Lipocalin 2 prevents intestinal inflammation by enhancing phagocytic bacterial clearance in macrophages

Lipocalin 2 (Lcn2), also called neutrophil gelatinase B-associated lipocalin (NGAL), is an anti-microbial peptide originally identified in neutrophil granules. Although Lcn2/NGAL expression is increased in the inflamed intestinal tissues of patients with inflammatory bowel disease, the role of Lcn2/NGAL in the development of intestinal inflammation remains unclear. Here we investigated the role of Lcn2/NGAL in intestinal inflammation using a spontaneous mouse colitis model, interleukin-10 knock out (IL-10 KO) mice. Lcn2 expression in the colonic tissues of IL-10 KO mice increased with the development of colitis. Lcn2/IL-10 double-KO mice showed a more rapid onset and development of colitis compared to IL-10 KO mice. Lcn2 enhanced phagocytic bacterial clearance in macrophages in vitro after infection with Escherichia coli. Transfer of Lcn2-repleted macrophages prevented the development of colitis in Lcn2/IL-10 double-KO mice in vivo. Our findings revealed that Lcn2 prevents the development of intestinal inflammation. One crucial factor seems to be the enhancement of phagocytic bacterial clearance in macrophages by Lcn2.

Inflammatory bowel disease (IBD) is a chronic relapsing-remitting disorder characterized by recurrent intestinal inflammation. Although the pathophysiology of IBD remains unclear, intestinal inflammation appears to result from a dysregulated mucosal immune response toward enteric commensal bacteria and their interactions with environmental triggers, including dietary elements 1 . The role of specific food components or nutrients in the pathophysiology of IBD is uncertain, but findings from recent clinical and experimental studies suggest that dietary iron contributes to the development of IBD by its effects on enteric bacterial flora [2][3][4] .
Lipocalin 2 (Lcn2), also called neutrophil gelatinase B-associated lipocalin (NGAL), is a member of the lipocalin superfamily of extracellular transport proteins 5 . In addition to the function of lipocalins as transporters of some lipophilic molecules, such as retinoids, fatty acids, and cholesterol, Lcn2/NGAL is involved in iron delivery, cell migration, apoptosis, and cell differentiation [6][7][8][9] . One of the best-characterised functions of Lcn2/NGAL is to deprive bacteria of iron essential to their growth by sequestering iron-laden siderophores, a diverse group of small non-peptide iron-binding chemicals produced by bacteria 6,10 . Indeed, Lcn2-deficient mice are more sensitive to bacterial infection than wild-type mice, and exhibit higher mortality rates after intraperitoneal challenge with Escherichia coli 11 . In this regard, Lcn2/NGAL plays an essential role in the innate immune response against bacterial infection.
While Lcn2/NGAL was originally identified as a component of neutrophil granules, it is also expressed in macrophages and epithelial cells of the respiratory and gastrointestinal tracts in response to inflammatory signals [12][13][14][15] . Lcn2/NGAL expression is increased in the colonic tissues of patients with active IBD 16,17 . Furthermore, Lcn2/NGAL levels are increased in the serum, urine, and feces of patients with active IBD [18][19][20] . The precise role of Lcn2/NGAL in IBD pathophysiology, however, remains to be elucidated. Therefore, in the present study,

Results
Lcn2 expression in the colonic tissues of IL-10 KO mice increased with the development of colitis. Lcn2 expression in the colonic tissues of 12-week-old C57BL/6 wild-type (WT) and IL-10 KO mice was evaluated by immunohistochemical analysis. Lcn2 expression was observed mainly in the epithelial cells and partly in infiltrating immune-cells in the colonic tissues of IL-10 KO mice, whereas little Lcn expression was detected in WT mice (Fig. 1a). Gene expression of Lcn2 in the colonic tissues of IL-10 KO mice increased in a time-dependent manner (Fig. 1b). Similarly, immunohistochemical analysis revealed that Lcn2 expression increased in the colonic tissues of IL-10 KO mice over time (Fig. 1c). Furthermore, the fecal concentration of Lcn2 in IL-10 KO mice remarkably increased within a time interval of 12 weeks of colitis induction in the IL-10 KO mice (Fig. 1d) and was significantly positively correlated with the degree of histological inflammation (Fig. 1e).
Both IL-1β and Toll-like receptor 4 signaling pathways play a crucial role in enhancing Lcn2 expression in the inflamed colonic tissues of IL-10 KO mice. To reveal how Lcn2 expression is regulated in the colonic epithelia, we performed an in vitro experiment using a murine colonic epithelial cell-line, CT26. Stimulation with IL-17A and IL-1β significantly increased Lcn2 secretion from CT26 cells, whereas stimulation with interferon-gamma (IFN-γ ) and tumour necrosis factor-alpha (TNF-α ) did not (Fig. 2a). In addition, lipopolysaccharides (LPS), a ligand of Toll-like receptor (TLR) 4, significantly increased Lcn2 secretion from CT26 cells, whereas ligands for TLR2 and TLR9 did not (Fig. 2a).
To confirm the influence of these stimulants in vivo, we examined the effect of neutralizing antibodies against IL-17 and IL-1β on the fecal concentration of Lcn2 in IL-10 KO mice. Administration of anti-IL-1β antibody, but not anti-IL-17 antibody, significantly decreased the fecal Lcn2 concentration in IL-10 KO mice at 4 weeks of age (Fig. 2b). We also examined the effect of TLR4 deficiency on the fecal concentration of Lcn2 in IL-10 KO mice using TLR4/IL-10 DKO mice. Fecal Lcn2 concentrations were significantly lower in TLR4/IL-10 DKO mice than in IL-10 KO mice at 4 weeks of age (Fig. 2b).
Lcn2 deletion elicits spontaneous colitis in IL-10 KO mice at an early age. To assess the role of Lcn2 in intestinal inflammation, we compared intestinal inflammation between IL-10 KO and Lcn2/IL-10 DKO mice, because Lcn2 KO mice did not develop spontaneous colitis ( Supplementary Fig. 1). Macroscopically, remarkable shrinkage of the cecum and stiffening of the colonic wall were observed in Lcn2/IL-10 DKO mice as early as 4 weeks of age (Fig. 3a). Similarly, histological scores were significantly higher in Lcn2/IL-10 DKO mice than in IL-10 KO mice, even at 4 weeks of age (Fig. 3b). Gene expression of pro-inflammatory cytokines, such as IFN-γ , IL-17A, IL-1β , and TNF-α , was significantly upregulated in the colonic tissues of Lcn2/IL-10 DKO mice compared with IL-10 KO mice (Fig. 3c). Also, colonic explants from Lcn2/IL-10 DKO mice secreted significantly higher levels of TNF-α and IL-12p40 than those from IL-10 KO mice (Fig. 3d). Mesenteric lymph node (MLN) cells from Lcn2/IL-10 DKO mice produced greater amounts of IL-17A than MLN cells from IL-10 KO mice after stimulation with cecal bacterial lysates (CBLs; Fig. 3e).
Lcn2 deficiency does not affect mucosal barrier function. The enhanced Lcn2 production in the colonic epithelia of IL-10 KO mice implied the involvement of Lcn2 in mucosal barrier function. Thus, we examined the effect of Lcn2 deficiency on enteric bacteria by comparing enteric bacterial flora between WT and Lcn2 KO mice. Terminal restriction fragment length polymorphism (T-RFLP) analysis, however, revealed no significant difference in the luminal bacterial composition between WT and Lcn2 KO mice at 4 weeks of age ( Supplementary Fig. 2a, Supplementary Table 1). We also examined the expression of intestinal barrier-related molecules in the colonic tissues of WT and Lcn2 KO mice. The number of Ki-67 positive colonic epithelial cells ( Supplementary Fig. 2b) and HES-1 gene expression in the colonic tissues ( Supplementary Fig. 2c) did not differ significantly between WT and Lcn2 KO mice. Moreover, expression of TLR genes (TLR2, TLR4, and TLR9), junctional molecules (ZO-1 and Claudin-1), mucin (MUC2), trefoil factor (TFF3), and anti-microbial peptides (mBD3, CRAMP, and REG3γ ) in the colonic tissues was not significantly different between WT and Lcn2 KO mice ( Supplementary Fig. 2d).

Lcn2-repleted macrophages prevented the development of colitis in Lcn2/IL-10 DKO mice.
Finally, we performed a macrophage transfer experiment to elucidate the role of macrophage Lcn2 in the development of spontaneous colitis. Transfer of IL-10KO-derived, but not DKO-derived, TEPMs prevented the development of colitis in Lcn2/IL-10 DKO mice and these mice exhibited a significantly lower histological score than control mice at 4 weeks of age (Fig. 7).

Discussion
The findings of the present study revealed that Lcn2 prevents the development of spontaneous colitis in IL-10 deficient mice by enhancing phagocytic bacterial clearance in macrophages. The results clearly demonstrated a crucial role for Lcn2/NGAL in the development of intestinal inflammation by its regulation of mucosal innate immune responses toward enteric commensal bacteria.  To determine the type of stimulus that regulates Lcn2 expression in the colon, we stimulated CT26 cells with representative pro-inflammatory cytokines involved in the pathophysiology of IL-10 KO mice 21,22 . We also stimulated CT26 cells with ligands for TLR2, TLR4, and TLR9, all of which are expressed by intestinal epithelial cells and recognise enteric bacterial components 23 . These in vitro experiments revealed increased Lcn2 production from CT26 cells after stimulation with IL-1β , IL-17A, and LPS. We further confirmed these results in vivo, and finally revealed that IL-1β and TLR4 signaling pathways are crucial for enhancing Lcn2 expression in the colonic tissues of IL-10 KO mice. In addition, previous studies using murine tracheal epithelial cells or a human lung epithelial cell-line reported that Lcn2 was upregulated by IL-1β and IL-17A, but not by TNF-α in vitro 14    We then generated Lcn2/IL-10 DKO mice and compared intestinal inflammation with that of IL-10 KO mice to reveal the role of Lcn2 in the development of colitis. While Lcn2 KO mice did not develop spontaneous colitis, a rapid onset and development of colitis was observed in Lcn2/IL-10 DKO mice soon after the weaning period compared to IL-10 KO mice, suggesting that the innate immune responses were disrupted in the colonic tissues of Lcn2/IL-10 DKO mice. Furthermore, the enhanced Lcn2 production in colonic epithelia in response to inflammatory signals suggests a role for Lcn2 in mucosal barrier function. Therefore, in this study, we focused on the effect of Lcn2 on mucosal defence systems, including mucosal barrier and underlying innate immune cells, especially macrophages. Regarding mucosal barrier function, we first compared the luminal bacterial composition between WT and Lcn2 KO mice by T-RFLP analysis to examine the effect of Lcn2 on enteric bacteria. We also examined the proliferation and differentiation of colonic epithelial cells by counting Ki-67 positive cells in the crypts and quantifying HES-1 gene expression in the colonic tissues of WT and Lcn2 KO mice 27 . Moreover, we quantified the gene expression of intestinal barrier-related molecules, such as junctional molecules, mucin, trefoil factor, anti-microbial peptides, and TLRs in the colonic tissues of WT and Lcn2 KO mice 23,28 . These components, however, did not significantly differ between WT and Lcn2 KO mice, indicating that Lcn2 does not affect mucosal barrier function, at least in non-inflamed colon. A substantial amount of Lcn2 produced in the inflamed milieu, however, might affect mucosal barrier function. In contrast to our results, Singh V et al. showed a distinct difference in fecal bacterial composition between WT and Lcn2 KO mice with increased Bacteroidetes and Proteobacteria phylums and decreased Tenericutes phylum in Lcn2 KO mice, although they did not find spontaneous intestinal inflammation in Lcn2 KO mice like our study 29 . Furthermore, they demonstrated the potential colitogenicity of this gut flora of Lcn2 KO mice by co-housing IL-10 KO mice with Lcn2 KO mice. These findings conflicting with our data might result from the different genetic background of mice or environmental factors in the different facility. Further study is necessary to elucidate the effect of Lcn2 on gut microbiota and mucosal barrier function in colon.
Next, we investigated the effect of Lcn2 deficiency on macrophage functions. We used TEPMs in this study, because a recent large-scale genetic analysis of the mouse immune system elucidated the genetic similarity between TEPMs and intestinal macrophages on the basis of macrophage-related gene signatures 30 . We first demonstrated that Lcn2KO-derived TMPMs produced higher mounts of pro-inflammatory cytokines, such as TNF-α and IL-12p40, after stimulation with Pam3CSK4 and LPS, whereas they produced lower amounts of IL-10 compared to WT-derived TEPMs. Zhang et al. also reported that pre-treatment with Lcn2 suppressed LPS-induced gene expression of pro-inflammatory cytokines in a murine macrophage cell-line, RAW264.7 31 . Furthermore, Warszawska et al. reported significantly higher production of keratinocyte-derived chemokine and lower production of IL-10 in Lcn2KO-derived bone marrow-derived macrophages compared to WT-derived bone marrow-derived macrophages after infection with Streptococcus pneumonia 32 . In their study, Lcn2 induced the deactivation of macrophages, characterised by the down-regulation of pro-inflammatory cytokines and the upregulation of anti-inflammatory cytokines, including IL-10. In the present study, however, cytokine production did not differ significantly between IL-10KO-and DKO-derived TEPMs. Thus, the effect of Lcn2 deficiency on cytokine production from macrophages could not be directly involved in the deteriorated colitis of Lcn2/IL-10 DKO mice.
We then assessed the effect of Lcn2 on the phagocytic function of macrophages using pHrodo dye-conjugated E. coli, which exhibit a dramatic increase in fluorescence inside the cells. The increased fluorescence intensity in DKO-derived TEPMs compared to other TEPMs after E. coli challenge indicated the enhanced uptake of bacteria in DKO-derived TEPMs. Furthermore, we examined the role of Lcn2 in the intracellular bacterial killing function of macrophages using a gentamicin protection assay. Intriguingly, DKO-derived TEPMs showed impaired intracellular bacterial clearance compared to other TEPMs, followed by increased TNF-α production from DKO-derived TEPMs. The increased TNF-α secretion from infected DKO-derived TEPMs, despite no significant difference in TNF-α production between DKO-derived and IL-10KO-derived TEPMs after stimulation with TLR ligands, suggested persistent macrophage activation due to increased numbers of surviving E. coli within the cells. In contrast, neither phagocytosis nor intracellular bacterial clearance differed significantly between WTand Lcn2KO-derived TEPMs. To further explore this discrepancy, we evaluated Lcn2 expression in WT-and IL-10KO-derived TEPMs infected with E. coli, and found that IL-10KO-derived TEPMs, but not WT-derived TEPMs, expressed Lcn2 after infection with E. coli. Namely, 1-hour infection with E. coli might be insufficient for WT-derived TEPMs to express Lcn2, and inadequate Lcn2 expression in WT-derived TEPMs might result in the same level of phagocytosis and intracellular bacterial clearance as Lcn2KO-derived TEPMs.
Autophagy is a cellular degradation system for numerous cytosolic contents, including long-lived proteins, mitochondria, and peroxisomes, and also plays an essential role in the elimination of intracellular bacteria 33 . Furthermore, recent genetic analyses demonstrated an essential role of autophagy in innate immunity, and implicated several autophagy-related genes, such as ATG16L1 and IRGM, in the pathophysiology of IBD [34][35][36] . Indeed, Lassen et al. revealed that Atg16L1 T300A, the coding mutation associated with an increased risk of Crohn's disease in humans, contributes to decreased antibacterial autophagy and increased IL-1β production from CD11b-positive colonic and splenic macrophages in mice 37 . Induction of autophagy leads to the fusion of bacterial phagosomes with lysosomes facilitated by autophagy-related proteins, such as LC3 and p62, and results in the degradation of bacteria within the formed autolysosome 33 . Therefore, we assessed the effect of Lcn2 deficiency on phagocytosis/autophagy pathways by comparing LC3-II formation and p62 degradation between IL-10-and DKO-derived TEPMs. We also examined the localization of Lcn2 and LC3 in macrophages after infection with E. coli by immunocytochemistry. Our data clearly demonstrated that Lcn2 enhances phagocytosis/autophagy pathways in macrophages after infection with E. coli. To further confirm the results in vivo, we performed a macrophage transfer experiment. Restoring Lcn2 expression in intestinal macrophages by transfer of IL-10KO-derived TEPMs attenuated colitis in 4-week-old Lcn2/IL-10 DKO mice. This finding indicates that Lcn2 in macrophages is crucial for preventing the development of colitis in IL-10 deficient mice, although the incompletely attenuated colitis in IL-10KO-derived TEPMs-transferred Lcn2/IL-10 DKO mice suggested that Lcn2 from other origins, including colonic epithelial cells, could also be involved in the development of colitis. Considering that exogenous Lcn2 enhanced autophagic bacterial clearance in macrophages in our in vitro experiments, an intriguing possibility is that Lcn2 secreted from colonic epithelial cells also affects intestinal macrophages to prevent the development of colitis in vivo.
In the present study, Lcn2 deficiency enhanced bacterial phagocytosis, but impaired intracellular bacterial clearance in macrophages. Recently, Bonilla et al. evaluated the role of autophagy in regulating phagocytosis in macrophages. They demonstrated that p62 accumulation due to impaired autophagy increases the cell-surface expression of class A scavenger receptors, which recognise a diverse group of bacteria, including E. coli, leading to enhanced phagocytosis of these bacteria in macrophages 38 . In our study, enhanced uptake of bacteria in DKO-derived TEPMs might attribute to impaired autophagy in DKO-derived TEPMs.
In conclusion, Lcn2 prevented the development of spontaneous colitis in IL-10 deficient mice. One crucial factor seemed to be the enhancement of phagocytic bacterial clearance in macrophages by Lcn2. Further studies of the interaction between Lcn2 and autophagy-related proteins will provide new insight into the role of Lcn2/ NGAL in the pathophysiology of IBD.

Methods
Ethics statement. This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Protection Committee of Kyoto University.
Microscopic assessment of colitis. Mice were killed at 4, 8, and 12 weeks of age. Rectums were dissected and transverse sections prepared. The sections were fixed in 10% formaldehyde, dehydrated in ethanol, and embedded in paraffin for histological analysis. Sections were stained with haematoxylin and eosin (HE), and histologically analysed in a blind manner. Histological damage was quantified by the previously described histological scoring system 39 . Immunohistochemistry. Immunostaining was performed as described prevously 39  Preparation of cecal bacterial lysates. CBLs were prepared from cecal contents of WT mice as described previously 40 and stored at − 20 °C until use. The sterility of the lysates was confirmed by culture.

Preparation of stool samples. Murine cecal contents were collected using Smart-Prep Fecal Sample
Extraction KIT (ALPCO Diagnostics, Salem, NH, USA). Stool samples were subsequently prepared using the extraction buffer from a PhiCal Calprotectin ELISA Kit (Immundiagnostik AG, Bensheim, Germany) and stored at − 80 °C until analysis.
Administration of neutralizing antibodies. The neutralizing goat anti-mouse IL-1β , rat anti-mouse IL-17, and their control antibodies (goat IgG control and rat IgG2a isotype control) were purchased from R&D Systems. Three-week-old IL-10 KO mice were intraperitoneally injected with 100 μ g of anti-IL-1β or goat control antibody on day 0 42 . Otherwise, 100 μ g of anti-IL-17 or rat control antibody was intraperitoneally injected on days 0, 2, 4, and 6 43 . The mice were killed on day 7 and fecal proteins were extracted as described above.
Terminal restriction fragment-length polymorphism analysis. T-RFLP analysis was performed as described previously 44 . The 16S rRNA gene was amplified from fecal DNA with labelled primers and treated with 10 U Bsl l (New England BioLabs, MA, USA). The digested products were fractionated using an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems, CA, USA) with DNA analysis software Gene Mapper. The major terminal restriction fragments of 1~3 bp were summarized as operational taxonomic units, and the corresponding bacterial groups were estimated by computer simulation based on a human intestinal microbiota database 45 . The percentage of each bacterial group to the total enteric bacteria was represented by the percentage of each summed operational taxonomic unit area to total operational taxonomic unit area 46 . Preparation of thioglycollate-elicited peritoneal macrophages. Mice were intraperitoneally injected with 4% thioglycollate (Eilken Chemical Co., Tokyo, Japan) and peritoneal cells were harvested on day 4. The TEPMs were prepared after removing erythrocytes with lysis buffer. In several experiments, the TEPMs were seeded at 1 × 10 6 cells/well in 6-well culture plates and incubated overnight in complete RPMI medium containing 40 ng/ml recombinant macrophage colony-stimulating factor (rM-CSF; Peprotech, Rocky Hill, NJ, USA) before stimulating with Pam3CSK4 (500 ng/ml), LPS (100 ng/ml), or ODN1585 (500 nM) for 24 hours. Supernatants were harvested and stored at − 80 °C until analysis.
Phagocytosis assay. The phagocytosis assay was performed using pHrodo Green E. coli BioParticles Conjugate (Invitrogen) according to the manufacturer's instructions. The fluorescence intensity was measured at the indicated times using a fluorescence microplate reader (Infinite F200 PRO; TECAN, Kawasaki, Japan). Phagocytic activity was evaluated by subtracting the mean fluorescence intensity of no-cell negative-control wells from all experimental wells and is indicated as relative value.
Gentamicin protection assay. A gentamicin protection assay was performed as previously described with some modifications 47 . Briefly, TEPMs were seeded at 5 × 10 5 cells/well in 12-well cell culture plates and incubated overnight in complete RPMI medium containing 40 ng/ml rM-CSF. The wells were challenged with E. coli (ATCC; 25922) at a multiplicity of inoculation of 10 bacteria per cell. In several experiments, cells were further incubated with 100 ng/ml recombinant Lcn2 (R&D Systems) for 4 hours before challenge with E. coli. Cells were incubated for 1 hour to internalise the bacteria, washed with phosphate buffered saline (PBS), and cultured in complete RPMI medium containing 100 mg/ml gentamicin. At several time points after infection, the cells were lysed with 1% Triton-X 100 in PBS, and surviving bacteria were counted on Brain-Heart Infusion agar plates. Bactericidal activity of macrophages was evaluated by calculating the ratio of colony forming units at each time point to that at 2 hours after infection.

Western blot analysis.
Western blot analyses were performed on whole-cell extracts as described previously 40 . Rabbit anti-mouse LC3, p62, and β -actin were obtained from Cell Signaling Technology (Beverly, MA, USA). The image was recorded using a chemiluminescent image reader (ChemiDoc XRS plus, Bio-Rad Laboratories).
Immunocytochemistry. TEPMs were seeded on culture cover glasses (Matsunami Glass, Osaka, Japan) and incubated overnight. After infection with E. coli, the cells were washed with PBS and fixed with 4% paraformaldehyde at 4 °C overnight. For antigen retrieval, the cells were boiled in citrate buffer (pH 6.0) in a microwave oven for 10 minutes. After blocking with 5% BSA in PBS containing 0.1% Tween20 (PBS-T), cells were incubated with goat anti-Lcn2 antibody and rabbit anti-mouse LC3 for 1 hour at room temperature. After washing with PBS-T, cells were incubated for 30 minutes with Alexa Fluor 488-conjugated anti-rabbit antibody (Invitrogen) and Alexa Fluor 594-conjugated anti-goat antibody. Immunofluorescence imaging was acquired using a fluorescence microscope (BIOREVO BZ-9000) with BZ-Analyzer version 2.1 software and composed in Photoshop Elements 12.
Macrophage transfer. Macrophage transfer was performed as described previously with some modification 48,49 . Briefly, 3 × 10 7 TEPMs were prepared from IL-10 KO or Lcn2/IL-10 DKO mice and suspended in 300 μ l PBS. Three-week-old mice were intraperitoneally injected with TEPM, or PBS as a control. The mice were killed at day 7, and histological inflammation was evaluated as described above.
Statistical analysis. Numerical data are expressed as means with SD or SEM, and the differences between groups were analyzed by one-way analysis of variance (ANOVA) with Bonferroni's correction for multiple comparisons if variables distribute normally. Numerical data are expressed as medians with interquartile ranges, and the differences between groups were analyzed by non-parametric Mann-Whitney U test or Kruskal-Wallis test with Games-Howell post hoc analysis for multiple comparisons if variables do not distribute normally. A P value of less than 0.05 was considered statistically significant.