A Reverse Genetics Approach To Study Feline Infectious Peritonitis

Abstract

Feline infectious peritonitis (FIP) is a lethal immunopathological disease caused by feline coronaviruses (FCoVs). Here, we describe a reverse genetics approach to study FIP by assessing the pathogenicity of recombinant type I and type II and chimeric type I/type II FCoVs. All recombinant FCoVs established productive infection in cats, and recombinant type II FCoV (strain 79-1146) induced FIP. Virus sequence analyses from FIP-diseased cats revealed that the 3c gene stop codon of strain 79-1146 has changed to restore a full-length open reading frame (ORF).

TEXT

Feline coronaviruses (FCoVs) are members of the Alphacoronavirus genus within the family Coronaviridae. They are widespread in the cat population, with seropositivity of up to 90% (13, 19), are horizontally transmitted via the fecal-oral route, and can cause persistent infection (1, 2, 7, 9, 10, 19). FCoVs can be divided into type I and type II serotypes (14, 15, 20), with type I FCoVs accounting for the majority (90%) of infections (2, 13, 17, 18). Type II FCoVs acquired the S gene and adjacent regions through recombination between type I FCoV and the closely related canine coronavirus (5, 11, 26). Despite this genetic difference, both serotypes can be differentiated into two biotypes. Feline enteric coronavirus (FECV) can cause mild enteritis and persistent infection. Feline infectious peritonitis virus (FIPV) causes the highly lethal immunopathological disease called feline infectious peritonitis (FIP) that is accompanied by systemic inflammation of serosal membranes, disseminated pyogranulomas in multiple organs, and, frequently, accumulation of exudate in the thorax and abdominal cavity (ascites) (8, 19). According to the widely accepted “internal mutation theory,” FIPVs develop via mutations from FECVs during persistent infection in individual infected cats (4, 8, 19). Changes in the S gene and the accessory genes 3c, 7a, and 7b have been reported to correlate with a biotype switch from FECV to FIPV (3, 4, 12, 16, 19, 21–23, 27). Particularly, accessory gene 3c has received wide attention, since it is mutated in the majority of FIPVs, whereas FECVs possess an intact 3c gene (3, 21, 22). This observation led to the hypothesis that an intact 3c gene is required for FECV replication in the gut but is dispensable for systemic FIPV replication.

Despite these compelling observations, it should be noted that samples are usually from field cases and it is difficult to accurately trace with which FECVs these animals have initially been infected. Thus, it remains to be experimentally verified which mutations indeed promote a biotype switch. Ideally, such experimental validation may be done using well-defined prototype viruses. However, FECV field isolates are hardly, if at all, cultivable, and to date, there are mainly only FIPV isolates available that can be grown in tissue culture (8, 19). Moreover, with very few exceptions, these FIPV isolates are known to lose their pathogenicity through cell culture adaptation (19).

In order to overcome this limitation, recombinant FCoVs generated from cloned full-length cDNA are desirable because they permit the generation of well-defined virus stocks which originate from a cloned sequence after a few passages in cell culture. Based on our recently established reverse genetics system for the type I FCoV strain Black (25), we therefore generated a set of recombinant FCoVs using vaccinia virus-mediated recombination of full-length FCoV cDNA cloned in vaccinia virus (6, 24, 25). The recombinant FCoVs used in this study comprised (i) the type I FCoV strain Black (25), (ii) a type I FCoV Black strain with restored accessory gene 7b, (iii) a chimeric type I/type II FCoV, recFCoV-S-3abcII, encoding the structural gene S and the accessory genes 3a, 3b, and 3c from the highly pathogenic type II FCoV strain 79-1146, (iv), a chimeric type I/type II FCoV, recFCoV-1b-3abcII, possessing a genome organization similar to that of natural type II FCoVs, and (v) the type II FCoV strain 79-1146 (Fig. 1).

Fig 1.

Genome structure of recombinant FCoVs. ORFs of recombinant FCoVs used in this study are shown. ORFs derived from type I FCoV strain Black are shown as white boxes, and ORFs derived from type II FCoV strain 79-1146 are shown as black boxes. The uppermost panel shows ORFs of recFCoV and recFCoV-ΔStop-7b. The 62-nucleotide (nt) deletion at the border of ORFs 3b and 3c, which leads to a fusion of truncated ORFs 3b and 3c to one ORF designated ORF 3bc, is represented by a triangle. The introduced change to convert the stop codon UAA in ORF 7b of recFCoV to a CAA codon encoding glutamine in ORF 7b of recFCoV-ΔStop-7b is depicted by an asterisk and shown at the nucleotide and amino acid level. The middle panels show ORFs of the chimeric type I/type II FCoVs recFCoV-S-3abcII and recFCoV-1b-3abcII. Junctions of type I and type II FCoV sequences are enlarged to depict the nucleotide sequences (regular letters for type I FCoV-derived nucleotides and bold letters for type II FCoV-derived nucleotides) and the corresponding amino acid sequences for the relevant FCoV ORFs (shaded in gray). Nucleotide numbers are shown according to the published genomic sequences of the type I FCoV strain Black (GenBank accession number EU186072) and type II FCoV strain 79-1146 (GenBank accession number NC_002306). The stop codon in the type II FCoV strain 79-1146-derived ORF 3c is indicated by an asterisk.

All recombinant FCoVs were successfully rescued after electroporation of full-length in vitro-transcribed RNA into BHK-FCoV-N cells (25), and their integrity was confirmed by sequencing analysis. Following infection of Felis catus whole-fetus 4 (FCWF-4) cells (25), the recombinant virus recFCoV-ΔStop-7b displayed growth kinetics indistinguishable from those of the recombinant type I FCoV strain Black (recFCoV) (25) and the corresponding isolate with peak titers of 10⁴ PFU/ml at 36 h postinfection (p.i.) (Fig. 2A). Similar to what was previously described for FCoVs encoding the type II FCoV strain 79-1146 S protein (24), viruses containing type II sequences (recFCoV-II) induced full cytopathic effect (CPE) on FCWF-4 cells faster than the type I FCoVs (data not shown), gave rise to significantly larger plaques (recFCoV-II, 0.80 mm ± 0.15 mm; recFCoV, 0.28 mm ± 0.07 mm; n = 31; P < 0.001), and reached peak titers in the range of 10⁶ PFU/ml within 24 h p.i. We reproducibly observed slightly higher peak titers for the type II FCoV strain 79-1146 isolate than for the recombinant recFCoV-II isolate (P < 0.001; 24 h p.i.), although their genomic sequences have only two silent nucleotide differences (Table 1).

Fig 2.

Analysis of recombinant FCoV infection in vitro. Growth kinetics of the type I FCoV strain Black, recFCoV, and recFCoV-ΔStop-7b (A) and recFCoV-S-3abcII, recFCoV-1b-3abcII, recFCoV-II, and the type II FCoV strain 79-1146 (B). FCWF-4 cells were infected with the indicated virus with a multiplicity of infection (MOI) of 0.01, and virus titers in cell culture supernatants were determined as numbers of PFU at the indicated time points postinfection (p.i.). Results are the means ± the standard errors of the mean (SEM) of two independent experiments done in triplicate. ***, P < 0.001 (unpaired Student t test; recFCoV-II versus the type II FCoV strain 79-1146 at 24 h p.i.).

Table 1.

Summary of nucleotide and amino acid differences compared to the published type II FCoV strain 79-1146 sequence^a

Published sequence	Type II FCoV strain 79-1146b	recFCoV-II	Ascites (cat 11 infected with the type II FCoV strain 79-1146)	Spleen (cat 13 infected with recFCoV-II)	Kidney (cat 13 infected with recFCoV-II)	Proteinc
T			T616C (silent)			1a-nsp1
G	G796T (Trp→Cys)	G796T (Trp→Cys)	G796T (Trp→Cys)	G796T (Trp→Cys)	G796T (Trp→Cys)	1a-nsp2
A			A4682G (Ile→Val)			1a-nsp3
A					A7544C (Ser→Arg)	1a-nsp3
C					C7933T (silent)	1a-nsp4
T					T7996C (silent)	1a-nsp4
G		G9943C (silent)		G9943C (silent)	G9943C (silent)	1a-nsp6
T		T10729C (silent)		T10729C (silent)	T10729C (silent)	1a-nsp6
G					G13089C (Met→Ile)	1a-nsp6
G	G17709A (silent)	G17709A (silent)	G17709A (silent)	G17709A (silent)	G17709A (silent)	1ab-nsp14
G			G20567T (silent)			S
T			T20657C (silent)			S
A			A20742G (Lys→Glu)			S
G			G21366T (Val→Leu)			S
G					G22365C (Val→Leu)	S
T				T22468C (Leu→Pro)	T22468C (Leu→Pro)	S
G			G23280T (Asp→Tyr)			S
T					T24155G (Cys→Trp)	S
T			T25353C (Stop→Gln)	T25353C (Stop→Gln)	T25353C (Stop→Gln)	3c
T					T25562C (silent)	3c
T			T26165C (Thr→Ile)			E
G					G27819T (Cys→Phe)	N

^aThe published type II FCoV strain 79-1146 sequence is in GenBank under accession number NC_002306.

^bThis sequence corresponds to the type II FCoV strain 79-1146 determined in our laboratory.

^c1a, open reading frame 1a; nsp; nonstructural protein; S, spike protein; 3c, 3c protein; E, envelope protein; N, nucleocapsid protein.

To assess viral pathogenicity, groups of two FCoV-seronegative, specific-pathogen-free (SPF) cats (Charles Rivers, France) at an age of 4 to 5 months were inoculated intraperitoneally (i.p.) with 10⁶ PFU of virus or phosphate-buffered saline (PBS) as a negative control. Cats were monitored daily for clinical signs, and blood and fecal samples were collected weekly over a period of 7 to 8 weeks. All FCoVs established productive infection in cats (Fig. 3), as determined by detection of viral RNA in feces by reverse transcription-PCR (RT-PCR) and induction of FCoV-specific antibody responses, measured in serum samples by endpoint dilution using FCoV-infected CRFK cells. Notably, viral RNA in fecal samples was not detectable by RT-PCR at all time points, a result which is in concordance with a recent report of experimental i.p. FCoV infection (22). None of the type I or chimeric FCoVs induced clinical signs of FIP, suggesting that the type I FCoV strain Black isolate used in our laboratory had already lost pathogenicity through cell culture adaptation. Furthermore, an intact 7b accessory gene carried by recFCoV-ΔStop-7b did not confer a detectable gain of pathogenicity in vivo, suggesting that additional attenuating mutations are elsewhere in the genome. The apathogenic phenotype of the chimeric FCoVs suggests that the type-II-derived sequences (open reading frame [ORF] 1b; genes S, 3a, 3b, and 3c) are not sufficient to increase pathogenicity and further supports that the remaining sequences derived from type I FCoV strain Black (5′ untranslated region [UTR]; ORF 1a; genes M, N, 7a, and 7b; 3′ UTR) may contain attenuating cell culture adaptations.

Fig 3.

Analysis of recombinant FCoVs in vivo. Groups of two FCoV-seronegative SPF cats were infected intraperitoneally (i.p.) with 10⁶ PFU of the type I FCoV strain Black (cat 1 and cat 2) (A), recFCoV (cat 3 and cat 4) (B), recFCoV-ΔStop-7b (cat 5 and cat 6) (C), recFCoV-S-3abcII (cat 7 and cat 8) (D), recFCoV-1b-3abcII (cat 9 and cat 10) (E), the type II FCoV strain 79-1146 (cat 11 and cat 12) (F), and recFCoV-II (cat 13 and cat 14) (G). FCoV-specific antibody titers were determined from serum, and fecal samples were collected and analyzed once a week until 7 weeks p.i. Arrows indicate the time points at which FCoV RNA could be detected by RT-PCR in fecal samples. For animals inoculated with recFCoV-II or the type II FCoV strain 79-1146, body temperatures (°C) were determined.

In sharp contrast, animals infected with the type II FCoV strain 79-1146 and recFCoV-II showed signs of clinical disease that started to develop at 2 to 3 weeks p.i. The animals showed high antibody titers, anorexia, loss of appetite and weight, conjunctivitis, and anemia, which were accompanied by recurring fever typical for FIP (Fig. 3F and G). In both groups, one cat had to be euthanized at 7 weeks p.i. when clinical manifestation of FIP became apparent (cats 11 and 13). In both of these cats, abdominal fluid (ascites) was found, while the liver, the spleen, the kidneys, and the serosa of the peritoneal cavity were covered by pyogranulomas. During the postmortem investigation, characteristic signs of FIP could also be detected in cats 12 and 14. These data demonstrate that the recombinant recFCoV-II derived from cloned cDNA, like the type II FCoV strain 79-1146 isolate, is highly pathogenic and can induce FIP in cats.

By amplifying a set of overlapping FCoV-specific PCR products using RNA isolated from ascites of cat 11 (infected with the type II FCoV strain 79-1146), we successfully determined the full-length genome sequence. We noticed nine nucleotide differences, with three nonsynonymous mutations in the S gene, one in nonstructural protein (nsp) 3 of the replicase gene, and one in the E gene (Table 1). Surprisingly, we found that the stop codon in the 3c accessory gene of type II FCoV strain 79-1146 (input virus) was changed to encode a glutamine residue (Table 1). We also determined the full-length genomic FCoV sequence by RT-PCR using RNA isolated from the spleen and kidney of the FIP-diseased cat 13 (infected with recFCoV-II). We detected 2 nonsynonymous nucleotide changes in RNA isolated from the spleen and 7 nonsynonymous and 3 silent changes in RNA from kidney, suggesting that in vivo, recFCoV-II further diversified in the kidneys (Table 1). Importantly, the restoration of the ORF 3c was also present in the viral RNA derived from the recFCoV-II-infected FIP-diseased cat. Moreover, by performing a 3c-specific RT-PCR, we also detected the restoration of ORF 3c in viral RNA in the gut and fecal samples from FIP-diseased cats 11 and 13, which had been infected with the type II FCoV strain 79-1146 and recFCoV-II, respectively (data not shown). Although this finding appears to contradict previous reports (3, 22), it remains to be determined at which phase during the infection in vivo, and in which specific target cells/organs, a functional 3c protein is promoting replication and when it becomes dispensable. It should be noted that previous studies reported mutated 3c genes mainly in the context of natural FCoV infection. This includes a first phase of replication of an FECV in the gut before pathogenic FIPV variants, often containing mutations in the 3c gene, may emerge.

In conclusion, we have generated a set of full-length FCoV cDNA clones that provides a valuable basis to study the molecular pathogenesis of FIP. In the long term, it will be important to extend this reverse genetics approach to pathogenic type I FCoVs. This will greatly help to identify and compare pathogenicity factors of type I and type II FCoVs that critically contribute to the development of FIP. Furthermore, we believe that this approach is exceptionally well suited to generate recombinant FECV that may be used to study early events of enteric FCoV replication and the possible emergence of pathogenic FIPV variants in individual cats. Finally, we expect that FCoV reverse genetics will significantly contribute to furthering our knowledge on the molecular biology and pathogenesis of FIP and, thus, provide the basis for the rational design of efficacious vaccines to prevent FIP.