The Distribution of Cytotoxic Necrotizing Factors (CNF-1, CNF-2, CNF-3) and Cytolethal Distending Toxins (CDT-1, CDT-2, CDT-3, CDT-4) in Escherichia coli Strains Isolated from Extraintestinal Infections and The Determination of Their Philogenetic Relationship by PFGE

Background: Determinants of extraintestinal infection in Escherichia coli (E. coli) remain unclear. Virulence factors making Extraintestinal pathogenic E. coli (ExPEC) different from other E. coli strains are the host cell adhesion, invasion, and two important factors/toxins, Cytotoxic Necrotizing Factor (CNF) and Cytolethal Distending Toxin (CDT) that are responsible for cell death. In the present study, prevalence of CNF-CDT genotypes was investigated in 645 E. coli strains isolated from patients. This prevalence was analyzed regarding clinical origins, phylogroups and putative phylogenetic relationships. Results: At least one virulence gene identied for ExPEC was found in 156 (24%) of 645 E. coli strains. 78, 12, 20 of ExPEC strains contained cnf1, cnf2, cnf3, respectively. Genes of cdt1, cdt2, cdt3 and cdt4 were detected as 20, 4, 4, 4. Finally, two factors were detected cnf1-cnf3 (n=6), cnf1- cdt1 (n=4), cdt1-cdt4 (n=4). These 156 strains were found to be distributed in 106 large clusters by Pulsed-Field Gel Electrophoresis (PFGE). Virulent ExPEC was primarily related to the groups B2 (60%) and D (32%). Conclusions: The CNF gene family is believed to enhance colonization of E. coli especially in the urinary system and the genes of E. coli gene pool gain the ability to survive in new environments, such as the human urinary tract. The and these genes specic coli.


Background
Escherichia coli (E. coli) is one of the major elements of the human colon micro ora. E. coli can easily transfer virulence and resistance genes from other ora bacteria in the intestinal ora. Some geno-serotypes of E. coli cause pathologies in the intestinal tract, ranging from simple secretory diarrhea pathologies to in ammatory diarrhea in the form of dysentery with mortality. However, their colonization in sterile areas in the body, outside of the intestinal system, is always considered to be the cause of clinical pathology. Extraintestinal Pathogenic E. coli (ExPEC) causes urinary tract infection, newborn meningitis, sepsis, osteomyelitis, pneumonia, surgical site infections and infections in other extraintestinal areas. Although infections caused by ExPEC have low morbidity, it is becoming an increasingly important endemic problem due to its fatal course and prolonged hospital stay. Important virulence mechanisms that distinguish ExPEC from other E. coli strains include adhesion, invasion, defect of host defense and direct intervention in host cellular functions by secreted effectors (11,12). These effectors are composed of a functional toxin class disrupting cell cycle, the basic cell process of the host cell. Especially toxins-factors in E. coli species are de ned under 2 classes: Rho-GTPase-targeting toxin class is cnf1-cnf3 (Cytotoxic Necrotizing Factor) and cdt1-cdt4 (Cytolethal Distending Toxin), which is genotoxin. While CNFs promote DNA replication without cytokinesis, CDTs block mitosis (13,14). These toxins-factors are encoded through mobile elements (island of pathogenicity, plasmids, bacteriophages, etc.) (15).
There have been a limited number of clinical-epidemiological studies revealing that ExPEC strains have different colonization and virulence factors and that polymorphism in genes encoding virulence factors affect the prognosis of the disease.
The aim of our study was to investigate the prevalence of CNF-CDT and the relationship between the clinical picture and genotypes in E. coli strains with different clinical origins. Thus, it was to detect a speci c marker for the diagnosis of non-commensal E. coli.

Isolation and Identi cation
A total of 645 clinical samples (i.e., blood, urine, bronchoalveolar lavage, sputum, and surgical wound) were collected from different clinics of Balcalı Hospital at Cukurova University and of Adana Regional Hospital, Turkey, between September 2014 and April 2016. These 645 ExPEC strains were isolated from 387 outpatients and 129 inpatients. Most of the specimens were urine (80%) and most were taken from the polyclinics of urology and pediatric ( Fig. 1-Figure 2). Other specimens taken in other medical units were wound, blood, sputum, aspirate, sterile body uid, vagina and urosepsis ( Fig. 1-Figure 2). The median age of patients with ExPEC was 45 years (range, new born to 82 years). While urine samples were obtained from 80 males and 442 females, other specimens including wound, blood etc. were obtained from 107 males and 16 females.
The samples were cultured on blood agar and Endo agar plates (Merck, Germany) at 37 °C for 24 hours. The isolates were rst evaluated by Gram staining to examine the morphology of colonies and biochemical test characteristics. Strains were con rmed phenotypically by the IMVIC test and genotypically by the PCR method in the gene region (uidA) encoding Beta-D-glucuronidase enzyme found structurally in E. coli (Fig. 3). E. coli isolates were kept at -70 °C in broth containing 20% glycerol.

DNA Extraction and PCR
In all isolates, DNA extraction was performed mechanically with a Mickle cell disruptor (The Mickle Lab. Engineering Co. Ltd., Gamshall, Surrey, UK). The amplicons were obtained by PCR using previously published primers (Table 1) uidA, cnf1, cnf2, cnf3, cdt1, cdt2, cdt3, cdt4 and extracted DNA samples. PCR ampli cations were performed in a thermal cycler.
The PCR ampli cation was carried out in a total volume of 25 µl. The PCR mixture was constructed as follows: 12.5 µl PCR master mix (1x without MgCI2, 4 µl MgCI2 (25 mM), 1 µl dNTP (200 µM each nucleotides), 1 µl each primer, Taq DNA polymerase 0.2 µl U (5 U/µl), template DNA 5 µl (approximately 50 ng) and distilled water by completing to 25 microliters. The uidA gene and cdt1, cdt2 cdt3, cdt4 genes were ampli ed in a single PCR reaction in one tube. Multiplex PCR protocol was used to determine cnf1, cnf2, cnf3 genes. The PCR processes were applied for the ampli cations of toxin genes ( Table 2). The PCR products were separated on 2% agarose gel for cdts and uidA, and on 1.8% agarose gel for cnfs. Then, they were monitored using a Kodak UV transilluminator (Kodak, New York, USA).
Phylogenetic Classi cation E. coli strains belonged to four groups (A, b1, b2, D) based on the presence of the chuA and yjaA genes and the DNA fragment (TSPE4.C2) ( Table 1) by multiplex PCR as previously de ned ( Table 2). The results allowed the isolates to be classi ed into any of the four major phylogenies (224).
XbaI Pulsed eld gel electrophoresis PFGE was performed to determine DNA pro les of CNFs-CDTs producing strains. In our study, we applied the protocol described in a previous study (10) for Klebsiella, E. coli, P.aeruginosa, Acinetobacter (KEPA), which are the most frequently isolated gram-negative bacilli from the clinical material. PFGE method was performed for the determination of the clonal association of ExPEC strains and as gold standard. Epidemiological relationships between strains were evaluated by studying the PFGE patterns of genomic DNA after restriction by XbaI. The XbaI-PFGE was performed as described in the previous study (10) (Fig. 4).
According to the clinical origins of 645 ExPEC isolates, the distribution of virulence factors was found to be 3% in the ExPEC strains of the cdt gene family isolated from the urinary system and 15% in the non-urinary system isolates. While the rate of Cnf gene family was 20% in the urinary system, the ratio was 3% for the cdt gene family (Table 3).
97 (20) 17 (13)   cdt1 20 (3) 12 (2)  8 (7) - In 156 E. coli isolates, at least one of whose virulence genes de ned for ExPEC were positive and therefore can be identi ed as ExPEC (cnf1, cnf2, cnf3-cdt1, cdt2, cdt3, cdt4), the highest prevalence was cnf1 among the virulence genes with a rate of 50% + 7.6% in binary combinations. This was followed by cdt1 with a rate of 12.8% + 2.5% in binary combinations (Table 4). In the distribution of virulence genes / gene combinations detected in these strains via the clinical materials from which the strains were isolated, it was observed that 121 (78%) of 156 strains, where at least 1 virulence gene was positive, were associated with the urinary system, and the remaining 35 (22%) were isolated from extra-intestinal urinary infection foci. The most common of the virulence genotypes in 121 clinical materials originating from the urinary system was the CNF gene family with a rate of 78%. The CDT gene family was 17%. In CNF-CDT positive isolates, it was 2%. (Table 5).
While 68 of 78 cnf1 positive strains originated from urinary system, 10 of them were isolated from other extraintestinal infection foci. Similar rates were observed in cnf2 and cnf3. However, among the positive genes of cdt1, cdt2, cdt3, cdt4, only cdt1 proved superiority in the urinary system with a rate of 10%, while cdt2, cdt3 and cdt4 were found in equal percentages. The cdtB gene family proved its superiority, especially in aspirated, sputum, peritoneal uid, urosepsis and CSF material. The strains with at least 2 gene regions were found to be predominant in strains isolated from non-urinary tract infection. (Table 5). - Of the 156 ExPEC strains with at least one virulence gene, group b2 was the mostly associated with virulence with a rate of 60.8%. This was followed by group D at a rate of 30.7%. The rates of groups A and b1 were 5.12% and 3.2%, respectively (Table 6).
It was observed that cnf1 was found to be signi cantly more frequent in the phylogroup b2 urinary tract strains than the phylogroup b2 non-urinary tract strains. While the same situation was valid for group D as well, the opposite was the case in group b1, and it was more common in non-urinary system isolates (Table 6).  . In response to this, 81 strains were distributed into single-member clusters (Fig. 8).
The strains collected in a single-member cluster and in the same subset were determined to have different virulence factors. ExPEC isolates isolated from blood and urine clinical materials from the patients with urosepsis were found to be distributed in different clusters and had different virulence factors (for EXPEC strains, while cdt1 was positive in the blood sample (Z), it was negative in the urine sample in terms of any virulence factors (Z10)) using the PFGE method (Fig. 9).

Discussion
ExPEC-related infections, which have a wide range of disease spectrum from urinary tract infections to deadly bacteremia, have increased from 17.8-65.3% in Turkey in the last 15 years (11). In this increase, it is important that ExPEC, which is under antibiotic pressure, gains new virulent properties with mobile genetic elements from different bacteria in order to survive under stress conditions. In addition to this, the use of irrational antibiotics is responsible for that increase. With a limited number of clinical-epidemiological studies; direct intervention to host cell functions, damage to epithelial cells and potent virulence factors that can lead to tissue pathologies such as stimulating the in ammatory response have been demonstrated by CNF-CDT positivity.
CDTs are genotoxins causing DNA damage in target cells. In our study, we found the prevalence of CDT as 6.3% (40/645) ( Table 3). While this rate is similar to 6.8% (222-223-9) having been reported in other studies; it is higher than the rate of 0.9%-2.5% (229-230). The variable prevalence of CDT genes can be explained by the fact that the test strains in our study and other studies were isolated from different extraintestinal infection areas.
CDT genes have been found to have a higher prevalence, especially in sepsis-related ExPEC strains compared to ExPEC strains isolated from other infection sites such as urinary tract infections (Table 3) (222-9). In our study, cdt gene was detected in 19 (3%) of 522 strains isolated from patients with urinary system infections and in 18 (15%) of 123 strains isolated from non-urinary system infections. The prevalence of urinary tract infections due to E. coli can also be explained by its being prone to easy colonization of the intestinal strains by invading into the urinary system through the neighborhood relationship and its capacity to easily create disease in the host under different stress factors without the need for an extra virulence factor. However, as a result of the disruption of the mucosal epithelial integrity of the mucosal epithelium in the intestinal lumen by CDT, which causes disruption of cellular integrity, ExPEC strains containing this factor easily translocate to the lamina propria prior to the intestines and then spreads to distant tissues with general circulation, making it possible to cause distant tissue infectionsespecially sepsis. This may explain why the prevalence of CDT is high in ExPEC strains isolated from patients with sepsis and pneumonia. Sequence analysis of the E. coli cdt operon indicated that this operon read cdtA, cdtB and cdtC; that is, it is necessary to read these 3 genes for the production of (231) active CDT. CdtB encodes enzymatically active protein subunits; polypeptides encoded by cdtA and cdtC are heterodimeric subunits required for binding of CDT to the target cell.
Four types of cdtB, which are de ned as cdt1, cdt2, cdt3, cdt4, were identi ed in E. coli. (232). We detected all cdtB alleles in our study (Table 4). These ndings revealed that in our ExPEC test strains, the prevalence of cdt1 and cdt4 was higher than the prevalence of cdt2 and cdt3 (Table 4). Similar to the other studies, the cdt1 and cdt4 alleles were found, but differently, the cdt2 and cdt3 alleles were not found (222, 223, 234, 229, 238). These ndings imply that the cdt1 and cdt4 alleles of the CDT genes show very common and close homology and may have been derived from a common ancestor by phage transduction. Unlike these groups, it can be explained by many factors such as our nding that the cdt2 and cdt3 allo-types, namely the richness of our alleles, had local stress differences as well as the variety of our case groups in which we isolated the test strains.
On the other hand, CNF is the rst toxin produced by the pathogen E. coli strains and reported to activate Rho GTPases dominantly. This toxin got this name because of its necrotizing effect on rabbit skin (134). We detected that 120 (18.9%) of the 645 E. coli isolates included in the study carried at least one cnf allele (Table 3). In previous studies, this rate has been between 5-34% and it can be asserted that the rates are similar to each other considering the number of test strains (240-241-242), and deviations may stem from unreported patient characteristics. cnf1 is responsible for tissue damage and impairs epithelial barrier function in tissue culture systems (149,150) and affects the function of immune cells by blocking phagocytosis (151). cnf1 induces the expression of cyclooxygenase-2 (COX-2), activates nuclear factor-kappa B (NF-κB), increases cell mobility and inhibits apoptosis. Strong evidence has been obtained that cnf1 may play a role in cancer development (152-154). cnf1 delete mutants have been shown to have a lower potential to colonize the urinary system (148). In our study, 78 (65%) of the necro-toxigenic E. coli isolates, detected as 120, were cnf1 positive. On the other hand, cnf1 ratio was above 50% in 156 ExPEC isolates with positive virulence genes. In other studies, this rate has varied between 16.5% and 61.9% (143-243-244-229-222). In a study conducted in Turkey by Bozcal et al; cnf1 ratio was 12% in ExPEC isolates isolated from blood samples (32). In our study, considering only blood samples, this rate was 11% (3/28) (Table 3) and it can be stated that the results are similar. These results indicated that cnf1 may be a speci c marker for virulent strains, especially for the urinary tract and sepsis-related ExPEC.
In our study, contrary to cnf1 (56%), cnf2 with a rate of 7,6% and cnf3 with a rate of 16% were found less (Table 4). This can be explained by the fact that cnf2 and cnf3 are of animal origin and are rare in human beings. However, as opposed to other studies, prevalence of cnf2 and cnf3 was high in our study (222,233,241,242,243). This rate may represent the prevalence closer to reality, which was also due to the high number of strains.
Phylogenetic analysis revealed that natural E. coli isolates were divided into four main phylogenetic groups: A, B1, B2 and D. With using a simple and fast phylogenetic grouping technique triplex PCR (9), the group being mostly associated with virulence was B2 with a rate of 60% among 156 ExPEC strains having at least one virulence gene. This was followed by group D with a rate of 30%. The rates for the groups A and B1 were 5.1% and 3.2%, respectively (Table 6). Considering other studies, it can be claimed that ExPEC originated predominantly from phylogroup B2 and group D to a less extent (230-229-224-248-242-222). However, in the study carried out by Bozcal et al., these rates were D (38.14%), A (29.89%), B2 (20.61%) and B1 (11.34%) in ExPEC isolates isolated from blood samples. The reason for this difference may be the high number of samples in this study, the diversity of our clinical materials (such as urine, blood, wound, sputum, aspirate etc.) and different regions. In our study, in CNF-CDT positive strains; the PFGE method was used to reveal whether E. coli strains evolved from deletion, recombination mutations or evolved from an ancestor strain and acquired the ExPEC feature. In determining phylogenetic relationships of 156 strains known to be ExPEC, based on 80% similarity in the whole genome DNA fragment polymorphism analysis with XbaI-PFGE, it was observed that the test isolates were distributed in 106 large clusters, the largest of which was 5-membered (Fig. 8). ExPEC isolates with the same phylogenetic properties may have different virulence characteristics (Fig. 9). ExPEC isolates isolated from the same patient but from different clinical materials may originate from evolutionarily different ancestors. This may explain that most extraintestinal E. coli infections were recovered from the community. Oloomi et al. (249) identi ed CNF-CDT (+) E. coli strains as highly heterogeneous. In a study conducted by Mora et al. (250) with 59 ExPEC strains, 85% similarity was taken as basis; 36 of the 59 strains were collected in 12 clusters; however, they found that 23 strains were distributed into single-member clusters and concluded that the XbaI-PFGE pro les of ExPEC isolates, which have recently been shown to move away from a common ancestor, form a homogeneous clonal group based on their signi cant similarities. Because of the redundancy in the number of clusters and the low rate of similarity between strains, our study may represent the heterogeneity which is closer to reality, resulting from the excess of our strain count.

Conclusion
The widespread diffusion of the cnf gene in E. coli can help to distinguish ExPEC from commensal strains and these genes can be a speci c marker for non-commensal E. coli. Especially, cnf-1 increases the E. coli colonization in the urinary system and it can be a characteristic virulence factor for the uropathogenic E. coli isolates. The CDT gene family support the colonization of E. coli, in extraintestinal areas outside the urinary system. Virulence factors can be distributed depending on the phylogenetic groups (A, B1, B2, and D) and these phylogroups can be used in separation of pathogen-non pathogen.
Considering the XbaI-PFGE results, the phylogenetic relationships of ExPEC isolates were very weak, so it was inferred that E. coli strains can acquire ExPEC feature by acquiring different virulence genes as a result of deletion and recombination mutations under adverse conditions including antibiotic suppression. For this reason, multiple drug resistance may be common among ExPEC strains and this may result in treatment failure. Moreover, infections caused by these strains are more likely to be community-acquired rather than hospital-acquired. In addition, the fact that ExPEC isolates belonging to the same clusters display different virulence characteristics is another indication that these strains are of community origin. The characteristics and spread of virulent ExPEC strains should be monitored by molecular surveillance and limited to vaccine studies.       Clustering of virulence genotyping

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.