Abstract
In addition to the transmembrane protein, GP~1,2~, the Ebola virus glycoprotein gene encodes the soluble glycoproteins sGP and Δ‐peptide. Two more soluble proteins, GP~1~ and GP~1,2__Δ__TM~, are generated from GP~1,2~ as a result of disulfide‐bond instability and proteolytic cleavage, respectively, and are shed from the surface of infected cells. The sGP glycoprotein is secreted as a disulfide‐linked homodimer, but there have been conflicting reports on whether it is arranged in a parallel or antiparallel orientation. Off‐line HPLC‐MALDI‐TOF MS (MS/MS) was used to identify the arrangement of all disulfide bonds and simultaneously determine site‐specific information regarding N‐glycosylation. Our data prove that sGP is a parallel homodimer that contains C53–C53’ and C306–C306’ disulfide bonds, and although there are six predicted N‐linked carbohydrate sites, only five are consistently glycosylated. The disulfide bond arrangement was confirmed by using cysteine to glycine mutations at amino acid positions 53 and 306. The mutants had a reduced ability to rescue the barrier function of TNF‐__α‐__treated endothelial cells—a function previously reported for sGP. This indicates that these disulfide bonds are critical for the proposed anti‐inflammatory function of sGP.