Murine coronavirus evolution in vivo: functional compensation of a detrimental amino acid substitution in the receptor binding domain of the spike glycoprotein

Abstract

Murine coronavirus A59 strain causes mild to moderate hepatitis in mice. We have previously shown that mutants of A59, unable to induce hepatitis, may be selected by persistent infection of primary glial cells in vitro. These in vitro isolated mutants encoded two amino acids substitutions in the spike (S) gene: Q159L lies in the putative receptor binding domain of S, and H716D, within the cleavage signal of S. Here, we show that hepatotropic revertant variants may be selected from these in vitro isolated mutants (Q159L-H716D) by multiple passages in the mouse liver. One of these mutants, hr2, was chosen for more in-depth study based on a more hepatovirulent phenotype. The S gene of hr2 (Q159L-R654H-H716D-E1035D) differed from the in vitro isolates (Q159L-H716D) in only 2 amino acids (R654H and E1035D). Using targeted RNA recombination, we have constructed isogenic recombinant MHV-A59 viruses differing only in these specific amino acids in S (Q159L-R654H-H716D-E1035D). We demonstrate that specific amino acid substitutions within the spike gene of the hr2 isolate determine the ability of the virus to cause lethal hepatitis and replicate to significantly higher titers in the liver compared to wild-type A59. Our results provide compelling evidence of the ability of coronaviruses to rapidly evolve in vivo to highly virulent phenotypes by functional compensation of a detrimental amino acid substitution in the receptor binding domain of the spike glycoprotein.

FIG. 1.

(A) Schematic of MHV-A59 Spike (S) showing the approximate position of hepatitis revertant hr2 virus amino acid substitutions relative to the receptor binding domain (RBD), cleavage signal, and heptad repeat domains (HR1 and HR2) of S. Spike is cleaved into two 90-kDa noncovalently associated subunits, S1 and S2. S1 contains the receptor binding domain (RBD) and S2 contains amphipathic heptad repeat sequences (HR1 and HR2) important to engage in coiled-coil formation. Q159L lies in the putative RBD of S, H716D within the cleavage signal of S (RRAHR), and E1035D is located in HR1. R654H substitution maps in a region of S in which functional domains have not been yet identified. (B) Scheme of targeted RNA recombination. Feline cells (FCWF) were infected with fMHV, a chimeric recombinant MHV virus expressing the FIPV spike, and then electroporated with pMH54-derived, in vitro transcribed RNA containing the engineered mutations in the spike gene. These infected and eletroporated FCWF cells were overlaid onto murine L2 cells, and recombinants viruses were selected on their ability to infect murine cells (as described in the text).

FIG. 2.

Time course of released (A and C) and cell-associated virus (B and D) production in L2 cells cultures. Replication kinetics of viruses with a WT cleavage site are shown in A and B, whereas viruses with the H716D amino acid substitution are depicted in C and D. Released and cell-associated kinetics of recombinant RA59 and Q159L-R654H-H716D-E1035D viruses are shown in all panels (A through D). The C12 isolate (Q159L-H716D) has been previously studied in vitro (17); C12 exhibits released and cell associated kinetics similar to hr2 virus (data not shown). L2 cells were infected in duplicate with recombinant viruses at a multiplicity of infection of 1 PFU/cell. The data shown represent the mean titer of duplicate samples. Two independent recombinant viruses were analyzed. At indicated times, virus titers were determined in cells and culture supernatants by plaque assay in L2 cells.

FIG. 3.

(A) Viral load in liver of C57BL/6 mice at 1, 3, 5, and 7 days p.i. after i.h. inoculation with 500 PFU of parental hr2 virus as well as recombinant viruses Rhr2-A, Rhr2-B, Q159L, and RA59. The highly hepatotropic MHV-2 (parental virus) and a recombinant A59 expressing the spike of MHV-2 (Penn98-1) were used as controls. Viral titers were determined by plaque assay and are presented as log₁₀ PFU/g of liver. Errors bars represents logarithmic standard deviation. The limit of detection was 200 PFU/g of liver. Five mice per day per virus were examined. Viruses expressing the hr2 spike (hr2 parental, and recombinants Rhr2-A, Rhr2-B) exhibited significant higher viral load in the liver of mice (P < 0.05). Viral load in liver (B), and brain (C) at 1, 3, 5, and 7 days p.i. from mice inoculated intracranially with 500 PFU of the above viruses. Parental and recombinant hr2 viruses exhibited higher viral titers in liver (B) than in brain (C) after i.c. inoculation (P < 0.05). No significant differences in viral titers among viruses were observed in the brain (C).

FIG. 4.

Viral load in liver of C57BL/6 mice at 5 day p.i. after intrahepatic inoculation with 500 PFU of recombinant viruses RA59, Q159L-H716D, Q159L-R654H-H716D-E1035D, Q159L-R654H-H716D, Q159L-H716D-E1035D, Q159L-R654H-E1035D, Q159L-E1035D, R654H-E1035D, Q159L-R654H, R654H-H716D, Q159L, R654H, E1035D, and H716D. Viral titers were determined by plaque assay and are presented as log₁₀ PFU/g of liver. The limit of detection was 200 PFU/g of liver. Ten mice were examined per virus, and two independent recombinant viruses were evaluated (only results from one independent recombinant per virus are shown).

FIG. 5.

(A) Susceptibility of C57BL/6 mice to recombinant H716D virus infection after i.c. (•), as well as i.h. (□) inoculations. Survival curves were determined as described in the text. (B) Viral load in liver and brain of mice inoculated after i.h and. i.c. inoculations with two independent recombinant H716D viruses (R# A, R# B).

FIG. 6.

Immunohistochemistry of liver sections of C57BL/6 mice infected with the recombinant viruses and mock-infected control, at day 5 p.i. MHV was detected by immunolabeling with a MAb against the N protein of MHV as described in the text. Viral antigen always colocalized with necrotic areas. No signs of viral antigen were found in a mock-infected control. Magnification, ×100.