A hemolytic-uremic syndrome-associated strain O113:H21 Shiga toxin-producing Escherichia coli specifically expresses a transcriptional module containing dicA and is related to gene network dysregulation in Caco-2 cells

Shiga toxin-producing (Stx) Escherichia coli (STEC) O113:H21 strains are associated with human diarrhea and some of these strains may cause hemolytic uremic syndrome (HUS). The molecular mechanism underlying this capacity and the differential host cell response to HUS-causing strains are not yet completely understood. In Brazil O113:H21 strains are commonly found in cattle but, so far, were not isolated from HUS patients. Here we conducted comparative gene co-expression network (GCN) analyses of two O113:H21 STEC strains: EH41, reference strain, isolated from HUS patient in Australia, and Ec472/01, isolated from cattle feces in Brazil. These strains were cultured in fresh or in Caco-2 cell conditioned media. GCN analyses were also accomplished for cultured Caco-2 cells exposed to EH41 or Ec472/01. Differential transcriptome profiles for EH41 and Ec472/01 were not significantly changed by exposure to fresh or Caco-2 conditioned media. Conversely, global gene expression comparison of both strains cultured in conditioned medium revealed a gene set exclusively expressed in EH41, which includes the dicA putative virulence factor regulator. Network analysis showed that this set of genes constitutes an EH41 specific transcriptional module. PCR analysis in Ec472/01 and in other 10 Brazilian cattle-isolated STEC strains revealed absence of dicA in all these strains. The GCNs of Caco-2 cells exposed to EH41 or to Ec472/01 presented a major transcriptional module containing many hubs related to inflammatory response that was not found in the GCN of control cells. Moreover, EH41 seems to cause gene network dysregulation in Caco-2 as evidenced by the large number of genes with high positive and negative covariance interactions. EH41 grows slowly than Ec472/01 when cultured in Caco-2 conditioned medium and fitness-related genes are hypoexpressed in that strain. Therefore, EH41 virulence may be derived from its capacity for dysregulating enterocyte genome functioning and its enhanced enteric survival due to slow growth.

Introduction 0.35) were inoculated in 4 mL of F or C medium and incubated for 3 h at 37˚C. There were four biological replicates for each bacterial strain. After this period the bacteria were recovered by centrifugation for 10 min at 5000 xg and the pellet was suspended in 600 μL of RNAprotect Bacteria Reagent (Qiagen cat. no. 76506, Valencia, CA) for RNA preservation. Bacterial cells were lysed using lysozyme (1 mg/mL) and proteinase K (2 mg/mL) and incubated for 10 min at 20˚C. Total RNA was obtained using the RNeasy Mini Kit (Qiagen cat no. 74104, Valencia, CA). RNA purity analysis and quantification was accomplished by using the NanoVue spectrophotometer. RNA integrity was assessed on the Agilent BioAnalyzer 2100 (Agilent, Santa Clara, CA). All samples presented RIN > 7.5 and were stored at -80˚C until use in hybridization experiments.

Enterocyte-STEC interaction assays
The Caco-2 cells were cultured in T flasks (25 cm 2 ) or in 4 wells cell culture plates containing glass coverslips (13 mm) and DMEM with FBS (10%) and penicillin-streptomycin (100 U/mL-100 μg/mL) in a 5% CO 2 at 37˚C. The cells were grown until confluence and the formation of a polarized epithelial cell monolayer (it occurs between 5-7 days). Twenty-four hours prior to interaction assays the cells were washed three times with 1X phosphate buffered saline and incubated with 4 mL of antibiotic-free DMEM containing 10% fetal bovine serum. Interaction assays were performed with 400 μL or with 50 μL (Abs 550 = 0.35) of bacterial culture respectively placed on a Caco-2 monolayer cultured in T flasks or in 4 wells cell culture plates. Subsequently, the cell cultures were incubated in 5% CO 2 at 37˚C for 3 hr. After this period, the Caco-2 cells that were exposed to bacteria and the uninfected controls-both in quadruplicates-were recovered for RNA extraction or for electron microscopy.

RNA extraction
After enterocyte-STEC interaction, the cells were gently washed 3 times with 1X phosphate buffered saline. Subsequently, Caco-2 cells were recovered from the culture flasks by vigorous pipetting with 1X phosphate buffered saline. The cells were then centrifuged for 5 min at 5000 xg and the pellet was resuspended in 600 μL of RNAlater Reagent (Qiagen cat. no. 76154, Valencia, CA) for RNA preservation. Total RNA was extracted from Caco-2 cells after lysing with RLT buffer and the RNeasy Mini Kit (Qiagen cat no. 74104, Valencia, CA). RNA purity analysis and quantification was accomplished by using the NanoVue spectrophotometer. RNA quality was assessed on the Agilent BioAnalyzer 2100 (Agilent, Santa Clara, CA). All samples presenting RIN > 7.5 were stored at -80˚C until use in hybridization experiments.

Data acquisition and processing
The microarray images were captured by the reader Agilent Bundle according to the parameters recommended for bioarrays and extracted by Agilent Feature Extraction software version 9.5.3. Spots with two or more flags (low intensity, saturation, controls, etc.) were considered as NA, that is, without valid expression value. The R software version 2.11.1 and an in house script were used for: i) sample grouping (the comparison groups are described in Table 1); ii) excluding transcript spots presenting three or more NAs per group; iii) converting gene expression values to log base 2 [14,15]. Through this procedure the valid transcripts were obtained for each of the comparison groups (Table 1). TMEV software version 4.6.1 and Significance Analysis of Microarrays (SAM) was used for obtaining the differentially expressed genes/ transcripts for all comparisons. All array data is available from the GEO database (https://www.ncbi.nlm.nih.gov/geo/, accession number GSE45979).

Gene co-expression networks (GCNs)
Gene co-expression networks (GCNs) were constructed by using Pearson's correlation. Data analysis, hierarchical network structure, and visualization were accomplished through Cytoscape (version 3.1.0, www.cytoscape.org). All networks were tested for scale-free status, i.e. power law distributions in empirical data [15]. The GCN correlation thresholds were chosen in order to ensure that most of nodes continued to be connected to the major component and that the network remained stable along a threshold range, i.e., maintaining network's topological structure [14].

Venn diagram analysis
We used a web tool Draw Venn Diagram-UGent (available at http://bioinformatics.psb.ugent. be/webtools/Venn/) for analyzing DE gene data sets.

PCR analysis for specific gene detection in a panel O113:H21 STEC strain
We used Primer-BLAST (Primer3 Input, version 0.4.0 and BLAST, available at http://www. ncbi.nlm.nih.gov/tools/primer-blast/) for designing specific primers (S2 Table) in order to search for selected genes in a panel of STEC O113:H21 strain encompassing the 12 STEC strains used in this study (S1 Table). Each PCR reaction was performed in a final volume of 25 μl containing 1.5 U of High-Fidelity Platinum Taq DNA Polymerase (Invitrogen, Carlsbad, CA, USA), 20 mM Tris-HCl (pH 8.4), 50 mM KCl, (1.0-2.0) mM MgSO4 (S2 Table), 150 μM each of dATP, dCTP, dGTP and dTTP, 0.3 μM of primers and 1 μl of a boiled bacterial

Validation of STEC microarrays
Amplification reactions were performed in a 20 μL final volume containing 1X RT Enzyme Mix and 1X RT-PCR Mix (Power SYBR Green RNA-to-C T 1-Step, Applied Biosystems, Carlsbad, CA), 5 pmol of primers and 100 ng of total RNA. We used the following cycling parameters: a RT step of 48 o C for 30 min, an enzyme activation of 95˚C for 1 min followed by 40 cycles of 95˚C for 15 s and 60˚C for 1 min. In order to normalize qPCR reactions, rpoA was included as reference gene. Relative gene expression was determined by the relative standard curve method and presented as relative expression using rpoA as endogenous control for STEC strains.

Validation of Caco-2 microarrays
Amplification reactions were performed in a 25 μL final volume containing 1X SYBR Green mix (Quantitec SYBR Green PCR kit, QIAGEN, Hilden, DE), 10 pmol of primers and 2 μL cDNA (1/10 dilution, synthesized from 1μg of total RNA). We used the following cycling parameters: an initial hot start of 95˚C for 15 min followed by 50 cycles of 95˚C for 15 s and 60˚C for 30s. In order to normalize qPCR reactions, GAPDH was included as reference gene. Relative gene expression was determined by the relative standard curve method and presented as relative expression using GAPDH as endogenous control for Caco-2 infected by STEC EH41 or by EC472 and for uninfected Caco-2.
Bacterial growth assessment for EH41 and Ec472/01 The bacterial growth was evaluated by absorbance readings at 550 nm (Multiskan MS Primary EIA, Labsystems, MA) just before and after 3h of growth in C medium.

Scanning electron microscopy (SEM) for Caco-2 cells
After Caco-2 cell-STEC interaction, the cells exposed to bacteria and the uninfected controls were gently washed 3 times with 1X phosphate buffered saline (PBS) and fixed with Karnovsky fixative solution for at least 24 h at 4˚C. After fixation, cells were washed 3 times with 0.1 M cacodylate buffer (10 min) and post-fixed with 1% osmium tetroxide (prepared in the same buffer) for 30 min. After being washed for 3 times with distilled water, preparations were dehydrated through a graded ethanol series (50%, 75%, 85%, 95%, and 100%). Subsequently, the preparations were dried (critical point method), mounted on stubs and sputter coated with gold. Specimens were then examined under SEM (QUANTA 250-FEI Company, Netherlands) at 12.5 kV.

Results
In this study (see Fig 1) we investigated the comparative global gene expression of two STEC strains-EH41, reference strain isolated from a patient with HUS, and Ec472/01, isolated from cattle feces-after bacterial growth in Caco-2 cells conditioned (C medium) or fresh media (F medium). Another comparative analysis was performed for Caco-2 cells after 3h of in vitro bacteria-host cell interaction. We obtained GCNs based on differentially expressed genes for Caco-2 cells and for both STEC strains, as described below.
Transcriptional analysis of STEC EH41 and Ec472/01 growth in C and F media Our initial analysis aimed at investigating if enterocyte soluble mediators modify STEC global gene expression. In order to accomplish this goal we performed a comparative analysis by growing the two STEC strains in C or F medium for 3 hours. Comparative gene expression analysis of STEC strains cultured in C or F medium revealed that 38 transcripts for EH41 and 53 transcripts for Ec472/01 are differentially expressed in C medium ( Table 1). The biological functions of these genes are listed in S4 and S5 Tables. Venn diagram analysis of these differentially expressed transcripts (Fig 2) showed that only two transcripts (ECs1070 and ECs4328, both functionally uncharacterized) are common between EH41 and Ec472/01 strains. Noteworthy, none of these transcripts is a known or putative virulence factor. On the other hand, it is well described that the expression of virulence factors is modulated by enterocyte mediators   [16]. Therefore, our results indicate that HUS-associated STEC (EH41) and Ec472/01 strains have their own "pre-exposure" differential transcriptomic profile. Consequently, we conducted a comparison between EH41 and Ec472/01 exposed only to the C medium. All differentially expressed (DE) and exclusively expressed (EE) transcripts (Table 1) and their biological functions are listed in S6 Table. The functional profile analysis of the DE and EE transcripts, except for hypothetical or unknown protein, are shown in Fig 3. Almost half of the DE transcripts (49%) are involved in metabolic process (Fig 3A), whereas 33% of EE transcripts in EH41 codify prophage-derived genes and three genes are involved in acid resistance (Fig 3B). Only seven Ec472/01 EE transcripts have a known biological function ( Fig 3C). Interestingly, the majority of DE transcripts are hypoexpressed in EH41 (S6 Table) and are involved in fitness. Furthermore, we evaluated bacterial growth in C medium. The EH41 strain showed a significant slow growth when compared with Ec472/01 (S1 Fig).

Gene co-expression networks (GCN) for EH41 and Ec472/01
This GCN analysis considered just the DE and EE genes obtained from the comparison between the transcription profiles of EH41 and Ec472/01 strains grown in C medium. The two networks were constructed considering only the transcripts codifying for known or putative proteins: 103 transcripts for EH41 and 89 transcripts for Ec472/01 (Table 2, Fig 4). A list of the most relevant hubs, i.e. those presenting high numbers of gene-gene links, and their biological function appears in Table 3.

EH41 network
The EH41 GCN ( Fig 4A) revealed a major transcriptional module constituted by genes exclusively expressed (EE) in STEC EH41 and containing most of the hubs in this network (10 out of 16). This module presents positive interaction between its constituent EE genes (diamond nodes) but negative interaction with the network's DE genes (circle nodes). Moreover, the EH41 hierarchical network (Fig 5) showed that the transcriptional regulators dicA and dicC are the first and the second hierarchical genes in the EE transcriptional module, here named dicA module. Moreover, dicC is linked, in the first level, with almost all the EE genes (20 out of 23, where 8 are hubs), whereas dicA is linked with 16 EE genes, of which nine are hubs, including dicC (these links are indicated by red lines in Fig 5B and 5C). Here is important to mention that the genes dicA and dicC are part of the Qin prophage [17] and that DicA is homologous to the virulence regulators RovA and SlyA [18,19]. This issue will be discussed further. The list of hubs and their respective biological functions is presented in Table 3. The dicA module encompasses 27 genes: 23 EE genes, of which nine are hubs, and four are DE genes, of which only one is a hub. In this module seven EE hubs may be involved in bacterial virulence (dicC, ECs1176, rusA, ECs1666, ydfZ, insA and insI) [20-22] and two EE hubs are associated with acid stress adaptation to survive in the host (gadB and ECs2098) [23].
The second major module encompasses 24 DE genes, of which seven are hubs and only one is an EE gene (ygbT). This EE hub codifies for a CRISP-associated protein related to antivirus immunity and DNA repair [24,25]. Three DE genes are also hubs in the Ec472/01 network: adiA is involved in acid stress adaptation [26]; metQ is involved in ABC-type transporter in E   in various metabolic processes, such as electron transfer, amino-acid synthesis or transcription.
PCR detection of the dicA module EE genes using a panel of STEC O113:H21 strains We used PCR for detecting the presence of nine EH41 EE genes-insI, ECs2098, dicA, dicC, fecC, gadB, Ecs1176, insA and rusA-of the dicA module in a panel constituted by Ec472/01, one Brazilian STEC strain isolated from beef meat and nine other Brazilian STEC strains isolated from the animal reservoir (S1 Table). This analysis revealed that five genes (dicA, fecC, Ecs1176, insA and rusA) are absent in all these strains. Two genes-ECs2098 and gadB-were detected in all strains (Table 4).

GCN analysis of Caco-2 cells after interaction with STEC strains
Three gene co-expression networks were constructed for DE genes ( Table 2): i) for Caco2-EH41 ( Fig 6A); ii) for Caco2-Ec472 ( Fig 6B); and iii) for Caco2-Control (Fig 7). The control network was based on all DE genes obtained in the two former comparisons (155 genes, S2 Fig). A list of the most relevant hubs according to the number of gene-gene links appears in Table 5.
Symbols: (+) for positive or (-) for negative PCR product b strain EH41 express these genes c strain Ec472/01 does not express these genes ND, not determined; same clonal group of the STEC strains isolated from an Argentine HUS-patient d and from a Germany patient with diarrhea e [11] https://doi.org/10.1371/journal.pone.0189613.t004 Additionally, two other hubs could also be involved in immune response, COL1A2 and THBS1. The first encodes a cellular matrix protein that can be stimulated through TLR4 by E. coli and induces a chronic inflammatory state in murine colitis [59,60]. The former codifies for an adhesive glycoprotein and its deficiency is associated with decreased phagocytosis and possibly bacterial clearance [61,62]. The remaining six hubs are involved in cytoskeletal organization, signaling and cell growth or codifies a G protein-coupled receptor and transcription factors. Two hubs-THBS1 and MAB21L2-are also hubs in Caco2-Ec472 network and RHOU is likewise hub in Caco2-Control network (Table 5).
Other four hubs also have interesting features: three of these hubs are involved in cytoskeleton organization (CLIC5, RND1 and RND3) and one (ARL5B) codifies for a protein transport regulator. In HeLa cells this protein is involved in Shiga toxin transport along the endosome to Golgi trafficking [77]. The remaining hubs are involved in other biological functions such as signaling, metabolism and ubiquitination.

Caco2-Control network
The Caco2-Control network analysis (Fig 7) revealed that all hubs belong to a single major transcriptional module and present positive gene-gene interactions. The majority of the hubs (9 out of 11) are involved in cell growth control (seven hubs) and epigenetics (two hubs). Other two hubs are related to regulation of cell morphology and ion transport (Table 5). However, the interaction with EH41 leads to a severe microvilli loss (Fig 8B), whereas Ec472/ 01 induced only modest microvilli loss (Fig 8C), but significant microvilli morphology change (elongation, orientation) when compared to Caco-2 control cells (Fig 8A).

Caco-2 cells SEM images
qPCR validation of microarray data STEC group. Three hyperexpressed (phoA, phoE and EcS1174) and two hypoexpressed (hycE and hycG) genes were selected for qPCR analysis. The fold-changes for each gene-comparing EH41 versus Ec472/01 groups' average of relative gene expression-confirmed DNA microarray gene expression results (S4 Fig).

Discussion
The association of O113:H21 STEC strains with hemolytic uremic syndrome (HUS) has been reported in several countries [11]. In Brazil O113:H21 strains are commonly found in the animal reservoir but, so far, were not isolated from HUS patients [7,8]. Interestingly, some Brazilian O113:H21 strains and strains isolated from HUS patients in Argentina belong to the same clonal group [11]. Several studies have been done in STEC strains isolated from animal and environmental sources aiming at identifying HUS-related virulence genes and characterizing the pathogenic potential of these strains. However there is no specific virulence genetic profile that enables one to distinguish between the pathogenic and environmental strains [6,11,102]. In this study we were able to show genomic and phenotypic differences between two O113: H21 STEC strains: EH41, a reference strain isolated from HUS patient in Australia, and Ec472/01, isolated from cattle feces in Brazil. We also characterized the differential enterocyte response after exposure to EH41 or Ec472/01.
First we investigated if Caco-2 cell soluble mediators could increase the expression of STEC virulence genes in EH41 and Ec472/01 bacteria. Comparative global gene expression analyses revealed, for both strains, that the expression of virulence genes was not significantly different: only one per cent of the transcripts were differentially expressed. Furthermore, the majority of these transcripts was not yet functionally characterized, as in EH41, or involved in metabolic processes, as in Ec472/01 (S4 and S5 Tables). Comparatively, in EHEC, more than 1,400 genes-most of them related to virulence-were differentially expressed after culturing in enterocyte conditioned medium [16]. These results indicate that EH41 and Ec472/01 virulence genes are not modulated by soluble mediators, i.e. their virulence phenotypes were largely set before infecting the human host. Therefore we conducted a comparative transcriptomic analysis between EH41 and Ec472/ 01 cultured in C medium. The results revealed that these two strains have different gene expression profiles. We found important differences in these profiles: i) 82 out of 97 DE transcripts are hypoexpressed in EH41 and almost half of these genes are involved in metabolic process (Figs 2 and 3); ii) in EH41 10 out of 29 EE genes are prophage-derived genes; iii) in Ec472/01 only seven out of 35 EE genes are functionally characterized an all related to metabolic or transcriptional control processes (S6 Table).
It is interesting to note that in EH41 six of the EE genes belong to the Qin prophage [17]: dicA, dicC, relE, ynfN, and two not yet functionally characterized. It is well known that cryptic prophages provide multiple adaptative advantages for the bacteria, such as antibiotic resistance, oxidative and acid stresses tolerance, enabling survival in adverse environmental conditions [103]. The dicA is an E. coli transcriptional regulator acting as a temperature sensor repressor for bacterial growth and, when mutated, is complemented by the adjacent gene dicC [34,104]. Additionally, DicA shares significant amino acid sequence similarity with the DNA-  binding domains of RovA and SlyA, which are regulators of pathogenic genes in Yersinia and Salmonella, respectively [17]. These two proteins regulate a wide range of physiological processes involved in survival, stress adaptation, and virulence [18,19]. In Salmonella, SlyA regulates virulence factors necessary for environmental adaptation and survival in mice [105]. In E. coli SlyA induces a cryptic hlyE gene (aliase clyA), which encodes for hemolysin E [106].
Hierarchical analysis of the EH41 network showed that dicA is the first hierarchical gene in the EE module and encompasses 22 out of the 24 EE genes in the network (Fig 5). The dicA transcriptional module is distinctive of the EH41 network and seems to be very important since it encompasses 11 out of the 16 network hubs. Moreover, most of these hubs are prophage-derived genes, including dicC, or involved in acid stress adaptation, what indicates that the dicA module could well be related to EH41 virulence.
The other two modules in EH41 network (Fig 4A) encompass only DE genes-many of them involved in metabolic process. The majority of these genes are hypoexpressed in EH41, except for the hub Z1902, which is a prophage-derived gene. It is interesting to note that metQ-also a hub in Ec472 network, but hypoexpressed in EH41 -codifies for a DL-methionine transporter subunit and may be involved in bacterial growth efficiency. Streptococcus pneumoniae metQ and metEF mutants show a decreased growth in methionine restricted conditions [107]. Here we confirmed that EH41 has a diminished growth in C medium when compared with Ec472/01 (S1 Fig). Altogether, EH41 GCN analysis indicates that dicA positively regulates other EE genes (probably virulence genes) in the dicA transcriptional module. Here it is worth to note that the dicA, B, C gene family exerts a negative regulation on bacterial growth [34,104]. These two regulatory roles of dicA could represent an adaptation for improving long-term bacterial survival in the enteric environment. Indeed, there are reports of long-term STEC shedding, i.e. well after the symptoms are resolved. The median duration of shedding has been shown to be 20 days; however, some patients were STEC PCR-positive up to 9 months after symptoms disappearance [108,109].
Ec472/01 network presented just six EE genes which are scattered in a few transcriptional modules (Fig 4B). In this network only two out of 15 hubs are EE genes, but all these hubs belong to a single main transcriptional module. As mentioned before, most of the DE genes are hyper-expressed in Ec472/01 and probably involved in bacterial growth (S6 Table). Five hubs may be related to virulence: two are phage-derived genes, one is involved in acid stress adaptation, one codifies for a recombinase and the last codifies for an antirepressor protein.
Recombinase and antirepressor proteins are associated with acquisition and expression of bacteriophage-virulence genes [42,46,110].
Our results clearly indicate that dicA transcriptional module, and especially the dicA gene, contributes to the distinctive phenotypic difference between EH41 and Ec472/01. Hence, we investigated the presence of dicA and other eight EH41 EE hub genes in a STEC panel. We selected STEC strains from different clonal groups, including strains belonging to same clonal group of the STEC strains isolated from an Argentine HUS-patient and from a German patient with diarrhea [11]. The results revealed that all STEC strains isolated from bovine, goat or buffalo feces, or from beef meat, lack dicA and four other genes present in EH41 (Table 4). This result indicates that STEC strains isolated from animal reservoirs do not have the complete gene repertoire to cause severe diseases, such as HUS. Hereafter, it would be necessary to extend this investigation to other STEC strains isolated from patients with HUS, hemorrhagic colitis, and diarrhea, as well as from animal and environmental sources. In this study, we could not test other O113:H21 STEC strains isolated from patients because: i) this serotype has not yet been isolated from HUS patients in Brazil, and ii) nowadays for biosafety reasons HUSassociated STEC strains from other countries is hard to obtain. Nevertheless, it is utterly important to undergo this kind of investigation in order to provide biomarkers for identifying STEC strains capable of causing HUS and related severe diseases.
The Caco-2 cell response after three hours of interaction with STEC was investigated by GCN analysis for DE genes. Distinct enterocyte responses to EH41 and Ec472/01 strains in DE networks are clearly evidenced by the pronounced differences in network topology (Fig 6A  and 6B). Caco2-EH41 GCN reflects global gene expression dysregulation, since this network has many positive and negative gene-gene covariation coefficients in the major module. Conversely, in Caco2-Ec472/01 and Caco2-Control networks most of the nodes have positive links and just a few genes-corresponding to the nodes at the transcriptional module border or nodes connecting two modules-have negative links. This is what is expected in normal cell functioning or just before the health-disease transition [111][112][113]. Nodes with positive links tend to cluster together, while nodes with negative links usually act as bridges between clusters of positively linked nodes [112]. This result suggests that EH41 is capable to induce, after three hours of cell-bacteria interaction, an intense genomic dysregulation in Caco-2 cells.
The majority of the hubs in Caco2-EH41 GCN were found to be related to inflammatory/ innate immune response (Table 5). It is possible to infer, based on KEGG data, that several hubs in this network are involved in the activation of inflammatory response and neutrophil recruitment via TNF signaling pathway (Table 5). Actually, CXCL2 and CXCL3 (aliase GROg) codify for chemoattractants involved in neutrophil migration [47, 50] and, TNF, an inflammatory cytokine. Additionally the hub PTN encodes a pleiotropin and induces production of inflammatory cytokines, such as TNF, IL-1b and IL-6 [52]. In mice, CXCL2 is involved in neutrophil migration into the kidney after exposure to EHEC O157:H7 (E. coli possessing stx/eae genes and associated with HUS) virulence factors [50]. The increased production of neutrophil chemoattractants such as IL-8, Stx-mediated ERK1/2 activation promotes inflammation and the systemic uptake of Stx, leading to the onset of HUS [114]. Moreover, it was described that flagellin is the major EHEC determinant contributing to chemokine production in human intestinal epithelium [115], suggesting that Stx is not a major participant in promoting intestinal inflammation. It is possible that in EH41 Stx exacerbates epithelial inflammation induced firstly by other virulence factors.
This scenario did not occur in Caco-2 cells during interaction with Ec472/01. Only one third of the hubs are involved in innate immunity and enrichment analysis based on KEGG molecular pathways showed that three hubs are related to TNF signaling pathway: i) cell adhesion (ICAM1); ii) inflammatory cytokine (LIF) and; iii) CFLAR that codifies an apoptosis inhibitor regulator [116]. Additionally, both networks have hubs involved in cystoskeletal organization.
Scanning electron microscopy (SEM) images revealed microvilli loss after Caco-2 interactions with EH41 or with Ec472/01 (Fig 8). Epithelial cell morphology alterations are clearly more severe after exposure to EH41, including brush border and microvilli destruction. Microvilli establish an electrostatic barrier to microbial adhesion [117] and, therefore, EH41-induced microvilli destruction may contribute to the enhanced persistence of this strain in the enteric environment.
There is evidence that acute inflammation plays a role in the development of HUS. Patients with HUS demonstrate a rise in c-reactive protein, neutrophilia and an increase in circulating proinflammatory cytokines, indicating that the impact of hemorrhagic colitis may be important for the subsequent development of severe complications, such as HUS and encephalopathy [78]. Moreover, Stx probably is not the central factor involved in enteric inflammation, but together with others bacterial effectors contribute to promote inflammation and neutrophil migration.
In conclusion, we used a gene co-expression network (GCN) analysis to investigate two STEC strains-one associated to HUS (EH41) and another isolated from cattle (Ec472/01)and also the enterocyte response after bacterial interaction. Comparative STEC GCN analysis revealed that STEC HUS-associated has a distinctive dicA transcriptional module and majority of the genes in others modules-possible related to bacterial growth-are hypoexpressed. This fact may be related to bacterial adaptation to survive for long period in intestinal environment and consequently cause severe disease in human. The PCR detection of nine hub genes belonging to dicA module in EH41 network indicated that STEC strains isolated from animal reservoirs could not have complete gene repertoire to cause severe disease, such as HUS. Moreover, comparative Caco-2 cells GCN analysis indicated that STEC HUS-associated induces pronounced inflammation response and gene network dysregulation. Finally, dicA and other four genes (fecC, insA, rusA and ECs1176) may be used as molecular markers to distinguish between HUS-associated O113:H21 STEC strains and other strains isolated from animal or environmental sources. This characterization is important since some O113:H21 STEC strains isolated from animal sources in Brazil belong to the same clonal group of STEC strains isolated from Argentine HUS-patients or from animal sources.