Novel Polymorphisms and Genetic Features of the Prion Protein Gene (PRNP) in Cats, Hosts of Feline Spongiform Encephalopathy

Prion diseases are fatal neurodegenerative disorders characterized by vacuolation and gliosis in the brain. Prion diseases have been reported in several mammals, and genetic polymorphisms of the prion protein gene (PRNP) play an essential role in the vulnerability of prion diseases. However, to date, investigations of PRNP polymorphisms are rare in cats, which are the major host of feline spongiform encephalopathy (FSE). Thus, we investigated the genetic polymorphisms of the cat PRNP gene and analyzed the structural characteristics of the PrP of cats compared to those of dog, prion disease-resistant animal. To investigate the genetic variations of the cat PRNP gene in 208 cats, we performed amplicon sequencing and examined the genotype, allele and haplotype frequencies of cat PRNP polymorphisms. We evaluated the influence of cat PRNP polymorphisms using PolyPhen-2, PANTHER, PROVEAN and AMYCO. In addition, we carried out structural analysis of cat PrP according to the allele of nonsynonymous single nucleotide polymorphism (SNP) (c.457G > A, Glu153Lys) using Swiss-PdbViewer. Finally, we compared the structural differences between cat and canine PrPs for SNPs associated with prion disease resistance in dogs. We identified a total of 15 polymorphisms, including 14 novel SNPs and one insertion/deletion polymorphism (InDel). Among them, Glu153Lys was predicted to affect the structural stability and amyloid propensity of cat PrP. In addition, asparagine at codon 166 of cat PrP was predicted to have longer hydrogen bond than aspartic acid at codon 163 of canine PrP. Furthermore, substitution to dog-specific amino acids in cat PrP showed an increase in structural stability. To the best of our knowledge, this is the first study regarding the structural characteristics of cat PRNP gene.

FSE occurred with similar temporal and geographical distributions as the BSE outbreak. The first FSE case was reported in domestic cats in 1990 [17]. From 1990 to 2007, FSE was reported in various feline animals, including domestic cats, cheetahs, lions, pumas,

Genetic Analysis
Genomic DNA extraction was performed from tissue samples and whole blood samples. In the tissue samples, genomic DNA extraction was performed using Labopass Tissue Genomic DNA Isolation Kit Mini (Cosmogenetech, Seoul, Korea). In the blood samples, genomic DNA extraction was performed using the Bead Genomic DNA Prep Kit (Biofact, Daejeon, Korea). Polymerase chain reaction (PCR) was performed with a specific primer set for the cat PRNP gene (forward: 5 -CCGAGTGGTTCCAACATGAA-3 , reverse: 5 -CTAAAGGGCTGTAGGTAGACAC-3 ). These primers were designed based on the DNA sequence of the cat PRNP gene (GenBank ID: EU341499.1) to amplify the open reading frame (ORF) of the cat PRNP gene. The PCR mixture was composed of 2.5 µL of 10× Taq polymerase reaction buffer (25 mM MgCl 2 mixed), 5 µL of 5× band helper, 1 µL of dNTP mixture (each 10 mM), 1 µL of each primer (10 µM), 0.2 µL of Taq DNA polymerase (5 U/µL, Biofact, Daejeon, Korea) and nuclease-free water to a total volume of 25 µL. The PCR experimental conditions were as follows: 1 cycle of predenaturation at 95 • C for 2 min, 34 cycles of denaturation at 95 • C for 20 sec, annealing at 60 • C for 40 sec, elongation at 72 • C for 1 min and 1 cycle of final elongation at 72 • C for 5 min. PCR products were purified using a FavorPrepGel/PCR Purification Mini Kit (FAVORGEN, Pingtung County, Taiwan). Purified PCR products were sequenced by using an ABI 3730 Capillary Sequencer (ABI, Foster city, CA, USA). The sequencing data were analyzed by Finch TV (Geospiza Inc., Seattle, WA, USA).

Statistical Analyses
Hardy-Weinberg equilibrium (HWE), LD and haplotype analyses of cat PRNP polymorphisms were performed using Haploview Version 4.2 (Broad Institute, Cambridge, MA, USA). The LD score was measured with r 2 value.

Sequence Alignment
The sequence alignment of cat and canine PrP was performed by Clustal Omega. Information on the canine PrP sequence (ACO71291.1) was obtained from GenBank at the National Center for Biotechnology Information (NCBI).

In Silico Analysis
We used the PROVEAN, PolyPhen-2 and PANTHER programs to evaluate the influence of nonsynonymous SNPs and insertion/deletion polymorphism (InDel) on cat PrP and AMYCO to evaluate alterations in the amyloid propensity of cat PrP according to alleles of nonsynonymous SNPs and InDel. PolyPhen-2 predicts the effect of amino acid substitution according to the position-specific, independent count (PSIC) score and indicates 3 types of functional changes: "benign", "probably damaging" and "possibly damaging". PANTHER uses a hidden Markov model (HMM) based on a statistical modeling method to evaluate Genes 2021, 12, 13 4 of 16 scores according to amino acid substitution. PANTHER predicts amino acid substitutions into 3 types: "probably damaging" (450 my < score), "possibly damaging" (200 my < score < 450 my) and "probably benign" (score < 200 my). PROVEAN predicts the influence on amino acid variations of protein with a score below −2.5 being "deleterious" and a score above −2.5 being "neutral". The predictions including "benign", "probably benign", "probably damaging" and "possibly damaging" by PolyPhen-2, PANTHER and PROVEAN indicate a degree of structural and functional changes of template protein according to the substitution of amino acid. AMYCO uses the combination of the PAPA score and pWALTZ score to evaluate protein amyloid propensity. An AMYCO score below 0.45 indicates low amyloid propensity, and an AMYCO socre above 0.74 indicates high amyloid propensity.

3D Structure Analysis in the Cat and Canine PrPs
Information on the cat PrP sequence was obtained from this study and the 3D structure of cat PrP was predicted by the SWISS-MODEL program. Information on the 3D structure of canine PrP (1XYK) was obtained from the Protein Data Bank (PDB). The structural change according to nonsynonymous SNPs found in this study was determined using the Pdb-Viewer program to analyze the 3D structure. We performed hydrogen bond analysis and the threshold of the hydrogen bond distance was 2.35 to 3.2 Å.

Investigation of the Influence on Nonsynonymous Polymorphisms of Cat PrP
We estimated the influence on polymorphisms of cat PrP using PolyPhen-2, PANTHER and PROVEAN (Table 4). The nonsynonymous SNP, c.128G>A (Gly43Glu) was estimated to be "probably damaging" with a score of 1.0 by PolyPhen-2. However, this SNP was estimated to be "neutral" with a score of -1.381 by PROVEAN ( Table 4). The InDel, c.214_240delCCCCACGCCGGCGGAGGCTGGGGTCAG was estimated to be "deleterious" with a score of −13.052 by PROVEAN. The SNP, c.457G>A (Glu153Lys) was estimated to be "probably damaging" and "possibly damaging" with scores of 0.998 and 361 by PolyPhen-2 and PANTHER, respectively. In contrast, this SNP was estimated to be "neutral" with a score of −1.691 by PROVEAN.
Others contain rare haplotypes with frequency <0.02.

Evaluation of Amyloid Propensity of Cat PrP according to of Polymorphism Alleles
We investigated the amyloid propensity of cat PrP according to alleles of nonsynonymous SNPs and InDel (Figure 3). The scores of the c.128G>A (Gly43Glu) SNP and the InDel were equal to that of Wt (0.36) by AMYCO. However, the score of the c.457G>A (Glu153Lys) SNP was lower (0.28) than that of Wt.
ous" with a score of −13.052 by PROVEAN. The SNP, c.457G>A (Glu153Lys) was estimated to be "probably damaging" and "possibly damaging" with scores of 0.998 and 361 by PolyPhen-2 and PANTHER, respectively. In contrast, this SNP was estimated to be "neutral" with a score of −1.691 by PROVEAN.

Evaluation of Amyloid Propensity of Cat PrP according to of Polymorphism Alleles
We investigated the amyloid propensity of cat PrP according to alleles of nonsynonymous SNPs and InDel (Figure 3). The scores of the c.128G>A (Gly43Glu) SNP and the InDel were equal to that of Wt (0.36) by AMYCO. However, the score of the c.457G>A (Glu153Lys) SNP was lower (0.28) than that of Wt.

Impact of Nonsynonymous SNPs on the 3D Structure of Cat PrP
To identify the influence of nonsynonymous SNP on cat PrPs, we performed 3D structure analysis using Swiss-PdbViewer (Figure 4). The amino acid sequence of the previously reported 3D structure of the cat PrP (1XYJ) was different from that of cat PrP found in the study. Thus, based on the amino acid sequence of cat PrP found in this study, we performed homology-based modeling using SWISS-MODEL and analyzed the hydrogen bonds of cat PrP according to alleles of the nonsynonymous SNP. There was a difference in the number of hydrogen bonds between cat PrP with the Glu153 allele and that with the Lys153 allele. In detail, Glu153 was predicted to have no hydrogen bond with other amino acids (Figure 4a), and Lys153 was predicted to have a hydrogen bond (3.04 Å) with Thr208 (Figure 4b).
structure analysis using Swiss-PdbViewer (Figure 4). The amino acid sequence of the pre-viously reported 3D structure of the cat PrP (1XYJ) was different from that of cat PrP found in the study. Thus, based on the amino acid sequence of cat PrP found in this study, we performed homology-based modeling using SWISS-MODEL and analyzed the hydrogen bonds of cat PrP according to alleles of the nonsynonymous SNP. There was a difference in the number of hydrogen bonds between cat PrP with the Glu153 allele and that with the Lys153 allele. In detail, Glu153 was predicted to have no hydrogen bond with other amino acids (Figure 4a), and Lys153 was predicted to have a hydrogen bond (3.04 Å ) with Thr208 ( Figure 4b).

Comparison of Tandem Repeat Regions of PrP in Several Mammals
The tandem repeat regions of feline species, including cats and tigers, are different from those of other mammals ( Figure 5). The tandem repeats of mammalians including humans, are octapeptide repeat; however, those of feline species, including cats, are nonapeptide repeats. The R1 repeat in the tandem repeat region is PQGGGGWGQ in humans, dogs, horses, cattle, sheep, tigers and cats. The tandem repeat region of the PrP sequence in the feline species has an additional alanine, similar to PHAGGGWGQ.

Comparison of Tandem Repeat Regions of PrP in Several Mammals
The tandem repeat regions of feline species, including cats and tigers, are different from those of other mammals ( Figure 5). The tandem repeats of mammalians including humans, are octapeptide repeat; however, those of feline species, including cats, are nonapeptide repeats. The R1 repeat in the tandem repeat region is PQGGGGWGQ in humans, dogs, horses, cattle, sheep, tigers and cats. The tandem repeat region of the PrP sequence in the feline species has an additional alanine, similar to PHAGGGWGQ.

A Structural Comparison of PrP between Asp163 in Dogs and Asn166 in Cats
In previous studies, Asp163 in canine PrP was associated with prion disease resistance [31]. Thus, we performed sequence alignment between cat and canine PrPs (Figure 6a). In cat PrP, Asn166 is homologous to Asp163 of canine PrP (Figure 6a). We performed 3D structural analysis of cat and canine PrPs using Swiss-Viewer. (Figure 6b,c). In canine PrP, Asp163 was predicted to have a hydrogen bond distance of 2.54 Å with the Met138 (Figure 6b). In cat PrP, Asn166 was predicted to have hydrogen bond distance of 3.16 Å with Met141 (Figure 6c). This result may indicate that the hydrogen bond distance between Asp163 and Met138 of canine PrP is shorter than that between Asn166 and

A Structural Comparison of PrP between Asp163 in Dogs and Asn166 in Cats
In previous studies, Asp163 in canine PrP was associated with prion disease resistance [31]. Thus, we performed sequence alignment between cat and canine PrPs (Figure 6a). In cat PrP, Asn166 is homologous to Asp163 of canine PrP (Figure 6a). We performed 3D structural analysis of cat and canine PrPs using Swiss-Viewer. (Figure 6b,c). In canine PrP, Asp163 was predicted to have a hydrogen bond distance of 2.54 Å with the Met138 (Figure 6b). In cat PrP, Asn166 was predicted to have hydrogen bond distance of 3.16 Å with Met141 (Figure 6c). This result may indicate that the hydrogen bond distance between Asp163 and Met138 of canine PrP is shorter than that between Asn166 and Met141 of cat PrP.

Investigation of the Influence according to Substitution of Cat PrP
We estimated the influence on dog-specific amino acid substitution, Asp166 a Asn166 of cat PrP using PolyPhen-2, PANTHER and PROVEAN. The Asp166 was est mated to be "benign" and "neutral" with scores of 0.000 and −1.173 by PolyPhen-2 an PROVEAN, respectively. However, the Asp166 was estimated to be "possibly damaging with a score of 220 by PANTHER (Table 5). In addition, the cat PrP with Asp166 showe a lower amyloid propensity than that with Asn166 ( Figure 7a). Finally, we performed 3 structural analysis to predict a structural change according to the substitution of amin

Investigation of the Influence according to Substitution of Cat PrP
We estimated the influence on dog-specific amino acid substitution, Asp166 at Asn166 of cat PrP using PolyPhen-2, PANTHER and PROVEAN. The Asp166 was estimated to be "benign" and "neutral" with scores of 0.000 and −1.173 by PolyPhen-2 and PROVEAN, respectively. However, the Asp166 was estimated to be "possibly damaging" with a score of 220 by PANTHER (Table 5). In addition, the cat PrP with Asp166 showed a lower amyloid propensity than that with Asn166 ( Figure 7a). Finally, we performed 3D structural analysis to predict a structural change according to the substitution of amino acids at codon 166 of cat PrP (Figure 7b,c). Asn166 was predicted to have a hydrogen bond with Met141 at a 3.16 Å distance (Figure 7b). Remarkably, Asp166 was predicted to have two hydrogen bonds with Met141 at a 3.16 Å distance and Gln167 at a 2.24 Å distance (Figure 7c).

Discussion
In the present study, we performed amplicon sequencing in 208 domestic cats and found 1 In/Del in the nonapeptide repeat region and 14 novel SNPs, including two nonsynonymous SNPs for the first time (Figures 1 and 2, Table 1). In a previous study, the polymorphisms of cat PRNP were mentioned. However, the authors did not provide the information on polymorphisms, including the location, genotype frequencies and base substitution [28]. Although 15 polymorphisms were found in this study, main breeds of cats analyzed in this study is Korean Domestic Shorthair (80.76%) and the numbers of remaining 18 breeds are not enough to represent genetic variations of each local cat breeds. Thus, further investigation in sufficient samples of cats from each local breed is highly desirable in the future. In addition, since polymorphisms shared with other species analyzed in previous studies provide clues to understand a physiology of the cat PRNP polymorphisms, further investigation of polymorphisms shared with other species is needed. Furthermore, we found several rare and common variants (Table 1). These results indicate that temporal difference in the occurrence of the variants [47]. We also identified the SNPs, c.-3G>A and c.171C>T showing strong LD with c.797G>A and c.201C>T, respectively. These results denote that there is distinct genetic structure among these SNPs via non-random heredity of the alleles.
Since polymorphisms of the PRNP gene can influence PrP structure and susceptibility to prion disease, we evaluated the impact of polymorphisms that cause amino acid

Discussion
In the present study, we performed amplicon sequencing in 208 domestic cats and found 1 In/Del in the nonapeptide repeat region and 14 novel SNPs, including two nonsynonymous SNPs for the first time (Figures 1 and 2, Table 1). In a previous study, the polymorphisms of cat PRNP were mentioned. However, the authors did not provide the information on polymorphisms, including the location, genotype frequencies and base substitution [28]. Although 15 polymorphisms were found in this study, main breeds of cats analyzed in this study is Korean Domestic Shorthair (80.76%) and the numbers of remaining 18 breeds are not enough to represent genetic variations of each local cat breeds. Thus, further investigation in sufficient samples of cats from each local breed is highly desirable in the future. In addition, since polymorphisms shared with other species analyzed in previous studies provide clues to understand a physiology of the cat PRNP polymorphisms, further investigation of polymorphisms shared with other species is needed. Furthermore, we found several rare and common variants (Table 1). These results indicate that temporal difference in the occurrence of the variants [47]. We also identified the SNPs, c.-3G>A and c.171C>T showing strong LD with c.797G>A and c.201C>T, respectively. These results denote that there is distinct genetic structure among these SNPs via non-random heredity of the alleles.
Since polymorphisms of the PRNP gene can influence PrP structure and susceptibility to prion disease, we evaluated the impact of polymorphisms that cause amino acid variation, including Gly43Glu, p.76_84del PHAGGGWGQ and Glu153Lys using in silico tools. Interestingly, these polymorphisms of cat PrP were predicted to be damaging by PROVEAN, PANTHER and PolyPhen-2. Since several neurodegenerative diseases have been related to the accumulation of amyloidized substrate proteins, we estimated the amyloid propensity of Gly43Glu, p.76_84del PHAGGGWGQ and Glu153Lys. Notably, cat PrP with Lys153 showed decreased amyloid propensity compared to that with Glu153. Glu153Lys is located on helix 1, which is known to be important in the conversion of PrP C to PrP Sc [48]. In addition, a previous study reported that mouse PrP with Glu145 showed a higher conversion rate of PrP C to PrP Sc than that with Lys145 in the ScN2A cell line. Codon 145 in mouse PrP is equivalent to codon 153 in cat PrP [49]. Thus, we estimated the impact of cat PrP on Glu153Lys using 3D structural analysis. Notably, we identified the distributional difference of hydrogen bonds according to Glu153 and Lys153 of cat PrP (Figure 4). In detail, cat PrP with Lys153 has additional hydrogen bonds compared to that with Glu153. Since hydrogen bonds can affect the stability of proteins [50], Glu153Lys may be related to resistance to FSE. However, since our analysis is based on single genetic variation, further analysis based on haplotypes of cat PRNP polymorphisms is highly desirable in the future.
Next, to identify structural differences that can affect the difference in resistance to prion disease, we compared the core region of the canine PrP related to resistance to prion diseases with the corresponding region of cat PrP. (Figure 6). The aspartic acid at codon 163 of canine PrP is known as a specific amino acid for the resistance of prion disease, and codon 166 of cat PrP is equivalent to codon 163 of canine PrP. In this region, both codon 163 of canine PrP and codon 166 of cat PrP have one hydrogen bond; however, these codons showed a difference in the length of hydrogen bonds. In detail, Asn166 of cat PrP has a longer hydrogen bond length with Met141 (3.16 Å) than Asp163 of canine PrP with Met 138 (2.54 Å). Since the length of the hydrogen bond is related to strength of the hydrogen bond [51], this result suggests that cats have weaker hydrogen bonds than dogs. Because the weak hydrogen bond is associated with the stability of protein [50] and the stability of PrP influences the conversion of PrP Sc [33,45], these results are remarkable.
Finally, to identify the impact on the substitution of dog-specific amino acids on cat PrP, we evaluated amyloid propensity and performed 3D structural analysis on cat PrP ( Figure 7, Table 5). Notably, the substitution of dog-specific amino acids resulted in a decrease in the amyloid propensity of cat PrP (Figure 7a). In addition, cat PrP with Asp166 has additional hydrogen bonds compared to that with Asn166. In a previous study, the canine PrP transgenic mouse with Asp163Asn substitution was shown to be infected with a prion strain, including sheep-BSE [29]. These results suggest that the Asp163 amino acid of canine PrP plays a pivotal role in the resistance of prion disease. Interestingly, we observed that dog specific amino acids in cat PrP affect the amyloid propensity and stability of prion protein. However, since we performed a quite simplified analysis with 3D snapshots of protein, there was limitation, which cannot reflect dynamics of PrP according to alleles of genetic variations. Further investigation based on molecular dynamics is highly desirable in the future [52,53]. Furthermore, these in silico analyses need to be verified using in vivo infection experiments. Thus, the infection study of prion agents, including scrapie strains in cat PrP transgenic mice according to nonsynonymous polymorphisms and dog-specific amino acids is highly desirable in the future.

Conclusions
In this study, we identified 15 genetic variations, including 14 novel SNPs, and found strong LD and 7 major haplotypes among 15 polymorphisms. Among these polymorphisms, the Gly43Glu, p.72_80delPHAGGGWGQ and Glu153Lys polymorphisms of cat PrP were predicted to be damaging. In addition, Lys153 of cat PrP showed a lower amyloid propensity than Glu153 and was predicted to have one hydrogen bond. However, The Glu153 was predicted to have no hydrogen bond. Furthermore, the hydrogen bond between Asn166 and Met141 of cat PrP was predicted to be longer than that between Asp163 and Met138 of canine PrP. Finally, Asp166 of cat PrP showed a lower amyloid propensity than the Asn166 of cat PrP, which is homologous to Asn163 of canine PrP. Interestingly, Asn166 and Asp166 of cat PrP were predicted to have one and two hydrogen bonds, respectively.