Comparative whole-genome analysis of diphtheria toxin-producing Corynebacterium ulcerans isolates from sheltered cats in Osaka, Japan

Abstract

Diphtheria toxin-producing Corynebacterium ulcerans is a zoonotic pathogen that causes human diphtheria-like symptoms. After performing whole-genome analysis of the five isolates from sheltered cats in Osaka, Japan, we compared them with genome sequences of 25 strains of C. ulcerans from a public database. The five isolates from cats harbored 14 genes encoding possible virulence factors in diphtheria-toxin-producing C. ulcerans. These isolates also had diphtheria toxin gene-encoding prophage in their chromosome, although differences were found in other prophages possession. Whole-genome single-nucleotide polymorphism analysis showed that cats’ isolates belonged to ST337 branch, as were strains from Japanese human patients, with 41 or more single-nucleotide polymorphisms variations. High-resolution single-nucleotide polymorphism analysis of C. ulcerans was sufficient to distinguish cats’ isolates clearly as not different by conventional genotyping methods.

Diphtheria toxin-producing Corynebacterium ulcerans is a zoonotic pathogen that can cause diphtheria-like infectious disease [20]. The bacterium is detected in widely diverse hosts such as cats, dogs, cows, monkeys, killer whales, lions, birds, squirrels, and shrew-moles around the world [3, 4, 6, 14]. It has been suggested that human C. ulcerans infection occur primarily from dogs and cats [3, 4, 11].

We earlier investigated the prevalence of diphtheria toxin-producing C. ulcerans among dogs and cats held in animal shelters, and in wild rats in Osaka, Japan, between 2011 and 2012. In the paper, the diphtheria-toxin producing C. ulcerans was detected in 5 of 137 cats’ throat swabs (3.6%), whereas was not detected in 125 dogs and 29 rats [19].

Recently, high-throughput sequencing analyses of C. ulcerans have progressed in many countries. Many sequences obtained by whole-genome sequencing (WGS) have been deposited into the public databases DDBJ/EMBL/GenBank. It has been reported that previous genomic analyses revealed genes encoding virulence factors other than diphtheria toxin gene (tox) and the distribution of various prophages, which play an important role in the acquisition of the pathogenicity [2, 13, 15, 18]. Sekizuka et al. recently reported whole-genome single-nucleotide polymorphism (SNP) analysis as being a higher-resolution and more accurate genotyping method than other conventional methods such as ribotyping [12].

To better understand the genetic characteristics of C. ulcerans isolates from sheltered cats obtained during our earlier investigation, WGS analyses of isolates were performed. Our analysis comprised of a reassessment of the genotypes through whole-genome SNP analysis, and an investigation into the presence of genes encoding virulence factors and prophages in the genome.

The five C. ulcerans isolates from cats used in this study (ID; 11021B-1, 11022B-1, 11030B-1, 11031B-1, and 12109B-1) produced diphtheria toxin, and showed identical biochemical properties, antibiotic sensitivity patterns, and genotypes by conventional methods of pulsed-field gel electrophoresis (PFGE) and ribotyping [19]. The information of the five host cats was described in our previous report [19]. Briefly, host cats were mature or old. The cats were mutually unrelated and all exhibited wounds or symptoms of weakness. Four of the five cats were strays.

Genomic DNA samples were extracted using a DNeasy Power Soil kit (Qiagen Inc., Hilden, Germany) according to the manufacturer’s instructions. The short-read library was prepared using a Nextera DNA Flex Library Prep Kit (Illumina Inc., San Diego, CA, USA) and was sequenced on iSeq 100 instrument (Illumina Inc.). Pair-end reads were assembled de novo using CLC Genomic Workbench software 10.0.2 (Qiagen Inc.). Automatic annotation was performed using DFAST pipeline (https://dfast.ddbj.nig.ac.jp/) [17] with subsequent manual modifications. Bacterial species were also confirmed by the average nucleotide identity (ANI) value output from the DFAST pipeline. The sequence type (ST) for multi-locus sequence typing (MLST) was identified by MLST 2.0 (https://cge.food.dtu.dk/services/MLST/) [9]. Prophage regions were searched using PHASTER (https://phaster.ca/) [1], followed by manual modifications according to the earlier reports [10, 13, 18]. Prophage regions were compared using EasyFig. 2.2.3 software [16]. As whole genome SNP analysis, SNP detection and phylogenetic inference were performed using CSI phylogeny 1.4 (https://cge.food.dtu.dk/services/CSIPhylogeny/) [5]. SNP analysis parameters were as follows: minimum depth at SNP positions, 10; relative depth at SNP positions, 10; minimum distance between SNPs (prune), 10; minimum SNP quality, 30; minimum read mapping quality, 25; minimum Z-score, 1.96.

The five isolates from cats were compared with genome sequences of 25 strains of C. ulcerans (Supplementary Table 1). Strain 0102 was used as the reference sequence. The phylogenetic tree was constructed using maximum-likelihood method with MEGA X software [8]. To calculate the maximum-likelihood tree, Tamura-Nei model was used.

Genomic information of the five isolates is presented in Table 1. To ascertain the whole-genome sequences of C. ulcerans isolates, the obtained short reads were assembled de novo. Then between five and seven scaffolds per isolate were constructed with more than 45 × genome coverages. The size of the genome assemblies varied between 2.47 and 2.52 Mbp. The guanine-cytosine contents were all 53%. The number of cording sequences was between 2,217 and 2,296. These genomic features were comparable to those of C. ulcerans strains described in earlier reports [10, 12, 13, 15].

Table 1. Genomic information of diphtheria toxin-producing Corynebacterium ulcerans isolates from cats in Japan.

Information	11021B-1	11022B-1	11030B-1	11031B-1	12109B-1
Genome size (bp)	2,520,008	2,520,022	2,471,752	2,471,864	2,472,441
Total read length (bp)	138,130,750	137,029,935	139,847,110	130,459,018	113,007,331
Genome coverage (×)	55.0	54.6	55.7	52.0	45.0
G-C content (%)	53.4	53.4	53.3	53.3	53.3
The number of scaffolds 1	6	7	5	6	6
N50 (bp)	544,356	433,376	816,060	543,501	552,369
Cording sequences	2,296	2,281	2,220	2,222	2,217
Average protein length (aa)	320.9	322.6	325.2	324.9	325.3
Cording ratio (%)	87.7	87.6	87.6	87.6	87.5
Ribosomal RNAs	3	3	3	3	3
Transfer RNAs	51	47	51	47	51
CRISPRs	3	3	3	3	3
ST	337	337	337	337	337
ANI (%) 2	98.6	98.6	98.7	98.6	98.7
Accession Nos.	BQFG01000001- BQFG01000006	BQFH01000001- BQFH01000007	BQFI01000001- BQFI01000005	BQFJ01000001- BQFJ01000006	BQFK01000001- BQFK01000006

G-C: guanine-cytosine, CRISPRs: clustered regularly interspaced short palindromic repeats, ST: sequence type, ANI: average nucleotide identity. ¹ Scaffolds are defined as various contigs joined together with undetermined sequences. ² ANI (%) was calculated reference as C. ulcerans strain NCTC7910.

Previous reports have described that 14 possible virulence factors other than tox gene are detected in C. ulcerans genomes [15, 18]. The scaffolds of the five isolates from cats contained the 12 genes encoding virulence factors; endoglycosidase (endoE), cell wall-associated hydrolase (cwlH), neuraminidase H (nanH), phospholipase D (pld), rpf-interacting protein (rfpI), five surface-anchored protein (spaB, spaC, spaD, spaE, and spaF), trypsin-like serine protease (tspA), and verom serine protease (vsp1). The five isolates lacked the remaining two genes, shiga toxin-like ribosome-binding protein (rbp) and an additional verom serine protease (vsp2), which were found in 809 tox-negative C. ulcerans strains [14] (Supplementary Table 2). The five cats’ isolates, harboring tox, were clarified to possess virulence factors the same as those in 0102 strains from a Japanese human patient with symptoms of dyspnea and fever in 2001 [7, 11, 13].

Typically, C. ulcerans carries one or more prophages related to acquisition of the tox gene [10, 13]. In addition, variation of prophages was observed among C. ulcerans isolates. It is expected to have correlation with virulence and zoonotic transmission [10]. Sekizuka et al. reported that the whole genome sequence of 0102 strain harbored three distinct prophages (ΦCUlC0102-I, II and III) [13]. One of these, ΦCUlC0102-I was tox-encoding, whereas ΦCUlC0102-II and III contained no gene encoding known virulence factors. The five isolates from cats had tox-positive prophage on their chromosomes, and the genetic compositions of tox-positive prophages were identical to those of ΦCULC0102-I. Also, ΦCULC0102-II existed only in 11021B-1 and 11022B-1. ΦCUlC0102-III was absent from all isolates. No other prophage was identified by PHASTER search (Table 2). The 0211 strain and FH2016 strain, which are from Japanese human patients in 2002 and 2016, included three prophages (ΦCUlC0102-I, II and III), similarly to 0102 strain. The five isolates from cats could be divided into 2 different types according to prophage possessions. Moreover, they were distinct from those of three strains from Japanese human patients.

Table 2. Prophage possession of Corynebacterium ulcerans isolates from cats in Japan.

Prophage 1	Virulence factor	Length (kbp)	11021B-1	11022B-1	11030B-1	11031B-1	12109B-1
CULC0102-I	diphtheria toxin (tox)	38.3	+	+	+	+	+
CULC0102-II		21.4	+	+	-	-	-
CULC0102-III		41.4	-	-	-	-	-
Other prophage 2			ND	ND	ND	ND	ND

¹ According to references [10, 13]. ² Prophage possession except for CULC0102-I, II, and III, searched using PHASTER program.

The phylogenetic tree was constructed based on whole-genome SNP analysis of 30 C. ulcerans strains: 5 in the present study and 25 from humans and animals from many countries available in the DDBJ/EMBL/GenBank database (Fig. 1A). The whole-genome SNP analysis has been the most effective genotyping method at present [12]. The five isolates from sheltered cats belonged to the ST337 branch, as were WGS of three Japanese strains from human patients (0102, 0211, and FH2016). To date, only Japanese strains have been reported as belonging to ST337 [12]. The ST337 strains are known to have low genetic variability by SNP analysis [12]. Zooming in the branch of ST337, 8 strains including 5 isolates from cats in Japan and 3 strains from Japanese human patients were shown branched off from each other (Fig. 1B). The comparison of SNPs variation within these 8 strains was shown in Fig. 1C. SNPs variation among the five isolates from cats in Japan, which were mutually unrelated but indistinguishable by PFGE, ribotyping, or MLST, ranged from 41 to 176. Also, SNPs variation between isolates from timewise and geographically unlinked cats and human patients in Japan, belonging to ST337, ranged from 122 to 162.

Fig. 1.

Phylogenetic tree based on whole-genome single-nucleotide polymorphism (SNP) analysis of Corynebacterium ulcerans. (A) Genome sequences of the five C. ulcerans isolates from cats (boldface) were analyzed with 25 strains from worldwide. Strain information is arranged in the order of country, source, ST and productivity of diphtheria toxin in parentheses. Accession numbers are presented in the Supplementary Table 1. SNPs positions were calculated to 79,788 positions: 22 minimum and 34,569 maximums. In the tree, the proportion of sites where at least 1 unambiguous base was present in at least 1 sequence for each descendent clade is shown next to each internal node. The bar at the bottom of the tree indicated the scale of the amount of nucleotide substitutions per site. (B) From the phylogenetic tree of whole-genome SNPs, eight Japanese strains belonging to ST337 are shown in a close-up view. Strain information is arranged in the order of source, province, and isolated year in parentheses. The bar at the bottom of the tree indicated the scale of the amount of nucleotide substitutions per site. (C) The numbers of SNP differences between among eight Japanese strains was calculated and shown in the table.

In summary, findings from WGS comparative analysis of diphtheria toxin-producing C. ulcerans isolates from sheltered cats suggest that the five isolates harbored 14 genes encoding known virulence factors. However, the five isolates mutually differed by their possession of prophages that do not encode a tox gene. Whole-genome SNP analysis showed that the five isolates from cats are classified to the ST337 branch as with strains from Japanese human patients, whereas were clearly differentiated by SNPs variations. These results suggested that WGS analysis is useful for C. ulcerans comparative genotyping because cats’ isolates which had identical genotype by conventional methods were distinguished by prophage possession and SNPs differences. Our WGS data represent the first examples obtained for Japanese cats, however the limitation of this study is the few isolates obtained in a limited area were analyzed. Further real-world investigations are necessary to confirm the genetic relationship between humans and animals in C. ulcerans transmission.

CONFLICT OF INTERESTS

The authors have no conflict of interest to disclose.