Yersiniabactin Siderophore of Crohn’s Disease-Associated Adherent-Invasive Escherichia coli Is Involved in Autophagy Activation in Host Cells

Background: Adherent-invasive Escherichia coli (AIEC) have been implicated in the etiology of Crohn’s disease. The AIEC reference strain LF82 possesses a pathogenicity island similar to the high pathogenicity island of Yersinia spp., which encodes the yersiniabactin siderophore required for iron uptake and growth of the bacteria in iron-restricted environment. Here, we investigated the role of yersiniabactin during AIEC infection. Methods: Intestinal epithelial T84 cells and CEABAC10 transgenic mice were infected with LF82 or its mutants deficient in yersiniabactin expression. Autophagy was assessed by Western blot analysis for p62 and LC3-II expression. Results: Loss of yersiniabactin decreased the growth of LF82 in competitive conditions, reducing the ability of LF82 to adhere to and invade T84 cells and to colonize the intestinal tract of CEABAC10 mice. However, yersiniabactin deficiency increased LF82 intracellular replication. Mechanistically, a functional yersiniabactin is necessary for LF82-induced expression of HIF-1α, which is implicated in autophagy activation in infected cells. Conclusion: Our study highlights a novel role for yersiniabactin siderophore in AIEC–host interaction. Indeed, yersiniabactin, which is an advantage for AIEC to growth in a competitive environment, could be a disadvantage for the bacteria as it activates autophagy, a key host defense mechanism, leading to bacterial clearance.


Introduction
Crohn's disease (CD) is a chronic, inflammatory disorder of the gastrointestinal tract caused by microbial and environmental factors in genetically susceptible hosts [1,2]. It has been observed that the intestinal mucosa of CD patients is abnormally colonized by specific Escherichia coli strains [3][4][5][6]. Characterization of these E. coli strains has showed that they are able to adhere to and to invade intestinal epithelial cells (IECs), as well as to survive and replicate within macrophages [3]. Thus, these strains have been designated as a new pathovar of E. coli and called adherent-invasive E. coli (AIEC) [3,7]. Once inside the cell, AIEC can proliferate and induce an inflammatory response [8,9]. Implication of AIEC in the etiology of CD has been then largely studied [10]. It has been revealed that AIEC are well adapted for the high nutrient competition with the commensal microbiota [11]. For example, bile salts activate many degradation pathways that promote the growth of AIEC, and this was not observed for commensal bacteria [11]. commensal E. coli strains isolated from healthy donors. We found a higher prevalence of PAI II in AIEC strains (67.1%, 49/73 strains tested) compared to commensal E. coli strains (45%, 18/40 strains tested) (* p = 0.022; Figure 1b). This result suggested that PAI II might confer an advantage for AIEC to efficiently colonize the intestinal tract of CD patients.

Loss of Functional Yersiniabactin Does Not Alter AIEC Morphology and Growth
As iron concentration controls numerous signaling pathways [30], we investigated whether PAI II could affect the morphology and growth of AIEC. To test our hypothesis, we constructed a genomic mutant of the AIEC reference strain LF82 that lacks the entire PAI II (LF82∆PAI II) and a mutant lacking YbtE (LF82∆YbtE), a gene required for yersiniabactin siderophore production [41]. The loss of the entire PAI II island or of the YbtE gene was verified using PAI II and YbtE internal primers, respectively (data not shown). Electronic microscopy was used to compare the morphology of wild-type LF82 with those of the mutants LF82∆PAI II and LF82∆YbtE. As shown in Figure 1c, the mutants LF82∆PAI II and LF82∆YbtE did not display any abnormalities in morphology compared with wild-type LF82. Furthermore, loss of PAI II did not alter the growth of LF82 bacteria either in LB or in DMEM/F12 used for IEC culture (Figure 1d,e). These results suggested that loss of PAI II does not affect AIEC LF82 morphology and growth.

Loss of Functional Yersiniabactin Impairs AIEC Growth and Colonization of IECs in Competitive Conditions
We next investigated the impact of the loss of the yersiniabactin siderophore of LF82 in competition experiments. As shown in Figure 2a, when the mutant LF82∆PAI II or LF82∆YbtE was mixed with the wild-type LF82 strain, their growth in LB was significantly decreased compared to that of the LF82 strain. Next, human intestinal epithelial T84 cells were infected with a combination of the same amounts of wild-type LF82 and LF82∆PAI II or LF82∆YbtE. After 3 h of infection, T84 cells were washed, and the number of associated bacteria, including adherent and invasive bacteria, was determined. To determine the number of invasive bacteria, after 3 h of infection, the cells were washed and incubated with gentamicin for 1 h to kill the extracellular adherent bacteria. As shown in Figure 2b, there was a decrease in the number of the mutant LF82∆PAI II or LF82∆YbtE associated with T84 cells compared to that of wild-type LF82. Furthermore, the number of the mutant LF82∆PAI II or LF82∆YbtE that invaded the cells was also decreased compared to that of wild-type LF82 (Figure 2c).
To validate these data, CEABAC10 transgenic mice were infected by gavage with a combination of the same amounts of wild-type LF82 and LF82∆PAI II. Bacterial colonization was assessed in the feces on days 2 and 3 post-infection and in the ileal tissues collected on the day of sacrifice (day 3 post-infection). We found higher numbers of wild-type LF82 compared to LF82∆PAI II, both in the feces and in the ileal tissue (Figure 2d,e). These results demonstrated that in a competitive environment, PAI II is necessary for AIEC growth and colonization of IECs.

Loss of Functional Yersiniabactin Increases AIEC Intracellular Proliferation and Decreases Autophagy in IECs
One of the particularities of AIEC is their ability to survive and to replicate inside host cells [8,9]. We next investigated whether the loss of functional yersiniabactin alters these properties. The number of intracellular LF82 at 12 h versus 4 h post-infection was determined using the gentamicin protection assay and considered as indicative of bacterial intracellular replication. We observed a significant increase in the replication of the mutants LF82∆PAI II and LF82∆YbtE compared to the wild-type LF82 strain in T84 cells (Figure 3a    As we have previously shown a crucial role for autophagy in the control of AIEC intracellular replication [21,[24][25][26][27], we next investigated whether autophagy is implicated in the increase of LF82 intracellular replication associated with the loss of yersiniabactin. Autophagy was assessed by Western blot analysis by evaluating the shift of LC3-I (the cytosolic form) toward LC3-II (the autophagosomal form and a marker of autophagy induction) and the level of p62, an autophagy receptor which is degraded by a functional autophagy [42]. As expected, autophagy was induced in T84 cells upon infection with the wild-type LF82 strain, as evidenced by increased LC3-II level compared with uninfected cells (Figure 3b). Furthermore, induction of a functional and degradative autophagy in LF82-infected cells was confirmed by a decrease in p62 level compared to uninfected cells (Figure 3c). Infection with the mutant LF82∆PAI II or LF82∆YbtE led to increased LC3-II level and decreased p62 level compared to uninfected cells, but with a less pronounced effect compared to infection with the wild-type LF82 strain (Figure 3b,c). These observations suggested a link between the presence of a functional yersiniabactin siderophore and the activation of autophagy in infected cells. The increase in LC3-II levels could result from the activation of autophagy or a blockage of autophagy flux, leading to the accumulation of undegraded LC3-II [42]. Thus, to examine whether the presence of yersiniabactin siderophore led to autophagy activation or autophagy flux blockage, we treated the cells with the inhibitor of autophagy flux bafilomycin A1 before infection. Under autophagy flux inhibition, we still observed the increase in LC3-II level upon infection with the wild-type LF82 strain, and a lower increase in LC3-II level was observed upon infection with the mutant LF82∆PAI II or LF82∆YbtE (Figure 3d). This demonstrated that autophagy was induced in IECs upon LF82 infection and this was dependent, at least in part, on the presence of yersiniabactin siderophore.

Loss of Yersiniabactin Inhibits AIEC-Induced HIF-1α Expression and Activation in IECs
It was reported that during infection with human pathogenic bacteria, bacterial siderophores are implicated in hypoxia-independent activation of the transcription factor HIF-1α [43]. As shown in Figure 4a, infection with the wild-type LF82 strain induced HIF-1α expression in T84 cells, and this was not observed for infection with the mutants LF82∆PAI II or LF82∆YbtE. Furthermore, LF82 infection induced an increase in the mRNA expression level of VEGF, one of the well-known target of HIF-1α [44], but infection with LF82∆PAI II or LF82∆YbtE failed to do so ( Figure 4b). These results suggested that yersiniabactin of AIEC is crucial for expression and activation of HIF-1α.

AIEC Induced HIF-1α Expression by Chelating Iron Rather Than by Inducing Hypoxia
It is known that HIF-1α can be activated by either hypoxia or iron deprivation [45,46]. In order to elucidate the mechanism underlying LF82-induced HIF-1α expression, T84 cells were grown on regular plastic or gas-permeable dishes. We observed a strong expression of HIF-1α upon LF82 infection in both conventional and gas-permeable cell culture dishes (Figure 5a), indicating that hypoxia was not involved in LF82-induced HIF-1α expression. To assess whether iron deprivation was responsible for LF82-induced HIF-1α expression, T84 cells were infected with LF82 in the presence or absence of FeCl 3 . As shown in Figure 5b, the presence of iron strongly diminished LF82-induced HIF-1α expression. Together, these results indicated that AIEC LF82 activated HIF-1α expression by chelating iron.

HIF-1α Controls AIEC Intracellular Replication by Modulating Host Autophagy
Finally, we sought to investigate the functional impact of AIEC-induced HIF-1α expression. To analyze the role of HIF-1α in AIEC-induced autophagy, before infection, T84 cells were transfected with an siRNA directed against HIF-1α mRNA or a scramble siRNA. As shown in Figure 6a, in cells transfected with the scramble siRNA, infection with LF82 induced HIF-1α expression and autophagy, as shown by the increased LC3-II level. HIF-1α siRNA inhibited LF82-induced HIF-1α expression and subsequently suppressed autophagy activation (Figure 6a). The decrease in autophagy activation in HIF-1α-siRNA-transfected cells was correlated with an increase in LF82 intracellular replication (Figure 6b). Together, our results suggest that HIF-1α controls AIEC LF82 intracellular replication by modulating host autophagy.
derophores are implicated in hypoxia-independent activation of the transcrip HIF-1α [43]. As shown in Figure 4a, infection with the wild-type LF82 stra HIF-1α expression in T84 cells, and this was not observed for infection with t LF82ΔPAI II or LF82ΔYbtE. Furthermore, LF82 infection induced an incre mRNA expression level of VEGF, one of the well-known target of HIF-1α [44] tion with LF82ΔPAI II or LF82ΔYbtE failed to do so (Figure 4b). These results that yersiniabactin of AIEC is crucial for expression and activation of HIF-1α.   As shown in Figure 5b, the presence of iron strongly dimin expression. Together, these results indicated that AIEC LF82 by chelating iron.

Discussion
In the present study, we observed that commensal E. coli could harbor the Yersinia HPI encoding yersiniabactin, confirming previously reported data [34]. Interestingly, we found that the yersiniabactin siderophore was over-represented in CD-associated AIEC compared to commensal E. coli. This result confirmed previous data showing increased prevalence of yersiniabactin iron transport genes in AIEC strains (presence in 6/8 strains) relative to non-pathogenic E. coli strains (presence in 1/12 strains) [40]. In the current study, our data, using 73 AIEC and 40 commensal E. coli strains, strongly support the overrepresentation of yersiniabactin siderophore-encoding genes among the AIEC strains, suggesting that this siderophore could be implicated in the pathogenic properties of the bacteria.
Given the overrepresentation of yersiniabactin among the AIEC strains and that CD patients suffer from chronic iron deficiency [47], we can speculate that yersiniabactin might confer the bacteria an advantage in the "iron competition" inside the intestinal tract. Indeed, we showed that the loss of yersiniabactin, while it did not affect the growth of AIEC LF82 in culture media, was a disadvantage for the bacteria when they were cultured with the wildtype LF82 strain having yersiniabactin. This supported the hypothesis that in a complex and hyper-competitive environment, such as the gastrointestinal tract, this siderophore could be crucial for AIEC colonization. This was confirmed by in vitro and in vivo experiments showing that the loss of yersiniabactin led to a decrease in the number of AIEC that adhered to and invaded IECs and to the reduction of AIEC colonization in the gut of CEABAC10 transgenic mice. This is in accordance with previous data showing that in Enterobacteriaceae, HPIs contribute to fitness and virulence of the bacteria in vivo [43,48,49]. Recently, the deleterious effect of yersiniabactin of the mouse AIEC strain NC101 was demonstrated [50]. Indeed, in germ-free il10 −/− mice, this siderophore promoted a profibrogenic host response, suggesting a direct link between intestinal E. coli and the induction of inflammationassociated fibrosis [50].
Y. enterocolitica has been shown to activate the transcriptional factor HIF-1α in vivo [43]. This activation is protective, since hif1a −/− mice are more susceptible to Y. enterocolitica infection than wild-type mice. Furthermore, yesrsiniabactin derived from Y. enterocolitica activates HIF-1α in vitro [43]. In our study, we also observed an intestinal epithelial response to yersiniabactin leading to HIF-1α activation. We observed that the presence of this siderophore in AIEC activated HIF-1α, leading to autophagy activation and consequently to a better control of AIEC intracellular replication. We have previously shown that autophagy plays a crucial role in the interaction between AIEC and the host. Upon AIEC infection, autophagy is induced in host cells via the activation of the metabolic stress response pathway EIF2AK4-EIF2A-ATF4 [24]. However, AIEC can subvert autophagy by up-regulating the levels of specific microRNAs that target autophagy and inhibit this process [26]. AIEC also inhibit autophagy by impairing host SUMOylation, a eukaryoticreversible post-translational modification [27]. These mechanisms consequently lead to increases in AIEC intracellular replication and AIEC-induced inflammation in vitro and in vivo [24][25][26][27]. However, in our previous reports, the bacterial virulence factor responsible for these effects was not explored. Here, we demonstrated that a functional siderophore yersiniabactin is implicated in the activation of autophagy upon infection with AIEC. Thus, a pathogenicity island could be, at some point, a disadvantage for the bacteria, since its presence may "alarm" the host defense mechanisms, such as autophagy.
In conclusion, our study highlights the complex crosstalk between AIEC and the host and the dual effect of PAI II-encoded yersiniabactin. We showed that yersiniabactin, which confers the advantage to AIEC to grow in a complex environment and gain access to competitive ecological niches, could be, at some point, a disadvantage when it activates signaling pathways leading to bacterial clearance, such as autophagy (Figure 7).

Autophagy
Bacterial clearance Wild-type AIEC

HIF-1α
Decreased bacterial clearance HIF-1α Figure 7. Hypothetical model for the involvement of PAI II in the interaction of AIEC with intestinal epithelial cells. Left panel: PAI II confers AIEC a high ability to compete with other bacteria for growth and adhesion to IECs. AIEC PAI II-encoded yersiniabactin activates autophagy via a HIF-1α-dependent mechanism, leading to bacterial clearance. Right panel: AIEC invalidated for PAI II (AIECΔPAI II) have a lower ability to compete with other bacteria for adhesion to intestinal epithelial cells. Autophagy in IECs infected with AIECΔPAI II is reduced, leading to impaired bacterial clearance, compared to that in IECs infected with wild-type AIEC. Thus, the presence of yersiniabactin, which is an advantage for AIEC to grow in a competitive environment, could be a disadvantage for the bacteria, as it activates autophagy-mediated bacterial clearance.

Bacterial Strains
The AIEC reference strain LF82 was isolated from the ileal mucosa of a CD patient [7]. Bacteria were grown in Luria-Bertani (LB) broth or on LB agar plates overnight at 37 °C.

Screening for the Presence of PAI II by PCR
Bacteria were grown in LB overnight, then boiled, and PCR was performed using GoTaq DNA Polymerase (Promega, Charbonnières-les-Bains, France) following the manufacturer's instructions. Two different sets of primers, each amplifying a different PAI II gene, were used.

Construction of AIEC LF82 Chromosomal Mutant Invalidated for PAI I
The entire PAI II was deleted in the AIEC LF82 strain by homologous recombination, as previously described [51,52]. Briefly, 540 bp of the LF82_p294 gene (the first gene of the PAI II) and 607 bp of the LF82_p308 gene (the last gene of the PAI II) were amplified by PCR using the following primers: • LF82_p294Forward: 5′-GGGCCCGAATTCGCTAAGTAATATGCGCCCCG-3′; LF82_p294Reverse: 5′-CGATGTTTCGGAGTCAGTCGCGGATACCTTCACGTTGCTG-3′; PAI II confers AIEC a high ability to compete with other bacteria for growth and adhesion to IECs. AIEC PAI II-encoded yersiniabactin activates autophagy via a HIF-1α-dependent mechanism, leading to bacterial clearance. Right panel: AIEC invalidated for PAI II (AIEC∆PAI II) have a lower ability to compete with other bacteria for adhesion to intestinal epithelial cells. Autophagy in IECs infected with AIEC∆PAI II is reduced, leading to impaired bacterial clearance, compared to that in IECs infected with wild-type AIEC. Thus, the presence of yersiniabactin, which is an advantage for AIEC to grow in a competitive environment, could be a disadvantage for the bacteria, as it activates autophagy-mediated bacterial clearance.

Bacterial Strains
The AIEC reference strain LF82 was isolated from the ileal mucosa of a CD patient [7]. Bacteria were grown in Luria-Bertani (LB) broth or on LB agar plates overnight at 37 • C.

Screening for the Presence of PAI II by PCR
Bacteria were grown in LB overnight, then boiled, and PCR was performed using GoTaq DNA Polymerase (Promega, Charbonnières-les-Bains, France) following the manufacturer's instructions. Two different sets of primers, each amplifying a different PAI II gene, were used.

Construction of AIEC LF82 Chromosomal Mutant Invalidated for PAI I
The entire PAI II was deleted in the AIEC LF82 strain by homologous recombination, as previously described [51,52]. Briefly, 540 bp of the LF82_p294 gene (the first gene of the PAI II) and 607 bp of the LF82_p308 gene (the last gene of the PAI II) were amplified by PCR using the following primers:
The two PCR fragments were assembled by PCR and cloned into the pST76 plasmid using BamHI and EcoRI restriction sites. This vector possesses a thermosensible Ori and an I-SceI restriction site. By increasing the temperature, we forced pST76 insertion inside the AIEC LF82 chromosome. This insertion involved a homologous recombination between PAI II and sequences cloned inside the vector. Bacteria were then transformed by an I-SceI enzyme-producing pDAI vector. Cleavage of the I-SceI site inside pST76 can induced homologous recombination, resulting in the loss of PAI II (AIEC∆PAI II). Deletion of PAI II was verified by PCR using the primers located inside the island: For YbtE gene deletion, we used the method described by Datsenko et al. [53] and modified by Chaveroche et al. [54], comprising the replacement of the YbtE gene by a selective antibiotic cassette (kanamycin) generated by PCR using the following primers: Briefly, LF82 was transformed with pKOBEG, a plasmid encoding the Red proteins that protect linear DNA from degradation in bacteria. The plasmid was maintained in bacteria at 30 • C with 25 µg/mL of chloramphenicol, and Red protein expression was induced using 1 mM of L-arabinose (Sigma-Aldrich, Saint-Quentin Fallavier, France). Then, the YbtEKana PCR product was electroporated in previously refrigerated glycerol-washed LF82. The resulting LF82∆YbtE isogenic mutant (Km R ) was selected on LB containing 50 µg/mL of kanamycin. The replacement of the YbtE gene by the kanamycin resistance cassette was confirmed by PCR.

Cell Culture
The intestinal epithelial cell line T84 (CCL-248; ATCC, Molsheim, France) was maintained in an atmosphere containing 5% CO 2 at 37 • C in the culture medium recommended by ATCC (Molsheim, France). T84 cells were cultivated using conventional polystyrene dishes or special gas-permeable dishes with a hydrophilic tissue culture-treated bottom membrane (Lumox; Sarstedt, Marnay, France) as previously used [43].

Infection and Gentamicin Protection Assay
T84 cells were seeded on 12-or 24-well plates and infected at a multiplicity of infection (MOI) of 10 bacteria per cell. Gentamicin protection assay was performed as previously described [26]. Briefly, after 3 h of incubation with the bacteria in the culture medium without antibiotics, the cells were washed with PBS and incubated with the culture medium containing 100 µg/mL of gentamicin for 1 h, 3 h, or 9 h. When indicated, the cells were pre-treated with Bafilomycin A1 (Sigma-Aldrich, Saint-Quentin Fallavier, France) at 50 nM for 30 min before infection, and then Bafilomycin A1 was maintained in the medium during infection. The cells were lysed with 1% Triton X-100 (Sigma-Aldrich, Saint-Quentin Fallavier, France) in deionized water. Samples were serially diluted and plated onto LB agar plates, and the number of bacteria was determined by counting the colony-forming units (CFU).
In the competitive experiment, T84 cells were infected with 10 wild-type LF82 bacteria and 10 mutant bacteria (LF82∆PAI II or LF82∆YbtE) per cell as described above.

Quantitative Real-Time RT-PCR (qRT-PCR)
Total RNAs were isolated using the TRIzol reagent (Invitrogen, Illkirch, France) following the manufacturer's instruction. In this experiment, 2 µg of total RNA were reversely transcribed using the first-strand cDNA synthesis kit (Euromedex, Souffelweyersheim, France) to quantify mRNA expression levels. Quantitative (q)RT-PCR was performed using SYBR Green Master Mix (Euromedex, Souffelweyersheim, France) on a Mastercycler Realplex 4 (Eppendorf, Montesson, France) using the following primers: RPLP0 was used as a house-keeping gene. Fold-induction was calculated using the comparative threshold cycle number (Ct) method as follows: ∆∆Ct = (Ct VEGF − Ct RPLP0 ) test condition − (Ct VEGF − Ct RPLP0 ) control condition , and the final data were derived from 2 −∆∆Ct .

Transmission Electron Microscopy
Bacteria were grown overnight at 37 • C in LB without shaking, placed for 1 min on carbon-Formvar copper grids (Electron Microscopy Sciences, Hatfield, PA, USA), and negatively stained for 1 min with phosphotungstic acid pH 6.0. Grids were examined with a Hitachi H-7650 transmission electron microscope (Hitachi, Krefeld, Germany).

Competition in Vitro
Bacteria (LF82, LF82∆PAI II and LF82∆YbtE) were grown overnight in LB at 37 • C. In total, 100 bacteria of wild-type LF82 and the mutant were mixed in 10 mL of LB and then incubated at 37 • C. At the indicated time, serial dilutions were spread on LB agar plates with or without antibiotics (to distinguish the different bacterial strains). All the inocula were spread to verify the quantity of bacteria mixed.

Competition In Vivo
Experiments were performed using 6-week-old CEABAC10 transgenic male mice housed in the specific pathogen-free animal facility at the University Clermont Auvergne. Mice were fed standard chow ad libitum throughout the experiments, had free access to sterile water, and were subjected to 12-12 light/dark cycles. They received oral gavage of a 200 µL suspension containing the same amounts of wild-type LF82 and LF82∆PAI II (10 9 CFU each). Feces and tissues were collected at the indicated times and crushed, and serial dilutions were plated either on LB agar plates containing ampicillin (100 µg/mL) and erythromycin (25 µg/mL) to detect LF82 or on LB agar plates containing ampicillin (100 µg/mL), erythromycin (25 µg/mL), and rifampicin (100 µg/mL) to detect AIEC∆PAI II.

Statistical Analysis
Values were expressed as means ± SEM. Statistical analysis between two groups was performed with GraphPad Prism version 5.01 software (San Diego, CA, USA) using a two-tailed Student's t-test analysis or a non-parametric Mann-Whitney test depending on D'Agostino-Pearson omnibus normality test. Statistical analysis between more than two groups was performed using one-way ANOVA, followed by a Bonferroni post-test. To analyze the prevalence of PAI II between AIEC strains and commensal E. coli strains, a chi-square test was used. A p value less than 0.05 was considered statistically significant.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.