SARS Press Release

Department of Microbiology, Immunology and Parasitology, LSU Health Sciences Center, and Department of Microbiology and Immunology, Tulane Health Sciences Center, New Orleans, Louisiana
wgalla@lsuhsc.edu; rfgarry@tulane.edu

Shown is a model, in helical net projection, of the 314 amino acids of the spike (S2) glycoprotein of the SARS-associated coronavirus (Ksiazek et al. 2003) just prior to membrane insertion into the viral envelope. The peptide sequence (NCBI sequence ref. NC_004718) can be fitted to a general scaffold of the gp41 transmembrane fusion glycoprotein of HIV-1 we first described in Gallaher et al. 1989, and is similar in a number of detailed respects. While lacking x-ray crystallographic or other biophysical data needed to confirm this model, this model is consistent with the proven structures of other viral fusion glycoproteins, beginning with the influenza hemagglutinin in 1981(Wilson et al. 1981), as well as with experimental data in other coronavirus systems from other laboratories (e.g. Luo et al 1999). Cartoon models of the coronavirus spike as a coiled coil have been proposed as early as 1987 (deGroot et al, 1987), but previous models have not previously been presented in this detail, or demonstrating the close parallels with the other fusion glycoproteins to this degree. The detailed model presented here has significant implications for avenues to develop antiviral drugs that function as fusion inhibitors of the SARS coronavirus.

First, beginning about amino acid 900 there is an extended heptad repeat region similar to HR1 of HIV-1. This differs from the retroviruses, Ebola and lymphochoriomeningitis virus (Gallaher 1996; Gallaher et al. 2000) principally in the extraordinarily length of the helix. While there are helix-breaking motifs present (e.g. TTTS), the helix may be stabilized in such areas by the very strong heptad repeat of hydrophobic amino acids shown in black along the right side of the helix projection. At 17 nm, this helix is overly long for the known dimensions of the coronavirus surface spike, but may reflect an extension that occurs upon binding or configurational alteration of the protein while in the process of becoming a fusion-active form. Peptide analogues or peptidomimetics of this structure are predicted to have significant antiviral activity, and we are investigating such antivirals.

Second, there is a short region bounded by cysteines, of comparable length to the C(x6)C motif in retroviruses and also containing charged amino acids on one side only, which is so similar to that of retroviruses and Ebola to prompt us to model it as a similar disulfide-stabilized apex.

Third, there is a region with several sites (shown by stick figures) for possible N-linked glycosylation that, like HIV-1, are only found after the disulfide-linked apex. This region is highly variable among coronaviruses, not unlike among the retroviruses.

Fourth, there is a region 80 amino acids prior to the point the viral protein is anchored in the viral envelope membrane, which has a high percentage of charged amino acids, a strong propensity to form a helix and also has a limited heptad repeat, so that it is comparable to the corresponding region of HIV-1 gp41. Indeed, the amino terminal end of this charged pre-insertion helix shows a peptide motif ELDKY, that is very similar to a biologically significant peptide, ELDKW, in the comparable region of HIV-1 gp41. The importance of this peptide in the coronavirus protein is underscored by the high degree of conservation of this peptide sequence among unrelated coronaviruses that otherwise vary widely in amino acid sequence. Peptide analogues or peptidomimetics of this structure are also predicted to have significant antiviral activity, and we are investigating such antivirals.

In HIV-1 this peptide ELDKWA is recognized as a neutralization epitope, for which a human monoclonal antibody has been developed (Muster et al.1993) and is currently in human clinical trials (Armbruster et al. 2002). Based on the probable cross-reactivity with the ELDKY region in SARS coronvirus, this monoclonal antibody is predicted to have some neutralizing activity against the SARS coronavirus. The ELDKW motif is also represented in the recently licensed peptide fusion inhibitor, Fuzeon™, that suppresses HIV-1 infection in the nanomolar range (Kilby et al. 1998). While the overall similarity of the SARS region to HIV-1 gp41 is low, Fuzeon would also be predicted to have some limited activity against the SARS coronavirus based on this limited similarity.

Finally, just prior to membrane insertion there is a region highly enriched in aromatic amino acids and extraordinarily highly conserved throughout the coronavirus family. This region lies in an identical location to comparable aromatic rich regions in the transmembrane proteins of HIV-1 and Ebola, which have been shown to be fusogenic in liposome systems (Suarez et al. 2000). An experimental octapeptide mimicking this region in feline immunodeficiency virus (FIV) has been found to inhibit fusion by that virus in cell culture (Gianecchini et al. 2003). Fuzeon also contains an enrichment in aromatic amino acids at its carboxy-terminal end, offering another predicted mechanism for some degree of inhibition of the SARS virus by this existing antiretroviral drug.

We have not modeled further toward the amino terminus of the protein, since there are no parallels established among other viruses for the structure of the protein prior to the first extended heptad repeat region. This region is only shown schematically as a large ellipse corresponding to the large globular head group that forms the top of the characteristic “lollipop” spike seen in electron micrographs of coronaviruses, giving it the “crown”-like appearance from which the virus family derives its name.

The structural parallel of the helical fibrous region of the SARS coronavirus S glycoprotein to HIV-1 and other members of the same superfamily of viral transmembrane entry proteins offers considerable support for the predicted fusion inhibitory effects of antiviral peptides modeled from the SARS amino acid sequence We are currently investigating and testing such peptides as antiviral fusion inhibitors against coronaviruses in a manner that does not pose a biohazard to human beings.

However, as noted above, the model suggests that clinical testing of antiviral activity may be bypassed in specific instances. We are also investigating and testing current reagents such as a human monoclonal antibody to the ELDKW peptide region, or an already licensed antiviral drug such as Fuzeon, to blunt the impact of SARS on infected human beings. This period of medical emergency in several countries worldwide has prompted us to release this model prior to formal publication.

Disclaimer: Drs. Gallaher and Garry have filed for invention rights emanating from the discoveries described in this release, but, to the best of their knowledge, do not have any financial interest in the 2F5 or Fuzeon drugs named herein, nor in the corporations owning or producing those drugs. Requests for further information should be directed to the Office for Information Services at LSU Health Sciences Center, (504)568-4806.

MODEL of the PRE-INSERTION REGION of the SPIKE (S2) FUSION GLYCOPROTEIN of the HUMAN SARS CORONAVIRUS: IMPLICATIONS FOR ANTIVIRAL THERAPEUTICS