Analysis of Human miRNA Derived From Bladder Epithelial Cells Infected With Uropathogenic Escherichia coli

TL;DR

This study explores how bladder cells infected with certain types of E. coli release microRNAs that may influence both human and bacterial genes during urinary tract infections.

Contribution

The study identifies miRNAs released by bladder cells infected with ESBL-producing E. coli and their potential targets in human and bacterial genomes.

Findings

402 unique miRNAs were identified, with 30 differentially expressed in ESBL019-infected cells.

Differentially expressed miRNAs target 747 human genes linked to immune and stress pathways.

Nine miRNAs were predicted to interact with bacterial genes from an ESBL-producing UPEC strain.

Abstract

MicroRNAs (miRNAs) have been shown to regulate many cellular processes and to play a role in host‐pathogen interactions. However, the role of miRNA in urinary tract infection (UTI) remains unclear. The aim of this study was to analyze and compare miRNAs from supernatants of human bladder epithelial cells infected with ESBL‐producing (ESBL019) and non‐ESBL‐producing (CFT073) uropathogenic
E. coli
(UPEC) strains and to identify miRNA target genes in human cells and uropathogenic bacteria. In total, 402 unique miRNAs were found. The statistical analysis showed differential expression of 30 miRNAs from bladder cells stimulated with ESBL019, while stimulation with CFT073 did not show any significantly expressed miRNAs when compared to unstimulated controls. The 30 differentially expressed miRNAs in ESBL019 stimulated cells showed 747 predicted individual human gene targets. KEGG and REACTOME pathways showed enrichments in pathways mainly connected to immune regulation and stress responses. Of the 30 differentially expressed host miRNAs, nine miRNAs were found to interact with predictive targets of the whole genome from the multi‐resistant, ESBL‐producing UPEC strain EC958. This study shows that ESBL019‐infected bladder epithelial cells release miRNAs with predictive targets in both human and bacterial genes, although their role in UTI cross‐species interactions remains to be clarified.

Figures

(A) Principal component analysis (PCA) of miRNA expression levels in supernatants from bladder epithelial cells stimulated with ESBL019, CFT073 or unstimulated control (overlaid as colors). Components 1–2 account for 65% of the variation. (B) Volcano plot of differential expression of miRNAs between ESBL019 and unstimulated control. Linear mixed effect model analysis with Benjamini‐Hochberg correction for multiple testing applied. p < 0.05 was considered statistically significant and a FC > 2 was considered biologically significant. Gray = no significance; blue = fold change (FC) or p‐value significant; red = both FC and p‐value are significant. (C) Heatmap of the 18 individual sample levels for the 30 differentially expressed miRNAs in ESBL019 stimulated cells.

The functional annotation and enrichments of the 30 differentially expressed miRNAs towards KEGG and REACTOME pathways. Pathways targeted by at least 10 of the 30 differentially expressed miRNAs are shown with size representing the number of individual miRNAs and color representing the statistical enrichment after Benjamini‐Hochberg correction for multiple testing.

Tables

TABLE 1. Prediction of bacterial gene targets for the differentially expressed host miRNAs.

miRNA	Targets	Definition
>hsa‐miR‐940	mukB (EC958_RS05245)	Chromosome partition protein MukB
AAGGCAGGGCCCCCGCUCCCC
>hsa‐miR‐3155a	pyrB (EC958_RS24425)	Aspartate carbamoyltransferase
CCAGGCUCUGCAGUGGGAACU
>hsa‐miR‐3679‐3p	pyrB (EC958_RS24425)	Aspartate carbamoyltransferase
CUUCCCCCCAGUAAUCUUCAUC
>hsa‐miR‐4462	mukB (EC958_RS05245)	Chromosome partition protein MukB
UGACACGGAGGGUGGCUUGGGAA
>hsa‐miR‐4653‐3p	pyrB (EC958_RS24425)	Aspartate carbamoyltransferase
UGGAGUUAAGGGUUGCUUGGAGA	cysI (EC958_RS15335)	Assimilatory sulfite reductase (NADPH) hemoprotein subunit
>hsa‐miR‐1238‐5p	pyrB (EC958_RS24425)	Aspartate carbamoyltransferase
GUGAGUGGGAGCCCCAGUGUGUG	cysI (EC958_RS15335)	Assimilatory sulfite reductase (NADPH) hemoprotein subunit
>hsa‐miR‐6769a‐5p	mukB (EC958_RS05245)	Chromosome partition protein MukB
AGGUGGGUAUGGAGGAGCCCU	cysI (EC958_RS15335)	Assimilatory sulfite reductase (NADPH) hemoprotein subunit
>hsa‐miR‐3158‐5p	mukB (EC958_RS05245)	Chromosome partition protein MukB
CCUGCAGAGAGGAAGCCCUUC	cysI (EC958_RS15335)	Assimilatory sulfite reductase (NADPH) hemoprotein subunit
>hsa‐miR‐7109‐5p	mukB (EC958_RS05245)	Chromosome partition protein MukB
CUGGGGGGAGGAGACCCUGCU

Full text

Introduction

1

Urinary tract infection (UTI), predominantly caused by uropathogenic Escherichia coli (UPEC), is one of the most common infectious diseases [1, 2]. The rapid spread of drug resistance among Gram‐negative organisms, including UPEC, causes a major challenge in the clinical management of UTI [3]. Production of extended‐spectrum β‐lactamases (ESBLs) is one of the most important resistant mechanisms in Gram‐negative bacteria, which give the bacteria the ability to resist therapy with many β‐lactam antibiotics. Co‐resistance to fluoroquinolones, tetracyclines, and other antibiotic classes is common, creating multidrug‐resistant isolates [4]. The multidrug‐resistant E. coli sequence type ST131 is a globally distributed pandemic lineage that has been associated with worse clinical outcomes and relapse in UTI [5, 6]. Thus, development of novel therapeutic strategies for management of UTI that do not depend on antibiotics is urgently needed.

MicroRNAs (miRNAs) are a group of short sequence (typically 20 to 22 nucleotides long), non‐coding RNA molecules that are present in a wide range of organisms [7]. These short RNA molecules have been implicated in regulating key cellular processes and play an active role in host‐pathogen interactions by modulating immune signaling pathways and influencing the outcome of bacterial infections. In recent years, the interest in miRNA in medicine has increased, and dysregulation of miRNA expression or function has been implicated in a variety of diseases, including cancer, cardiovascular disease, neurodegenerative diseases, metabolic disorders, and inflammatory/autoimmune diseases [8]. MiRNAs are not only present intracellularly but also as an important cargo in exosomes in different bio‐fluids including blood plasma and urine [9, 10]. Given their stability and resistance to RNase degradation in extracellular fluids, miRNAs have emerged as attractive biomarker candidates for diagnosis and prognostication of disease progression. In the urinary tract, miRNA from urine samples has been investigated as potential non‐invasive biomarkers for diagnosis and monitoring the progression of renal damage [11, 12].

Extracellular miRNA may be delivered to recipient cells and regulate translational activity of target genes. This regulatory mechanism involves miRNAs binding to their target mRNAs with perfect or partial complementarity, leading to mRNA cleavage or translational repression and ultimately altering the expression of protein‐coding genes [13]. Identifying miRNA interactions with target mRNA is of interest for understanding biological processes and to address pathological conditions involving miRNA [14]. Emerging evidence also suggests that host‐derived miRNAs may be internalized by bacteria and influence bacterial gene expression [15], and this cross‐species communication by miRNA may coordinate the interplay between host immune cell function and bacterial pathogenesis during infection. Thus, bacteria may evoke distinct miRNA signatures to subvert the key host cell pathways and/or to promote their own survival and virulence [16]. Host cell miRNAs have been recognized as regulators upon infection with lipopolysaccharide (LPS) or bacterial pathogens e.g., * Salmonella enterica, Helicobacter pylori, Mycobacterium* spp. *and Listeria monocytogenes

• [16, 17]. In the early work by Taganov and colleagues three endotoxin‐responsive miRNAs (miR‐146a/b, miR‐132, miR‐155) were identified in LPS‐stimulated human monocytes [17]. Recent studies have begun to explore how UPEC specifically interacts with host miRNA responses, revealing that UPEC infection can modulate miRNA expression in epithelial and immune cells. Interestingly, UPEC strain CFT073 has been found to induce expression of miR‐146b in mouse kidney and macrophages during the early phase of UPEC infection [18].

Targeting miRNA could represent a non‐antibiotic strategy for management of UTI caused by multidrug‐resistant ESBL‐producing bacteria. To our knowledge, no study has yet evaluated if human miRNA can target UPEC‐associated virulence or resistance genes. A recent study demonstrated that UPEC‐infection stimulated isolated mouse bladder epithelial cells to secrete miRNA‐enriched exosomes that were efficiently absorbed by macrophages [19]. The miRNA‐enriched exosomes regulated macrophage function by exacerbating UPEC‐mediated tissue damage, suggesting that targeting exosome release during UPEC infection may be beneficial [19]. In this study, we identified miRNAs that are expressed by UPEC‐infected human bladder epithelial cells and analyzed miRNA‐targeted genes in human cells and uropathogenic bacteria. The present study is primarily descriptive and represents an initial mapping of miRNA responses in a controlled in vitro model. Our intention was to establish a first, unbiased overview of extracellular miRNAs released from human bladder epithelial cells following infection with clinically relevant UPEC strains, including a multidrug‐resistant ESBL‐producing isolate.

Materials and Methods

2

Cell Culture

2.1

The spontaneously immortalized human bladder epithelial cells HBLAK (CELLnTEC, Advanced Cell Systems AG, Bern, Switzerland) was cultured in CnT‐58 cell culture medium (CELLnTEC) supplemented with 100 U/mL penicillin and 100 μL/mg streptomycin (Invitrogen Ltd., Paisley, UK) in a humidified atmosphere at 37°C with 5% CO_2_. At confluency the culture was differentiated into umbrella cell‐like cells during 24 h using CnT‐21 differentiation medium (CELLnTEC) supplemented with 1 mM CaCl_2_. Gentamicin (50 μg/mL; Sigma‐Aldrich, St. Louis, MO, USA) was added to cells 24 h before infection but was excluded during the experiments. During infection experiments, CnT‐21 differentiation medium (CELLnTEC) supplemented with 1 mM CaCl_2_ was used.

Bacterial Strains

2.2

The UPEC strain CFT073 was originally isolated from a patient with pyelonephritis [20] and the ESBL‐producing UPEC strain (designated ESBL019) has previously been characterized and determined to belong to the ST131 clone and to be multi‐resistant (CTX‐M‐15) [21]. The bacteria were stored for diagnostic purposes in the bacterial biobank of the hospital and was provided by the hospital from frozen stock. The bacteria were prior to experiments grown in Luria broth (Lennox; Franklin Lakes, NJ, USA) at 37°C aerobically on a shaker overnight.

RNA Isolation

2.3

The HBLAK cells were stimulated with CFT073 or ESBL019 for 6 and 24 h at a multiplicity of infection (MOI) of 10. An MOI of 10 has previously been used in many of our UPEC‐bladder epithelial cell infection studies [22, 23] and ensures that most of the bladder cells get infected. No cytotoxicity was observed in the bladder epithelial cells under the experimental conditions used, as assessed by morphological inspection and cell adherence throughout the experiments. Unstimulated cells served as controls. After infection, supernatants were collected and centrifuged at 5000 g for 5 min to remove cells and bacteria. RNA was then isolated from the supernatants using the Urine cell‐free circulating RNA purification mini kit (Norgen Biotek Corp, Ontario, Canada) and Urine Exosome RNA isolation kit (Norgen Biotek Corp) according to the manufacturer's instructions. The RNA was then pooled from both kits for each sample.

miRNA Analysis With 3D‐Gene Microarray

2.4

miRNA expression in supernatants was analyzed using the 3D microarray platform (Toray) at TATAA Biocenter (Gothenburg, Sweden). The presence of small RNA in samples was verified using capillary electrophoresis (Fragment Analyzer, Advanced Analytical) with the Standard Sensitivity RNA kit (DNF‐471‐0500). For labeling and hybridization, each sample was combined with a miRNA spike‐in (TRT‐XR304, Toray) and processed according to the Toray H‐M‐R miRNA 4‐Plex V3 protocol, using the Toray miRNA Labeling kit (TRT‐XE211) and Human miRNA Oligo chip 4plex, based on miRBase 21 (TRT‐XR520).

Scanning was conducted using the 3D‐Gene Scanner 3000 (Toray) with automated gain, focus, and analysis settings. Quality control was performed on the instrument's QC reports, and all samples passed criteria for successful hybridization and detection. Background subtraction was applied to spots with intensity exceeding the mean noise plus two standard deviations. Global normalization was performed on background‐subtracted spots using a factor derived from dividing 25 by the median intensity. Each normalized value was computed by multiplying background‐subtracted data by this factor.

Data Processing and Statistical Methods

2.5

Global normalized data was filtered to exclude miRNAs below limit of detection in > 50% of the samples. Remaining miRNAs were log2‐transformed prior to statistical and bioinformatic analyses. Samples from 6 and 24 h within respective stimulation were grouped together for statistical analysis to generate a composite miRNA expression profile per condition. This approach increased statistical power and reduced the influence of transient, time‐specific fluctuations, thereby enriching for miRNAs consistently regulated across both time points and improving comparability with unstimulated controls. Unsupervised dimensionality reduction of miRNA expression levels was performed with principal component analysis (PCA) in R (version 4.2.3) to analyze the distribution of miRNA variation with regard to the stimulus groups.

Univariable analyses were performed with linear mixed effects models evaluating the effect of individual miRNA expression levels and stimulus group, fitted for fixed effects of stimulus group, with distribution to repeated experiments as random effect (lme4 and lmerTest package in R). Benjamini‐Hochberg correction for multiple testing was subsequently performed. Differentially expressed miRNAs were analyzed for human gene targets with the miRDB miRNA target prediction tool [24], based on support vector machines (SVMs) predictions and miRBase V22. Gene targets with a prediction score above 60 were further used to evaluate overlapping targets between miRNAs. Functional annotation and enrichments of differentially expressed miRNAs to KEGG [25, 26, 27] and REACTOME pathways were analyzed with miRPath v4.0 [28] using genes union with TarBase v8.0 targets and miRBase‐v22.1 annotations. P‐value threshold after FDR correction was set to 0.05 for pathway enrichments.

While CFT073 has a fully sequenced and annotated genome available, ESBL019 itself has not been fully sequenced. Therefore, for bacterial target prediction we used the closely related and fully sequenced ST131 ESBL strain EC958 as a representative genome. Prediction of the regulatory gene targets in the genome of EC958 was performed using the machine learning tool TargetRNA3 [29]. Only candidate regulatory targets with a probability greater than or equal to 0.5 and a p‐value less than or equal to 0.05 were selected.

Results

3

Identification of Host‐Derived miRNA

3.1

To investigate whether infection with UPEC strains alters host miRNA expression, miRNA expression in supernatants from human bladder epithelial cells stimulated with UPEC strains was profiled. In total, 402 unique miRNAs with expression levels above the limit of detection in > 50% of the samples were found (Table S1). The multivariable analysis of these using PCA showed a slight separation, especially in principal component 2, of the ESBL019 samples from the CFT073 and unstimulated control samples with regard to distribution of variation in expression levels (Figure 1A). The univariable statistical analysis showed a differential expression of 30 miRNAs in supernatants from cells stimulated with ESBL019 when compared to unstimulated controls (Figure 1B,C). The stimulation with CFT073 did not show any significantly expressed miRNAs when compared to unstimulated controls.

(A) Principal component analysis (PCA) of miRNA expression levels in supernatants from bladder epithelial cells stimulated with ESBL019, CFT073 or unstimulated control (overlaid as colors). Components 1–2 account for 65% of the variation. (B) Volcano plot of differential expression of miRNAs between ESBL019 and unstimulated control. Linear mixed effect model analysis with Benjamini‐Hochberg correction for multiple testing applied. p < 0.05 was considered statistically significant and a FC > 2 was considered biologically significant. Gray = no significance; blue = fold change (FC) or p‐value significant; red = both FC and p‐value are significant. (C) Heatmap of the 18 individual sample levels for the 30 differentially expressed miRNAs in ESBL019 stimulated cells.

miRNA Targeted Genes in Human Genome

3.2

The functional implications of the differentially expressed miRNAs through their predicted human gene targets were investigated. The 30 differentially expressed miRNAs in ESBL019 stimulated cells show 747 predicted individual human gene targets with at least 3 individual miRNAs targeting the same gene. Several of the gene targets were targeted by 10 or more of the 30 individual miRNAs (Table S2). The functional annotation and enrichments of the 30 differentially expressed miRNAs towards KEGG and REACTOME pathways show enrichments in pathways mainly connected to immune regulation and stress responses. Pathways targeted by at least 10 of the 30 differentially expressed miRNAs are shown in Figure 2.

The functional annotation and enrichments of the 30 differentially expressed miRNAs towards KEGG and REACTOME pathways. Pathways targeted by at least 10 of the 30 differentially expressed miRNAs are shown with size representing the number of individual miRNAs and color representing the statistical enrichment after Benjamini‐Hochberg correction for multiple testing.

miRNA Targeted Genes in Bacterial Genome

3.3

One hypothesis tested in the present study was that bladder cell miRNA, released in the extracellular space upon infection, may be internalized by bacteria and influence bacterial gene expression. We therefore investigated potential interactions of the differentially expressed host miRNAs to bacterial gene targets by mapping to potential binding sites in the bacterial genome of the multi‐resistant, ESBL‐producing UPEC strain EC958. Of the 30 differentially expressed host miRNAs, nine of them were found to interact with predictive targets of the whole genome from UPEC strain EC958 according to the TargetRNA3 machine learning prediction tool (Table 1).

Discussion

4

The knowledge on the role of miRNA in the pathogenesis of infection and immune responses is emerging rapidly, including the potential of miRNAs as therapeutic targets and diagnostic tools [30]. The potential of cross‐species miRNA interactions was of particular interest in this study, and therefore the focus was analysis of miRNA expression from the supernatant of infected cells. Building on this, our study investigated whether infection with ESBL‐producing and non‐ESBL UPEC strains induces a distinct host miRNA response in the supernatant of human bladder epithelial cells, and whether these miRNAs potentially could influence both host and bacterial gene regulation. We identified 30 miRNAs that were differently expressed and upregulated in supernatants of bladder epithelial cell infected by the ESBL‐producing UPEC strain ESBL019 compared to uninfected control cells. MiR‐4726‐5p, miR‐1238‐5p and miR‐4485‐5p were the top three most upregulated miRNAs. In a study with a similar experimental set up, the UPEC‐infected mouse bladder epithelial cell line (MB49) was found to express 326 miRNAs enriched in exosomes [19]. Interestingly, one of the most significantly expressed miRNAs identified in mouse bladder epithelial cells, miR‐18a‐5p, was also demonstrated to be significantly higher in exosomes isolated from urine of patients with UTI compared to healthy controls [19]. This highlights the potential of extracellular urine miRNAs as a diagnostic tool and raises the possibility that specific miRNA signatures could be developed as non‐invasive biomarkers to distinguish infection severity. The differently expressed miRNA in our study did not overlap with any of the miRNAs identified in mouse bladder epithelial cells, possibly related to species differences or different virulence properties of the used UPEC strains. However, it is important to note that the identities of the detected miRNAs in our study need to be validated by qPCR experiments to confirm their expression. Other important areas for future investigations are e.g., time‐dependent miRNA regulation, release mechanisms and ratio of vesicle‐bound miRNA vs. non‐vesicular miRNA. The presence of a common set of miRNAs (e.g., miR‐155, miR‐146, let‐7) has been reported in human dendritic cells as a core cellular response to infection shared among many pathogens [16, 31]. However, a specific virulence‐dependent variation in the miRNA dynamic response among closely related mycobacterial strains was also demonstrated [31].

Differentially expressed host cell miRNAs were only detected following infection with ESBL019 and not after infection with CFT073. CFT073 is a well‐characterized, non‐ESBL‐producing UPEC reference strain originally isolated from pyelonephritis, whereas ESBL019 belongs to the globally disseminated ST131 clone and produces CTX‐M‐15 ESBL. The strains differ substantially in resistance profile and immune‐modulatory capacity, which we believe contributes to the distinct miRNA responses observed. The control of miRNA homeostasis is complex as both the pathogen itself and the cellular inflammatory response it elicits may influence and regulate miRNA expression. UPEC strains have evolved different mechanisms to deviate host recognition and signaling, which may enhance their survival in the urinary tract, including stabilization of IkB and a subsequent inhibition of cytokine expression [32, 33], secretion of TcpC and inhibition of MyD88 downstream signaling [34, 35] and an α‐hemolysin‐mediated inhibition of cytokine production [36, 37]. In a previous study from our group [38], we found that the α‐hemolysin‐positive strain CFT073 evoked a lower cytokine response from isolated human kidney epithelial cells than the α‐hemolysin‐negative strain ESBL019. Thus, the fact that the miRNA response was lower in CFT073 may suggest that this strain has a more immune subverting phenotype that can control and suppress expression of host cell miRNAs. At present, we cannot directly link specific bacterial genes to the induction of particular host miRNAs. However, the differential miRNA response elicited exclusively by the ESBL/ST131 strain suggests that strain‐specific virulence or immune‐evasion mechanisms may influence host miRNA regulation. This forms a strong rationale for future studies combining bacterial genomics, virulence factor mutants, and host miRNA profiling to dissect causality. Also, a broader validation using additional UPEC isolates is necessary to assess generalizability. It is not only the identity and virulence of the pathogens that regulate host miRNA, but also the cell type (e.g., monocytes/macrophages or epithelial cells) and potentially the extent of bacterial internalization [16, 39]. UPEC is mainly considered to be an extracellular pathogen, but UPEC may also invade bladder epithelial cells [40] and establish intracellular protected niches for colonization and proliferation. A limitation of the present study is that it is not known if there were any differences between CFT073‐ and ESBL019‐infected bladder cells regarding intracellular localization or whether the extent of bacterial internalization has any impact overall on the expression of host bladder miRNAs. Further investigations using gentamicin protection assays combined with intracellular bacterial quantification and miRNA profiling may address whether bacterial invasion influences miRNA release patterns.

The possible functional role of the released miRNA in host cells was addressed by analysis of human gene targets with the miRDB miRNA target prediction tool. MiRNAs often target multiple mRNAs simultaneously, and conversely, a single mRNA can be regulated by multiple miRNAs, creating complex networks for control of gene expression [14]. Hence, characterization of downstream miRNA targets may uncover and bring to light novel molecular factors and pathways relevant to host defense mechanisms and bacterial virulence in UTI. In our study, we identified 747 predicted interactions of the differentially expressed miRNA with human mRNA. Nectin1, SH3PXD2A and KIF21B, that encodes a cell adhesion protein [41], tyrosine kinase substrate 5 [42] and a kinesin protein [43], respectively were among the top targeted genes. However, none of these genes have, at least to our knowledge, been studied in the context of UTI pathogenesis and thus presents new knowledge for exploring the functional consequences of miRNA regulation in urinary tract infections. Pathway analysis of miRNA‐mRNA interactions from infected human bladder epithelial cells revealed most enrichment in p53, TGF‐β and oxidative stress signaling pathways. The pathway “Pathogenic E. coli infection” was, not surprisingly, also among the top enriched pathways. Analysis of miRNA‐enriched exosomes isolated from UPEC‐infected mouse bladder epithelial cells [19] revealed that the genes targeted by the most significantly enriched miRNAs were related to e.g., MAPK, ErbB and FoxO signaling pathway. Interestingly, our study also detected enrichment in the ErbB och FoxO signaling pathways. ErbB and FoxO signaling pathways are implicated in a broad range of cellular physiological events, and in the urinary tract these pathways have mainly been investigated related to molecular events underlying bladder cancer [44].

Beyond modulating host responses, we also investigated the potential for host miRNAs to directly target bacterial genes. miRNA can be transmitted from one cell to another even in a cross‐species manner [15], and this mode of cross‐species communication may regulate the pathogenic relationships between bacteria and their host [45]. Bacteria lack true miRNA [46], but bacterial infection may evoke a distinct miRNA synthesis in host cells with a subsequent transmission and interaction with bacterial mRNA. Indeed, experimental studies have reported that synthetic miRNA can penetrate the membrane of E. coli and regulate the expression of bacterial genes [15]. Our gene target analysis revealed that one third of the significantly expressed host cell miRNAs from cell supernatants interact with predictive mRNA targets in the genome of the multi‐resistant, ESBL‐producing UPEC strain EC958. Overall, only a small percentage (approximately 2%) of E. coli genes are predicted to be effectively targeted by human miRNA [47]. In our study, three bacterial genes, mukB, pyrB and cysI, were targeted by several of the nine human miRNAs. MukB is a structural maintenance of chromosome‐like protein that is required for DNA condensation in E. coli [48]. MukB inhibitors have been investigated as therapeutic agents to affect bacterial pathogenicity because of their ability to inhibit cell division and growth of E. coli [49]. PyrB encodes aspartate carbamoyltransferase of E. coli , the first enzyme of the pyrimidine biosynthetic pathway. Experimental starvation of the pyrimidine pool in E. coli revealed the importance of pyrimidine biosynthesis pathways in controlling virulence mechanisms, including biofilm formation and expression of type‐1 fimbriae [50]. Information on the role of the bacterial gene cysI in E. coli virulence mechanisms seems limited, at least to our knowledge. CysI encodes a component of the sulfite reductase complex that catalyzes electron reduction required for the biosynthesis of L‐cysteine in E. coli [51]. However, it is not yet known if the miRNAs derived from UPEC‐infected bladder cells in our study functionally affects the expression of the three identified bacterial target genes. Future studies may include testing whether exposure to miRNA mimics, i.e., synthetic RNA molecules designed to mimic the activity to endogenous miRNA, affects the expression of the identified genes. Moreover, assessing bacterial growth, biofilm formation and virulence traits following miRNA mimic exposure would allow direct functional validation of the computational predictions presented in the present study.

Rocket‐miR is a software tool designed to enable prediction of interactions of 630 human miRNA with various pathogens, with one application to identify miRNA‐based antimicrobial targets [47]. For E. coli , 22 human miRNAs are predicted to effectively target 10% or more of all genes in a non‐pathogenic E. coli K12 strain [47]. However, none of the 30 identified miRNAs in our study were represented in the Rocket‐miR group of 22 human miRNAs predicted to target 10% or more of all E. coli genes. The worldwide increase of ESBL‐producing E. coli , including UPEC, is mainly due to the spread of E. coli clone ST131 that within the last years has emerged as a major pandemic clone [52]. The reason for the successful spreading of the ST131 clone may be a unique balance between virulence and resistance mechanisms that favor the ability to cause persistent infections [53]. It would be of interest in the future to explore miRNA‐based therapy against virulence or resistance mechanisms in ESBL‐producing UPEC. Although our identified miRNAs did not overlap with those predicted by Rocket‐miR to broadly affect E. coli genes, several known virulence‐related genes were predicted as targets, suggesting selective therapeutic potential against UPEC. The identification of host miRNAs predicted to target immune and stress‐response pathways, as well as bacterial genes involved in replication and metabolism, suggests potential for miRNA‐based interventions. Approaches such as miRNA mimics or anti‐miRNA oligonucleotides [54] could, in principle, be used to modulate host responses or attenuate bacterial virulence without relying on antibiotics. The fact that one miRNA has several and diverse target genes may be an advantage for miRNA‐based antimicrobial drug strategies to limit tolerance development and restrict antimicrobial resistance mechanisms.

In the present study we used human spontaneously immortalized bladder epithelial cells, which arise from normal cells, instead of more commonly used bladder cancer epithelial cell lines. While in vitro epithelial models lack the full complexity of the urinary tract, the use of spontaneously immortalized non‐malignant human bladder epithelial cells provides a biologically relevant human system compared to cancer‐derived cell lines or non‐human models. However, isolated cells of course lack the complexity of the in vivo urinary tract, including immune and microbiota interactions, that mouse models offer. In addition, target predictions with high probabilities give a good measure of potential interactions of miRNA towards host and microbe genes, but they were computational and not experimentally validated, limiting functional interpretation. Thus, functional independent validations of the identified miRNAs and their predicted targets are needed in future studies to confirm the in silico target predictions.

In conclusion, this study highlights that infection of human bladder epithelial cells with an ESBL‐producing UPEC strain induces a distinct release of miRNAs predicted to target host pathways associated with immune regulation and stress responses and key bacterial genes involved in replication, metabolism, and virulence. Much remains to be learned about the roles of miRNA in interspecies interactions during infection before we can understand the significance of exogenous miRNA‐mediated cross‐species communication.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Differential expression of miRNAs between human bladder epithelial cells HBLAK infected with ESBL019, CFT073 and uninfected Control. Table S2: Predicted genes targeted by at least 3 of the 30 differentially expressed miRNAs in ESBL019 stimulated cells (miRNA target prediction: miRDB v6 and miRBase v22).