Our findings show differences in 2 alternative codons of the FCoV S gene that correlate with the feline infectious peritonitis disease phenotype in >95% of cases. Besides providing a realistic basis for diagnostic discrimination of the 2 FCoV pathotypes, our findings also support the mutation hypothesis. Thus, we propose that alternative mutations in the S protein of FECV give rise to a tropism change that allows the virus to escape from the intestine into body tissues, where it causes feline infectious peritonitis. Proof of this hypothesis will require introduction of these mutations into the FECV genome and demonstration of the virulence switch by infection of cats. However, this is a formidable challenge in the absence of a reverse genetics system and a proper cell culture system to generate and propagate these viruses.
Our findings relating the S protein to FCoV pathogenicity are not surprising, given earlier explorations into the involvement of various genes (7a, 7b, M, and 3c) (5–7,10,11,19–22,24,25
). One of the most notable consequences of the presumed mutation in FECV is the acquisition of monocyte/macrophage tropism by the resulting virus (26
). Thus, whereas replication of FECV is restricted to the epithelial cells lining the gut, the virulence mutation enables FIPV to efficiently infect and replicate in macrophages and spread the infection systemically (26
). Such tropism change corresponds most logically with a modification in the S protein. An earlier study, using serotype II FCoVs, indicated a virulence role for the S protein (21
); however, identification of the mutation(s) was not pursued because of the controversial nature of the FECV strain used in the study (14
As for the serotype II viruses, the putative virulence mutations detected in the serotype I FCoV spike occur in the membrane-proximal domain of the protein. In coronaviruses, the S protein functions in cell entry; it is responsible for receptor attachment and membrane fusion. While the receptor binding site is located in the N terminal part of the protein, fusion is mediated by its membrane-proximal part. Coronavirus S proteins are class I fusion proteins, which typically contain domains instrumental for this process: 2 heptad repeat regions and a fusion peptide (27
). The fusion peptide is located just upstream of the membrane-distal heptad repeat region, but it remains to be proven that it functions as a fusion peptide. The 2 putative virulence mutations identified in our study, M1058L and S1060A, map to this characteristic hydrophobic domain. Both changes are subtle and do not give clues as to their functional consequences. We assume, however, that these alternative mutations have a similar effect, and we speculate that the mutations in the remaining 4% of cases might also occur in the fusion peptide of the S protein.
If these mutations are all that is needed to convert a nonvirulent FECV into a lethal FIPV, the question arises as to why feline infectious peritonitis occurs so infrequently. For example, simple calculations based on a 10−4
frequency and a stochastic occurrence of RNA polymerase errors across the genome (28
) predict that the M1058L mutation, for which 2 alternative substitutions of A23531
(to T or G) occur, would statistically arise once in every 1.5 × 104
genomes produced. In experimental FECV infection of kittens, we showed that up to 108
genome equivalents of the virus are shed per microliter of feces (18
); thus, typical FECV infections would be expected to generate thousands of progeny carrying 1 of the critical mutations. However, the virulence phenotype supposedly associated with the mutation is not observed to any proportional extent. We can only speculate as to the reasons.
One likely possibility is that additional mutations (1 or more, perhaps alternative mutations) are required to generate the virulent pathotype. Such mutations would most probably involve the accessory gene 3c, which is intact in FECVs but severely affected in about two thirds of FIPVs (7,10–12
). The 3c protein apparently is essential for replication of FECV in the gut but becomes nonessential once virulence mutation(s) elsewhere in the genome (e.g., in the S gene) enable the virus to infect macrophages and spread systemically. As we suggested earlier, loss of 3c function may not only be tolerated, it may even enhance the fitness of the mutant virus in its new biotope and, as a consequence, hamper its return to the gut. If the mutant virus is absent in the gut, it will not be shed in feces, providing an explanation for the seemingly rare incidence of feline infectious peritonitis outbreaks. Our discoveries of the critical differences between FECVs and FIPVs are clearly only a small step toward understanding the pathogenetic phenomena of feline coronavirus infections.