IL-10 produced by macrophages regulates epithelial integrity in the small intestine

Macrophages (Mϕs) are known to be major producers of the anti-inflammatory cytokine interleukin-10 (IL-10) in the intestine, thus playing an important role in maintaining gastrointestinal homeostasis. Mϕs that reside in the small intestine (SI) have been previously shown to be regulated by dietary antigens, while colonic Mϕs are regulated by the microbiota. However, the role which resident Mϕs play in SI homeostasis has not yet been fully elucidated. Here, we show that SI Mϕs regulate the integrity of the epithelial barrier via secretion of IL-10. We used an animal model of non-steroidal anti-inflammatory drug (NSAID)-induced SI epithelial injury to show that IL-10 is mainly produced by MHCII+ CD64+ Ly6Clow Mϕs early in injury and that it is involved in the restoration of the epithelial barrier. We found that a lack of IL-10, particularly its secretion by Mϕs, compromised the recovery of SI epithelial barrier. IL-10 production by MHCII+ CD64+ Ly6Clow Mϕs in the SI is not regulated by the gut microbiota, hence depletion of the microbiota did not influence epithelial regeneration in the SI. Collectively, these results highlight the critical role IL-10-producing Mϕs play in recovery from intestinal epithelial injury induced by NSAID.

processes are microbiota-dependent 14,15 . Consistent with this, we have previously reported that IL-10-producing Mφs in the SI are regulated by dietary factors 16 . The IL-10-producing Mφs in the SI are less abundant and consequently produce less IL-10 in a total parenteral nutrition mouse model in which mice do not experience enteral stimulation by dietary antigens 16 . Notably, this animal model is associated with gut translocation and impairments of the intestinal barrier 16 . As such, the dietary antigen-dependent population of Mφs in the SI may also be involved in intestinal homeostasis. Specifically, we hypothesized that IL-10-producing Mφs might contribute to the regulation of epithelial barrier integrity in the SI.
In order to study the role Mφs play in maintaining homeostasis in the SI, we employed the non-steroidal anti-inflammatory drug (NSAID)-induced epithelial injury model. Administration of NSAIDs, such as indomethacin (IND), is well known to induce epithelial injuries in the upper gastrointestinal tract, including the SI, a serious clinical event in both human and mouse applications 17 . The mechanism and appropriate treatment of NSAID-induced SI injury have not been completely elucidated. Using this mouse model, we determined that IL-10-producing MHCII + CD64 + Ly6C low Mφs play an essential role in the recovery from acute SI injury.

Results
Indomethacin induces epithelial injury in the SI. A single dose of IND (15 mg/kg body weight) was adequate to induce SI epithelial injury in C57BL/6 wild-type (WT) mice. Maximum body weight loss was observed on day 3 and by day 7 the animals largely managed to regain their initial weight (Fig. 1a). Body weight loss corresponded with increased intestinal permeability, as quantified by a fluorescein isothiocyanate (FITC) dextran assay as well as endoscopic findings of ulceration and hemorrhage (Fig. 1b,c). Thus, IND can cause recoverable SI injury. To clarify the immune mechanism underlying the recovery from SI injury, we next assessed gene expression changes in the SI mucosa. Mucosal tissue samples were obtained from there different regions of the SI (proximal, middle and distal SI) during the early phase of SI injury (24 hours post IND injection). mRNA levels of pro-inflammatory cytokines, such as tumor necrosis factor-alpha (TNF-α), IL-6, and IL-1β, were significantly increased in all regions of the SI mucosa following IND injection (WT + IND) (Fig. 2a). Notably, the expression of anti-inflammatory cytokine IL-10 was likewise elevated in the SI mucosa during the early phase of SI injury (Fig. 2a). These results suggest that IL-10 may play a compensatory role in the early phase of injury, paving the way to recovery from epithelial damage and body weight loss.
Deficiency of IL-10 exacerbates indomethacin-induced SI injuries. Next, we examined the role of IL-10 in the recovery from IND-induced SI injuries using IL-10-deficient (Il10 −/− ) mice. Compared to WT mice, Il10 −/− mice showed impaired recovery from SI injury, thereby resulting in higher mortality (Fig. 2b). The SI weight-to-length ratio was significantly increased in IND-treated Il10 −/− mice, indicating that Il10 −/− mice develop more severe intestinal inflammation as a result of IND treatment (Fig. 2c). These results suggest that in the absence of IL-10, SI epithelial tissue repair does not occur, in part due to unhindered effects of pro-inflammatory cytokines.
IL-10 is primarily produced by macrophages during the acute phase of injury. We next examined the source of IL-10 in vivo by using IL-10 reporter mice (IL-10 Venus mice 15,18 ). Based on the IND kinetics data (Fig. 1a), body weight loss peaked on day 3 post injection and then gradually returned to normal levels. In order to focus on IL-10 responses during this phase, we analyzed mice on day 3 post IND injection. Lamina propria mononuclear cells (LPMCs) were isolated from the SI of IND-treated IL-10 Venus mice on day 0 and day 3 following IND injection. IL-10-producing cells (Venus + ) found in the total population of CD45 + SI LPMCs were analyzed by flow cytometry. We first gated on IL-10(Venus) + cells within total CD45 + leucocytes ( Fig. 3a top panel), and then further characterized by staining T cells and myeloid cells (Fig. 3a middle and bottom panels). Within the IL-10(Venus) + CD45 + cells, helper T cell populations were gated as CD3 + CD4 + T cells (Fig. 3a middle panel). Non-T cell populations were further distinguished into Mφs (CD64 + ), dendritic cells (DCs; CD64 − CD11c + ) and other cells (CD64 − CD11c − ) ( Fig. 3a bottom panel). At steady-state, CD4 + T cells and Mφs were the major producers of IL-10 in the SI mucosa (day 0 IND) ( Fig. 3b). During the acute phase of SI inflammation (day 3 IND), CD64 + Mφs became the predominant source of IL-10 ( Fig. 3b). Although the abundance of IL-10-producing CD3 + CD4 + T cells was lower than that of Mφs during SI injury, it is possible that IL-10 produced by T cells also contributed to recovery. T and B cell-deficient Rag1 −/− mice were used to exclude the involvement of T cell-derived IL-10 in this injury model (Fig. 3c). Rag1 −/− mice also experienced maximum body weight loss on day 3 following IND injection and then recovered normally (Fig. 3c). Consistently, intestinal permeability was increased on day 1 following IND administration but was restored by day 3 and day 7 in both WT and Rag1 −/− mice (Fig. 3d). These data indicate that Mφs are the major source of IL-10 during SI inflammation and play a predominant role in the recovery from SI injury.
MHC-II + CD64 + Ly6C low macrophages are responsible for IL-10 production. Next, we further characterized the IL-10-producing cell population during IND-induced SI injury. It is well-documented that intestinal inflammation promotes the infiltration of Ly6C hi CCR2 + monocytes from peripheral blood into the lamina propria. Recruited Ly6C hi monocytes then give rise to Mφs locally. We sought to identify whether the newly recruited Ly6C hi monocytes or resident/terminally differentiated monocyte-derived Mφs are the major producers of IL-10 during SI injury. To this end, we isolated SI LPMCs from IL-10 Venus mice on day 0, day 3 and day 7 following IND injection. CD64 + monocyte/Mφ lineage cells were then further separated into Ly6C hi MHC-II −/+ newly recruited monocytes and Ly6C lo MHC-II + resident/terminally-differentiated Mφs (Fig. 4a). CD11c + CD64 − DCs were also analyzed as a control subset. At steady-state (day 0), a limited number of Ly6C hi MHC-II −/+ monocytes were present in the SI LPMCs ( Fig. 4a,b). At the peak of SI inflammation (day 3), a significant number of Ly6C hi MHC-II −/+ monocytes were recruited into the SI mucosa, but these cells disappeared as SI inflammation resolved (day 7) (Fig. 4a,b). We next analyzed IL-10 expression in each subset. Although Ly6C hi MHC-II −/+ monocytes produced a notable amount of IL-10, Ly6C lo MHC-II + resident/terminally-differentiated Mφs remained the predominant sources of IL-10, even during inflammation (Fig. 4c,d,e). CD11c + CD64 − DCs were a minor source of IL-10, both at steady-state and the inflamed SI mucosa (Fig. 4c,d,e). To address the role IL-10-producing Mφs  serve in the recovery from SI injury, we next attempted to deplete these cells. To this end, we used CCR2 DTR mice, as the CD64 + Ly6C lo MHC-II + resident/terminally-differentiated Mφs arise from CCR2 + monocytes 4 (Fig. 4f). Continual depletion of CCR2 + monocyte-derived Mφs during IND-induced injury demonstrated a decreased ability to recover from epithelial damage as evidenced by the inability to return to initial body weight (Fig. 4g). In order to confirm that the IL-10-mediated recovery is Mφ-dependent, we depleted Mφs from CCR2 DTR mice and adoptively transferred SI Mφs isolated either from WT or Il10 −/− mice (Fig. 4f). The animals that received WT intestinal Mφs managed to regain most of their initial weight, indicating recovery from intestinal injury (Fig. 4g).
On the other hand, the adoptive transfer of Il10 −/− intestinal macrophages resulted in a more significant weight loss and an inability to recover from intestinal injury, similar to mice lacking intestinal Mφs altogether (Fig. 4g).
The gut microbiota is not required for epithelial regeneration in the SI. Colonic Mφs and IL-10 production by these cells are known to be regulated by the gut microbiota and its metabolites 4,10,11,19 . Hence, depletion of the gut microbiota significantly impairs epithelial recovery in the colon, in part due to insufficient activation of colonic Mφs 12,20 . In contrast, we have previously demonstrated that IL-10-producing Mφs in the SI are not regulated by the gut microbiota 16 . Thus, we thought it likely that Mφ-mediated epithelial recovery in the SI is independent of the gut microbiota. To address this hypothesis, we depleted the gut microbiota by treating mice with a cocktail of four antibiotics (Abx; vancomycin, neomycin, metronidazole, and ampicillin) prior and during IND treatment. When compared to Abx-negative control mice, Abx-treated mice experienced similar body weight loss and recovery (Fig. 5a). Correlating with this observation, there was no significant difference in intestinal permeability during intestinal injury or recovery between control and Abx-treated mice (Fig. 5b).
Consistently, the number of Mφs and IL-10 production in the SI LPMCs were not affected by Abx-treatment (Fig. 5c,d). These results indicate that the gut resident microbiota does not regulate IL-10-producing Mφs in the SI and the absence of the microbiota does not influence recovery from SI epithelial injury. In this study, we have shown that Mφs contribute to recovery from SI injury in an IL-10-dependent manner. IL-10 is constitutively produced by Mφs at steady-state as well as during inflammation, and the intestine contains the largest pool of IL-10-producing Mφs in the body 2 . Using an animal model of drug (IND)-induced SI injury, we found that a significant number of Ly6C hi monocytes are recruited to the SI mucosa. Since depletion of CCR2 + cells, a receptor abundantly expressed on Ly6C hi monocytes, impairs recovery from IND-induced epithelial injury, recruitment of Ly6C hi monocytes is crucial for re-epithelialization in the SI. However, Ly6C hi monocytes are not the major source of IL-10 during the recovery phase. Thus, it is likely that the recruited Ly6C hi monocytes do not directly promote epithelial repair. It has been reported that Ly6C hi CCR2 + monocytes down-regulate the expression of Ly6C and CCR2 in the gut 4,5 . Subsequently, they give rise to tissue resident Mφs and express MHC-II and CX 3 CR1 4 . Thus, it is plausible that the recruited Ly6C hi monocytes undergo terminal differentiation to become IL-10-producing MHCII + Ly6C low CD64 + Mφs during SI injury and that these cells ultimately drive re-epithelialization in the SI. Notably, although MHCII + Ly6C low CD64 + Mφs constitutively produce large amounts of IL-10 compared to monocytes or DCs, IL-10 production did not change drastically during SI injury (Fig. 4c). These data suggest that constitutive IL-10 production, unlike inflammation-induced responsive IL-10 production, by MHCII + Ly6C low CD64 + Mφs might play an important role in epithelial regeneration. However, we are unable to dissociate constitutive and responsive IL-10 production in this model, and hence it is difficult to conclude which type of IL-10 production (constitutive vs responsive) plays the more important role in epithelial regeneration.

Discussion
The importance of IL-10-producing MHCII + Ly6C low CD64 + Mφs in the regulation of intestinal homeostasis, particularly epithelial regeneration, has been described in different types of intestinal injury and/or inflammation. A recent report has shown that IL-10, produced by MHCII + Ly6C low CD64 + Mφs, promotes epithelial re-generation in a colon biopsy-induced injury model 14 . In addition to murine models of intestinal inflammation and injury, IL-10-producing monocyte-derived Mφs are important for the maintenance of intestinal homeostasis in humans. In the human intestinal mucosa, the HLA-DR high CD14 + CD163 high CD160 high Mφ subset, which is thought to arise from peripheral blood CD14 + monocytes, has been identified as a significant source of IL-10 that is able to suppress the inappropriate proliferation of Th1/Th17 cells 21 . Interestingly, this subset is known to be diminished in patients with ulcerative colitis (UC) 21 . Since UC patients present with chronic mucosal ulceration, it is possible that the dysregulation of IL-10-producing intestinal Mφs may lead to compromised epithelial regeneration in these patients.
The mechanisms by which IL-10-producing MHCII + Ly6C low CD64 + Mφs promote re-epithelialization in the SI is unclear. In the colon biopsy-induced injury model 14 , IL-10 acts on colonic epithelial cells and activates CREB signaling and its downstream target WISP-1, which then promotes β-catenin/TCF signaling, epithelial cell proliferation and wound closure. Given this evidence, it is plausible that IL-10, produced by MHCII + Ly6C low CD64 + Mφs, directly promotes the regeneration of epithelial cells in the SI. However, it remains possible that IL-10 primarily modulates immune cells, thus making its effect on epithelial healing indirect. In this regard, a recent report has demonstrated that IL-10 modulates CD11c + myeloid cells in both the SI and colonic mucosa 22  signaling in CD11c + mononuclear phagocytes, in turn, regulates T cell activation and the proliferation of intestinal crypt cells. Hence, the loss of IL-10 signaling in CD11c + cells leads to augmented T cell activation as well as crypt hyperplasia 22 . Thus, IL-10, produced by the SI MHCII + Ly6C low CD64 + Mφs, may also modulate other immune cells, including CD11c + mononuclear phagocytes, and indirectly promote epithelial regeneration during SI injury. Accumulating evidence has highlighted the importance of the gut resident microbes for epithelial regeneration. An earlier report has demonstrated that acute epithelial injury (dextran sulfate sodium (DSS)-induced Additional studies have demonstrated that specific bacterial species or their metabolites elicit re-epithelialization in the colon. For example, Akkermansia muciniphila has been shown to be enriched in the proximity of damaged epithelium and promote wound healing in the colon 23 . Likewise, a microbial metabolite deoxycholate regulates the regeneration of colonic crypts, thus playing an important role in colonic epithelial repair 20 . The gut microbiota, in concert with intestinal macrophages, is clearly essential for epithelial regeneration. In this regard, activation of MyD88 signaling, elicited by the gut microbiota, is required for IL-10 production in colonic macrophages 10 . This notion is consistent with the importance of intestinal Mφs and MyD88 signaling for epithelial recovery in the colon 12 . Unlike in the colon, we have demonstrated that the gut microbiota is dispensable for epithelial injury recovery in the SI. Depletion of the gut microbiota did not alter IL-10 production by SI Mφs or the restoration of SI barrier function. This is consistent with our earlier finding that IL-10 production in the SI is independent of the gut microbiota and is instead regulated by dietary antigens 16 . It is noteworthy to mention that the role of the gut microbiota in SI injury remains controversial. Earlier studies suggested that the gut microbiota was crucial for the induction of NSAID-mediated epithelial injury in germ-free and gnotobiotic animals 24,25 . In contrast, more recent studies have demonstrated that the gut microbiota and its metabolites protect the host from SI injury and, therefore, depletion of the gut microbiota by antibiotics significantly exacerbates SI injury 26,27 . In our study, depletion of the gut microbiota did not affect susceptibility to SI injury. Heterogeneity of the gut microbiota composition across various animal facilities may account for these differences. Thus, further studies are needed to elucidate the impact of IL-10-producing Mφs on epithelial recovery and its relationship with the gut microbiota. Another limitation is related to the background strain of mice used in this study. We used an inbred mouse strain (C57BL/6) in this study due to the availability of genetically engineered strains. However, immune responses and the resident microbiota may differ in outbred mouse strains. Further studies that use outbred mouse strains may provide more insight.
Dietary nutrients have been reported to modulate the development of regulatory immune cells and potentiate gut barrier function 28 . Since the abundance of some nutrients, particularly food-derived micromolecules, such as amino acid, is higher in the SI compared with the colon, the impact of dietary factors on the regulation of homeostasis should be more pronounced in the SI. In the context of Crohn's disease, exclusive enteral nutrition (EEN) has been found to effectively induce remission with mucosal healing. Several clinical studies have shown this response to be particularly robust in the SI 29 . However, the precise mechanism by which EEN attenuates inflammation and improves Crohn's disease symptoms remains largely unknown. Our previous study has demonstrated that IL-10 production by Mφs in the SI is regulated by dietary amino acids 16 . Dietary amino acids activate the mTOR signaling pathway, which regulates the secretion of IL-10 from SI Mφs through as yet unidentified mechanism 16 . Given this evidence, it is conceivable that amino acids included in EEN promote mucosal healing through the activation of IL-10 production by SI Mφs.
Taken together, Mφs and IL-10 produced by these cells largely contribute to the maintenance of epithelial barrier integrity in the SI. Since the development and function of anti-inflammatory Mφs in the SI appear to be regulated by dietary factors, optimal dietary approaches could be used to promote mucosal healing in the SI, thus providing a more targeted option for a subset of Crohn's disease patients.

Methods
Mice. Wild-type (WT) C57BL/6 and IL-10-deficient (Il10 −/− ) mice (C57BL/6 background), originally obtained from Jackson Laboratories (Bar Harbor, ME), were bred and kept under SPF conditions in the University of Michigan Animal Facility. IL-10 Venus reporter (Il10 Venus ) mice (C57BL/6 background) were generated as previously described 15 . CCR2 depleter (CCR2 DTR ) mice were kindly provided by Dr. Eric Pamer 30 . 8 to 16-week-old female and male mice were used in all experiments. SPF mice were fed a sterilized laboratory rodent diet 5L0D (LabDiet). All animal studies were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Michigan.

Induction of indomethacin-induced small intestinal epithelial injuries.
A single dose of indomethacin (IND) (15 mg kg −1 , dissolved in 5% NaHCO 3 ) (Sigma-Aldrich), injected subcutaneously 31 , was used to induce SI epithelial injury. Control mice were injected with the vehicle alone (5% NaHCO 3 ). The following parameters were used to quantify the severity of SI injuries: body weight, survival rate, weight-to-length ratio of the SI, histology, cytokine mRNA levels in experimental tissues. Endoscopic examination of the murine small intestine. Mice were euthanized and SI tissues were harvested. Luminal contents were flushed with ice-cold PBS. A high-resolution (1,024 × 768 pixels), miniaturized endoscopic system (Colorview Veterinary Endoscope; Karl Stortz) was used to view the images of the SI mucosa 13 .

Chemicals and Reagents. Diphtheria toxin (DT) was obtained from List Biological Laboratories
Quantitative real-time PCR. Mucosal scrapings were collected from the proximal, middle and distal regions of the SI, and then placed in TRIzol reagent (Life Technologies). Purified RNA isolated from the tissues was reverse-transcribed into cDNA and then used for quantitative real-time PCR (qPCR) using SYBR Green Supermix (BioRad, Hercules, CA). The relative expression of target genes was calculated using β-actin as a reference gene. Intestinal permeabilization assay. Mice were denied access to food for 4 hrs prior to gavage with FITC-labeled dextran (500 mg kg −1 , dissolved in PBS, MW 3000 -5000; FD4, Sigma-Aldrich). Blood samples were collected 1 hour later and fluorescence intensity in serum was detected (excitation, 485 nm; emission, 525 nm; BioTek). FITC-dextran concentration was determined using a standard curve generated by serial dilution of FITC-dextran.

Isolation of lamina propria mononuclear cells.
Lamina propria mononuclear cells (LPMCs) were isolated from SI specimens using a modification of a previously described protocol 5,32 . Briefly, the dissected mucosal tissue was incubated in calcium and magnesium-free Hank's balanced salt solution (HBSS) (Life Technologies, Carlsbad, CA) containing 1.5% heat-inactivated fetal bovine serum (FBS) (Life Technologies) and 1 mM dithiothreitol (Sigma-Aldrich) to remove mucus. Epithelial cells were removed by incubation in HBSS containing 1 mM EDTA (Quality Biological, Gaithersburg, MD) at 37 °C, repeated 2 times (first incubation; 10 min, second incubation; 30 min). The tissues were then collected and incubated, with agitation, in HBSS containing 400 U ml −1 of type 3 collagenase and 0.01 mg ml −1 of DNase I (Worthington Biochemical, Lakewood, NJ) for 90 min at 37 °C. The insoluble fraction was pelleted, re-suspended in a 40% Percoll solution (GE Healthcare Life Sciences, Pittsburgh, PA), layered on top of a 75% Percoll solution and centrifuged at 2,000 r.p.m. for 20 min at room temperature. Viable LPMCs were recovered from the discontinuous gradient interface.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.