Gut microbiota regulates gut-lung axis inflammatory responses by mediating ILC2 compartmental migration

Abstract

Gut microbiota is increasingly linked to the development of various pulmonary diseases through a “gut-lung” axis. However, the mechanisms by which gut commensal microbes impact trafficking and functional transition of immune cells, remain largely unknown. Using integrated microbiota dysbiosis approaches, we uncover that the gut microbiota directs the migration of group 2 innate lymphoid cells (ILC2s) from the gut to the lung through a gut-lung axis. We identify proteobacteria as a critical species in the gut microbiome to facilitate (natural ILC2) nILC2 migration, and increased proteobacteria induces IL-33 production. Mechanistically, IL-33-CXCL16 signaling promotes the nILC2 accumulation in the lung, whereas IL-25-CCL25 signals augment (inflammatory ILC2) iILC2s accumulation in the intestines upon abdominal infection, parabiosis and cecum ligation and puncture (CLP) in mice. We reveal that these two types of ILC2s play critical but distinct roles in regulating inflammation, leading to balanced host defense against infection. Overall results delineate that proteobacteria in gut microbiota modulates ILC2 directional migration to the lung for host defense via regulation of select cytokines (IL-33), suggesting novel therapeutic strategies to control infectious diseases.

Introduction

The intestine is an important digestive organ and the largest immune organ in humans and its health is vital for human wellbeing. Intestinal microorganisms (microbiota) refer to the large number of microorganisms present in the intestine (1), which can help the host to perform a variety of physiological and biochemical functions including immune regulation, host defense and homeostasis (2). Disturbance of the symbiotic microbiota may have untoward consequences, such as allergic type 2 immunity and exacerbated inflammation (3). Gut microbiota is critical for the functional programming of immune systems, but the underlying mechanism remains unclear (4). Moreover, the gut microbiota is reported to influence the lung health via altering pulmonary immune defense. The interaction between the intestines and the lungs has been described as a gut-lung axis (5), but the underlying mechanism is poorly defined.

The innate lymphoid cells (ILCs) are divided into three major categories: ILC1, ILC2, and ILC3 with multiple sub-populations (6). Recent studies indicate that intestinal microbiota may drive lung immune responses of ILCs in newborn mice to combat pneumonia by promoting the production of IL-22-derived group 3 innate lymphocytes (ILC3s) into the lung of newborn mice (7). This work underscores a concept of ILCs’ interaction with intestinal microbiota. In addition, ILCs are found to play important roles in lung injury (8), pulmonary infections (8, 9), and inflammatory bowel disease (IBD) (10). Importantly, ILCs protect lung endothelial cells from death in sepsis induced by cecum-ligation and puncture (CLP) or endotoxins (LPS) (11). However, roles of distinct ILC populations in a gut-lung axis remains untested.

ILC2 is involved in the “type 2 immunity” and can be induced by cytokines IL-25 and IL-33 to produce a large amount of IL-5 and IL-13 (12). Studies show that IL-25 and IL-33 act on two types of ILC2: iILC2 (inflammatory ILC2: with ST2⁻IL-17Rb^hi and response to IL-25) and nILC2 (natural ILC2: with ST2⁺IL-17Rb^low and response to IL-33). iILC2 expresses IL-17 and plays an immune defense role in lung infection (13). Hence, we postulate that ILC2 serves as a signal connector between the lung and gut immunity. A recent report suggests that iILC2s could be recruited from the gut to the lung and other organs in response to helminthic infection and that CCR9 is highly expressed in iILC2 and acts as a chemokine for cell migration (14). However, how the migration occurs, such as whether microbiota plays a role in migration of iILC2 and nILC2 to respond to bacterial infection and inflammatory stimuli, is not clear.

Li et al. recently reported that although ILC2s express both CCR9 and CXCR6, CXCL16 rather than CCR9, the ligand of CXCR6, exhibits a direct chemoattractant effect on nILC2 and IL-33 facilitates nILC2s migration in an asthma model (15). Herein, we posit that CCR9-CCL25 and/or CXCR6-CXCL16 may play roles in driving the migration of ILC2s (nILC2) to the lung in inflammatory disease (enteritis) and sepsis models. As CLP is a widely used and clinically relevant sepsis model exhibiting serious celiac inflammation and systemic inflammatory response syndrome (SIRS) (16), we set out to determine the traffic of ILC populations between the lung and gut using several inflammatory models including IBD, CLP-induced sepsis and bacterial infection.

Here, we delineate the dynamic of nILC2 (lineage-CD45⁺ST2⁺CD90.2⁺) and iILC2 (lineage-CD45⁺ST2⁻KLRG1⁺) population in the lung and the gut in mouse models. Our data demonstrate that IL-33 and CXCR6 may play a pivotal role in nILC2 migration, which is also phenocopied in in vitro models after respective chemokine stimulation. Using two-photon microscopy analysis, we find that ILCs migrate to the lung to respond to the abdominal infection in a mouse parabiosis model. Through disturbance of gut microbiota, we discover that the proteobacteria is critical for releasing gut signals to impact nILC2 migration because the increased proteobacteria can promote the expression of IL-33. Hence, specific commensal bacteria are identified important to influence the migration of ILC2s in distinct organs in infectious diseases.

Materials and methods

Mice

C57BL6/J (WT) mice were obtained from Envigo (Indianapolis, IN). Rag1 knockout mice, EGFP knock-in mice and all other mice were housed in University of North Dakota. All animal studies were performed according to the guidelines of NIH and approved by the University of North Dakota (UND) IACUC (17).

CLP

Prior to the procedures, mice were anesthetized by injecting buprinex (2 mg /kg), ketamine (80 mg/kg) and xylazine (10 mg/kg) via intramuscular (i.m.) injection. Then, the abdominal area was disinfected and the cecum was exposed, ligatured at its external third, and punctured with 27-gauge needle. The abdominal musculature and abdominal skin were closed by applying simple #4 suture (Surgical Instruments). NaCl solution (0.9%, 1 mL) was injected i.p. to supplement the lost moisture. After surgery, the mice were put on a 37°C pad during recovery. After recovery, the animals were allowed to drink and eat freely. In the Sham group, the cecum was only exposed but not punctured, and was then returned to the abdominal cavity (18, 19). Animal studies were performed according to the guidelines of NIH and approved by the UND IACUC.

Antibiotic induced gut microbiota dysbiosis

8–12 weeks old C57BL6/J female mice were fed 5 days with drinking water alone or with different antibiotics individually or cocktail (1 mg/ml colistin, 5 mg/ml streptomycin, 1 mg/ml ampicillin and 0.5 mg/ml vancomycin) before constructing inflammatory models. Animal studies were performed according to the guidelines of NIH and approved by the UND IACUC.

Parabiosis

A pair of females (or male) mice of similar weight and age is co-housed for 2 weeks before the surgery. Buprenorphine SR (2 mg/kg) was subcutaneous injected 3 hours prior surgery to ensure effective analgesia. The mice are smeared pre-operative eye lubricant to prevent dry eyes and anesthetized with isoflurane (isoflurane vaporizer, (4–5% v/v) for inducing anesthesia, and (1.5–2% v/v) for maintaining anesthesia). The depth of anesthesia can be confirmed using a toe pinch. The mice were on a heated pad (37°C) in a supine position. The surgical areas on the opposite sides of the two mice are shaved and hair removal cream is applied to remove the hair. Matching skin incisions were made from the elbow (front legs) to the knee (back legs) and then the skins of the two mice were connected with a continuous suture. Buprenorphine (0.02 mg/kg) was administrated immediately after surgery for pain control and repeated once in 24 hours and daily for 3 days. Parabiotic mice were carefully monitored after surgery and maintained under antibiotic-containing water (2.5 mg/liter polymyxin B sulfate) for 2 weeks after surgery and then returned to regular drinking water (20). Cytokine injections (IL-25 or IL-33) were performed on parabiotic mice at least 2 weeks after surgery. 3 days after cytokine injection, mice were euthanized by CO₂ inhalation and cervical dislocation and harvested tissue or observed the cell migration directly with a two-photon microscope (Zeiss).

I.P. induced abdominal infection

8–12 weeks old C57BL6/J mice were i.p. injected 1×10⁷ of P. aeruginosa PAO1. PAO1 was grown in Lysogeny broth (LB) in 37°C in 220 rpm overnight. Next day, 50 μl of bacterial suspension was taken to another fresh 5 ml LB and cultured 2–4 h until the optical density (OD) at 600 nm was 0.6–0.8. Then the cell number was adjusted for infection (1 OD unit = 1X10⁹ cells/ml). Before infection, mice were randomly grouped (n=3 and anesthesia by ketamine. 1X10⁷ bacteria were intranasally (strain in 30 μl of saline) (21). Tissues were harvested for analysis at different times post infection. Animal studies were performed according to the guidelines of NIH and approved by the UND IACUC.

Two-photon imaging mouse cecum

Mice were injected IL-33 for three days, and anesthetized with i.p. ketamine (80 mg/kg) and xylazine (10 mg/kg) after using buprenorphine (2 mg/kg, SQ). Once the mouse was anesthetized, it was secured with a gas mask over the nose with tape. Then i.v. injection of 2μg antibody of ST2 to stain ILC2 cells was performed before surgical process. We opened the abdomen and took out the cecum to soak in 37°C 0.9% sodium chloride drops and maintaining physiological environment while taking images (22). Animal studies were performed according to the guidelines of NIH and approved by the UND IACUC.

qRT-PCR

Total RNA of lung and small intestine was extracted by TRIzol per the manufacturer’s instruction. SYBR Green and gene-specific primers (CCL25-FAAGGCTAGTCCACTGGAAGAGC; CCL25-R GTGGCACTCCTCACGCTTGTAC; CCR9-FGCAGGCTGTTGACGCTTATG; CCR9-RCCTTCGGAATCTCTCGCCAA; GAPDH-FCATCACTGCCACCCAGAAGACTG; GAPDH-R ATGCCAGTGAGCTTCCCGTTCAG; MIL-33FGATGGGAAGAAGCTG ATG GTG; MIL-33RTTG TGA AGG ACG AAG AAG GC; CXCR6-F ATGCCATGGATGGGC;CXCR6-RCTACAATTGGAACATACTGG; MCXCL16FTCCATTCTTTATCAGGTTCCAGTTG; MCXCL16RTGAAAACTCTTCCCATGACCAGT; GATA3-FAACCACGTCCCGTCCTACTA; GATA3-RGGTGGATGGACGTCTTGGAG; IL-25FCGGAGGAGTGGCTGAAGTGGAG; IL-25R ATGGGTACCTTCCTCGCCATG) were used in a CFX Connect System (Bio-Rad). The mRNAs were expressed as the fold change from GAPDH and all detected genes showed fold differences between controls and experimental groups in each experiment (23).

Cell isolation from the lung, spleen, and small intestine.

Spleen, small intestine, and lung tissues were harvested after cold PBS washing and were cut into small pieces with razor blades, then were digested for 30 min at 37 °C with collagenase D. Homogenized cells were obtained by passing 70 μM strainer. Then the cells were treated with ACK solution to break the red cells.

The small intestine was opened and cut into 1 cm pieces and washed in ice-cold PBS buffer and removed fat. The tissues were incubated in PBS containing 5 mM EDTA, 10% FBS and 1 mM DTT at 37 °C for 40 min to separate intraepithelial cells. Homogenized cells were obtained by passing 70 μM strainer. The remained parts were washed twice with PBS and then were digested for 45 min at 37 °C in HBSS buffer containing collagenase D and DNase I. The cells were treated with ACK solution to break the red cells (24).

Analysis of ILC2 cell recruitment by flow cytometry

Cell suspension in PBS were incubated with fluorochrome-conjugated antibodies at room temperature for 30 min. Free fluorescent dye was washed out and cells were fixed by 4% paraformaldehyde for 15 min. The cells were washed with PBS for three times and cells suspension were analyzed on symphony flow cytometer (BD Biosciences) and data were analyzed with FlowJo software (25).

H&E staining

Tissues were fixed in 10% formalin for 48 h and then soaked in 30% until sink to the bottom. The tissues were embedded with OTC compound in liquid nitrogen with the isopentane as a medium. Then the tissues were cut as 8μm and applied to slides. The slides were stored in −80 °C and moved to room temperature for 30 min before staining. The slides were fixed in acetone for 5 min and washed in the running water for another 5 min then followed the commonly used H&E protocol to perform the hematoxylin and eosin stain (Pu et al, 2017). The slides were imaged under a 20X objective in a conventional light microscope. The lung injury was scored for each sample determined by neutrophil accumulation in the alveolar, alveolar wall thickening and the alveolar integrity. The intestinal injury was scored by thickening of small intestinal wall, mucosal inflammatory cell infiltrates and ileum intactness. Each of these parameters was scored on a scale of 0 (absent) to 3 (severe) and summed up to generate injury scores.

16s metagenomic sequencing

Dual-indexed 16S rRNA sequencing was performed on isolated fecal DNA by QIAGEN stool RNA extraction kit. The V3 and V4 regions (26) of the16S rRNA gene was PCR-amplified, clean-up, pooled, normalized, and sequenced using the Illumina MiSeq platform.

Two -photon imaging

Mice were i.v. injected with Alexa 647-labeled CD90.2 antibody for 20 min before sacrifice. Tissues were removed and fixed in the 4% agarose for imaging (27). Cecum showed that the mice were i.v. 5 mg/kg Dextran –TRITC 1 h for blood vessel staining before sacrifice and i.v. BV421-labeled ST2 antibody before the mice were anesthetized with i.p. injection of a mixture of ketamine and xylazine. The cecum was removed similarly to the CLP model but without puncture or ligation and soaked in PBS for imaging.

Migration

1× 10⁵ nILC2s in RPMI1640 medium were placed in the upper chamber of an 8-μm pore-size, 24-transwell plate. 600 μl of test chemotaxis consisting of various concentrations of rmIL-33, rmIL-25, rmCXCL16, and rmCCL25 diluted in RPMI 1640 were placed to the lower chambers and cultured in a cell incubator for 24 h. Migrated cells from the upper chamber to the lower chambers were stained with 1% crystal violet and counted (15).

Statistical analysis

Statistical analysis was performed by one-way ANOVA with Tukey post hoc tests or student t test. The results were expressed as the mean ± SD and significance difference between two groups was determined as p < 0.05 (28).

Results

Microbial dysbiosis mediates disturbance in ILC2 populations.

Intestinal commensal microbiota is shown to influence immunity as disruption of the symbiotic microbiota leads to exacerbated type 2 immunity, inflammation (29–31) and celiac infection (32); however, whether the intestinal commensal microbiota facilitates ILC2s expansion and migration remains unknown. Herein, we investigated the critical role of ILC2 in regulating immune response to control infection and maintain lung homeostasis by a microbiota-mediated “gut-lung” axis. We fed female mice with water containing 1 mg/ml colistin, 5 mg/ml streptomycin, 1 mg/ml ampicillin and 0.5 mg/ml vancomycin individually or cocktail (all 4 antibiotics) for 3 days. Analysis of immunohistochemistry-stained section showed antibiotics administration induced intestinal tissue damage (Figure 1A) and colon shortening (Figure 1B). Populations of ILC2 in the lung and gut were analyzed by fluorescent-activated cell sorting (FACS). The data showed that both nILC2 and iILC2 were disrupted in the lung and in the small intestine after drinking antibiotics (individual or cocktail) containing water compared to drinking water alone and showed different responsive profiles with various antibiotic treatments (Figure 1C–F). We observed a general pattern: inhibition in ILC2 populations in the lungs but increase in ILC2 populations in the gut (except ampicillin treatment). Colistin did not induce significant changes in inflammation and cell populations due to its absorption capacity by oral administration. Vancomycin just affected iILC2 in the gut while ampicillin and streptomycin have same effects in the gut but not in the lung, and these differences may be due to multiple factors including the intrinsic characteristics of different antibiotics, various immune cells, and diverse virulence factors within bacteria. Hence, further investigation may help determine relevant microbial species in response to various antibiotics treatment. Additionally, the decrease in ILC populations in the lung is consistent with the decreased GATA3 expression (Figure S1A–B). nILC2 population changes were also observed in the groups of P. aeruginosa intraperitoneal injection and cecum ligation and puncture (CLP)-induced sepsis (Figure 1G–H). Moreover, after returning normal drinking water with the ampicillin-treated mice, intestinal morphological alterations and lower ILC2 populations in the lung were partially restored (Figure S1C–D). These comprehensive analyses with controls, CLP models and various combination of antibiotic treatments suggest that the accumulation of ILC2 in the lung is impacted by gut microbiota. To test the roles of T and B cells, we employed rag^−/- mice to generate CLP intestinal disease model and examined ILC2 population and found that there was no significant difference between rag^−/- mice and WT mice (Figure 1I–J). This result implies that ILC2 disturbance is independent of T and B cells. Taken together, these data demonstrate that microbiota potentially modulates the balance of ILC2 population via a “gut-lung” axis upon immune reaction.

Figure 1.

Microbiological disorder-induced intestinal inflammation disrupts ILC2 populations in the lung. (A) Mice were fed 5 days with drinking water alone or with different antibiotics individually or cocktail (1 mg/ml colistin, 5 mg/ml streptomycin, 1 mg/ml ampicillin and 0.5 mg/ml vancomycin). Tissues injury were detected by H&E staining (n=6). (B) Morphological changes of cecum were observed directly. The colon lengths were directly observed and measured (n=3). (C-F) nILC2 (lineage-CD45+ ST2+ CD90.2+) and iILC2 (lineage-CD45+ ST2- KLRG1+) in the gut(c-d) and lung (E-F) were detected (n=3). (G-H) nILC2 population changes in the lung from 48 h post infection or surgery mice. (Infection: i.p. 1×10⁷ P. aeruginosa, Surgery: CLP). Rag−/- mice and WT mice were subjected to CLP surgery (n=3). (I-J) nILC2 and iILC2 populations were detected in the lung at 48 h post-surgery. The results were expressed as the mean ± SD from three independent experiments and significant difference between two groups was defined by one-way ANOVA with Tukey post hoc tests *p < 0.05 (Except I-J, all treatment groups were compared with controls).

Proteobacteria activation in gut microbiota promotes nILC2 lung accumulation by inducing IL-33.

We next attempted to identify the critical microbes required for maintaining ILC2 homeostasis as microbiota is implicated in modulating immune cell function (33, 34). Specifically, we elucidated how microbiota may affect ILC2 function. The stools from the small intestines after antibiotics treatment or CLP procedures were collected to perform 16s metagenomic sequencing. In terms of alpha diversity, the microbiota from CLP mice or antibiotic-treated mice was more diverse than un-treated mice (P < 0.05) (Figure 2A). For beta diversity, UniFrac analysis demonstrated two separate clusters reflecting sharp differences in microbial compositions between CLP and control mice (Figure 2B). As to taxonomical differences, there were also significant alterations in the abundance of families, genera, and even specific Amplicon Sequence Variant (ASV) between the treatment group and control group (Figure 2C–E). Both CLP procedures and ampicillin treatment increased the nILC2 population in the lung as shown above (Figure 1), and another common point in these two groups in abundances of families was noted: proteobacteria especially Escherichia shigella is significantly increased after the CLP procedures and ampicillin treatment (Figure 2E–F). Different from the ampicillin treatment, streptomycin treatment did not induce nILC2 increase in the lung but significantly caused iILC2 increase in the gut. Consistently, the ASV results showed the common changes, increased proteobacteria for CLP procedures and ampicillin groups (Figure 2F). Using KEGG signal pathway (Figure S2A) and functional orthologs analysis (Figure S2B), we also noticed similar changes between the treatments. After challenged mice with E. coli (one species of the Escherichia) or P. aeruginosa (control), we observed that E. coli induced high levels of IL-25 and IL-33 (Figure S3A–B). Taken together, these results showed that CLP procedures and antibiotics treatments remarkably altered the gut microbiota, with the proteobacteria being the most highly altered species that may be related to changes in quantity and inter-organ traffic of ILC2 populations.

Figure 2.

Microbiota is disrupted by CLP procedures and antibiotics treatment. 16S Metagenomic Sequencing was performed as described in Methods. (A) Alpha diversity by ASVs for fecal collected from was determined by Observed, Chao1, Shannon, and Simpson indices in mice. Data are presented by box plots with horizontal line indicating the median for each data set. *P<0.05, by t test. (B) PCoA plots representing beta diversity by ASVs of bacterial communities from feces. (C-D) Distribution of bacterial communities at phylum level by ASVs in feces. (E-F) Differential abundance by ASVs in colon expressed as Log2FC each point represents an ASV at the genus level P-value<0.05. These data were generated from animal experiments described in Figure 1.

The bacteria in proteobacteria including E. coli and Enterococcus sp. were reportedly associated with cytokine production (i.e., IL-33) during intestinal infection (35). We tested several cytokines and noted that IL-33 was pronouncedly elevated in the peritoneal lavage fluids (PLF) of CLP mice while cytokines IL-4, IL-5 and IL-25 were not (Figure 3A–D). These results underscore that the type 2 immunity may be not activated at 48 h post CLP surgery. In the antibiotic-treated mice, IL-25 and IL-33 expression followed the same pattern of the ILC2 cell population: Ampicillin treatment induced the highest expression of IL-33 in the lung and small intestine tissues (Figure 3E–H), which is consistent with the proteobacteria depletion in the microbiota sequencing in figure 2D. By i.p. injecting IL-33 and IL-25 to mice, respectively, we examined these cytokines’ effects on nILC2/iILC2 populations, and found that IL-25 preferentially induced iILC2, while IL-33 induced pronounced nILC2 population in the lung (Figure 3I) and in the MLN (Figure 3J–K), consistent with the GATA3 expression in the small intestine (Figure 3L). To further confirm the role of gut microbiota in inducing IL-33 or IL-25, we fed mice with ampicillin-containing water for three days and then returned them to normal sterile water for 6 days. Then IL-25 and IL-33 cytokines from bronchoalveolar lavage fluid (BALF) and peritoneal lavage fluid (PLF) or ILC2 populations in the gut and lung were detected. The results showed after the microbiota recovered, IL-25 and IL-33 returned to nearly the levels prior to treatment (Figure S3C–F). These data illustrate that the ILC2 population changes are associated with microbial dysbiosis in an IL-33- or IL-25-dependent manner after specific antibiotic treatment.

Figure 3.

Microbial dysbiosis leads to aberrant levels of IL-25 and IL-33. (A-D) Mice were subjected to CLP procedures. 48 h post-surgery or injection, mice were sacrificed for analyzing IL4, IL-5, IL-25 and IL-33 levels in peritoneal lavage fluids and serum using ELISA (n=3). (E-H) Tissue’s IL-25 and IL-33 levels were detected by qPCR (n=3). (I-K) Mice were i.p. injected with 2μg/mouse of IL-25 or IL-33. At 24 h and 48 h post-CLP procedures or cytokine injection, mice were sacrificed for analysis. Representative flow cytometric analysis of nILC2 and iILC2 populations in the lung (I) and mesenteric lymph nodes (MLN) (J-K) (n=3). (L) GATA3 expression in the small intestine tested by qPCR (n=3). The results were expressed as the mean ± SD from three independent experiments and significant difference between two groups was defined by one-way ANOVA with Tukey post hoc tests *p < 0.05.

IL-33 accelerates nILC2 migration in vitro and in vivo.

ILC2 cells generate substantial effector cytokines in response to IL-25 and IL-33 stimulation (36, 37). Our data demonstrate that microbiota-mediated regulation of IL-25 and IL-33 would increase the growth of the ILC2 population. However, whether the similar mechanism also drives ILC2 migration from the gut to the lung is not tested. To probe the impact of the increased IL-25 and IL-33, we isolated nILC2s and cultured them in the upper chamber of a 24-well transwell plate for 24 h and the lower chamber containing IL-25 or IL-33 to investigate ILC2 trafficking. We found that IL-33 attracted nILC2 migrating from the upper chamber to lower chamber at a concentration of 15 ng/ml. However, IL-25 only induced moderate migration until the concentration reaching 120 ng/ml (Figure 4A–B). These data indicate that rather than a static cell, nILC2 may be migratory cells towards a specific compartment to protect hosts from inflammatory or injury sites. To further assess the migration in vivo, we first detected blood samples from the CLP model and the mice received IL-25/IL-33 priming, respectively. Increased nILC2 were noted in the blood of CLP mice or cytokine-challenged mice (Figure S4B), suggesting that nILC2 may migrate through the blood from one tissue to another. Importantly, we utilized parabiosis mice to verify the migration of ILC2 by surgically combining two mice (one mouse expressing EGFP, another WT C57BL/6N mouse) to share the blood. After the EGFP expressing mouse received IL-33 by injection, we isolated ILC2 cells from the WT mouse to perform flow cytometry analysis, which was gated as EGFP+ lin- ST2+ or as EGFP+ lin- KLRG1+ (Figure 4C). Remarkably, we found that the lung of WT mice contained the most abundant nILC2 migrated from the EGFP mouse (Figure 4D), which may be corresponding consequences of inflicted inflammatory responses in the recipient mouse. These data support our hypothesis that the tissue-resident nILC2 cells possess the migration ability to move between tissue sites. Moreover, iILC2s migrate to diverse tissues while the majority reside in the gut (Figure 4E), consistent with a previous report that iILC2s as circulating cells reside in intestinal lamina propria (38). We also injected Alexa594-conjugated CD90.2 (a cell surface marker of nILC2) through the tail vein 30 min before separating the WT mice in the parabiosis model to observe the cells under the two-photon microscopy. Intravital imaging showed that EGFP expressing cells also displayed nILC2 markers (stained with CD90.2), indicating that these nILC2 cells trans-migrated from the EGFP-expressing mouse to the WT recipient mouse, which transmitted the activated immunity to the naïve mouse (Figure 4F). To validate this observation, we injected BV421-conjugated ST2 and dextran (39) via the tail vein in the normal mouse after stimulated with IL-33 and removed the cecum to perform two photon imaging analysis to observe ST2 positive cells (activated iILC2) in the blood vessel (Figure 4G). The cell proliferation marker ki67 staining also showed that IL-33 augmented the proliferation of nILC2s and iILC2s (data not shown). To better understand the mechanism of cell movement, we isolated nILC2 from mice and stimulated them with IL-33 to observe the cell motility under microscopy. The video showed the nILC2 moved much faster after IL-33 stimulation in vitro while control cells did not (data not shown). Overall, these data demonstrate that IL-33 not only accelerated nILC2/iILC2 proliferation, but also facilitated nILC2 migration.

Figure 4.

IL-33 drives ILC2 migration in vivo and in vitro. (A-B) nILC2s were isolated by a commercial kit as described in the Methods and cultured in RPMI1640 medium in the upper chamber of a transwell plate. 600 μl of test chemotaxis consisting of various concentrations of rmIL-33 and rmIL-25 in RPMI 1640 were placed in the lower chambers of the transwell plate and cultured in normal cell incubator for 24 h. Migrated cells from the upper chambers to the lower chambers were stained with 1% crystal violet and evaluated and representative cells were counted (n=3). (C-G) Pairs of female or male mice (EGFP +/+ and WT) of similar weight and age were subjected to parabiosis surgery. Two weeks after procedures, the EGFP mice were injected with 2 μg IL-33 and the cells were isolated and analyzed. Showmen diagram of the parabiosis (C). Relative number of nILC2 or iILC2 cells among 30000 EGFP positive cells in different organs (D-E). The results were expressed as the mean ± SD from two independent experiments and significant difference between two groups was determined by one-way ANOVA with Tukey post hoc tests *p < 0.05. Images of the WT lung and small intestines after surgery and IL-33 induced. The CD90.2 antibody was injected from tail 20 min before sacrificing the mice to stain nILC2. Tissues were observed by a two-photon microscopy (F). Images about blood vessel & nILC2 cell in small intestines after IL-33 induction (G) (n=3). Statistical analysis was performed by one-way ANOVA with Tukey post hoc tests. The results were expressed as the mean ± SD from three independent experiments and significant difference between two groups was determined by one-way ANOVA with Tukey post hoc tests *p < 0.05.

IL-33-CXCL16-CXCR6 axis regulates the dynamic of nILC2 population.

Besides cytokines and their receptors, chemokine-mediated signals also contribute to cell migration (40). To gain the mechanistic insight into the effects of ILC2 population migration, we profiled cell traffic proteins in the small intestine of CLP mouse and control mouse with a RT² Profiler™ PCR Array Mouse Cytokines & Chemokines kit and found that 17 genes were upregulated among the tested 96 (Figure 5A), including CXCL16, the ligand of CXCR6. It is notable the ILC2s primarily express CCR9 and CXCR6 (15), so the increase of CXCL16 is in agreement with changes in the ILC2 population. The expression of chemokine receptors and ligands was also detected in the intestine of the antibiotics-treated mice. CXCL16 was found to be highly expressed in the lung and gut after ampicillin-treatment (Figure 5B–C), consistent with the increase in nILC2 population (Figure 1). The CXCR6 receptor was increased in the gut while not in the lung after ampicillin treatment (Figure S4A–B) and both CCR9 and its ligand CCL25 were decreased in the lung but increased in the gut (Figure S4C–F). The transwell analysis showed that CXCL16 attracted nILC2 migration from the upper chamber to the lower chamber at the concentration of 15 ng/ml. However, CCL25 only induced mild migration at the dose of 60 ng/ml (Figure 5D). To further understand the relationship of CXCR6/CCR9 with the ILC2s, we injected mice with IL-25 or IL-33 and performed flow cytometry analysis at 3 days. The data showed that both nILC2 and iILC2 augmented the expression of CCR9 and CXCR6 after the stimulation. We observed that IL-25 increased the expression of CCR9 in iILC2 while IL-33 upregulated CXCR6 expression in nILC2 (Figure S4G–J). Then we isolated nILC2 cells from the mouse with an ILC2 enrichment kit and co-cultured nILC2 with IL-25 or IL-33 at the indicated concentrations. The expression of CCR9 and CXCR6 was increased vs. the controls as observed by microscopy, and the result is in good agreement with the flow cytometry analysis (Figure 5E–F). These data indicate that CXC16 and IL-33 may be involved in the nILC2 migration.

Figure 5.

CCR9, CXCR6 and their ligands are required for ILC2 migration. (A) Array analysis showed the chemokines and receptors changes in the small intestine tissue in the mice subjected to CLP surgery. (B-C) qPCR showing the CXCL16 levels in the lung and small intestine. (n=3). (D) rmCXCL16, and rmCCL25 diluted in RPMI 1640 were placed in the lower chambers of transwell plates and cultured in normal cell incubator for 24 h. Migrated nILC2 cells from the upper chambers to the lower chambers of transwell plates were stained with 1% crystal violet and evaluated and representative cell images of three experiments were shown. Then the stained cells were dissolved with 95% ethyl alcohol and detected OD590 to reflect the migrated cell numbers of three experiments. (E-F) nILC2 cells were isolated by a commercial kit as described in the Methods. Cells were cultured in 1640 medium with indicated concentrations of IL-25, IL-33 for 24 h. CXCR6, CCR9 and IL-13 were stained with corresponding antibodies and imaged by confocal microscopy (n=3). Statistical analysis was performed by one-way ANOVA with Tukey post hoc tests. The results were expressed as the mean ± SD from two independent experiments and significant difference between two groups was defined as *p < 0.05.

Blockade of ST2 and CCR9 impedes ILC2 circulation and aggravates tissue damage in a sepsis model.

To evaluate whether ILC2s circulation affects inflammatory progression in animals, we detected the levels of IL-33 and IL-25 in the models of abdominal inflammation by ELISA and found that IL-33 and IL-25 reached the highest levels at 6 h post-CLP procedures or cytokines’ injection and thereafter petered off (Figure 6A–B), suggesting that ILC2 response to the stimulation is dynamic and rapid. After blocking CCR9 and IL-33 receptor ST2, we failed to observe an increase in ILC2 in the lung or gut in the CLP model (Figure 6C–D) and the tissues exhibited more severe damage due chiefly to the loss of ILC2 (Figure 6E–F), indicating that ILC2s may play a role in attenuating tissue injury. Hence, these data confirmed that ILC2 are protective from tissue injury during aberrant inflammatory responses. To further dissect the regulators of the signal pathways for the migration of nILC2, we performed IL-33 induced transwell assay using a series of chemical inhibitors: JNK inhibitor II SP600125, potent PI3K inhibitor LY294002, TGF-β receptor inhibitor, NF-κB inhibitor JSH-23 and STAT3 inhibitor. The data showed that these signaling pathways appear to be involved in nILC2 migration (Figure 6G), but the NF-κB signal seems to be most relevant due to the most profound change. Our assessments suggest that although the classical cell growth and inflammation signal regulators are related to the nILC transmigration, the master transcription factor NF-κB for inflammatory cytokines is a major player. Taken together, these findings unveil a potential role of IL-33/CXCL16-CXCR6 and IL-25/CXCL25-CCR9 signaling pathways in augmenting ILC2s function during severe sepsis and abdominal inflammatory diseases.

Figure 6.

Blockade of IL-33R and CCR9 results in more severe tissue damage. (A-B) Mice were subjected to CLP surgery or i.p. injected 2 μg/mouse of IL-25 or IL-33. At 24 h and 48 h post-surgery or injection, mice were sacrificed for analysis. IL-25 and IL-33 levels in the serum evaluated by ELISA (n=3). (C-F) Mouse was injected with 2 μg of ST2 or CCR9 antibody 20 min before performing CLP surgery. 12 h post surgery, mice were sacrificed for analysis. nILC2 population changes (C-D). The phenotype changes in different groups (evaluating macroscopic aspects of ceca) (E). H&E staining showing the injury of the small intestines and lung. For tissue injury extent, a score was made in each of these samples using standards as described in methods (F). (G) nILC2 was isolated from the WT mice and treated with different inhibitors (JNK inhibitor II SP600125 (10 μM), PI3K inhibitor LY294002 (10 μM), NF-κB inhibitor JSH-23 (20 μM), TGF-β receptor inhibitor (2 μM), and STAT3 inhibitor (10 μM)) 1 h before adding 60 ng/ml IL-33 for transwell plate experiments. Cells in the lower chamber and bottom of the upper chamber were fixed with 4% paraformaldehyde and stained with 1% crystal violet for 10 min. After washed the excessive dyes by PBS, the stained cells were dissolved with 95% ethyl alcohol and detected OD590 to reflect the migrated cell numbers (n=3). Statistical analysis was performed by one-way ANOVA with Tukey post hoc tests. Assays were performed three times and data expressed as the mean ± SD and significant difference between two groups was indicated as *p < 0.05.

Discussion

In this study, we identify the evidence that disruption of the intestinal microbiota affects the population and migration of ILC2s in the lung and gut, which is required for a robust host defense during inflammation and sepsis. The immune system constitutes the innate immunity and adaptive immune system. ILCs, belonging to innate immunity, are located in mucosal surfaces to initiate early immune responses and promote lymphoid organogenesis (41). However, excessive, or poorly controlled activation of ILCs may cause overzealous inflammatory responses, leading to sepsis, septic shock, and ultimately death. Increasing evidence suggests that microbiota influences host immunity and inflammation. Elucidating the role that microbiota may modulate ILCs expansion and migration can shed light on critical pathways and key molecules that are involved in inflammatory diseases (e.g., severe inflammatory response syndrome), infection, and sepsis.

We further demonstrate that the proteobacteria species in gut microbiota promotes abundance and traffic of nILC2s through the IL-33/CXCL16 axis, supporting a concept that tissue-resident nILC2s are not static and can have the ability of migration. Moreover, the IL-25/CCR9 axis is important for inflammatory ILC2s (iILC2s) accumulation dependent on diseases or model systems. Pharmacological blockade of IL-33R or CCR9 signaling pathway intensified inflammatory response and tissue injury during sepsis, indicating that cell migration is a critical event in host defense and may have therapeutic potential for inflammatory diseases. Our work provides a mechanistic framework of how commensal bacteria influence the ILC2 compartmental specific accumulation and migration during microbiota dysbiosis.

Microbiota is the most important external environment within the human body. Various microorganisms form an ecological balance in quality and quantity within the body. At present, firmicutes, bacteroidetes, proteobacteria, actinobacteria, and microbacteria and fusobacteria were identified to be the main constituents, in which Firmicutes and Bacteroidetes are most abundant (> 92.6%)(42). However, not all of the microorganisms are friends of human beings. According to the characteristics, bacteria can be divided into three categories, namely, good, bad and neutral bacteria. When the proportion of beneficial bacteria in the human intestines decreases and the number of bad bacteria increases, the body’s immunity is compromised, referring to dysbiosis that is associated with various diseases. In our experimental models, the proteobacteria species was found to be highly increased, consistent with a previous finding: the increase of E. coli and Enterococcus sp was positively correlated with the IL-6, IL-22, IL-17 and IL-33 levels (35). Other studies have shown that microbiota is closely related to the occurrence of a variety of diseases, such as digestive system diseases (43), mental disorders (44) and infectious diseases (45). To date, numerous studies have confirmed that the intestinal Proteobacteria including Enterobacteraceae, Vibrionaceae, Pseudomonadaceae, Salmonella and Yersinia may reflect micro-ecological disorders or unstable intestinal microflora structure (46, 47). The healthy mammalian gut contains abundant symbiotic bacteria, proteobacteria, whereas their loss causes inflammation in dysbiosis condition. Therefore, it is believed that a better understanding of proteobacteria is the key for maintaining the gut health of mammals. Nevertheless, the potential roles of the proteobacteria are not fully understood, such as the function of induced cytokines including IL-22, IL-33, IL-6, IL-17 and IL-10 (35, 48), which mediates a variety of physiological and pathological responses of immune cell maturation, activation, proliferation and immune regulation. Hence, the underlying principle of proteobacteria dysbiosis-induced abnormality is largely unknown.

Ampicillin is a semi-synthetic broad-spectrum penicillin. The antibacterial mechanism is to prevent the cell wall synthesis of bacteria, so it not only inhibits bacterial proliferation, but also directly kills bacteria. Neisseria gonorrhoeae, Meningococcus, Influenza bacillus, Pertussis, Typhoid, Paratyphoid bacillus, Proteus mirabilis and E. coli are sensitive to ampicillin but susceptible to resistance. Klebsiella pneumoniae, Proteus-positive Proteus, and P. aeruginosa are not sensitive to ampicillin (49, 50). Colistin sulfate has a strong inhibitory effect on Gram-negative bacteria (especially E. coli, Salmonella, P. aeruginosa, Proteus and Haemophilus), but has no effect on Gram-positive bacteria. Colistin sulfate is difficult to absorb orally and has low toxicity (51). Vancomycin can inhibit the synthesis of bacterial cell wall, and has a strong effect on Staphylococcus aureus, Streptococcus pyogenes, Streptococcus pneumoniae, etc., and has a good effect on Bacillus licheniformis, Bacillus anthracis, and Diphtheria (52, 53). Streptomycin against many Gram-negative bacilli such as E. coli, K. pneumoniae, Enterobacter, Salmonella, Shigella, Brucella, Pasteurella, and Neisseria gonorrhoeae (54, 55). Combining the characteristics of the above antibiotics, the main difference between ampicillin and streptomycin treatment may be that the sensitivity to E. coli and Enterococcus sp, rendering them different ability to induce IL-33 and driving ILC2 moving to the lung (Figure 1C–F). However, certain antibiotics exhibit not only anti-microbial properties but also unwanted effects on the immune system, which may contribute to complexity and potentially compromise the data of the current study. However, controls without antibiotics, individual antibiotic and different combination of antibiotics, and CLP sepsis model (for created infection), were all used to demonstrate that not only antibiotics but also additional factors can impact microbiota in our experimental systems. These comprehensive analyses at least partially support that antibiotics can influence microbiota, and in turn impact immunity against infection. In addition, multiple approaches including E. coli infection, Pseudomonas infection, parabiosis animal models as well as cytokine agonists and antagonists (antibodies and inhibitors) were employed to differentiate and validate the specific ILC2 cell migration and the pathophysiological sequelae.

With improved understanding of ILC2’s classification, function, development and differentiation, these cells are implicated in therapeutic potential for human diseases. However, in order to fully assess the therapeutic value of ILC2, some major issues remain to be resolved. The most difficult and controversial issue is to confirm that nILC2 and iILC2 are unique at the different stages and may transdifferentiate (change of the phenotypes between them). As both of them are regulated by GATA3 differentiation, their surface makers are also quite similar. Additionally, the role of ILC2 in the pathological response of intestinal diseases needs to be further elucidated, especially the relationship between ILC2 and other cells. In our study, the accumulation of ILC2 is independent of T cells and B cells. However, the Locksley team reported that Tuft cells can also promote ILC2 activation and secrete IL-13 by adding IL-25. Hence, ILC2 have the ability to interact with other cell types to exert the immune defense function.

An important finding of this work is that there is strong relationship between ILC2 migration and microbiota by identifying proteobacteria as the critical species in the gut microbiome to upregulate nILC2 migration. Importantly, this species from microbiota may regulate ILC2 trafficking through the organ-specific production of select cytokines dependent on the experimental conditions and external stimuli. Another highlight is that IL-33-CXCL16 signaling guides the nILC2 accumulation in the lung, whereas the IL-25-CCL25 axis facilitates iILC2 migration into the intestines, which only occur upon severe abdominal infection and CLP-induced sepsis in mice. Moreover, we extended previous observations that nILC2 can also migrate from the gut to the lung. Furthermore, we attempted to dissect the mechanism of distribution of different ILC2 populations between the lung and gut during infection or microbiota perturbation; however, they are not the only organs linked to trafficking of ILC2s. Other organ specific homing factors and regulating networks remain largely unknown and a number of other points are still unclear. For instance, what are the driving forces that direct ILC cell trafficking and how the microbiota regulates the transcription and secretion of select cytokines (IL-25 and IL-33)? Finally, some of our data might have not shown a drastic level of alterations in treated groups vs. the controls, but multiple approaches including knockout mice, infection, parabiosis animal models and microbiota perturbation as well as various cytokine agonists and antagonists (antibodies and inhibitors) were employed to comprehensively analyze the interorgan migration of distinct ILC populations and pathophysiological sequelae. Collectively, the current study establishes a framework to understand the migration and functional role of ILC2 across remote organs and the relationship with gut microbiota, indicating potential therapeutic targets for sepsis-like inflammatory diseases.

Key points.

• Gut microbiota involves in innate immune cells’ regulation.

• IL-33 plays important role in nILC2 migration.

• nILC2 and iLC2 migrate to respond to infection and microbial dysbiosis.