Pathogenicity Islands Distribution in Non-O157 Shiga Toxin-Producing Escherichia coli (STEC)

Shiga toxin-producing Escherichia coli (STEC) are foodborne pathogens associated with outbreaks and hemolytic-uremic syndrome. Cattle and meat foods are the main reservoir and infection source, respectively. Pathogenicity islands (PAIs) play an important role in STEC pathogenicity, and non-locus of the enterocyte effacement(LEE) effector (nle) genes present on them encode translocated substrates of the type III secretion system. A molecular risk assessment based on the evaluation of the nle content has been used to predict which STEC strains pose a risk to humans. The goal was to investigate the distribution of the PAIs OI (O-island)-36 (nleB2, nleC, nleH1-1, nleD), OI-57 (nleG2-3, nleG5-2, nleG6-2), OI-71 (nleA, nleF, nleG, nleG2-1, nleG9, nleH1-2) and OI-122 (ent/espL2, nleB, nleE, Z4321, Z4326, Z4332, Z4333) among 204 clinical, food and animal isolates belonging to 52 non-O157:H7 serotypes. Differences in the frequencies of genetic markers and a wide spectrum of PAI virulence profiles were found. In most LEE-negative strains, only module 1 (Z4321) of OI-122 was present. However, some unusual eae-negative strains were detected, which carried other PAI genes. The cluster analysis, excluding isolates that presented no genes, defined two major groups: eae-negative (determined as seropathotypes (SPTs) D, E or without determination, isolated from cattle or food) and eae-positive (mostly identified as SPTs B, C, or not determined).


Introduction
Shiga toxin-producing Escherichia coli (STEC) are heterogeneous foodborne pathogens associated with outbreaks and hemolytic-uremic syndrome (HUS) [1]. Cattle are the main reservoir of STEC and human infections are acquired mainly by ingesting food or water contaminated directly or indirectly with cattle feces. E. coli O157:H7 is the serotype most associated with diseases; however, more current studies have shown that the number of non-O157 STEC infections sometimes surpasses the number of STEC O157 infections [2].
Determinants of bacterial virulence are predominantly encoded by or associated with mobile genetic elements (MGEs) such as phages, plasmids, insertion elements, or transposons. A large number of these determinants are located within pathogenicity islands (PAIs), and can be exchanged among different bacterial species, and assembled and stabilized by selective pressure, leading to emerging pathogenic variants [3]. Increasing evidence shows that differences in virulence between pathogenic and non-pathogenic bacterial strains can be attributed in part to virulence genes located in PAIs [4,5].
Some pathogenic Escherichia coli use a type III secretion system (T3SS), encoded in the locus of the enterocyte effacement (LEE) pathogenicity island, to translocate a wide repertoire of effector proteins The genetic markers were detected at different frequencies in the STEC studied group. The prevalence of individual virulence genes is shown in Figure 1. Among all strains, the gene Z4321 (pagC) encoded in OI-122 was the most prevalent, resulting in both LEE-positive and LEE-negative isolates. In contrast, nleC encoded on OI-36 was the least prevalent, found in only four isolates of O26:H11 (from human), O145:NM (2 from humans), and O165:NM (from cattle).
Genes 2018, 9, x FOR PEER REVIEW 4 of 12 The genetic markers were detected at different frequencies in the STEC studied group. The prevalence of individual virulence genes is shown in Figure 1. Among all strains, the gene Z4321 (pagC) encoded in OI-122 was the most prevalent, resulting in both LEE-positive and LEE-negative isolates. In contrast, nleC encoded on OI-36 was the least prevalent, found in only four isolates of O26:H11 (from human), O145:NM (2 from humans), and O165:NM (from cattle). A group of forty-six isolates presented no gene. Forty-four strains (from cattle, meat food, and vegetables) out of these 46 strains were eae-negative, as expected, but, strikingly, two were eaepositive. Both were O145:NM strains isolated from humans. The serotypes with absence of the genes studied were: O8:H19, O20:H7, O20:H19, O25:H19, O79:H19, O117:H2, O120:H19, O141:H7, O175:H8, and ONT:H19. A second group comprised of 106 isolates carried only Z4321 (Table S1). Serotypes of these Z4321-positive strains are detailed in the legend of Figure 2.
The O26:H11 strains are among the strains that have the largest number of virulence genes. They showed 14 to 18 of the 20 genes. Those with more detected genes (16 and 18) were from humans. The strain with 18 genes only lacked nleD and Z4321 (OI-57 and 71 were complete). Notably, two cattle isolates belonging to O118:H16 and O165:NM carried 16 of the 20 analyzed genes, with the presence of genes of each OI.
The cluster analysis (which did not include isolates that presented no genes), presented in Figure 2, divided the studied set of strains into two major groups: (A) eae-negative, and (B) eaepositive isolates. Within the eae-negative group, the isolates were determined as belonging to SPTs D, E, or without determination and were isolated from cattle or meat food. Within the eae-positive group, the isolates were mostly identified as SPTs B, C, or not determined. Exceptionally, two isolates were characterized as SPT D (O146:H21 and O146:NM, both from cattle). Among this group, four principal branches could be differentiated (represented in four different green shades in

Discussion
The presence of nle genes and the number of genes carried by an E. coli strain are important criteria for estimating its virulence potential [7]. Previous evidence shows that virulence genes located on PAIs can be used to identify new and emerging pathogenic bacteria [4,5]. In this study, 18 isolates carried more than 60% of the 20 OI-genes analyzed. These isolates belong to the O5:NM, According to the criteria described by Karmali et al. [4] based on the detection of COI-122 that requires testing for four virulence putative genes (Z4321; Z4326; Z4332; Z4333), 13 of 204 (6.4%) isolates studied had a complete OI-122, 143 (70.1%) had an incomplete OI-122, and OI-122 was absent in 48 (23.5%) strains.

Discussion
The presence of nle genes and the number of genes carried by an E. coli strain are important criteria for estimating its virulence potential [7]. Previous evidence shows that virulence genes located on PAIs can be used to identify new and emerging pathogenic bacteria [4,5]. In this study, 18 isolates carried more than 60% of the 20 OI-genes analyzed. These isolates belong to the O5:NM, O26:H11, O38:H39, O118:H2/H16, O121:H19, O145:NM, O146:H21, and O165:NM serotypes. From these, four isolates, belonging to O26:H11 from humans, and O118:H16 and O165:NM from cattle, were classified as "virulence top ranking", presenting between 80 and 90% of the genes. The O26:H11 strains represent a group with a high virulence potential for humans. This is also corroborated by their serotype, which is associated with the classical EHEC serotype.
This study detected several profiles for each PAI, especially in different serotypes, and that one or more PAIs virulence genes could be absent in STEC non-O157. In agreement with Ju et al. [5,23], we postulate that PAI could be unstable in STEC. However, as highlighted in the results summarized in Table 2, the presence or absence of particular genes´intra-serotype is not random; instead, there are specific arrays of virulence factors. Further studies using whole-genome sequencing might identify additional virulence markers and increase the understanding of their contribution to human disease.
In spite of the fact that LEE is important, it is not essential for STEC pathogenesis, and sporadic cases and small outbreaks of STEC infections have been caused by LEE-negative strains [24]. Although the absence of eae was associated with the absence of many of the virulence genes, unlike previous publications, such as Franz et al. [22], this study detected some unusual eae-negative cattle and food strains belonging to serotypes O8:H16, O39:H49, O141:H8, O162:H7, and O171:H2/NM, which carried genes nleH1-1, nleG2-3, nleG9, Z4326, Z4332, or Z4333. Such strains, unlike other LEE-negative strains, may cause human diseases. Indeed, some of these serotypes, for example, O171:H2 and O171:NM, have been associated with HUS and/or bloody diarrhea in children in Argentina [25]. Therefore, the risk represented by them to public health should be monitored. On the contrary, two eae-positive strains, O145:NM isolated from a human, lacked all of the studied PAI genes.
It has been proposed and agreed that COI-122-positive strains belong to SPTs associated with outbreaks (A and B) and SPTs associated with HUS (A, B, and C) [4,14,15]. However, in this study, some COI-122-positive strains identified with different SPTs or not determined, O20:HNT (SPT not determined), O146:H21 (D) strains from cattle, and O111:H2 and O118:H2 (SPT not determined) from humans were detected. Also, in relation to OI-122, and coincidently with the results from Konczy et al. [8], results showed that when LEE was absent, only module 1 (gene Z4321) was present. These strains that carry the pagC-like gene exclusively (modules 2 and 3 absent) would have a non-synonymous substitution in (His→Gln) [8]. Wickham et al. [14] postulated that there was a significant association between the presence of a combination of OI-122 markers, such as the pagC-like gene and sen, nleB, and efa-1, and HUS after infection in non-O157 E. coli. Four strains with this profile: three human strains (one O111:H2, one O118:H2, and one O121:H19), and, interestingly, one O146:H21 strain isolated from the bovine group were found.
Newton et al. [26] proposed that NleE and NleB (OI-122) contribute to pathogenesis by inhibiting an initial host inflammatory response (contribute to the suppression of innate signaling pathways) to allow the bacteria to persist in the early stages of infection. NleE appears to obstruct nuclear translocation of Rel family transcriptional activators, while allowing the nuclear import of a transcriptional repressor, resulting in the suppression of IL8 expression. NleB prevents translocation of the immune regulator nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB) to the cell nucleus [26]. Gene nleB was detected in O26:H11, O111:H2, O118:H2, O121:H19, O128:NM, and O145:NM strains isolated from patients, and in O145:NM, O146:H21, O165:NM, and O177:NM cattle strains. The O26:H11 strains from humans presented nleB and nleE, but cattle O26:H11 strains presented only nleE. Unlike a study of Karmali et al. [4] in which they postulated that in cases where a serotype was represented by more than one strain, all strains belonging to the same serotype had identical patterns of distribution of OI-122 genes, in this work different virulence profiles for OI-122 in isolates belonging to the same serotype were found (see Figure 2). In previous work, in which a subset of these strains was analyzed using multilocus sequence typing (MLST), different sequence types (STs) in the same serotype were detected [27]. Further studies should determine if different STs correspond to different PAI-virulence profiles.
According to the results from Ju et al. [5], some of OI-122 and OI-57 nle (nleB, nleG2-3, nleG5-2, and nleG6-2) were primarily associated with eae-positive STEC strains and associated with SPTs that cause severe diseases (SPT A, B, C), offering an important basis for STEC MRA. But, based on the MRA framework, which uses nle genes as sole markers, all eae-negative virulence STEC strains, including HUS-associated O113:H21 and O91:H21, would be categorized as harmless. Therefore, we agree that additional markers, especially for eae-negative STEC, are needed to absolutely predict the potential of an STEC strain to cause disease. Recently, a novel PAI named Locus of Adhesion and Autoaggregation (LAA), exclusively present in a subset of LEE-negative STEC strains, with the mostly clinically relevant ones including O91:H21, O113:H21, and O174:H21, has been proposed. Several virulence factors, among them Hes, which participates in colonization-associated phenotypes, and other ones participating in adhesion and autoaggregation such as Pag-C, Iha, and, Agn43 are encoded in it [28].