Feline Coronavirus Infection of Domestic Cats Causes Development of Cross-Reactive Antibodies to SARS-CoV-2 Receptor Binding Domain

The current study was initiated when our specific pathogen-free laboratory toms developed unexpectedly high levels of cross-reactive antibodies to human SARS-CoV-2 (SCoV2) receptor binding domain (RBD) upon mating with feline coronavirus (FCoV)-positive queens. Multi-sequence alignment analyses of SCoV2 Wuhan RBD and four strains each from FCoV serotypes 1 and 2 (FCoV1, FCoV2) demonstrated amino acid sequence identity of 11.5% and similarity of 31.8% with FCoV1 RBD, as well as 12.2% identity and 36.5% similarity for FCoV2 RBD. The sera from all three toms and three mated queens cross-reacted with SCoV2 RBD and reacted with FCoV1 RBD and FCoV2 spike-2, nucleocapsid, and membrane proteins of FCoV2 whole-virus, but not with FCoV2 RBD. Additionally, the plasma from all six FCoV2-inoculated laboratory cats reacted with FCoV2 and SCoV2 RBDs, but not with FCoV1 RBD. In another study, eight group-housed laboratory cats from a different lineage had a range of serum cross-reactivity to SCoV2 RBD even 15 months later. Such cross-reactivity was also observed in FCoV1-positive group-housed pet cats. The SCoV2 RBD at a high non-toxic dose and FCoV2 RBD at a 60-400-fold lower dose blocked the in vitro FCoV2 infection of the feline cells, demonstrating their close structural conformations essential as vaccine immunogens. Furthermore, such cross-reactivity to SCoV2 RBD was also detected by the peripheral blood mononuclear cells of both transient and chronically FCoV1-infected cats. Overall, the cross-reactivity with SCoV2 RBD by the sera from both serotypes of FCoV-infected cats also suggests that the cross-reactive epitope(s) on FCoV1 and FCoV2 RBDs may be similar to those of SCoV2 RBD and provides essential insights to developing a pan-CoV vaccine. Author Summary To date, there are no reports on the sera from feline coronavirus (FCoV)-infected cats cross-reacting with either SARS-CoV-1 or SARS-CoV2 (SCoV2) receptor binding domains (RBDs). This report describes the presence of cross-reactive antibodies to SCoV2 RBD in the sera of FCoV-infected laboratory cats, even though SCoV2 RBD and each FCoV serotype (FCoV1, FCoV2) RBD had minimal sequence similarity. However, this observation of serum cross-reactivity to SCoV2 RBD was confirmed by more stringent antibody-based assays and viral assays. Furthermore, both serotypes of FCoV-infected cats, including FCoV1-infected pet cats, produced the cross-reactive antibodies, and such cross-reactivity to SCoV2 RBD was also detected, most likely, by the T cells in peripheral blood mononuclear cells of both transient and chronically FCoV1-infected cats. Since SCoV2 RBD is essential component for current vaccines against COVID-19 disease, our findings should provide essential insights to developing a pan-coronavirus vaccine that induces full-scale immunity to completely prevent SCoV2 infection in humans and pet animals.


Abstract (265)
The current study was initiated when our specific pathogen-free laboratory toms developed unexpectedly high levels of cross-reactive antibodies to human SARS-CoV-2 (SCoV2) receptor binding domain (RBD) upon mating with feline coronavirus (FCoV)positive queens. Multi-sequence alignment analyses of SCoV2 Wuhan RBD and four strains each from FCoV serotypes 1 and 2 (FCoV1, FCoV2) demonstrated amino acid sequence identity of 11.5% and similarity of 31.8% with FCoV1 RBD, as well as 12.2% identity and 36.5% similarity for FCoV2 RBD. The sera from all three toms and three mated queens cross-reacted with SCoV2 RBD and reacted with FCoV1 RBD and FCoV2 spike-2, nucleocapsid, and membrane proteins of FCoV2 whole-virus, but not with FCoV2 RBD. Additionally, the plasma from all six FCoV2-inoculated laboratory cats reacted with FCoV2 and SCoV2 RBDs, but not with FCoV1 RBD. In another study, eight group-housed laboratory cats from a different lineage had a range of serum cross-reactivity to SCoV2 RBD even 15 months later. Such cross-reactivity was also observed in FCoV1-positive group-housed pet cats. The SCoV2 RBD at a high non-toxic dose and FCoV2 RBD at a 60-400-fold lower dose blocked the in vitro FCoV2 infection of the feline cells, demonstrating their close structural conformations essential as vaccine immunogens. Furthermore, such cross-reactivity to SCoV2 RBD was also detected by the peripheral blood mononuclear cells of both transient and chronically FCoV1-infected cats. Overall, the cross-reactivity with SCoV2 RBD by the sera from both serotypes of FCoV-infected cats also suggests that the cross-reactive epitope(s) on FCoV1 and FCoV2 RBDs may be similar to those of SCoV2 RBD and provides essential insights to developing a pan-CoV vaccine.

Introduction
The global prevalence of feline coronavirus (FCoV) infection ranges from 6.6%-95% in multi-cat households and catteries [1][2][3]. FCoV is distributed into two phylogenic lineages of serotypes 1 and 2 (FCoV1, FCoV2). Both serotype FCoVs infect predominantly epithelial cells of the gastrointestinal tract and cause a mild gastrointestinal disease (e.g., diarrhea, vomiting, and transient weight loss) in domestic cats, especially in kittens. Often, upon chronic infection, these viruses can also mutate into pathogenic and fatal variants. These variants, called feline infectious peritonitis viruses (FIPVs), infect monocytes and macrophages, spreading throughout the body [4][5][6]. No known cases of FCoV infection in humans have been reported. However, the FCoV sequence sections have been found in recombinant coronaviruses infecting humans [7][8][9]. In comparison, many cases of SARS-CoV-2 (SCoV2) infection of pet cats have been reported worldwide through transmission from COVID-19-positive owners to their pet cats. These cats displayed mild symptoms or remained asymptomatic, even when found positive for SCoV2 by RT-PCR [10][11][12]. Experimental inoculation of laboratory cats with a human isolate of SCoV2 caused infection, ranging from mild to asymptomatic symptoms. These infected cats were diagnosed as RT-PCR positive through nasal and/or fecal swabs [13][14][15]. SCoV2 transmission from inoculated cats to non-inoculated cats through contact exposure has been demonstrated by multiple research groups worldwide [13,15].
Similarly, transmissions of SCoV2 from COVID-19-positive owners to their symptomatic and asymptomatic pet dogs have been reported [11,12,16]. Experimental infection of laboratory dogs also confirmed that dogs could be infected with SCoV2 [13,14].
Studies have shown that FCoV-infected cats develop antibodies that cross-react with SARS-CoV1 (SCoV1) nucleocapsid [17] but not with SCoV1 spike 1 (S1) glycoprotein [18], where the SCoV1 receptor binding region (RBD) is located [19]. SCoV1 infection in humans was first discovered in November 2002 in a patient with atypical pneumonia in Guangdong, China [20]. By February 2003, this virus was rapidly transmitted to a large population in Hong Kong, causing severe acute respiratory syndrome (SARS). The intermediate host for SCoV1 was suspected to be a masked palm civet [21]. SCoV1 preparation isolated from a patient, was inoculated intratracheally into six laboratory cats. This resulted in asymptomatic infection in all four cats, based on positive virus isolation and viral RT-PCR of the pharynx, trachea, and lungs [22]. Furthermore, two non-inoculated cats housed together with the SCoV1 inoculated cats became PCR positive starting day 2 of contact exposure with a peak titer at day 6. By day 28, they had seroconverted with a virus-neutralizing antibody (NAb) titer of 40 and 160.
One pet cat living in an apartment block with over 100 residents positive for SCoV1 was also positive for SCoV1 infection [22].
Both SCoV1 and SCoV2 belong to genus Betacoronavirus of family Coronaviridae, while FCoV1 and FCoV2 belong to genus Alphacoronavirus based on phylogenetic analyses [23]. Both SCoV1 and SCoV2 RBDs bind with human angiotensin converting enzyme-2 (hACE2) to infect hACE2-expressing human cells and serving as the major target of the NAbs generated by the infected hosts [24]. In contrast, neither FCoV1 nor FCoV2 use ACE2 as their host cell receptor [25,26]. The cell receptors for FCoV2 and canine coronavirus serotype 2 (CCoV2) are reported to be feline and canine aminopeptidase-N (fAPN and cAPN), respectively, which is like the human cold Alphacoronavirus (HCCoV) 229E, using human APN (hAPN) as its primary cell receptor [25,27,28]. The cell receptor for FCoV1 is still unknown and this is complicated by the fact that FCoV1 isolates do not readily replicate in cell cultures [6,29]. FCoV1 infection is more prevalent worldwide, including the US, while the FCoV2 infection is found predominantly in Southeast Asia [1-3, 6,30]. FCoV2 is reported to be a recombinant FCoV1 backbone with CCoV2 spike (S) glycoprotein [29,31]. Since hACE2 is the major cell receptor for SCoV2, SCoV2 infections of cats and dogs are reported to be mediated by the feline ACE2 (fACE2) and canine ACE2 (cACE2), respectively, based on their amino acid (aa) sequence similarity of fACE2 and cACE2 to hACE2 and the binding analyses of SCoV2 to species-specific ACE2 [23,24,32,33].
The aa sequence comparison between SCoV1 and SCoV2 S proteins demonstrates a considerably high sequence identity of 76%, and their RBDs similarly show a high sequence identity of 73% [34][35][36]. Hence, the aa sequence similarity should be much higher. Since the sera from FCoV1 and FCoV2 infected cats cross-react with SCoV1 nucleocapsid but not with SCoV1 S1 glycoprotein [17,18], it can be inferred from those reports that sera from FCoV1 and FCoV2 infected cats can cross-react with SCoV2 nucleocapsid but not with SCoV2 S1 glycoprotein, and, consequently, not with the SCoV2 RBD which resides on S1 glycoprotein. Furthermore, a few aa sequence analysis studies demonstrate that the Wuhan SCoV2 S glycoprotein is distinctly different from S glycoproteins of FCoV1 and FCoV2 [23,31]. The location for FCoV1 and FCoV2 RBDs has yet to be determined by biological analysis, but it has recently been predicted to be around residues 526-676, based on RBD sequence locations of porcine enteric diarrhea virus (PEDV) RBD at B residues 510-640 and transmissible gastrointestinal enteric virus (TGEV) RBD at D3 residues 500-651 [37][38][39]. The TGEV RBD has been reported to bind to the porcine APN as its primary host cell receptor [27,39], but whether PEDV RBD binds to pAPN as its host cell receptor is still controversial [38,40]. The RBD sequence prediction of both PEDV and TGEV was based on monoclonal Ab (MAb) studies, identifying the most potent neutralizing MAb(s) to these porcine Alphacoronaviruses reacting to B and D3 residue regions, respectively [38,39]. Although neutralizing MAb (nMAb) studies have been performed against FCoV2 [41,42], the location and the native conformation of FCoV1 and FCoV2 RBDs have yet to be clearly defined and confirmed by biological analyses required for use as vaccine immunogens.
Our laboratory has recently discovered that, shortly after specific pathogen-free (SPF) toms mated with FCoV-positive queens, they developed minor episode(s) of diarrhea with only low production of FCoV antibodies (Abs) but unexpectedly high levels of cross-reactive Abs to human SCoV2 RBD. Consequently, as a confirmation of the original finding, these sera from FCoV infected cats were tested for their cross-reactivity to SCoV2 RBD by more stringent ELISA, immunoblot analyses, SCoV2 RBD blocking assay, and SCoV2 RBD stimulation of peripheral blood mononuclear cells (PBMC).
Lastly, we further characterized the nature of FCoV transmission in cats at inducing crossreacting Abs to SCoV2 RBD.

Initial serological studies in FCoV naturally infected laboratory cats
FCoV whole-virus (WV) ELISA and SCoV2 ELISA with overnight serum incubation: Our studies on FCoV were initiated when four laboratory queens donated by the University of Georgia (UGA) were mated with three SPF inbred toms from our laboratory at the University of Florida (UF) (Fig 1A). The four juvenile laboratory cats born from two UGA queens (UGAQ3, UGAQ4) had FCoV Abs at 12 and 16 weeks of age based on FCoV2-WV ELISA. The UGA queens were all seropositive for FCoV2 (Fig 1B), and the sera from the toms, after mating, were weakly seropositive for FCoV2 but had no neutralizing Abs (NAbs) to FCoV2 (Fig 1B). Since these juvenile cats were to be used for a SCoV2 inoculation study at UGA, their serum was collected before shipment to UGA and tested by SCoV2 RBD ELISA. Three of the four juvenile cats had a serum that cross-reacted moderately with the SCoV2 RBD ( Fig 1C).

Stringent FCoV-WV ELISA and SCoV2 RBD ELISA:
The sera from all four UGA queens and three toms were incubated at the same dilution for only one hour instead of overnight, and PBS was used instead of bicarbonate buffer for coating of the antigen on the ELISA plate. In addition, ELISA with bovine serum albumin (BSA) antigen was included since the UGA queens were vaccinated three years ago with commercial vaccines. The veterinary vaccines were often contaminated with BSA. This BSA control was also important because the FCoV2-WV preparation contained about 5% BSA, whereas the RBDs were highly purified and devoid of BSA. Two UGA queens (UGAQ1, UGAQ4) had high levels of serum reactivity with FCoV but without any reactivity with BSA and SCoV2 RBDs, while the other two queens (UGAQ2, UGAQ3) had high serum reactivities with FCoV, which were slightly higher than the serum reactivities with BSA ( Fig 1D). The sera from UGAQ2 also had substantial cross-reactivities with both SCoV2 RBDs (UF and MassBiologics (MB) RBDs) (S1A Fig), whereas the sera from UGAQ3 had a modest cross-reactivity with MB-RBD and below the threshold cross-reactivity with UF-RBD. All UGA queens were group housed together upon arrival on Oct. 25, 2019. The earliest serum collected from the queens was on Dec. 19, 2019 from UGAQ1 ( Fig 1D) As expected, the sera from all toms before mating had no reactivity with FCoV, both UF/MB-RBDs (Fig 1E), and BSA (S1B Fig). However, post-mating sera from all toms had significant reactivity with SCoV2 RBDs but not with FCoV, compared to the corresponding pre-mating control results. The highest titers to SCoV2 RBDs were observed at the earliest timepoint of serum collection closest to the first exposure to the FCoV-positive queen. Subsequent sera showed declines that were still significantly different from the pre-serum titers (p<0.05; pre-and post-exposure paired T-test). The decline suggested that the development of cross-reactive Abs to SCoV2 RBD could have occurred during active FCoV infection. A slight conflict was observed between the substantial reactivity with FCoV for the toms (Fig 1B) and no reactivity with FCoV for the toms (Fig 1E). This may be due to the technical difference as detailed above. Overall, the stringent ELISA confirmed the sera from all three toms cross-reacted strongly with SCoV2 RBD. Only one long-term FCoV-infected queen UGAQ2 also had a high titer of crossreactive Abs to SCoV2 RBD ( Fig 1D).

The lack of S1 and RBD aa sequence identity/similarity between SCoV2 and FCoVs
Sequence analyses of SCoV2 and FCoV structural proteins: The aa sequence comparison between Wuhan SCoV2 versus FCoV1 and between SCoV2 versus FCoV2 demonstrated SCoV2 structural proteins (spike, envelope, membrane) were shorter than FCoV1 and FCoV2, with the exception of nucleocapsid protein and S2 (S2A Fig). The least aa identity and similarity were observed between SCoV2 and FCoV1/FCoV2 at the S1 glycoprotein among the five structural proteins composing these viruses. The aa sequence identity and similarity between S1 glycoproteins of SCoV2 and FCoV1 had 13.0% and 40%, respectively, whereas, between SCoV2 and FCoV2, 16.3% and 43.4%.
When comparing the cleavage sites separating S1/S2 among SCoV2, FCoV1 and FCoV2, SCoV2 S1/S2 cleavage site appears closer to the FCoV1 S1/S2 cleavage site than that of FCoV2 [44,45]. In contrast, the S2 glycoprotein between SCoV2 and FCoV1/FCoV2 had the highest aa identity of 30.0% and 33.5% with the second highest similarity of 62.2% and 64.5%, respectively, among all structural proteins. Hence, S1 glycoprotein was the most distinctly different structural protein between SCoV2 and FCoV1/FCoV2 among all five structural proteins.
When FCoV1 and FCoV2 spike glycoprotein sequences were compared, the S1 glycoprotein sequence had 29.5% and 31.6% identity and 62.9% and 64.5% similarity, which were much lower than the S2 sequence that had 60.6% and 68.7% identity and 84.0% and 89.9% similarity (S2A Fig). The S1/S2 cleavage site for the FCoV2 is proposed to be at the S2' cleavage site (another cleavage site on S2 glycoprotein) next to the fusion peptide because S1/S2 cleavage motif is absent in the location usually observed for FCoV1 and other coronaviruses [44][45][46]. Note that most coronaviruses, including SCoV2 and FCoV1, have the S2' cleavage site, but, for brevity, only S1/S2 cleavage site results were used in our analysis. The difference in the aa sequence between FCoV1 and FCoV2 at the envelope (Env), NC, and M ranged from 92.8%-96.3% identity and 97.6%-98.5% similarity, which demonstrated the high conservation between FCoV1 and FCoV2 for those structural proteins. The molecular weight (MW) of the structure proteins without glycosylation was predicted (S2B Fig) to analyze FCoV2-WV immunoblot described later.
Sequence analyses of SCoV2 and FCoV RBDs: Next, the Wuhan RBDs of UF and MB were aligned with four known FCoV1 S1 sequences and another set with four known FCoV2 S1 sequences using Clustal O (1.2.1) multiple sequence alignment of JustBio alignment server (https://justbio.com/) (Fig 2A and 2B). The goal of this alignment analysis was to identify the potential FCoV1 and FCoV2 RBD sites for use as the vaccine immunogens. Our logic for using such a comparison is based on the fact that speciesspecific ACE2 is used by SCoV2 to infect cats, dogs, and humans. The Wuhan SCoV2 RBD sequence is distinctly different in aa sequence from the RBD sequences of FCoV1 (11.5% & 12.3% identity; 31.8% & 33.6% similarity) and FCoV2 (12.2% & 12.3% identity; 36.5% & 37.7% similarity) (Fig 2C). The first value is based on UF-RBD, and the second value is based on MB-RBD. As shown, the carboxyl-end of the UF-RBD sequence (residues 319-529) is shorter by 12 aa from the MB-RBD sequence (aa residues 319-541) (Fig 2A and 2B and S1A Fig). The FCoV2 RBD had slightly more aa similarity with the Wuhan SCoV2 UF-and MB-RBDs than those between the FCoV1 RBD and SCoV2 RBDs ( Fig 2C). Thus, the aa sequences of the two FCoV2 RBDs (36.5% and 37.7%), more than the two FCoV1 RBDs (31.8% and 33.6%), had slightly more similarity to SCoV2 UF-and MB-RBD sequences.
The full-length aa sequence comparison of S glycoproteins of the SCoV2 Wuhan strain and FCoV2 79-1146 strain displays the S1/S2 cleavage site for SCoV2 at a location different from the S1/S2 cleavage site for FCoV2 (S3A Fig) [36,44,45]. In addition, the SCoV2 RBD sequence alignment pattern with a single FCoV2 RBD sequence of the fulllength S protein sequence differs slightly from the one aligned with four FCoV2 RBD sequences ( Fig 2B), even though both analyses used the JustBio alignment server. The full-length S sequences between SCoV2 and FCoV2 show the RBD has the least sequence similarity followed by the N-terminal domain (NTD) and then C-terminal domain (CTD) with the most similarity or conservation among two S1 domains (S3B Fig). This pattern was the same whether the S1/S2 cleavage site for SCoV2 or FCoV2 was used.  (Fig. 2A). The S1/S2 cleavage site for SCoV2 is only 46 aa plus five gaps away from the counterpart S1/S2 cleavage site for FCoV1 (S4A Fig) [36,44].
Conversely, the FCoV1 S1/S2 cleavage site is 37 aa plus 14 gaps from the counterpart SCoV2 cleavage site [45]. Thus, the S1/S2 cleavage sites are closer to SCoV2 and FCoV1 than those between SCoV2 and FCoV2. The most aa sequence conservation is observed at the S2 sequence, followed by S1 CTD, S1 NTD, and S1 RBD, which is identical to those observed between SCoV2 and FCoV2.
The comparison of the proposed FCoV1 and FCoV2 RBDs show three major gaps from the mid-to-carboxyl-end, which overlaps our predicted receptor binding motif (RBM), and has only 27.3% aa identity and 62.5% aa similarity (Fig 2D and 2E). The FCoV1 and FCoV2 at the proximity to our RBMs are quite different which may explain why FCoV1 does not use fAPN as a host cell receptor. The aa mutation sites that lower or eliminate the neutralization activity of the reported six nMAbs are shown with three color-coded symbols for each nMAb below the FCoV2 RBD sequence (Fig 2D). The aa mutation sites are clustered at mid-to-carboxyl-end of our proposed RBD which indicate that FCoV2 neutralizing epitopes are at or in the proximity of our proposed RBM.

Immunoblot analyses of sera from queens and toms
The proposed FCoV2 RBD, with the most cross-reactive sectional regions overlapping with SCoV2 RBD, as highlighted in the magenta sequence ( Fig 2B), were 1) produced in the same cell expression system with the same plasmid, 2) purified similarly as SCoV2 UF-RBD in PBS, and 3) used to develop the immunoblot strips. Two sera (UGAQ2, UGAQ4) from the four queens and the serum from all three toms strongly cross-reacted with the SCoV2 UF-RBD ( Fig 3A), but none of them reacted with FCoV2 RBD (Fig 3B).
All sera from the queens reacted with FCoV2-WV immunoblot strips strongly at the M (28-32kDa), NC (43kDa), degraded spike S (100-125kDa), and entire spike S (190kDa) ( Fig   3C), with weaker bands at 20kDa, 22kDa, 40kDa, and 55kDa except for UGAQ1. The single serum available from UGAQ1 was collected when she was severely sick from uteritis and on antibiotics. Her serum only reacted to FCoV2-WV at 90kDa and higher ( Fig   3C) but not to SCoV2 (Fig 3A), FCoV2 (Fig 3B), and FCoV1 (Fig 3D) RBDs. The toms had reactivity to NC, M, and proteins at 10kDa, 20kDa and 22kDa but almost no reactivity to FCoV2-WV S proteins and degraded S glycoproteins above 90kDa. The same sera from the toms did not react to FCoV2 RBD except for one tom (HOGT3), which reacted weakly with FCoV2 RBD (Fig 3B) and without FCoV2 NAb titer ( Fig 1B) but strongly with FCoV1 RBD (Fig 3D). These results demonstrate that the serum from three UGA queens (UGAQ2, UGAQ3, UGAQ4) and all three toms reacted with FCoV1 RBD (Fig 3D). The nil-to-weak cross-reactivity to FCoV2 RBD and strong reactivity to FCoV1 RBD suggest that our toms and queens were infected with FCoV1 which is the most common serotype in the US. Their sera also had nil-to-minimum titers of FCoV2 NAbs ( Fig 1B) which further supports our theory of FCoV1 infection of our queens and toms. One serum from a UGA cat (UGA1.4) ( Fig 3B) with FCoV2 NAb titer of 378 was our weak control serum that reacted to the FCoV2 RBD.

Characterization of Abs to SCoV2 RBD
Only sera from UGAQ2, UGAQ4 (10 months post-arrival) and all three toms had strong cross-reactivity to SCoV2 RBD ( Fig 3A). Both 5HQT1 and HOGT3 were exposed to two FCoV-positive (FCoV + ) queens over two breeding cycles (Fig 1A), whereas HOJT2 was exposed to three different FCoV + queens and retained the strongest SCoV2 RBD reactivity even at 13 months post-first exposure to the first FCoV + queen ( Fig 3A). This observation suggested that this tom was actively producing SCoV2 RBD cross-reactive Abs perhaps by sequential exposure to multiple FCoV + queens. The queen UGAQ4 was exposed to three different FCoV + toms during the three breeding cycles (Fig 1A) and had a moderate reactivity to SCoV2 RBD at 10 months post-first exposure (HOJT2), which was also 3 months post-second exposure (5HQT1) (Fig 3A). Nevertheless, UGAQ4 did not retain the cross-reactive Abs to SCoV2 RBD even after the third exposure to the FCoV + tom (HOGT3) 13 months post-first exposure ( Fig 3A). This observation may suggest that the FCoV + toms, upon exposure to FCoV + queens, developed only low titers of FCoV infection based on the low reactivity of their FCoV2-WV Abs and nil-to-negligible serum reactivity to the bands above the predicted FCoV2 S2 band of 80-90 kDa (Fig 3C   and 3E2). The rabbit polyclonal Abs to SCoV2 S protein and to S2 protein, but not to S1 protein, cross-reacted with FCoV2 S2, degraded S, and S glycoproteins on FCoV2-WV immunoblot ( Fig 3E1). These results confirmed the FCoV2 S, degraded S, and S2 MWs are detected by the sera from most UGAQ cats (Fig 3C and 3E2). Thus, the sera from the three toms with strong cross-reactive Abs to SCoV2 RBD and negligible Abs to FCoV2 S and degraded S may suggest low FCoV1 infection in the toms and the possibility of cross-reactive SCoV2 RBD Abs in lowering their level of FCoV1 infection.

Determining the FCoV2 infection blocking activity of FCoV2 RBD
If the proposed FCoV2 RBD is indeed the RBD site for FCoV2, our FCoV2 RBD should be able to block the FCoV2 infection of the feline cell line (Fc9 cells). The duplicate wells, starting at 2.86 μg/mL, were serially diluted three-fold for each well, up to the titer of 0.106 μg/mL (fourth duplicate wells) and presented <50% cytopathic effect (CPE), whereas all wells for virus control had 100% CPE including the PBS controls ( Fig 4A). Thus, 1 TCID50 titer, the titer with 50% CPE, is between 0.035-0.106 μg/mL. Another assay demonstrated 1 TCID50 titer of 50% blocking observed at about 0.100 μg/mL of FCoV2 RBD ( Fig 4B).

RBD Abs
Eight laboratory cats living together in group housing were all seropositive for FCoV infection by ELISA and immunoblot analyses (S5A and S5B Fig). Since these cats share the litter pans and FCoV transmits by oral-fecal route [6,29,47], once the immunity to the first infection subsides, they can be susceptible to reinfection at different times based on the immune constitution of the cat. Cat G-2, with the highest FCoV Ab titer, had the strongest cross-reactive Abs to SCoV2 UF-RBD immunoblot, followed by cats G-1, G-5, and G-7, with modest reactivity to SCoV2 RBD (S5B1 Fig). The remaining cats, G-4 and G-6, had negligible-to-no reactivity during the study, and cat G-3 showed no reactivity initially but 15 months later developed strong cross-reactivity to SCoV2 RBD. All eight cats were negative for FCoV2 RBD Abs throughout the study (S5B2 Fig Interferon-gamma (IFNγ) mRNA production of PBMC from FCoV1-infected cats upon stimulation with SCoV2, FCoV1, or FCoV2 RBD The PBMC from a transiently FCoV1-infected cat (4GC) and chronically FCoV1-infected cat (G-3) from group-housed laboratory cats referenced above, developed substantial level of IFNγ mRNAs in response to SCoV2 RBD stimulation ( Fig 4H). The levels were close to those stimulated by T-cell mitogen (concanavalin A, ConA). Chronically infected cat G-3 also responded with slightly lower, but substantial, IFNγ mRNA level in response to FCoV1 RBD stimulation, whereas transiently infected cat 4GC had no response to FCoV1 RBD. Both infected cats had no response to FCoV2 RBD stimulation. As expected, the PBMC from two uninfected SPF cats (2FB, 4GA) developed no IFNγ mRNA response to all three RBDs, while developing a robust IFNγ mRNA level to ConA stimulation.

Discussion
Our novel finding of cross-reactive Abs to SCoV2 RBD developing in cats during active FCoV infection was unexpected (Fig 1C,1D and 1E) based on low aa sequence similarity at RBD of SCoV2 and FCoVs (Fig 2A and 2C). Hence, the original observation using conventional SCoV2 RBD ELISA was further analyzed and confirmed by more stringent SCoV2 RBD ELISA, immunoblot analysis, in vitro FCoV2 infection blocking assay with SCoV2 RBD, and, finally, cellular IFNγ immune responses to SCoV2 RBD. The peak sera from UGAQ2 and all three toms cross-reacted strongly with sensitive SCoV2 RBD ELISA and immunoblot but not with FCoV2 RBD (Fig 3A and 3B). Perhaps the most striking observation was the sera from the toms cross-reacting to SCoV2 RBD with nil-to-minimal reactivity to FCoV2 S glycoprotein on the FCoV2-WV immunoblot (Fig 3C). These results indicate that cross-reactive Abs to SCoV2 RBD appeared without major development of cross-reactive Abs to FCoV2 S glycoprotein, but such results may be different if FCoV1 immunoblot was available for our use. None of our cats (queens, toms, their kittens) developed major NAbs to FCoV2 based on FCoV2 79-1146 NAb assay (Fig 1B). This result further supports our theory that our cats were infected with FCoV1, which was later confirmed by their sera reacting strongly with FCoV1 RBD immunoblot strips ( Fig 3D). In addition, the FCoV1 RBD immunoblot strips only reacted with plasma from FCoV1 KU-2 experimentally infected cats but not with plasma from FCoV2 79-1146 inoculated cats ( Fig 4E). Conversely, FCoV2 RBD strips only reacted with sera from FCoV2-infected cats but not with plasma from FCoV1-infected cats ( Fig 4F).
Unexpectedly, the plasma from FCoV1 KU-2 infected cats did not cross-react to SCoV2 RBD, which was not due to the use of plasma, since all plasma from FCoV2 79-1146 inoculated cats strongly cross-reacted with SCoV2 RBD (Fig 4G). This observation suggests that not all FCoV1-infected cats develop cross-reactive Abs to SCoV2 RBD or the timing of the plasma collection was possibly too late (60 days post-inoculation). These factors, combined with a) low virus inoculation dose (1-mL/cat of 1 TCID50), b) oral/nasal inoculation route, and/or c) low S1/RBD immunogenicity of the infection may have contributed to the lack of SCoV2 RBD cross-reactivity, similar to the lack/loss of crossreactivity of UGAQ1 and UGAQ3 sera to SCoV2 RBD (Fig 3A).
The fact that our immunoblots are developed under reducing conditions supports our contention that the cross-reacting epitopes on SCoV2 RBD are either linear aa epitopes and/or glycosylated epitope(s). There are a sufficient number of sectional aa sequences similar between SCoV2 and FCoVs (Fig 2 and S3  The majority of these cats had cross-reacting Abs to SCoV2 RBD, ranging from low to high titers (S5B1 Fig), which were maintained 15 months later. Although these cats were unrelated to our inbred cats, SCoV2 RBD cross-reacting Abs developed in laboratory cats purchased from the same commercial cat vendor as the UGA queens, but of different cat lineages. This supports the view that the laboratory cats sold by this commercial vendor have lineage(s) with FLA genetic makeup, which may sustain higher titers of crossreactive Abs to SCoV2 RBD, like cats UGAQ2 and G-2 (Fig 3A and S5B1 Fig). Grouphousing appears to sustain the FCoV infection by reinfection without clinical episode. Our preliminary results indicate that FCoV1-infected pet cats in the field possess crossreactive Abs to SCoV2 RBD without manifesting clinical signs. These observations may also explain why only some SCoV2-infected pet cats display clinical signs while others remain asymptomatic. Such variation in clinical signs was also observed during SCoV2 infection of humans [52,53].
The lack of serum cross-reactivity with FCoV2 S glycoproteins by two of three toms was unexpected since all four queens, including UGAQ2, reacted with FCoV2 S glycoprotein (Fig 3C). This is based on a rationale derived from the fact that both rabbit polyclonal Abs to S2 protein and to S protein cross-reacted with S, degraded S, and S2 of the FCoV2-WV (Fig 3E). Since S2 glycoprotein has high aa sequence identity and similarity, the likelihood of the cross-reactivity with S2 rather than S1 component of the S glycoprotein is a strong possibility based on the study by Zhao et al. 2019 [18]. Zhao et al. did not observe any serum from 137 FCoV-infected laboratory cats cross-reacting with SCoV1 S1 or other Betacoronaviruses S1 glycoproteins by ELISA. Their report shows serum from 15 FCoV infected cats (10.9%) cross-reacting with HCCoV 229E S1, and the sera from two cats (1.5%) cross-reacting with HCCoV NL63 S1. Both HCCoV 229E and NL63 are human Alphacoronaviruses, and, as stated before, HCCoV 229E uses hAPN as primary cell receptor [27,28], whereas HCCoV NL63 uses hACE2 as primary cell receptor like SCoV1 and SCoV2 [24,54]. None of the 137 sera from FCoV-infected cats cross-reacted with S1 of human SCoV1, MERS-CoV, HCCoV OC43, and HCCoV HKU1 [18]. HCCoV OC43 and HKU1 uses 9-O-acetyylated sialic acid and MERS-CoV2 uses dipeptidylpeptidase 4 (DPP4) as primary cell receptor [55,56]. This study, published in 2019, did not evaluate using the FCoV-WV immunoblot analysis or the cross-reactivity to SCoV2 RBD [18]. SCoV1 RBD has been reported to have 73% aa sequence identity to SCoV2 [34]. Our SCoV1 and SCoV2 S sequence analysis also confirmed their finding of 73% identity and further determined the aa sequence similarity of 90% at RBD and 91. showed no cross-reactivity to SCoV1 S1 glycoprotein.
The use of SCoV2 RBD to block the in vitro infection of SCoV2 has already been performed and demonstrated by others, blocking the SCoV2 infection of Vero CCL-81 cells with SCoV2 RBD-His tag at IC50 of 21 μg/mL [57]. The novelty of our study is that SCoV2 UF-RBD at a high concentration was able to cross-block or cross-protect against FCoV2 infection of feline cell line at a RBD dose of 48-68 μg/mL without any cellular toxicity ( Fig 4A). This is slightly more than twice the dose used by Shin et al. to block SCoV2 infection of Vero cells with SCoV2 RBD-His [57]. The ability for SCoV2 RBD to cross-protect against in vitro FCoV2 infection, although weaker than that of FCoV2 RBD, further supports our finding that SCoV2 RBD and FCoV2 RBD may be structurally and antigenically similar. We were unable to perform similar blocking study against FCoV1 infection with SCoV2 and FCoV1 RBDs due to the lack of FCoV1-susceptible cell line available [6,29]. Instead the PBMC from transiently and chronically FCoV1-infected cats were used to evaluate whether SCoV2 and  Table). Pfizer SCoV2 RBD mRNA vaccine was not as effective as their S mRNA vaccine perhaps due to the shorter RBD sequence and the T4 foldon used [59,60]. Since FCoV infection remained in domestic cats for over three decades [4,5], the FCoV RBDs must have endured the evolutionary pressure and are unlikely to undergo further mutations at RBD to retain its infectivity. Based on these findings, we propose that pan-CoV vaccine against SCoV2 infection in cats, dogs, and hamsters can be developed by combining FCoV1 RBD (gp52) and FCoV2 RBD (gp59) together with a larger SCoV2 RBD (gp40) for stability, potentially without addition of CCoV1/CCoV2 RBDs. However, the addition of conserved pan-CoV CTL epitopes from highly conserved SCoV2/FCoV protein(s) (e.g., RNA-dependent RNA polymerase with minimal mutations among SCoV2 variants) to the triple-RBDs may be required to induce sterilizing immunity (i.e., immunity against infection) in the vaccinated animals. Such pan-CoV vaccine tested in laboratory cats may provide insights on how to develop sterilizing immunity against SCoV2 in humans, while also preventing future zoonotic variant infections and benefiting companion animals.

Animal populations
The studies used laboratory cats that were bred and cared for under the UF IACUC The eight group-housed laboratory cats were initially purchased from Liberty Research, Inc. Their animal code did not show the same lineages as the UGA queens and were considered different cat lineages, unrelated to both UGA queens and UF toms. The second blood collection was obtained 15 months later exception for cat G-5, which was euthanized shortly after the first blood collection for unrelated medical reasons. All cats purchased from Liberty Research, Inc. were vaccinated against rabies using the RABVAC3 vaccine, and feline panleukopenia, calici, and rhinotracheitis viruses, along with hemorrhagic feline calicivirus strain, using the Fel-O-Vax PCT + Calicivax vaccine.
Our animal workers, including the Animal Care Service workers, were confirmed negative by RT-PCR for SCoV2 and COVID-19 when working with our cats.

SARS-CoV-2 RBD peptides
Two different versions of SCoV2 were used: the University of Florida-RBD (UF-RBD) and MassBiologics-RBD (MB-RBD). The UF-RBD was produced using Harvard Wuhan RBD plasmid kindly provided by Dr. Aaron G. Schmidt (Ragon Institute) and expressed in EXPI293F cells. UF-RBD had HRV cleavage site, 8x His tag, and streptavidin-binding peptide tag to assist in the purification of the RBD [68]. The MB-RBD was kindly provided by MassBiologics (Boston, MA) under UF/MB MTA and constructed from pcDNA with c-Myc (EQKLISEEDL) and 6x His tags [69].

Production and partial purification of FCoV2 whole-virus
The CrFK cells were infected with

FCoV-WV and SCoV2 RBD ELISAs with overnight serum incubation
Corning ELISA plate wells were coated with 100µL of 100 μg/mL FCoV2 whole virus antigen or 100 μg/mL SCoV2 RBD or FCoV RBD antigen in sodium bicarbonate ELISA coating buffer at pH 9.5 (BioLegend, San Diego, CA) and incubated overnight at RT. The next day, the plates were washed three times with phosphate-buffered saline tween

Stringent FCoV-WV and SCoV2 RBD ELISAs
To ensure that the serum ELISA reactivity was specific to the FCoV2-WV, FCoV-RBD, or SCoV2-RBD antigen, sera were incubated individually at the same dilution for only 1 h, instead of overnight. PBS was used instead of bicarbonate buffer for the coating of the antigen on the ELISA plates. Additionally, a BSA antigen control was included since veterinary vaccines used at the time in UGA queens often contained contaminating BSA from the cell cultures used during the manufacture of viral vaccines. This control was also important because the FCoV2-WV preparation contained about 5% BSA.

Gel and Immunoblot analyses
The purified proteins were analyzed by SDS-PAGE and immunoblot. Briefly, the FCoV2-WV or RBD proteins (100 μg) were boiled individually at 95°C for 5 min in a sample buffer.
The protein(s) and prestained marker were loaded into individual wells, separated by 10% Tris-HCL gel with 30% or 40% acrylamide/bis, and stained with Coomassie Blue solution for direct MW analysis or transferred to nitrocellulose membrane. Subsequently, the membrane was treated with Penta-His MAb (Qiagen, Germantown, MD) followed by HRP-conjugated goat anti-mouse IgG (Invitrogen) and HRP substrate for degraded RBD protein/peptide distribution. In the serum reactivity studies, the FCoV2-WV or RBD proteins were evenly loaded into a 7-cm wide well of the stacking gel, with one 4-mm wide well at the end for the prestained marker, separated by 10% Tris-HCL gel with 30% or 40% acrylamide/bis, and transferred to nitrocellulose membrane. Each nitrocellulose blot was cut vertically into 26 3.2-mm width strips using a Novex Model NZ-1CIS strip cutter (Novel Experimental Technology, San Diego, CA). Each immunoblot strip was incubated individually with FCoV-infected cat serum or plasma at a dilution of 1:50 or 1:100 in a blocking buffer and incubated overnight at RT on a rocker. After three washes, the strips were then incubated with alkaline phosphatase (AP)-conjugated goat anti-cat IgG (1:1,000) (SouthernBiotech) for 2 h at RT. Subsequently after three washes, the reactive bands were visualized with freshly mixed AP substrate from AP-Conjugate Substrate Kit (Bio-Rad).

FCoV2 NAb assay against FCoV2
The Fc9 cells were used for both FCoV2 NAb studies and FCoV2/SCoV2 RBD blocking studies against live FCoV2 infection. The FCoV2 NAb assay was a modification of FCoV2 NAb assay as described previously (41). Briefly, the diluted FCoV2 preparation (EMEM culture media with 5% heat-inactivated FBS) at 2 TCID50 were plated in a 96-well roundbottom microculture plate and then incubated with an equal volume of serially three-fold diluted cat serum. The plates were incubated at 37°C in a 5% CO2 incubator for 45 min.
The first row of the 12 wells was not used to prevent the drying effect. Thus, the second rows were the most concentrated serum dilutions of 1:6 with the virus preparation. The remaining serum dilutions in the wells were as follows: third rows with 1:18, fourth rows with 1:54, fifth rows with 1:162, sixth rows with 1:486, seventh rows with 1:1458, and eighth rows with 1:4374. Subsequently, 0.1 mL of the mixture of each well was transferred to the flat-bottom wells of Fc9 cell monolayer with 95-97% confluency and incubated for 24 h at 37°C and 5% CO2. The spent culture fluids were discarded. Each well was aliquoted with 0.1 mL of 0.25% sterile methyl cellulose in EMEM with 5% heat-inactivated FBS, and incubated as before for 18 h or until the 35-50 virus plaques per well were observed. The FCoV2 plaques in the wells were inactivated and stained with 0.1 mL/well of 1% crystal violet in 100% methanol at RT for 10 min. Each well received an additional 0.15 mL of 1% crystal violet in 20% methanol, incubated at RT for 24 h, decanted, and removed the stain with water.

RBD blocking assay against FCoV2
The RBD blocking assay against FCoV2 is a modification of the above FCoV2 NAb assay and differs from the NAb assay by the following three features: 1) A set amount of RBD is used instead of cat serum. 2) FCoV2 dose was 4 TCID50 instead of 2 TCID50 to assure 95-100% cytopathic effect (CPE) on the Fc9 cells.
3) The flat-bottom plates with Fc9 cells at 98-100% confluency instead of 95-97% confluency were used when adding the virus mixture. The remaining procedure is identical to the NAb assay.

RBD stimulation of PBMC from FCoV1-infected cats
The PBMC of transiently (4GC) and chronically (G-3) FCoV1-infected cats and two SPF cats (2FB, 4GA) were stimulated with 5 μg/mL of either FCoV1, FCoV2, or SCoV2 RBD  (B,C) that were also harem mated with toms 5HQT1 and HOJT2 respectively but did not give birth to kittens in the first mating as well as in the pair mating for UGAQ2 (A). The serum from FCoV2seropositive cat UGA1.4 (B) and SCoV2 inoculated cat UGA4.1 (C) were used as positive control serum. The serum from SPF cats HOE (B) and BDN (C) were used as negative control serum. The FCoV2 NAb titers shown below the cat identification code (B) were those measured at first arrival for queens and post-first mating for toms. FCoV2 NAb assays were performed thrice for repeatability. The sera from the later timepoints described in Figures 1D and 1E were tested twice and had no detectable NAbs (data not shown). Additionally, the serum Ab cross-reactivity of four FCoV + queens (D) at post-UF arrival in months (mo) and three toms (E) at pre-mating and postfirst mating with the FCoV + queens was tested by stringent ELISA. Their sera were tested for reactivity to FCoV2-WV (grey bar), SCoV2 UF-RBD (light blue bar), SCoV2 MB-RBD (blue bar), and BSA (white bar). Significant difference between the pre-mating and post mating sera of the respective bars are shown as p<0.05 (*) or p<0.005 (**) (E). The serum from SPF cat HOE was used as a negative control for all three antigens (D,E). The same sera from the toms tested negative for anti-BSA Ab titers before this set of ELISA assays (S1B Fig).    Figures 1D and 1E, the time of serum collection from four FCoV + queens are shown as time post arrival to UF, and those from the toms are shown as time post-first exposure to a FCoV + queen. The MW size range for the RBD bands SCoV2, FCoV2, and FCoV1 are shown next to the control bands. The bands for FCoV2 structural proteins in FCoV2-WV immunoblot (C) are 43 kDa for NC (long arrow), 28-32 kDa for M (long arrow), and 80-90 kDa for spike-2 glycoprotein (S2) (arrow head). The 190 kDa band (short arrow) in the FCoV2-WV immunoblot strips is the S glycoprotein confirmed by a cross-reactivity analysis using rabbit polyclonal antibodies (pAb) to SCoV2 S2 (Sigma ABF1063) and to SCoV2 S (Sigma ABF1066) which crossreacted with FCoV2 S2 and S, respectively (E1). The rabbit pAb to SCoV2 S1 (Sigma ABF1065) did not cross-react with FCoV2 S1 on immunoblot. Consequently, it served as rabbit control pAb to detect non-specific trapping of the rabbit pAb to the FCoV2 S2 band which had the highest antigen load on the immunoblot. The same analysis was performed at three serum dilutions for UGAQ4 and three toms (HOJT2, HOGT3, 5HQT1), with UGAQ2 only at 1:1000 using the immunoblot of the same batch (E2). All immunoblot photographs were adjusted for consistency to 10% brightness and 5% contrast.