Genotyping Based on CRISPR Loci Diversity and Pathogenic Potential of Diarrheagenic Escherichia coli

Diarrheagenic Escherichia coli (DEC) can cause epidemic diarrhea worldwide. The pathogenic potential of different strains is diverse and the continuous emergence of pathogenic strains has brought serious harm to public health. Accurately distinguishing and identifying DEC with different virulence is necessary for epidemiological surveillance and investigation. Clustered regularly interspaced short palindromic repeats (CRISPR) typing is a new molecular method that can distinguish pathogenic bacteria excellently and has shown great promise in DEC typing. The purpose of this study was to investigate the discrimination of CRISPR typing method for DEC and explore the pathogenicity potential of DEC based on CRISPR types (CT). The whole genome sequences of 789 DEC strains downloaded from the database were applied CRISPR typing and serotyping. The D value (Simpson’s index) with 0.9709 determined that CRISPR typing had a higher discrimination. Moreover, the same H antigen strains with different O seemed to share more identical spacers. Further analyzing the strains CRISPR types and the number of virulence genes, it was found that there was a significant correlation between the CRISPR types and the number of virulence genes (p < 0.01). The strains with the largest number of virulence genes concentrated in CT25 and CT56 and the number of virulence genes in CT264 was the least, indicating that the pathway potential of different CRISPR types was variable. Combined with the Caco-2 cell assay of the laboratory strains, the invasion capacity of STEC strains of different CRISPR types was different and there was no significant difference in the invasion rate between different CRISPR type strains (p > 0.05). In the future, with the increase of the number of strains that can be studied experimentally, the relationship between CRISPR types and adhesion and invasion capacities will be further clarified.


INTRODUCTION
Diarrhea illness is one of the global healthcare problems. Diarrheagenic Escherichia coli (DEC) is an important agent of epidemic diarrhea worldwide especially in some underdeveloped areas with poor sanitation among bacterial pathogens (Gomes et al., 2016). DEC can be divided into five categories according to its virulence characteristics: enterotoxigenic E. coli (ETEC), enteroaggregative E. coli (EAEC), enteroinvasive E. coli (EIEC), enteropathogenic E. coli (EPEC), and Shiga toxin-producing E. coli (STEC) (Shahbazi et al., 2021). Strains can be identified according to the characteristic virulence genes of different DEC categories: stx1, stx2, and eae in STEC; aggR and astA in EAEC; invE and ipaH in EIEC; est and elt in ETEC. Furthermore, typical (eae + , bfp + ) EPEC and atypical (eae + , bfp − , stx1 − , stx2 − ) EPEC can be distinguished (Fujioka et al., 2013;Rodrigues et al., 2016;Spano et al., 2021). Some studies have shown the association between EPEC and diarrhea. Moreover, EPEC was the main cause of death in children with diarrhea in developing countries (Scaletsky et al., 2002;Afset et al., 2004;Hernandes et al., 2009). ETEC was the most common cause of travelers' diarrhea followed by EAEC in developed and developing countries (Daniels, 2005;Croxen and Finlay, 2010). Large outbreaks of diarrheal diseases were caused by EIEC and proved to be indeed associated with diarrhea (Manuel et al., 1992). STEC was first associated with hemolytic uremic syndrome (HUS) and hemorrhagic colitis (HC) in the 1980s, and subsequently with uncomplicated diarrhea (Karmali et al., 1983;Riley et al., 1983;Pai et al., 1988). Both stx1 and stx2 carried by STEC are the most common in HUS patients. The stx were proved to reach endothelial cells and induce cascade thrombosis and inflammatory changes, leading to HUS in kidney and other organs (Keir et al., 2012). Both eae and ehxA carried by STEC also can cause serious diseases in humans (Shen et al., 2015). STEC can lead to typical colonic attachment and elimination of lesions by adhesion to intestinal epithelial cells, which can cause diarrhea or bloody diarrhea (Bruyand et al., 2018). This feature is primarily related to the intima of a protein encoded by the eae carried by the chromosomal locus of enterocyte effacement (LEE). Ruminants such as cattle and sheep are the main hosts of STEC. STEC infections are mainly foodborne, and from bovine feces and water contaminated by animal feces (Boerlin et al., 1999;Muniesa et al., 2006). Different E. coli pathogenic categories have different characteristic virulence genes, and virulence genes can be transferred horizontally in pathogenic E. coli, allowing pathogenic strains to acquire new biological characteristics and even lead to the generation of new strains (Kyle et al., 2012). Horizontal gene transfer (HGT) between bacteria by Mobile Genetic Elements (MGE) carrying virulence genes has resulted in the continuous emergence of highly virulent strains, which has brought serious harm to public health (Xiong et al., 2012;Yan et al., 2015). Therefore, it is an important task for epidemiological surveillance and investigation to accurately distinguish and identify diarrheagenic E. coli with different virulence. Serotypes can be used as a characteristic indicator of bacteria and long term for DEC typing in view of most of the pathogenic strains contained in the known O:H serotypes. For example, O157:H7 caused severe disease harming to patients. In addition, some studies have shown that O26:H11, O45:H2, O103:H2, O111:H8(NM), O121:H19, and O145:NM are also the main serotypes of major foodborne diseases (Karmali, 2018;Spano et al., 2021). Consequently, O:H serotyping plays an important role in bacterial classification, epidemiological monitoring, epidemic detection, and other aspects, and it can also provide information directly related to antigen response (Joensen et al., 2015). However, serotyping still has limitations, so it is necessary to use new typing methods to compensate for the shortcomings of isolates serotyping.
Clustered regularly interspaced short palindromic repeats (CRISPR) is one of the most rapidly evolving components of the genome, mainly composed of almost identical repeats and highly specific spacers. The insertion and deletion of spacers in CRISPR arrays reflect the polymorphism of bacterial evolution (Deveau et al., 2010;Makarova et al., 2011). CRISPR and CRISPR-associated sequence (CAS) proteins together form the CRISPR/CAS system (Díez-Villaseñor et al., 2010;Toro et al., 2014), which can acquire immunity for prokaryotes through invading phages and plasmids, thereby resisting phage infection and limiting the HGT Sontheimer, 2008, 2010). Some researchers found that spacers are derived from foreign genetic elements such as exogenous plasmids or phages (Bolotin et al., 2005;Mojica et al., 2005;Pourcel et al., 2005), which is related to the mechanism of action of CRISPR/CAS system. The CRISPRs of different strains are divergence and the high polymorphism of spacers can be used as high discrimination biomarkers for genotyping. Four CRISPR loci have been identified in E. coli, namely CRISPR1, CRISPR2, CRISPR3, and CRISPR4 (Díez-Villaseñor et al., 2010;Yin et al., 2013;Toro et al., 2014). A recent study (Long et al., 2019) classified 413 E. coli strains into 23 types according to CRISPR3 and CRISPR4 loci. In addition, CRISPR typing method has been applied to Salmonella, Campylobacter jejuni, Cronobacter, and other foodborne pathogens (Li et al., 2018;Zeng et al., 2019;Yeh and Awad, 2020), proving that CRISPR typing method has become popular. However, pathogenicity potential of DEC based on CRISPR types seldom focus on and needs further studies.
The purpose of this study was to investigate the distinguishing ability of CRISPR typing for DEC and explore the pathogenicity potential of DEC strains. We performed CRISPR typing of DEC strains and analyzed virulence genes of different CRISPR types. In addition, we combined the results of Caco-2 cell assay of laboratory E. coli to further explore the adhesion and invasion capacities of STEC strains based on CRISPR types.

Sequence Collection
The whole genome sequences of DEC were downloaded from the National Center for Biotechnology Information (NCBI) database (updated before May 19, 2021). The strains were classified according to the characteristic genes of DEC.

Laboratory Strains
Twelve STEC strains were screened in the laboratory. A total of 8 strains were isolated from ground beef and 4 strains were isolated from cattle feces. The whole genomes of the strains have been sequenced.

Serotyping
SerotypeFinder 2.0 Web-based tool 1 was applied to analyze the serotype of the whole genome sequence of laboratory strains and DEC downloaded by NCBI with the following parameters: 85% threshold for %ID and 60% minimum length.

Clustered Regularly Interspaced Short Palindromic Repeats Typing
The identification of CRISPR loci and the spacers were extracted using CRISPRfinder (Grissa et al., 2007) and BLAST. The comparison of spacer sequences and obtaining unique spacers were identified by ClustalX (Larkin et al., 2007). Each unique spacer was recorded associated a single number beginning with 1. Then, every CRISPR array with multiple spacers was assigned a number as a spacer code. CRISPR typing was performed by combining CRISPR1, CRISPR2, CRISPR3, and CRISPR4 into one allele and displayed this as an arrangement of CRISPR spacers. The CRISPR type (CT) of each strain was defined using a specific number beginning with 1. The discrimination index (D) was calculated based on the Simpson's index of diversity with the equation as previously defined (Hunter and Gaston, 1988;Yeh and Awad, 2020).

Identification of Virulence Genes
VirulenceFinder 2.0 Web-based tool 2 was used to determine virulence genes of DEC genomes with the following parameters: 90% threshold for %ID and 60% minimum length.

Adhesion Assay
In vitro, the adhesion capacity of DEC to Caco-2 cells was studied, and DEC in this study belonged to STEC. The STEC isolates grow overnight at 37 • C in Luria-Bertani (LB) culture. Bacteria suspension (1 ml) was added to each well of the tissue culture plates containing monolayers of Caco-2 cells and the precise number of bacteria in the inoculum (10 7 -10 8 CFU/ml) added to monolayers was determined retrospectively by serial dilutions and plate counting. After 2 h of incubation at 37 • C, the Caco-2 cells were washed three times with PBS to remove non-attached bacterial cells (Olesen and Jespersen, 2010). One milliliter 1% Triton X-100 (Applichem, Darmstadt, Germany) was added to each well to loosen and lyse the Caco-2 monolayers. Appropriate serial dilutions of the Caco-2/DEC mixtures were plated on LB agar plates to determine the number of adhered Caco-2 cells. For each assay, the mean value of adherent bacteria was determined and the SEM from triplicate experiments was calculated. The adhesion index was calculated as the ratio of the number of adhered cells to the amount of inoculation × 100%. Results were expressed as% bacteria adhered relative to inoculum.

Invasion Assay
To determine bacterial invasion, the bacteria solution was added to each well of a tissue culture plate containing Caco-2 monolayer cells, incubated at 37 • C for 2 h and then washed with PBS 3 times. Then 1 ml 100 µg/L penicillin-streptomycin solution was added and incubated for 1 h to kill remaining viable extracellular bacteria. The cells were lysed with 1% Triton X-100 for 5 min after washing the penicillin-streptomycin solution and a serial dilution method was used to quantify viable intracellular bacteria. LB agar plates were used to determine the number of invasive bacteria. The mean value of invasive bacteria was determined and the SEM from triplicate experiments was calculated. Results were expressed as% of invasive bacteria relative to inoculum.

Statistics
Results were analyzed using GraphPad Prism 8 and SPSS 25. The ANOVA analysis followed by a Duncan test at 95% confidence limits were applied to determine the differences in the STEC adhesive and invasion capacities, and the difference in the number of virulence genes among strains of different CRISPR types. P-values of <0.05 were considered statistically significant. Pearson test was used to analyze the correlation between CRISPR typing and virulence gene number of strains.

Sequence Collection
Complete genome sequences of 789 DEC strains were downloaded from NCBI database, including 521 strains of STEC, 116 strains of EAEC, 105 strains of EPEC, 30 strains of EIEC, and 17 strains of ETEC. The NCBI accession numbers of all genomes analyzed in the current work were listed in Supplementary Table 1.

Identification of Clustered Regularly Interspaced Short Palindromic Repeats Loci
The CRISPR1, CRISPR2, CRISPR3, and CRISPR4 loci of DEC were identified and extracted. A total of 733 DEC strains had CRISPR loci. The upstream and downstream sequences of CRISPR loci were relatively conserved. The CRISPR1 locus was located between iap and cysH, the CRISPR2 locus was located between ygcE and ygcF, and CRISPR3 and CRISPR4 [named CRISPR4. 1 and CRISPR4.2, respectively, by Díez-Villaseñor et al. (2010)] were located between clpA and infA. The results are consistent with a previous study (Díez-Villaseñor et al., 2010). The number of CRISPR loci varies with DEC types, as shown in Figure 1.
The number of CRISPR1 and CRISPR2 loci was the largest with 713 respectively, the number of CRISPR3 loci was 94, and the number of CRISPR4 loci was 16. STEC, EAEC, EPEC, EIEC, and ETEC had more CRISPR1 and CRISPR2 loci than CRISPR3 and CRISPR4, and the number of CRISPR4 loci was the least.

Serotyping
Among 789 DEC strains, the serotypes of 747 strains were identified. The strains were divided into 212 serotypes, of which O157:H7 accounted for a large proportion with 182 strains. The isolates of O157:H7 were mainly isolated from the United States, Netherlands, Sweden, and other places, which were associated with some food-borne disease outbreaks caused by STEC. O157:H7, O111:H8, O26:H11, O145:H28, and O121:H19 were the dominant serotypes of STEC. These serotypes most frequently implicated in outbreaks and sporadic cases of HC and HUS. The dominant serotypes of EAEC were O6:H1, O153:H30, and the dominant serotypes of EPEC were O55:H51 and O157:H16. The serotypes of EIEC and ETEC were dispersed. In addition, 119 strains had 119 serotypes, individually. The D value (Simpson's index) for DEC strains serotyping was 0.9316.

Clustered Regularly Interspaced Short Palindromic Repeats Typing
The downloaded genome sequences of DEC strains were classified by CRISPR typing method, and a total of 1,878 unique CRISPR spacers were obtained through alignment. The distributions of spacer numbers from each location seemed to be random. According to the different arrangement of these spacers to determine the CRISPR classification, the bacteria were divided into 397 CRISPR types (CT). CRISPR typing results of DEC are shown in Supplementary Table 2. CT7 (n = 86) was the most prevalent followed by CT48 (n = 85) and two kinds of CT were STEC serotype O157:H7. The strain number including CT6, CT32, and CT287 were 17, 16, and 10, respectively. The strains of other CTs were less than 10. 325 types of CRISPR typing were the least (n = 1). The D value (Simpson's index) with 0.9709 determining the discriminatory power of CRISPR typing method indicated that CRISPR typing method had a higher discriminatory power than serotyping for strains, which was related to the polymorphism of CRISPR. In addition, the same serotype can be divided into different CRISPR types (Figure 2).  O157:H7 serotype was divided into 8 CRISPR types including CT7, CT21, CT44, CT48, CT49, CT60, CT365, CT366; O111:H8 contained CT6, CT32, CT58, CT62, and CT79; O104:H4 was separated by CT129, CT131, CT155, CT156, and CT157; O121:H19 included CT40, CT42, CT55, CT66, CT67, CT68, CT77, CT84, and CT360. Otherwise, O26:H11, O145:H28, O113:H21, and O103:H2 contained more than 10 CRISPR types. CRISPR typing method had a higher ability to distinguish strains than serotypes. The study suggests the possibility of serotype prompting through CRISPR arrays. It is worth mentioning that the same H antigen strains with different O seemed to share more identical spacers, while the same O group with different H could not share spacers or had fewer spacers in CRISPR. For example, O26:H11 and O116:H11 had the same CRISPR typing (CT145), which means that spacers were arranged exactly the same. O111:H11 and O69:H11 had 6 identical spacers in CRRSPR1  Table 2). Similar results in the study of STEC were found by another study (Toro et al., 2014), indicating that H antigen loci are more evolutionarily stable than O antigen because of the diversity of O antigen due to the O antigen transformation caused by gene level transfer (Rump et al., 2010;Wang et al., 2012). CRISPR1, CRISPR2, CRISPR3, and CRISPR4 loci and spacers of STEC screened by our laboratory were identified and extracted, as shown in Table 1. A total of 156 spacers in 12 STEC strains were checked, including 107 unique spacers. Also, 18 unique spacers from 12 STEC strains were not found in the 789 DEC genomes downloaded from the database. Therefore, the CRISPR typing of strains was acquired after spacers of STEC were sorted based on assigning new unique spacers. The spacer arrangement is shown in Figure 3. All 3 strains (MRL380004, MRL380001, and MRL380003) were the same CRISPR typing as the downloaded strains, and two belong to CT48 and one belongs to CT7. In addition, all the strains of CT7 and CT48 were O157:H7 serotype.
The results indicated that the spacers of STEC are relatively diverse. The spacers of CRISPR1 loci of MRL380009, MRL380010, MRL380011, and MRL380012 strains were completely consistent, and the spacer sequences of MRL380009 and MRL380012 strains were consistent at CRISPR2 loci. However, the distribution of spacers at CRISPR3 loci was inconsistent. Similarly, MRL380010 and MRL380011 strains had the same spacer sequences at CRISPR2 loci, while the spacer sequences at CRISPR3 loci were different. MRL380009, MRL380010, MRL380011, and MRL380012 belonged to O136:H12 serotype, but belonged to four different CTs.

Relationship Between Diarrheagenic Escherichia coli Virulence Gene Quantity and Clustered Regularly Interspaced Short Palindromic Repeats Typing
The type and number of virulence genes of foodborne pathogenic bacteria represent the pathogenic potential of the strain, and the strain with strong pathogenic factors is often more likely to bring pathogenic risk to human. For example, STEC has caused many foodborne disease outbreaks and public health challenges (Karmali, 2017) mostly from stx, the main virulence factor of STEC. In this study, 123 strains were positive for stx1 only and 237 were positive for stx2 only. There was difference in CRISPR types between strains carrying only stx1 or stx2. The virulence genes were analyzed and it was found that stx2 was more prevalent than stx1 in STEC. In STEC strains isolated from animals, food, and clinical samples by Kruger et al. (2015), stx2 was more popular, while other studies have shown that the prevalence of stx1 was higher than that of stx2 (Wani et al., 2007;Michel and Kase, 2009). These results may be caused by the differences between the isolation place and source.
Typical virulence genes of EAEC strains are astA, pic, and aggR, and these virulence factors are commonly used as important indicators for EAEC isolation and identification (Hebbelstrup Jensen et al., 2017). Researchers have investigated Frontiers in Microbiology | www.frontiersin.org the prevalence of diarrheagenic Escherichia coli pathotypes among diarrhea and healthy children up to 5 years of age in Brazil, and found that EAEC was the most common pathological type, indicating that the pathogenicity of EAEC should not be underestimated (Dias et al., 2016). In a Brazilian study that investigated childhood diarrhea, the toxin gene astA was associated with acute diarrhea (Pereira et al., 2007). In our study, astA was more popular than pic and aggR in EAEC (n = 116); the clinical strains containing astA that were downloaded in this study can cause diarrhea (not mentioned whether children are involved), indicating that EAEC with astA possibly possessed high potential pathogenicity.
A potential contributor to the less reports than other DEC strains, is that EIEC is often observed as a rare cause of diarrhea compared with other DECs (Vieira et al., 2007). However, EIEC is a leading cause of bacterial dysentery with fever, abdominal cramps, diarrhea, and other symptoms in places with poor sanitation (Farajzadeh-Sheikh et al., 2020). Therefore, its pathogenic risk should not be ignored, and it is necessary that the research of the distribution of virulence genes is more conducive to understand the pathogenic potential of bacteria.
Enterotoxigenic E. coli is a global diarrhea pathogen that causes disease in mammalian hosts infected by adhesins and secreted enterotoxins (Crofts et al., 2018). ETEC was first recognized as a cause of human illness in the 1960s (Bradley et al., 1971). ETEC is the most common cause of traveler's diarrhea and rarely found in meat products, and contaminated food and water have been implicated as vehicles for transmission of ETEC infection. Moreover, ETEC has been studied in India, the United States, and other regions (Daniels, 2005;Crofts et al., 2018;Mondal et al., 2022). While the amount of ETEC collected in this study was limited and originated from Bangladesh, the United States, China, Chile, and other countries, it was impossible to evaluate the geographical specificity of strains in each region.
Combined with the results of CRISPR typing of DEC strains, there was a significant correlation between CRISPR typing and the number of virulence genes (p < 0.01). The number of virulence genes between strains of different CTs was significantly different (p < 0.01), indicating that the pathogenicity potential of different CT strains was different. The distribution of virulence gene quantity of DEC strains with different CTs is shown in Figure 4. Virulence genes can not only be used for characteristic identification of strains, but also can make prospective judgments on the potential pathogenic properties of strains. To investigate the virulence potential of different CT strains, we analyzed the distribution of virulence gene quantity in different CTs of DEC strains downloaded (Figure 4). Figure 4 shows the CT of more than ten strains. The strains with the largest number of virulence genes were concentrated in CT25 (H11, H16) and CT56 (H11). CT264 (H14) was the least prevalent followed by CT213 FIGURE 4 | Number of virulence genes among different CRISPR types of DEC strains. The same letter in the figure indicates that there is no significant difference between each other, and the absence of the same letter indicates that there is a significant difference between each other (p < 0.01).
(H21). CT25 and CT56 only had CRISPR1 and CRISPR2 locus, while CT264 had CRISPR3 locus, and the number of spacers in CRISPR1 locus of CT264 is more than that of CT25 and CT56. CT48 (O157:H7) had significant difference with CT25 (H11, H16), CT56 (H11), CT131 (H4), CT213 (H21), and CT264 (H4) (p < 0.01), but had no significant difference from other CTs. It can also be found that the number of virulence genes varied with different CRISPR types, and the CRISPR types that were significantly different from other types were H4, H11, H14, and H21 serotypes, indicating that CRISPR typing can distinguish the number of virulence genes of these serotype strains excellently.
The distribution of virulence genes in different CRISPR types of laboratory STEC strains is shown in Figure 5. CT398 (O26:H11) was the most prevalent followed by CT7 (O157:H7), and CT402 (O105:H8) had the least number of virulence genes. As the number of strains increases in the future, it is expected to obtain a more accurate relationship between CT and the number of virulence genes.

Caco-2 Cell Assay
The virulence potential of the strain is not only reflected in the distribution of virulence genes. To investigate the pathogenic potential of the STEC more comprehensively, the adhesion and invasion capacities of STEC to Caco-2 cells were studied (Figures 6, 7).
The serotype of MRL380002 was O26:H11, which also carried eae, stx1, and stx2 virulence genes. The serotype of MRL380007 was O178:H19, which carried stx2 but did not carry eae and FIGURE 6 | Adherence to of Caco-2 cells by different strains of STEC. The same letter indicates that there was no significant difference between each other, and the absence of the same letter indicates that there was a significant difference between each other (p < 0.05). Results were expressed as percent of adherent bacteria with respect to inoculum. stx1. MRL380010 had the lowest adhesion rate, and its serotype was O136:H12, which did not carry eae and stx2 genes, but carried stx1 genes. We found that there were also differences in adhesion ability between strains of the same serotype. The serotypes of MRL380001, MRL380003, and MRL380004 were O157:H7, among which MRL380001 and MRL380003 showed significant differences in adhesion ability to cells. Kobayashi et al. (2016) measured the cell adhesion capacity of 11 STEC strains with serotype O103:H2, and the results showed that there were significant differences in adhesion capacity among strains of the same serotype, which was similar to our results.
MRL380002 had the greatest invasive capacity and it was significantly higher than that of MRL380003, MRL380005, MRL380009, MRL380010, and MRL380012 (p < 0.05). The invasion rates of MRL380004, MRL380007, MRL380008, and MRL380011 were not significantly different from other strains.
FIGURE 7 | Invasion of Caco-2 cells by different strains of STEC. The same letter indicates that there was no significant difference between each other, and the absence of the same letter indicates that there was a significant difference between each other (p < 0.05). Results were expressed as percent of invasive bacteria respect to inoculum. MRL380001, MRL380003, and MRL380004 belonging to O157:H7 showed no significant difference in invasive capacity. Serotypes of MRL380009, MRL380010, MRL380011, and MRL380012 were O136:H12. There was no significant difference in the invasion rate of Caco-2 cells. The adhesion and invasion abilities of MRL380002 were the highest among strains, but the order of adhesion and invasion abilities of other strains was not consistent, indicating that there was not necessarily a correlation between the adhesion ability and invasion ability. In addition, the outstanding adherence to and invasion of Caco-2 epithelial cells by O26:H11 confirmed that the virulence of non-O157 strain might not be underestimated. The serotypes O26:H11 and O103:H2 were used for comparison together with O157:H7 because these serotypes are considered the most important non-O157 STEC serotypes associated with increasing frequency in patients with bloody diarrhea and HUS (Jelacic et al., 2003;Alexander et al., 2005).
Combined with the CRISPR typing results of STEC, this study also analyzed the differences of adhesion and invasion capacities among strains with different CRISPR type. The distribution of cell adhesion capacity of different CT type STEC strains is shown in Figure 8.
The mean adhesion rate of CT398 was the highest followed by CT401, and adhesion rates of CT398 and CT401 strains to Caco-2 cells were significantly higher than that of other CRISPR types (p < 0.05). CT399 adhesion ability was significantly higher than CT400, CT402, CT403, CT404, CT405, and CT406 (p < 0.05), but with CT7 and CT48, there was no significant difference between strains. Whether the results of CRISPR typing can predict the adhesive capacity of the strains needs to be further studied by increasing the number of strains.
CT398 had the largest quantity of virulence genes, including eae and stx1, and CT402 and CT404 with the lowest adhesion rate did not have eae because eae was a factor related to adhesive capacity (Kobayashi et al., 2016), the lack of eae had a certain impact on the adhesive capacity of the strain. Including intimin FIGURE 8 | Distribution of adhesive capacity of different CRISPR typing STEC strains. The same letter indicates that there was no significant difference between each other, and the absence of the same letter indicates that there was a significant difference between each other (p < 0.05). The red lines represent the median, and the blue lines represent the quartile. (eae), translocated intimin receptor (tir), the type III secretion apparatus (espB and espD), and homolog adhesion (iha) are also associated with adherence to the intestinal epithelium (Tarr et al., 2000;Kobayashi et al., 2016). CT398 with the highest adhesion rate carried tir and espB. CT7 and CT48 carried tir and espB, and iha had a higher adhesion rate. CT402, CT403, CT404, CT405, CT406 tir, espB, and iha were negative, and their adhesive capacities were inferior. The largest number of spacer sequences were identified in CT402, but CT402 had the lowest number of virulence genes and a low adhesive capacity. How the spacer affects the adhesive capacity of strains is indistinct, and further investigation is needed by increasing the number of strains.
The invasion capacity of STEC strains of different CRISPR types is shown in Figure 9. CT398 showed the highest invasion value followed by CT400, and CT404 was generally lower than other CT types. CT398 carried eae and stx1; moreover, the number of virulence genes and adhesion rate are both the highest. CT404 had the least virulence genes quantity and adhesion rate carried stx1 without eae. There was no significant difference in the invasion rate between different CT type strains (p > 0.05), which was different from the adhesion capacity.

CONCLUSION
In conclusion, the existence rate of CRISPR1 and CRISPR2 loci in DEC strains was generally higher than that of CRISPR3 and CRISPR4 loci. The strains with the same H antigen shared more spacers, while the same O group did not share spacer sequences or shared spacers less. The CRISPR spacers polymorphism showed the potential for DEC typing, providing clues for inferring virulence of strains. The number of virulence genes was different among DEC strains with different CRISPR types, indicating that the pathogenicity potential of CT strains was different. For STEC, the adhesion capacity of different CT strains was significantly different, which provided a basis for CRISPR typing to distinguish the pathogenicity of strains. As a whole, with the increase in the number of DEC strains and the popularization of CRISPR typing method, it is expected to judge the pathogenic potential of the strain by CRISPR type. Moreover, paying attention to CT strains with strong virulence potential in advance is beneficial to reduce the occurrence of food safety problems.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.