mutations (Smut1: Y442C and L472F) the incorporation of S protein into virions was shown to be
were critical for determining the incorporation efficiency (Fig. critically affected by two amino acids at residues 442 and
2C). However, a similar lower level of incorporation of S 472. We also investigated the distribution of intracellular S
protein into virions was also observed in the single amino protein in SWT- or S#21-transfected cells together with M, N
acid mutation (Smut3: Y442C and Smut4: L472F) (Fig. 2C). Thus, and E expression vectors. As shown in Fig. 3, SWT protein
was located mainly in the plasma membrane at 48 h post- are necessary to confirm this prediction.
transfection, whereas S#21 protein was detected not only in It was reported that the epitope on the RBD that bound to
the plasma membrane but also in the cytoplasmic region. This the anti-S human monoclonal antibody (80R) overlaps very
tendency was also observed at 72 h post-transfection (Fig. 3). closely with the ACE2 binding site, and further, that the S-
Binding affinity of mutant S proteins with receptor ACE2 and S-80R interfaces share many common S amino
ACE2, as well as with an anti-S neutralizing antibody: acid residues, including Tyr442 and Leu472 (15). In addition,
Structural analysis of the S protein using a computer pro- the overall structure of the other neutralizing antibody (m396)-
gram indicated that Y442C and L472F substitutions might RBD interface was also shown to be not significantly differ-
reduce the binding affinity between the S protein and recep- ent from that of the ACE2-RBD interface (16). Neutralizing
tor ACE2 (Fig. 4A). In particular, Y442C substitution might determinants of the m396 antibody are located contiguously
change the orientation of the side-chain, leading to weak S- in one major segment of the 6- 7 loop, whereas receptor
ACE2 (His34) binding (Fig. 4A). In addition, extra Cys442 may ACE2 has determinants over most of the same extended loop
generate a novel S-S bond in the S protein, resulting in dras- appearing at the top of the RBD (16). Thus, we tried to inves-
tic conformational change. Furthermore, it has been observed tigate the binding affinity of mutant S proteins with an anti-S
using another computer model that a naturally occurring monoclonal antibody. The result showed that the anti-S mono-
L472P mutation may contribute to the attenuation by reduc- clonal antibody (3A2), possessing neutralizing activity (Fig.
ing the S-ACE2 contact surface (8). Further structural studies 4B), reduced the binding efficiency to the S#21 protein by
Western blotting (Fig. 4C). By using four types of chimeric
proteins (Smut1 to Smut4), we demonstrated that this antibody
recognized an epitope containing Tyr442 and that the substitu-
tion of Tyr by Cys reduced the binding efficiency to 3A2
(Fig. 4C). These results suggest that a conformational change
in the RBD caused by the Y442C substitution led to a reduc-
tion of the binding affinity to this anti-S neutralizing anti-
body, which is consistent with the data obtained using the
above computer analysis (Fig. 4A).
According to the genome sequence data of the structural
genes S, E, M and N in #21-derived SARS-CoV, we identi-
Fig. 3. Distribution of SWT or S#21 protein in transfected cells. SWT- or fied a total of four and one amino acid substitutions in the S
S21-transfected cells together with N, M and E (V5-tag [-]) expres-
sion vectors at 48 and 72 h post-transfection were subjected to and M proteins, respectively, and no mutation in the N and E
IFA using anti-V5 (S protein-specific) antibody with FLUOVIEW proteins. Among them, two amino acid substitutions, Y442C
FV1000 (Olympus) confocal microscopy. and L472F, of the S protein were shown to be related to the
Fig. 4. Binding affinity of mutant S proteins with receptor and neutralizing antibody. (A) S protein and receptor ACE2
structures are colored in magenta and cyan, respectively. Only residues around the positions of 442 and 472 are shown as a
stick model. (B) Vero E6 cells were infected with wild-type virus (104 TCID50) which had been incubated overnight at 4°C
with 3A2 (29 g) or normal mouse serum (100 g), as a control. After incubation for 48 h, the cytopathic effects appeared
were observed under phase-constrast microscope. (C) Vero E6 cells were transfected with SWT, S#21 and Smut1-4 expression
plasmids as described in Materials and Methods. These proteins were detected by Western blotting with 3A2 or the anti-V5
antibody, as a control. Lanes 1 - 6 are summarized in Fig. 2C lower panel. Lanes 1 and 2 indicate SWT and S#21, respectively.
Lanes 3 - 6 indicate Smut1, Smut2, Smut3 and Smut4, respectively.
reduced incorporation of this protein into virions that could for giving us the Frankfurt strain of SARS-CoV through Dr. Fumihiro
Taguchi, National Institute of Infectious Diseases, Tokyo, Japan. We thank
affect the #21 phenotype, i.e., the reduced affinities to the Dr. Makiko Yamashita and Mr. Masanobu Yamate for helpful discussions.
receptor ACE2 and anti-S neutralizing antibody, and the dif- This work was supported in part by a Grant-in-Aid for Scientific Research
ferent intracellular distribution. from the Ministry of Education, Culture, Sports, Science and Technology
The SARS-CoV M protein is exclusively N-glycosylated (MEXT), the Program of the Founding Research Center for Emerging and
in asparagine at amino acid residue 4, and a nonglycosylated Reemerging Infectious Diseases launched by a project commissioned by the
MEXT (K. I. and T. N.), and the project for the International Research Center
M is selectively incorporated into virions (17). It was also for Infectious Diseases, Research Institute for Microbial Diseases, Osaka
shown that glycosylation of the CoV M protein is not required University from the MEXT of Japan (T. N.).
for virus assembly, nor for the interaction between M and S
proteins to occur (18). Our results obtained using the VLP REFERENCES
system were consistent with the above observations, although
1. Drosten, C., Gunther, S., Preiser, W., et al. (2003): Identification of a
the function of the glycosylation of the SARS-CoV M pro- novel coronavirus in patients with severe acute respiratory syndrome.
tein is still not known, as has been described (18). N. Engl. J. Med., 348, 1967-1976.
Mutational analysis using the VLP system demonstrated 2. Rota, P.A., Oberste, M.S., Monroe, S.S., et al. (2003): Characterization
that the S protein itself determined the efficiency of its incor- of a novel coronavirus associated with severe acute respiratory syn-
drome. Science, 300, 1394-1399.
poration into the pseudoparticles, as shown by the signifi- 3. Marra, M.A., Jones, S.J., Astell, C.R., et al. (2003): The genome sequence
cant effect on the incorporation efficiency of the S protein of the SARS-associated coronavirus. Science, 300, 1399-1404.
into virions by amino acid substitutions Y442C and L472F. 4. Huang, Y., Yang, Z.Y., Kong, W.P., et al. (2004): Generation of synthetic
It is not clear why a small amount of N protein was detected severe acute respiratory syndrome coronavirus pseudoparticles: implica-
in VLPs in our system. Extra amino acids, including a V5- tions for assembly and vaccine production. J. Virol., 78, 12557-12565.
5. Hsieh, P.K., Chang, S.C., Huang, C.C., et al. (2005): Assembly of severe
tag at the C termini of the N protein, may be one reason for acute respiratory syndrome coronavirus RNA packaging signal into
the above. Otherwise, it is possible that the V5-tag of the M virus-like particles is nucleocapsid dependent. J. Virol., 79, 13848-13855.
protein affects the N protein assembly. 6. Lin, S., Lee, C.K., Lee, S.Y., et al. (2005): Surface ultrastructure of
Interestingly, Tyr442 and Leu472 were shown to be located SARS coronavirus revealed by atomic force microscopy. Cell Microbiol.,
in the RBD (8). Therefore, the Tyr442 to Cys442 substitution in 7. Li, W., Moore, M.J., Vasilieva, N., et al. (2003): Angiotensin-converting
the S#21 protein might affect the RBD structure in #21- enzyme 2 is a functional receptor for the SARS coronavirus. Nature,
derived viral particles. In fact, the substitution impaired the 426, 450-454.
binding affinity of S#21 protein and the anti-S neutralizing 8. Li, F., Li, W., Farzan, M., et al. (2005): Structure of SARS coronavirus
monoclonal antibody (3A2). In addition, the Leu472 to Pro472 spike receptor-binding domain complexed with receptor. Science, 309,
substitution often appeared in SARS-CoV isolates from 9. He, Y., Lu, H., Siddiqui, P., et al. (2005): Receptor-binding domain of
patients during the 2003 - 2004 outbreak, and this substitu- severe acute respiratory syndrome coronavirus spike protein contains
tion was shown by conformational analysis to contribute to multiple conformation-dependent epitopes that induce highly potent
the viral attenuation by reducing the S-ACE2 contact surface neutralizing antibodies. J. Immunol., 174, 4908-4915.
10. Liu, L., Fang, Q., Deng, F., et al. (2007): Natural mutations in the
(8). Consequently, the two amino acid substitutions Y442C receptor binding domain of spike glycoprotein determine the reactivity
and L472F identified in #21 in this study could alter the anti- of cross-neutralization between palm civet coronavirus and severe acute
genic structure, as well as the S-ACE2 binding affinity. How- respiratory syndrome coronavirus. J. Virol., 81, 4694-4700.
ever, the molecular mechanism by which these two substitu- 11. Yamate, M., Yamashita, M., Goto, T., et al. (2005): Establishment of
tions lead to reduced incorporation of S protein into virions Vero E6 cell clones persistently infected with severe acute respiratory
syndrome coronavirus. Microbes Infect., 7, 1530-1540.
remains unclear. As for influenza virus HA protein, a single 12. Ivanov, K.A., Thiel, V., Dobbe, J.C., et al. (2004): Multiple enzymatic
amino acid substitution at HA1 resulted in the reduction of activities associated with severe acute respiratory syndrome coronavirus
and slowed rates of transport to the medial Golgi apparatus helicase. J. Virol., 78, 5619-5632.
and cell surface due to partial impairment of folding (19). 13. Mizushima, S. and Nagata, S. (1990): pEF-BOS, a powerful mammalian
expression vector. Nucleic Acids Res., 18, 5322.
In this study, the intracellular location of SWT was slightly 14. Nakaya, T., Cros, J., Park, M.S., et al. (2001): Recombinant Newcastle
different from that of S#21 (Fig. 3). Thus, the two amino acid disease virus as a vaccine vector. J. Virol., 75, 11868-11873.
substitutions found in #21-derived SARS-CoV may have led 15. Hwang, W.C., Lin, Y., Santelli, E., et al. (2006): Structural basis of
to improper conformation of the S protein in the endoplasmic neutralization by a human anti-severe acute respiratory syndrome spike
reticulum/Golgi network, and this could have contributed to protein antibody, 80R. J. Biol. Chem., 281, 34610-34616.
16. Prabakaran, P., Gan, J., Feng, Y., et al. (2006): Structure of severe acute
the decrease in S proteins on the cell surface and viral mem- respiratory syndrome coronavirus receptor-binding domain complexed
brane. with neutralizing antibody. J. Biol. Chem., 281, 15829-15836.
Based on this evidence, the modification of SARS-CoV S 17. Voss, D., Kern, A., Traggiai, E., et al. (2006): Characterization of severe
structure may be related to the smaller amount of S protein in acute respiratory syndrome coronavirus membrane protein. FEBS Lett.,
the released virions. It has been reported that M protein plays 18. Oostra, M., de Haan, C.A., de Groot, R.J., et al. (2006): Glycosylation
a pivotal role in virion assembly through its ability to interact of the severe acute respiratory syndrome coronavirus triple-spanning
with S protein (4). Although an average of 65 trimeric spikes membrane proteins 3a and M. J. Virol., 80, 2326-2336.
per SARS-CoV particle was estimated by cryo-electron 19. Simpson, D.A. and Lamb, R.A. (1991): Influenza virus ts61S hemag-
microscopy (20), the detailed machinery of SARS-CoV S glutinin is significantly defective in polypeptide folding and intracellular
transport at the permissive temperature. Virology, 185, 477-483.
protein incorporation into particles remains to be clarified. 20. Beniac, D.R., Andonov, A., Grudeski, E., et al. (2006): Architecture of
the SARS coronavirus prefusion spike. Nat. Struct. Mol. Biol., 13, 751-
We are grateful to Dr. John Ziebuhr, University of Würzburg, Germany