Public health implications of Yersinia enterocolitica investigation: an ecological modeling and molecular epidemiology study

Epidemic profile of Yersinia during 2007–2019

A total of 9031 samples were monitored from 2007 to 2019, with the detection rate of Yersinia ranging from 0.9% to 7.6% (Table 1). The highest detection rate was in 2014 (7.6%), eightfold higher than in 2013 (0.5%). The difference in positivity rates between years was statistically significant (χ2 = 40.282, P < 0.001). No significant upward or downward trend in detection rates between years. Samples from different hosts were detected from 0.3% to 5.9%, with the highest detection rate of 5.9% for samples from poultry and livestock sources (Table 2). The detection rates of samples from different hosts were statistically different (χ2 = 22.636, P < 0.001).

Table 1 Detection rates of Yersinia, 2007–2019 in Ningxia Hui Autonomous Region, ChinaTable 2 Detection rates of Yersinia in samples from different hosts, 2007–2019 in Ningxia Hui Autonomous Region, China

The isolates were obtained from five prefectures in Ningxia Hui Autonomous Region [Yinchuan (n = 111), Shizuishan (n = 7), Wuzhong (n = 1), Guyuan (n = 3), and Zhongwei (n = 148)], from 2007 to 2019 (Fig. 1 and Additional file 8: Table S5). In total, 208 (77.0%) were of animal origin, 49 (18.2%) were of food origin, and 13 (4.8%) were of patient origin. Animal hosts included pig (150/208, 72.1%), sheep (32/208, 15.4%), rat (15/208, 7.2%), cattle (6/208, 2.9%), chicken (3/208, 1.4%), and hamster (2/208, 1.0%). The source of animal samples was mainly feces (n = 102), pharyngeal swabs (n = 39), anal swabs (n = 7), and intestinal contents (n = 59) of animals. Food was derived from meat products, comprising beef (n = 25), pork (n = 10), chicken (n = 9), lamb (n = 3), and fish (n = 2). Food (n = 49) comprised fresh meat (n = 18) and frozen meat (n = 31). Human samples were of fecal origin, with the majority (n = 9) coming from children (Fig. 1 and Additional file 8: Table S5). Using traditional phenotypic methods, 187/270 (69.3%) isolates were serotyped, with the most common serotypes being O:3 (n = 84), O:5 (n = 52), O:8 (n = 24), and O:9 (n = 20). In total, 83 isolates were reported as O: unidentifiable because the O-antigen reacted with more than one antiserum or with none of the antisera (Fig. 1).

Fig. 1figure 1

Number of Yersinia spp. isolates isolated from the Ningxia Hui Autonomous Region between 2007 and 2019. a Host distribution of isolates. b Serotype distribution of isolates. NA, not applicable; Nag, nonagglutinative

Distribution of Y. enterocolitica by biotype, host, and serotype

Of the 187 Y. enterocolitica isolates, 81.3% (n = 152) were of animal origin, the food source was 12.3% (n = 23) and the patient source was 6.4% (n = 12). Isolates of animal origin included 42.8% biotype 1A (n = 65), 50.0% biotype 4 (n = 76), 2.6% biotype 3 (n = 4) and 4.6% biotype 5 (n = 7). Of these, all isolates of biotype 5 were from sheep. In total, biotype 4 of isolates (n = 84) were from pig hosts. The dominant serotype of the isolates of animal origin was O:3. Food-derived isolates were 87.0% biotype 1A (n = 20) and 13.04% biotype 2 (n = 3). Patient-origin isolates included 33.3% biotype 1A (n = 4) and 66.7% biotype 4 (n = 8). There was no biotype 1B (Table 3 and Additional file 8: Table S6).

Table 3 Serotype and biotype distribution of Y.enterocolitica isolatesPhylogenetic analysis

Total of 270 Yersinia genomes were evaluated according to the standard 95–96% ANI [27]. Twelve species were delineated using a 95% ANI cut-off value: Y. enterocolitica (187/270, 69.3%), Y. intermedia (30/270, 11.1%), Y. massiliensis (30/270, 11.1%), Y. mollaretii (7/270, 2.6%), Y. pekkanenii (5/270, 1.9%), Y. proxima (4/270, 1.5%), Y. alsatica (2/270, 0.7%), Y. frederiksenii (1/270, 0.4%), Y. kristensenii (1/270, 0.4%), Y. hibernica (1/270, 0.4%), Y. canariae (1/270, 0.4%), and Y. rochesterensis (1/270, 0.4%) (Fig. 2). The clustering analyses was consistent with the results of the ANI analysis and identical separation into 12 distinct species as determined by BAPS (Bayesian analysis of population structure) (Fig. 2).

Fig. 2figure 2

SNP-based maximum likelihood (ML) trees of Yersinia spp. isolates. a The SNP-based ML tree was built from a recombination-filtered alignment of the whole genome SNP (wgSNP) present in 270 isolates. The ML tree was built using the GTR + F + ASC + G4 model, with 1000 bootstraps based on 1,563,073 SNPs, with Y. enterocolitica 8081 (GCA_000009345.1) was used as the reference sequence. b The cgSNP-based ML tree was built from 187 Y. enterocolitica isolates of this study and 88 public data. Y. enterocolitica 8081 (GCA_000009345.1) was used as the reference sequence. Y. pseudotuberculosis (GCA_900637475.1) was used as outgroups

The relatedness clustering presented by the Y. enterocolitica ML tree showed a direct relationship with the biotype. Biotypes 1A, 1B, and 5 isolates formed discrete clusters, whereas biotypes 2, 3, and 4 isolates consisted of closely related but distinct lineages, confirmed by BAPS clustering. Pathogenic isolates from Ningxia were genetically more distant from the reference genome than isolates from other countries and regions (no biotype 1B isolates in Ningxia). Isolate LC20 from Zhejiang had the longest genetic distance with the reference genome compared to the other isolates in China. Biotype 1A isolates exhibited no geographical differences and a broader range of host species and serotypes. The hosts of biotypes 4 and 5 isolates in Ningxia were pigs and sheep, respectively, whereas the hosts in the other regions were humans and hares. Biotype 3 isolates were O:3 in Ningxia, compared to O:5,27 and O:6,30 in the other countries (Fig. 2).

Associations between ecological factors and pathogenicity

Temperature, precipitation, altitude, and NDVI were highly statistically significant to pathogenicity (P < 0.001) (Fig. 3a–d). The ambient temperature of the collection locations for pathogenic isolates was lower than that of the non-pathogenic isolates (12.70 ± 6.39 ℃ vs. 16.85 ± 6.05 ℃); the precipitation of the collection locations for the pathogenic isolates was lower than that of the non-pathogenic isolates (1.34 ± 1.29 mm vs. 2.54 ± 1.42 mm). The median elevation of the collection locations for pathogenic and non-pathogenic isolates was 1111.00 m and 1816.00 m, respectively; the median NDVI was 0.35 and 0.19, respectively. Also, except for the host of the isolates in precipitation factors, the ecological factors and pathogenicity were statistically highly significant with biotype, serotype of the isolates, and host of the isolates (Fig. 3, Additional file 1: Fig. S1, Additional file 2: Fig. S2, Additional file 3: Fig. S3).

Fig. 3figure 3

Differences in pathogenicity of Yersinia enterocolitica isolates and ecological factors and the development of predictive models. a Temperature. b Precipitation. c Altitude. d NDVI. e Training dataset. f Testing dataset

Then, we used three models to predict the pathogenic of Y. enterocolitica with ecological environment factors. The AUC of the training and test sets of all three models exceeded 0.9, particularly the AUC of RF, and XGBOOST, indicating good evaluation performance (Fig. 3e, f). The contributions of ecological factors to the risk of Y. enterocolitica pathogenicity were shown in Additional file 4: Fig. S4, which indicated the prediction of Y. enterocolitica pathogenicity was the combination of all ecological factors. Finally, we used those performed models to evaluate the risk of Y. enterocolitica pathogenicity among our study regions (Additional file 5: Fig. S5). The results indicated that the highest risk area was concentrated in Yinchuan City and Shizuishan City in northern Ningxia, Zhongning City, and Zhongwei County in central Ningxia, and the southeastern part of Guyuan City and Haiyuan County in southern Ningxia. The predicted risk also varied between sources, with animals and humans suggesting a wider range of risk areas.

Virulence profiles

Y. enterocolitica isolates were annotated with 130 virulence genes in 5 categories: flagella, invasion, O-antigen, Yersiniabactin, and T3SS (Fig. 4). There were 53 flagella genes, most of which were che, flh, flg and fli genes. The O-antigen was represented by 27 genes, mostly mrk and wbc genes. The Yersiniabactin was made up of 9 genes, the majority of which were ybt gene. T3SS included 41 genes. The major ones were ysc, yop, icr, and syc genes. The distribution of virulence factors was closely related to the biotype. Most biotype 1A isolates lacked the virulence factors of invasion, O-antigen, Yersiniabactin, and T3SS. In contrast to pathogenic isolates of other biotypes, biotype 1B isolates were presented in Yersiniabactin.

Fig. 4figure 4

Distribution of virulence genes in Yersinia enterocolitica isolates. The heat map on the right depicted the host, serotype, STs, location, and virulence genes of the isolates

ST and cgMLST typing

In 187 Y. enterocolitica isolates, 54 STs were detected using multilocus sequence typing. ST429 was the most common, accounting for 42.3% of all STs (79/187) (Fig. 5 and Additional file 8: Table S6). ST429 isolates included biotype 4 (n = 79) and biotype 3 (n = 2), and they were closely related to serotype O:3. ST3 (n = 9), ST278 (n = 6), ST178 (n = 5), ST637 (n = 5), ST640 (n = 4), ST643 (n = 4), and ST216 (n = 3) were biotype 1A isolates with serotypes O:5, O:8, O:9. ST13 isolates (n = 7) were biotype 5, with serotypes O:5, O:8, and O:9. ST3, ST12, ST14, ST18, and ST13 were the principal STs of foreign isolates, which differed from those in the Ningxia region (Additional file 8: Tables S3, S6, and Fig. 5). The biotypes and serotypes corresponding to ST12 and ST4 were bioserotypes 1A/O:6,30 and 2,3,4/O:9 (Additional file 6: Fig. S6).

Fig. 5figure 5

The minimum spanning trees (MST) of Yersinia enterocolitica isolates. The circle size was proportional to the number of isolates. Links between circles were represented according to the number of allelic differences between STs. a Isolates from this study. b Isolates from this study and the public database

An NJ tree and an MST based on cgMLST analysis of 187 Y. enterocolitica isolates were constructed. These 1,553 cgMLST target genes were randomly distributed across the genome. Y. enterocolitica isolates were divided into 125 cgMLST types (CTs) (Additional file 8: Table S6 and Fig. 6). CgMLST analysis revealed the core genome diversity of isolates with the same ST from 0 to 84 allelic differences. The NJ tree of 187 Y. enterocolitica isolates from Ningxia indicated the names of two microclades of the HC1490 (Hierarchical clustering) cluster (Fig. 6). HC1490_10 and HC1490_2 were the primary phylogenetic branches. These two microclades of the HC1490 cluster were consistent with the results of the Y. enterocolitica ML tree. HC1490_2 was strongly associated with biotypes 3, 4, and 5 isolates. HC1490_10 was closely related to biotype 1A isolates. The 125 CTs present in the 187 isolates from the Ningxia region clustered to form 54 microclades of HC100. Of these, HC100_2571 was the principal microclades, and the isolates were all from this study. The NJ tree constructed from isolates from the Ningxia region and those publicly available in the database showed that several microclades of the HC100 cluster were significantly associated with serotypes, hosts, and countries. HC100_2571 isolates were all obtained from this study and were of bioserotype 4/O:3 and 3/O:3, hosts were pigs and humans. HC100_406 isolates were 4/O:3, mostly from patients in New Zealand. The serotype of HC100_397 isolates was O:5,27, with hosts of pigs, humans, and food. HC100_2 and HC100_111 isolates were 3/O:9 and 4/O:9, respectively, mainly from pigs and humans in the UK and New Zealand. HC100_4570 isolates were biotype 5, obtained from sheep in Ningxia. HC100_466 isolates were biotype 1B, which were isolated from patients. HC100_150 isolates from pigs and patients in Ningxia were 1A/O:5. The isolates HC100_1273 were 1A/O:5 and obtained from pigs, cattle, and poultry in the Ningxia region. The mostly hosts of the biotype 1A isolates were pigs, cattle, sheep, rats, poultry, and food. The most common serotypes were O:5, O:8, O:9, and O:6,30 (Fig. 6 and Additional file 7: Fig. S7).

Fig. 6figure 6

The neighbor-joining (NJ) tree of Yersinia enterocolitica isolates based on cgMLST. The circle size was proportional to the number of isolates. Clusters generated using the hierarchical clustering method from EnteroBase and using a 100 cgMLST allele distance (HC100) were represented by circles. a Isolates from this study. b Isolates from this study and the public database

留言 (0)

沒有登入
gif