Subtilase cytotoxin induces a novel form of Lipocalin 2, which promotes Shiga-toxigenic Escherichia coli survival

Shiga-toxigenic Escherichia coli (STEC) infection causes severe bloody diarrhea, renal failure, and hemolytic uremic syndrome. Recent studies showed global increases in Locus for Enterocyte Effacement (LEE)-negative STEC infection. Some LEE-negative STEC produce Subtilase cytotoxin (SubAB), which cleaves endoplasmic reticulum (ER) chaperone protein BiP, inducing ER stress and apoptotic cell death. In this study, we report that SubAB induces expression of a novel form of Lipocalin-2 (LCN2), and describe its biological activity and effects on apoptotic cell death. SubAB induced expression of a novel LCN2, which was regulated by PRKR-like endoplasmic reticulum kinase via the C/EBP homologous protein pathway. SubAB-induced novel-sized LCN2 was not secreted into the culture supernatant. Increased intracellular iron level by addition of holo-transferrin or FeCl3 suppressed SubAB-induced PARP cleavage. Normal-sized FLAG-tagged LCN2 suppressed STEC growth, but this effect was not seen in the presence of SubAB- or tunicamycin-induced unglycosylated FLAG-tagged LCN2. Our study demonstrates that SubAB-induced novel-sized LCN2 does not have anti-STEC activity, suggesting that SubAB plays a crucial role in the survival of LEE-negative STEC as well as inducing apoptosis of the host cells.


Results
SubAB induces lipocalin 2 (LCN2) expression. To understand the effect of STEC O113:H21-produced SubAB on defense factors generated by host cells, we focused on LCN2, a protein that was induced by ER stress 38 and acts as an anti-microbial defense factor by binding to a subset of bacterial siderophores 25 . In this study, we primarily used HeLa cells mainly as in our previous our reports 10,12,14,22 . Both Caco2 cells and HCT116 cells were difficult to analyze for SubAB-induced apoptosis. To assess the influence of SubAB on the production of LCN2 in HeLa cells, we used real-time quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) and Western blotting. Transcription of lcn2 mRNA was significantly increased by purified wild-type (wt) SubAB compared to catalytically inactivated mutant (mt) SubAB. PERK (RNA-dependent protein kinase (PKR)-like ER kinase), a key ER stress sensor of the unfolded protein response, is responsible for SubAB-induced apoptosis 14 . SubAB-increased lcn2 mRNA expression was suppressed in PERK-knockdown cells (Fig. 1A).
We detected wt SubAB-enhanced PARP cleavage (cPARP), and CHOP and LCN2 protein expression, which were not seen in PERK-knockdown cells (Fig. 1B). Further, to show that BiP cleavage by SubAB is essential for LCN2 expression, we investigated the effect of Brefeldin A (BFA), which inhibits retrograde transport by disrupting the Golgi apparatus, on SubAB-increased LCN2. BFA is known to inhibit SubAB-mediated BiP cleavage 7 . HeLa cells were pretreated with or without 10 mM BFA for 30 min, and then SubAB was added to cells. After a 24 h incubation, LCN2 expression was detected by Western blotting analysis (Fig. 1C). SubAB-increased LCN2 was significantly suppressed in the presence of BFA. BFA alone slightly increased low molecular weight of LCN2 expression compared to SubAB-increased LCN2, suggesting that BFA may affect ER stress signaling pathway by disrupting the Golgi apparatus.
We next investigated whether STEC O113:H21-secreted SubAB participated in the regulation of LCN2 expression using a co-culture system (Fig. 1D). To show that SubAB is a critical inducer of LCN2, HeLa cells were incubated with three types of STEC O113:H21 (e.g., wild-type, subAB-knockout, stx2-knockout). After a 24 h co-culture of HeLa cells with STEC O113:H21 strains, bacteria were collected by centrifugation; expression of Transcription factor C/EBP homologous protein (CHOP) is involved in SubAB-induced LCN2 expression. CHOP, which is known as an ER-stress marker, plays an essential role in cell cycle arrest and apoptosis, and also acts as a transcriptional regulator 39 . A previous study reported that LCN2 is a CHOP target gene that mediates ER stress-induced apoptosis in A549 cells 40 . We investigated whether SubAB-increased CHOP is involved in LCN2 expression in HeLa cells. When cells were co-cultured with the three types of O113:H21 strains as shown in Fig. 1D, the level of chop mRNA was increased in wild-type and stx2-deficient O113:H21 co-cultured with HeLa cells (Fig. 2A). Further, we examined whether CHOP is a crucial inducer of LCN2 expression; CHOP-knockdown cells were incubated for 24 h with mt or wt SubAB. We found that . The mRNA levels of lcn2 was measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells. (B) Cell lysates were subjected to immunoblotting with the indicated antibodies. GAPDH served as a loading control. Quantification of LCN2 in HeLa cells was performed by densitometry. Data are presented as mean ± SD of values from three independent experiments and significance is *P < 0.05. (C) HeLa cells were preincubated for 30 min with 10 mM BFA, and then incubated for 24 h with control PBS (−), 400 ng ml −1 of mt or wt SubAB in the presence or absence of the BFA. Cell lysates were subjected to immunoblotting with the indicated antibodies. GAPDH served as a loading control. (D) Schematic drawing of co-culture system. Confluent HeLa cells were plated on apical side (Apical), which has a semipermeable membrane. The wild-type, ∆subAB, or ∆stx2 STEC O113:H21 strain (1-2.5 × 10 5 cfu) was plated on the basolateral side (Baso) and the system cultured for 24 h. (E) HeLa cells were lysed with 1xSDS sample buffer for immunoblotting with the indicated antibodies. After centrifugation of STEC culture medium on the basolateral side, bacterial body (BD) or culture supernatant (sup) was collected and then lysed with 1xSDS sample buffer for immunoblotting with the indicated antibodies. GAPDH or RNAPα was used as an internal control. (F) HeLa cells were co-cultured for 24 h with the indicated STEC strains as shown in (D). The lcn2 mRNA levels were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells.
Overexpression of CHOP has been demonstrated to cause cell cycle arrest and/or apoptosis 41 . We next examined the effect of overexpressed CHOP on SubAB-induced LCN2. Cells transfected with FLAG-tagged CHOP expression plasmid were incubated for 24 h with mt or wt SubAB. SubAB-stimulated lcn2 mRNA was not altered in CHOP-overexpressing cells (Fig. 3A). We also tested whether SubAB-stimulated CHOP and overexpressed CHOP localized to the nucleus. CHOP was mainly localized in the nucleus ( Supplementary Fig. S3). In CHOPoverexpressing cells, SubAB-induced PARP cleavage was significantly increased after 8 h incubation, but LCN2 was not detected. After a 24 h incubation, SubAB-induced PARP cleavage and LCN2 expression were found at similar levels in control and CHOP-expressing cells (Fig. 3B). These findings suggest that SubAB-induced  Fig. 1D. The chop mRNA levels were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells. (B) Control (NC) or CHOP siRNAtransfected cells were incubated for 24 h with mt SubAB or SubAB (400 ng ml −1 ). The lcn2 mRNA levels were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells. (C) Control (NC) or CHOP siRNA-transfected cells were incubated for 24 h with mt or wt SubAB (400 ng ml −1 ). Cell lysates were subjected to immunoblotting with the indicated antibodies. GAPDH served as a loading control. (D) Quantification of cPARP or LCN2 in the transfected cells with mt or wt SubAB was performed by densitometry. Data are presented as mean ± SD of values from three independent experiments and significance is *P < 0.05. (E) Cells were incubated for 24 h with mt or wt SubAB (400 ng ml −1 ). The mRNA levels of cebpA, cebpB, or cebpG were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). * P < 0.05, versus mt SubAB-treated control cells. (F) The indicated siRNA-transfected cells were incubated for 24 h with mt or wt SubAB (400 ng ml −1 ). The lcn2 mRNA levels were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells. (G) Control (NC), C/EBPB, or C/EBPG siRNA-transfected cells were incubated for 24 h with mt or wt SubAB (400 ng ml −1 ). Cell lysates were subjected to immunoblotting with anti-LCN2 antibodies. GAPDH served as a loading control.

LCN2 participates in SubAB-induced apoptosis.
Several studies have shown that LCN2 plays a crucial role as a protector or enhancer of stress-induced apoptosis in various tissues or cells [42][43][44][45] . We investigated next whether LCN2 participated in SubAB-induced apoptosis. We assessed the efficiency of LCN2 siRNA by RT-qPCR. lcn2 mRNA expression was not increased by SubAB in LCN2-knockdown cells. The catalytically inactive SubAB (mt SubAB) had no effect (Fig. 4A). We investigated if knockdown of LCN2 by siRNA affected SubAB-induced apoptosis. In LCN2-knockdown cells, SubAB-induced LCN2 expression was suppressed, and PARP cleavage, a hallmark of apoptosis, was significantly enhanced compared to control siRNA-transfected cells (Fig. 4B). Previous studies reported that LCN2 attenuated NF-κB subunit p65 activation under hypoxic conditions 46 and regulated macrophage polarization and activation of NF-kB/STAT3 signaling 47 . We next investigated the effect of LCN2 knockdown on chop mRNA expression by RT-qPCR. SubAB-induced chop mRNA expression was decreased in LCN2-knockdown cells (Fig. 4C). Consistent with the mRNA expression, SubABinduced CHOP was significantly attenuated in LCN2-knockdown cells by Western blot analysis (Fig. 4D). Next, we tested the effect of LCN2 overexpression on SubAB-induced CHOP expression (Fig. 5A). We used tunicamycin (TM), a chemical ER stress inducer and N-glycosylation inhibitor, as a positive control 48 . After transfection for 48 h with control or FLAG-tagged LCN2 plasmid, cells were incubated for 24 h with TM, or mt or wt SubAB. CHOP expression by TM or wt SubAB was not different between control and FLAG-tagged . The lcn2 mRNA levels were measured by RT-qPCR as described in "Methods". GAPDH was used as an internal control. Data are mean ± SD (n = 3). *P < 0.05, versus mt SubAB-treated control cells. Ns: not significant. (B) Cell lysates from 8 h or 24 h incubation with toxins were subjected to immunoblotting with the indicated antibodies. GAPDH served as a loading control. Experiments were repeated three times with similar results. ND: not detected.

Scientific Reports
| (2020) 10:18943 | https://doi.org/10.1038/s41598-020-76027-z www.nature.com/scientificreports/ LCN2-transfected cells. Interestingly, we observed three different molecular sizes of LCN2 in the presence of TM or wt SubAB in FLAG-tagged LCN2-overexpressing cells. Thus, TM caused a shift to a lower molecular weight form of unglycosylated LCN2 (LCN2**), while SubAB-induced LCN2 was observed in two bands, which consist of normal size and a slightly lower molecular size LCN2 (LCN2*). Since LCN2 is a secreted protein 49 , we investigated whether the stimulated LCN2 was secreted into the culture supernatant. Although the normal size of FLAG-tagged LCN2 was detected in the culture supernatant, SubAB-induced low molecular weight LCN2 (LCN2*) was not observed. Of note, TM-induced FLAG-tagged unglycosylated LCN2 (LCN2**) was detected in the culture supernatant (Fig. 5B). Since overexpression of LCN2 was protective from inflammation-associated cell death by LPS 50 , we next examined the effect of LCN2 overexpression on SubAB-induced apoptosis (Fig. 5C).
In FLAG-tagged LCN2 transiently transfected cells, SubAB-induced PARP cleavage in control cells was not significantly different from that seen in LCN2-overexpressing cells after a 24 h incubation. We next investigated the localization of SubAB-increased novel FLAG-LCN2 by immunostaining. In mt SubAB-treated cells, FLAG-tagged LCN2 was observed in the cytosolic compartment and partially co-localized with ER marker, protein disulfide isomerase (PDI). However, SubAB-induced novel LCN2 (LCN2*) was also mainly localized in the cytosolic compartment after an 8 h incubation, and distributed at cell membrane after a 24 h incubation (Fig. 5D, Supplementary Fig. S4). We tested the protective effects of extracellular LCN2 on SubAB-induced cell death. Cells were transfected with siRNA for LCN2, and then incubated for 24 h with SubAB in the presence or absence of purified recombinant human LCN2 (rLCN2). SubAB-induced PARP cleavage was enhanced in LCN2-knockdown cells, which was not altered even in the presence of purified LCN2 (Fig. 5E). These results www.nature.com/scientificreports/ suggest that suppression of LCN2 or excess presence of rLCN2 in cells promotes an ER stress-induced apoptotic pathway. In addition, SubAB-increased novel LCN2 (LCN2*), was not secreted into culture supernatant, but was accumulated at the cell membrane. Our next question is whether intracellular iron level affects SubAB-induced cell death signaling. We tested whether SubAB-induced PARP cleavage was prevented by delivery of iron by using holo-transferrin (holo-Tf) or ferric chloride (FeCl 3 ). The results of Fig. 5F show that addition of either holo-Tf or FeCl 3 suppressed SubAB-induced PARP cleavage. The iron chelator deferoxamine (DF) had no effect on SubAB-induced PARP cleavage after 12 h incubation. These findings suggest that the SubAB-stimulated apoptotic pathway was suppressed by increased intracellular iron concentration.
LCN2 suppressed STEC O113:H21 growth. LCN2 secreted from host cells captures bacterial siderophores and inhibits iron reuptake into bacteria 51 . Further, LCN2 inhibited growth of a nonpathogenic clinical isolate of E. coli and protected mammals from E. coli infection 29,52,53 . Here, we investigated the effect of ER stressinduced LCN2 on growth of STEC O113:H21, a pathogenic clinical isolate. We collected the supernatants from control or FLAG-tagged LCN2-transfected cells in the presence or absence of SubAB or TM, incubated STEC O113:H21 with supernatants, and then measured bacterial growth at the indicated time points. At 24 h incubation, cultured STEC O113:H21 was centrifuged to separate bacteria and supernatant; FLAG-tagged LCN2 was detected in the supernatant. As shown in Fig. 5B, SubAB-induced FLAG-tagged normal-sized LCN2 (LCN2) was barely detectable in the supernatant compared to control or mt SubAB-treated cells. TM-induced unglycosylated LCN2 (LCN2**) was clearly detected. (Fig. 6A). The supernatants from control plasmid-transfected cells did not prevent STEC growth, even in the presence of SubAB or TM. However, STEC O113:H21 growth was suppressed with the supernatant containing normal-sized FLAG-tagged LCN2 from untreated or mt SubAB. However, the growth was not inhibited with TM-and wt SubAB-treated supernatant from FLAG-tagged LCN2-transfected cells (Fig. 6B). We next investigated the effect of purified recombinant human LCN2 (rLCN2) on STEC growth. As shown in Fig. 6C, the growth of all STEC O113:H21 strains (e.g., wild-type, ΔsubAB, Δstx2) was suppressed in the presence of rLCN2. We also investigated the effect of rLCN2 on STEC O113:H21-produced toxins; rLCN2 did not affect toxin transcription and production (Fig. 6D, Supplementary Fig. S6). These findings indicate that

SubAB induces mouse LCN2 in RAW264.7 cells and mouse intestinal cells. The lcn2 gene in
the human genome exhibited only 62 percent similarity with the mouse orthologue 54 . To assess if mouse LCN2 (mLCN2) was also induced by SubAB, we investigated whether SubAB induces mLCN2 in RAW264.7 cells by RT-qPCR. SubAB-increased mouse lcn2 (m lcn2) mRNA expression was observed after a 24 h incubation (Fig. 7A). We next cloned the mLCN2 gene from RAW264.7 cells and then constructed a FLAG-tagged mLCN2 expression vector. After HeLa cells were transfected with control, FLAG-tagged human LCN2 (hLCN2), or FLAG-tagged mLCN2 for 24 h, cells were treated for 24 h with TM, mt or wt SubAB. The molecular size of FLAG-tagged mLCN2 in control and in mt SubAB-treated cells was slightly lower compared to FLAG-tagged hLCN2. FLAG-tagged mLCN2 was also significantly shifted to low molecular weight by TM, and SubAB-caused a novel FLAG-tagged mLCN2 as well as FLAG-tagged hLCN2 (Fig. 7B). We next investigated whether SubAB induces expression of lcn2 mRNA in SubAB-injected mouse intraperitoneally. Mouse duodenum, ileum and colon were isolated after SubAB intraperitoneal injection for 0, 24 and 48 h, and then mRNA was purified. The mRNA level of mouse lcn2 was measured by RT-qPCR. SubAB-stimulated mouse lcn2 was increased in duodenum, ileum and colon after a 24 h injection (Fig. 7C).

Discussion
Herein we found that STEC O113:H21-produced SubAB, not Stx2, increased LCN2 expression. SubAB-induced BiP cleavage activates PERK, and leads to phosphorylation of eukaryotic translation initiation factor 2α (eIF2α), followed by increased ATF4 and CHOP transcription 13,14 . CHOP plays an important role in ER stress-induced apoptosis 55 , with overexpression of CHOP causing cell cycle arrest and/or apoptosis 41 . In agreement, knockdown www.nature.com/scientificreports/ of CHOP reduced TM-induced apoptosis in HCC cells 56 . Overexpression of FLAG-tagged CHOP in HeLa cells enhanced SubAB-induced PARP cleavage after an 8 h incubation, not after a 24 h incubation. Interestingly, CHOP knockdown increased SubAB-induced PARP cleavage at 24 h. Consistent with our results, cigarette smoke extract-induced apoptosis was increased in CHOP-knockdown cells 57 . These findings indicate that the change in the amount of CHOP may be an important factor regulating apoptosis. Knockdown of PERK and CHOP, but not ATF4, failed to induce LCN2, suggesting that the PERK/CHOP signaling pathway regulates LCN2 expression during ER stress. Analysis of LCN2 promoter region showed that the presence of transcriptional factor-binding sites (e.g., estrogen response element, NF-κB, C/EBPG, Vitamin D receptor, Stat1, Stat3, glucocorticoid response element). One previous study reported that LCN2 is a CHOP target gene, which mediates ER stress-induced apoptosis in A549 cells 40 . They suggested that thapsigargin-triggered ER stress-induced LCN2 required both CHOP and C/EBP binding to the LCN2 promoter. Heterodimer formation between CHOP and C/EBP proteins or activating transcription factor 4 (ATF4) is important in regulating transcriptional activity 58 . We could not detect SubAB-increased CHOP directly binding to LCN2 promoter by Chromatin Immunoprecipitation (ChIP) assay. Knockdown of C/EBPB and C/EBPG reduced SubAB-induced LCN2 and CHOP. LCN2 was not increased by CHOP overexpression alone. Taken together, our results suggest that the SubAB-increased C/EBPB or C/EBPG leads to CHOP increase, followed by formation of a complex with CHOP and C/EBP to stimulate LCN2 expression.
Previous study showed that tunicamycin (TM), N-glycosylation inhibitor and ER stress inducer, induced expression of an unglycosylated LCN2, which had no effect of its secretion into culture supernatant 48 . Consistent with the report, tunicamycin treatment induced shift of LCN2 to a smaller size (LCN2** in Fig. 5A), which was secreted into supernatant (Fig. 5B, right panels). SubAB-induced novel LCN2 was shifted on Western blots (LCN2* in Fig. 5A) with the size of LCN2* being slightly different from LCN2**. In the supernatant of SubABtreated FLAG-LCN2 transfected cells, normal-sized LCN2 was decreased and LCN2* was difficult to detect (Fig. 5B), indicating that LCN2* in SubAB-treated cells might not be secreted due to abnormal post-translational modification. N-Glycosylation in eukaryotic organisms is generally mediated through a membrane protein complex located in the ER that transfers an oligosaccharide to individual Asn residues in Asn-X-(Ser/Thr) sequons 59 . From the amino acids sequence, human LCN2 possesses only one sequon (Asn85), suggesting that the difference of LCN2* and LCN2** was not due to the number of glycosylation sites. To investigate LCN2 glycosylation, we tested the effect of N-glycosidase F on FLAG-LCN2. FLAG-LCN2 overexpressing cells were incubated for 12 h with TM, or mt or wt SubAB, and then the cell lysates were treated with N-glycosidase F. As shown in Figure S5A, the FLAG-LCN2 from control-or mt SubAB-treated cells (LCN2) showed a significantly decreased molecular weight (LCN2**) by N-glycosidase F and was found at the same size as those seen in TM-treated cells. TMinduced LCN2** was not changed by N-glycosidase F. SubAB-stimulated LCN2* was slightly shifted down by the enzyme. We next looked at whether SubAB directly cleaves LCN2. SubAB was incubated for 2 h with rLCN2 or cell lysate at 37 °C. We detected LCN2 or BiP as a positive control. The molecular weight of LCN2 was not changed the molecular weight by SubAB. BiP (78-kDa) was not seen in the presence of SubAB ( Supplementary  Fig. S5B). Further, we confirmed that the lcn2 mRNA sequence from mt and wt SubAB was the same. Thus, the carbohydrate structure of LCN2* induced by SubAB may affect secretion. SubAB induces various disruptions to cell pathways (e.g., protein synthesis inhibition, cell cycle arrest, apoptosis, stress granules formation) 14,15,19 . Thus, another possibility is that SubAB causes failure of the secretion system or exosome formation. These events might affect the secretion of LCN2*. In the case of SubAB-produced LEE-negative STEC infection, expression of normal form LCN2 is decreased in infected tissues because of the formation of a novel LCN2. Therefore, it might be difficult to detect as a biomarker or to have a significant antimicrobial effect.
Since LCN2 is an antimicrobial, siderophore-binding protein 29 , the prevention of LCN2 secretion by SubAB may promote the ability of STEC O113:H21 to survive in the host. On the other hand, the recombinant-human LCN2 treatment of STEC O113:H21 or O157:H7 Sakai strains could significantly suppress growth without induction of Stx2 or SubAB expression (Fig. 6, Supplementary Fig. S6). The gene for Stx is encoded on lambdoid bacteriophages integrated into the STEC genome, and phage induction is connected with the bacterial SOS response 60 . Thus, production of Stx2 in O157:H7 strain was stimulated by treatment with DNA-damaging agents, such as antibiotics [61][62][63] , H 2 O 2 64 or nitric oxide 65 , which is consistent with our result as shown in Supplementary Figure S6B. Thus, antimicrobial activity by LCN2 for STEC did not stimulate bacteriophage-mediated Stx2 production in STEC O157:H7 and O113:H21. Interestingly, we did not detect enhanced production of Stx2, which is encoded on bacteriophage integrated into the genome, or SubAB, by treatment of LEE-negative O113:H21 strain with H 2 O 2 ( Supplementary Fig. S6B).
Iron is an essential nutrient for all microorganisms and is used to catalyze various enzymatic reactions essential for growth 66 . LCN2 captures iron from microorganisms, thereby inhibiting their growth 25 . We investigated the effect of iron on STEC O113:H21 growth. As expected, the growth was promoted in an iron concentrationdependent manner (Supplementary Fig. S7A). Iron increased growth of O113:H21 strain, whereas a specific iron chelator 2,2′-dipyridyl (DIP), significantly suppressed growth. The growth was returned to control level in the presence of iron and DIP mixture (Supplementary Fig. S7B). Further, in the presence of ferric chloride, the levels of subAB and stx2 mRNA were significantly increased ( Supplementary Fig. S7C). These findings suggest that iron is an important factor for growth and toxin transcription in LEE-negative STEC O113:H21.
SubAB-induced apoptosis was promoted in LCN2-knockdown cells (Fig. 4B). LCN2 knockout by CRISPER/ Cas9 plasmid decreased cell proliferation and increased sensitivity to cisplatin in prostate cancer cells 67 . LCN2deficent mice showed increased intracellular labile iron, and were highly sensitive to LPS-induced mortality, associated with increased cell apoptosis and upregulation of proinflammatory gene expression 68 . Previous study suggested that decreased intracellular iron level induced cell death, which was prevented by delivery of iron 69 . SubAB-stimulated PARP cleavage was suppressed by addition of holo-Tf or FeCl 3 (Fig. 5F), suggesting that intracellular iron level is an important factor for the SubAB-induced apoptosis pathway.  Fig. 8. LEE-negative STEC O113:H21-produced SubAB is endocytosed into the ER, cleaves BiP, followed by activation of PERK. These events cause C/EBPB, and C/EBPG transcription, which further activates CHOP transcription. Overexpression of CHOP alone did not affect LCN2 expression and cell death. These transcriptional factors induce LCN2 transcription and translation of a novelsized LCN2, which might not be secreted due to abnormal post-translational modification of carbohydrate structure and accumulated in cells. Increased intracellular iron level by holo-Tf or FeCl 3 suppressed SubAB-induced PARP cleavage. However, over-expressed normal LCN2 and recombinant LCN2 prevented the siderophore iron-acquiring strategy of STEC, resulting in inhibition of STEC growth. SubAB-stimulated novel LCN2 did not show the inhibitory activity to growth of STEC. H 2 O 2 treatment of LEE-positive STEC O157:H7 Sakai strain caused Stx2 expression, which was not stimulated by LCN2. In contrast, in LEE-negative STEC O113:H21 strain, the expression of toxins by treatment with LCN2 or H 2 O 2 was not induced. STEC strains and growth conditions. STEC O157:H7 Sakai and O113:H21 were grown overnight in LB broth, which has 1.0-2.5 × 10 8 cfu ml −1 . The cultures were diluted at 10 -3 or 10 -5 in RPMI1640 medium and then grown at 37 °C, followed by addition of LCN2 or 0.1 mM H 2 O 2 . Bacterial growth was monitored at 595 nm optical density (OD 595 ) using iMark Microplate reader (BioRAD) for the indicated time points at 37 ℃.

Methods
Immunoblots analysis. The immunoblotting analysis was performed as described previously 22 . Briefly, proteins from cells were separated by SDS-PAGE and transferred to PVDF membranes, which were incubated with the indicated primary antibodies. Detection was performed with either horseradish peroxidase-labeled goat anti-mouse or anti-rabbit secondary antibodies (R&D systems), followed by enhanced chemiluminescence (EzWestLumi One, ATTO). Bands were visualized using Las 1000 (FUJIFILM). Densitometric analysis was performed by Image Gauge software (FUJIFILM) on the scanned blots, with proteins levels normalized to GAPDH.
Immunofluorescence confocal microscopy. Immunofluorescence analysis of LCN2, CHOP and FLAGtagged LCN2 was performed as described previously 19,74 . HeLa cells (2 × 10 5 cells) on glass (Matsunami Glass Industries, Japan) were incubated with 400 ng ml −1 SubAB or mt SubAB for the indicated time. Cells were fixed with 4% paraformaldehyde (PFA) (Fujifilm WAKO Pure Chemical Corporation) for 30 min. The cells are then rinsed three times with PBS and incubated with blocking buffer (5% goat serum, 0.3% Triton X-100 in PBS) at room temperature for 1 h. To visualize FLAG, PDI, or CHOP, cells were further incubated with primary antibodies in 0.4% BSA/PBS buffer at 4 °C overnight, washed twice with PBS and incubated with anti-mouse 488 (Cell Signaling) or anti-rabbit Cy3 (Rockland) antibodies at room temperature for 1 h in the dark. After washing with PBS three times, cells were mounted on glass slides using Prolong Gold Antifade reagent with DAPI (Thermo Fisher Scientific). The stained cells were visualized by FV10i-LIV confocal microscopy (Olympus). The images were arranged with Adobe Photoshop CS4.
Animal treatments. Male BALB/c mice (Japan SLC, Japan), 8 weeks old, were injected intraperitoneally with 10 µg/100 µl PBS of purified mt or wt SubAB, and mouse duodenum, ileum and colon were collected from mice at the indicated time points. As a control, mice received 100 µl of PBS. Total mRNA was purified by ISO-GEN II (NIPPON GENE) from the tissues.

Statistics.
The P values for densitometric analysis and RT-qPCR assays were determined by Student's t test with Graphpad Prism software (Graphpad, San Diego, CA, USA). P values of < 0.05 were considered statistically significant.
Ethical approval. Animal experiments were approved by Chiba university committee for animal experiments. All experiments were performed in accordance with relevant guidelines and regulations.