Abstract
In obesity, IL-13 overcomes insulin resistance by promoting anti-inflammatory macrophage differentiation in adipose tissue. Endogenous IL-13 levels can be modulated by the IL-13 decoy receptor, IL-13Rα2, which inactivates and depletes the cytokine. In this study, we show that IL-13Rα2 is markedly elevated in adipose tissues of obese mice. Mice deficient in IL-13Rα2 had high expression of IL-13 response markers in adipose tissue, consistent with increased IL-13 activity at baseline. Moreover, exposure to the type 2 cytokine-inducing alarmin, IL-33, enhanced serum and tissue IL-13 concentrations and elevated tissue eosinophils, macrophages, and type 2 innate lymphoid cells. IL-33 also reduced body weight, fat mass, and fasting blood glucose levels. Strikingly, however, the IL-33–induced protection was greater in IL-13Rα2–deficient mice compared with wild-type littermates, and these changes were largely attenuated in mice lacking IL-13. Although IL-33 administration improved the metabolic profile in the context of a high fat diet, it also resulted in diarrhea and perianal irritation, which was enhanced in the IL-13Rα2–deficient mice. Weight loss in this group was associated with reduced food intake, which was likely related to the gastrointestinal effects. These findings outline both potentially advantageous and deleterious effects of a type 2–skewed immune response under conditions of metabolic stress, and identify IL-13Rα2 as a critical checkpoint in adipose tissues that limits the protective effects of the IL-33/IL-13 axis in obesity.
This article is featured in In This Issue, p.1231
Introduction
Obesity is a growing epidemic, with high body mass index (BMI) contributing to disease burden worldwide (1). Emerging evidence suggests that obesity is supported and maintained by immune activation in metabolic tissues. The lean state is characterized by a protective anti-inflammatory milieu in adipose tissue, consisting of eosinophils, type 2 innate lymphoid cells (ILC2), T regulatory cells (Treg), and anti-inflammatory macrophages. In contrast, obesity is associated with a shift toward reduced eosinophilic inflammation, with elevated proinflammatory macrophages and CD8+ T cells (2, 3). In mouse models of obesity, the Th2 cytokines IL-4 (4), IL-13 (5), and IL-5 (6) can promote glucose tolerance and insulin sensitivity. IL-4 and IL-13 polarize macrophages to an anti-inflammatory phenotype through the IL-13Rα1/IL-4R receptor (2), whereas IL-5 signaling drives the development and recruitment of eosinophils (7). The association of type 2 cytokines with metabolic regulation suggests that glycemic control may be achieved through cytokine manipulation (8).
Whereas IL-13Rα1 can mediate cellular responses to both IL-4 and IL-13, an additional receptor form, IL-13Rα2, interacts with IL-13 but not IL-4. It is thought to act primarily as a decoy, binding IL-13 with higher affinity than IL-13Rα1, and sequestering IL-13 from the signaling receptor, thereby reducing Stat6-mediated signaling activity (9). IL-13Rα2 also mediates efficient internalization and depletion of extracellular IL-13 (10). Although IL-13Rα2 lacks any known cytoplasmic signaling motif, its ability to mediate cellular activation responses is debated (11). In addition to the membrane-bound receptor, mice but not humans have a soluble form of IL-13Rα2 (sIL-13Rα2) that is abundant in the circulation (12, 13). In contrast to the IL-13Rα1/IL-4Rα signaling receptor, which is widely expressed, tissue expression of IL-13Rα2 is normally very low, and is induced under conditions of high IL-13 release (14). In the absence of IL-13Rα2, circulating levels of IL-13 are reduced, suggesting that the soluble form acts as a carrier of IL-13 in murine blood (15). Mice lacking IL-13Rα2 have elevated tissue IL-13, consistent with impaired cytokine clearance, and have enhanced IL-13 bioactivity due to impaired cytokine neutralization (15), leading to exacerbated fibrotic responses (16, 17), increased smooth muscle contractility and epithelial resistance (18), and aggravated atopic responses (19, 20) driven by IL-13. Although the role of IL-13 in metabolic homeostasis is becoming appreciated (21–23), the contribution of IL-13Rα2 to regulation of this response has not been explored.
The alarmin IL-33 is stored in epithelial cells, endothelial cells, and fibroblasts, and is released upon tissue damage to mediate local immune activation (24). A key effector mechanism is the IL-33–induced release of cytokines, principally IL-5 and IL-13, from ILC2, Th2, mast cells, basophils, and other cell types expressing the cell surface IL-33 receptor, ST2 (24). Administration of IL-33 to mice triggers a range of IL-13–dependent responses (25–28), but may also induce production of IL-4 and IL-5 (25, 28–31), either of which could modulate inflammation in adipose tissue (4, 6, 32). IL-33 has been shown to protect mice from the metabolic consequences of obesity (29), but the extent to which this is dependent on induction of IL-13, as opposed to other cytokines, has not been evaluated.
In this study, we manipulated tissue IL-13 levels by blocking depletion through IL-13Rα2, both in untreated mice and in mice driven to express high levels of endogenous IL-13 by administration of IL-33. We examined effects on glucose homeostasis under conditions of a normal chow diet and high fat diet (HFD), and explored the IL-13 dependence of these responses. Our findings support that IL-33 regulation of metabolic homeostasis is largely IL-13 dependent, and that extreme type 2 skewing following IL-33 administration in the presence of IL-13Rα2 deficiency ameliorates the metabolic consequences of HFD. Although these observations suggest that manipulation of tissue IL-13 responses could be beneficial in restoring glycemic control on HFD, these effects were accompanied by increased risk of gastrointestinal (GI) toxicity.
Materials and Methods
Mouse model of diet-induced obesity
Male il13rα2−/− and il13−/−
Human adipose tissue RNA
Human adipose tissue RNA was obtained from ZenBio (Research Triangle Park, NC), from tissue isolated under Institutional Review Board–approved protocols. Omental visceral adipose tissue (VAT) was collected from subjects undergoing gastric bypass surgery. s.c. white adipose tissue (scWAT) was collected from abdomen, thighs, or hips of subjects undergoing elective liposuction or abdominal plasty.
Flow cytometry
Adipose tissue was collected in ice cold HBSS + 0.5% BSA, minced, and digested by 1 mg/ml collagenase type I (Worthington Biochemical, Lakewood, NJ) for 50 min at 37°C. For the final 5 min of incubation, 2 mM EDTA was added. Cells were washed, passed through a 100 μm filter, and centrifuged to separate the stromal vascular fraction (pellet) from the floating adipocytes as described previously (33). The stromal vascular fraction was resuspended in RBC lysis buffer (8.3 g/l NH4Cl, 1 g/l KHCO3, 0.09 g/l EDTA, pH = 7.3), incubated for 5 min at room temperature then washed with FACS buffer (PBS + 0.5% BSA + 2 mM EDTA). Cells were treated with rat purified anti-mouse CD16/CD32 (mouse Fc block) for 5 min on ice then stained for surface markers with combinations of the following conjugated Abs in Brilliant Stain Buffer: anti-mouse CD45-BUV395, Siglec F-PE, CD11b-BB515, CD3e-PE, CD5-PE, CD19-PE, CD4-PE, NK1.1-PE, CD11c-PE, CD11b-PE, F4/80-PE, FcεRlα-PE, ST2(IL-33R)-biotin, CD25-BV421, and streptavidin–allophycocyanin (secondary). Staining Abs were purchased from BD Biosciences (San Jose, CA). Stained cells were acquired on an LSRFortessa (BD Biosciences) and data analyzed using FlowJo software version 10 (Tree Star, Ashland, OR). ILC2s were gated as CD45+Lin−CD25+ST2+ as previously described (34) and lineage negative cells were defined as CD3e−CD5−CD19−CD4−NK1.1−CD11c−CD11b−F4/80−FcεRlα−. Eosinophils were gated as CD45+CD11b+SiglecF+.
Analysis of circulating lipids
Blood was collected by cardiac puncture and immediately following euthanasia by carbon dioxide. For serum measurements, blood was allowed to clot at room temperature for 2 h before centrifugation to separate the serum for further analysis. Total cholesterol and high-density lipoprotein (HDL) fraction were quantified in serum using the Siemens Advia 1800 Chemistry Analyzer (Malvern, PA). Whole blood in EDTA was rocked at room temperature to mix thoroughly, then total cell differential was determined using the Siemens Advia 2120 hematology system.
Protein quantification
15).
RNA isolation and quantitative RT-PCR
Spleen and peritoneal cell differential
Histopathology
Samples of esophagus, stomach, duodenum, jejunum, colorectum, pancreas, spleen, and/or bone marrow were fixed in 10% neutral buffered formalin, processed, and embedded in paraffin, and sectioned and stained with H&E for analysis. H&E-stained slides were evaluated by a board-certified veterinary pathologist and endpoints were scored according to the following scale: 0 = background, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked, 5 = severe.
Statistical analysis
Data were analyzed using GraphPad Prism 7 software (San Diego, CA) and are presented as mean + SEM. Box and whisker plots represent median, high, and low values. Group comparisons used the two-tailed Student t test or a one-way ANOVA followed by Tukey multiple comparison test. Statistical significance was set at p < 0.05.
Results
IL-13Rα2 expression in adipose tissue is elevated in mouse models of obesity
IL-13Rα2 is expressed at low or undetectable levels in the absence of tissue inflammation, but is induced in the presence of IL-4 or IL-13, in combination with TNF-α. Levels may be further enhanced by IL-17 or other inflammatory cytokines (14, 35). Obesity is characterized by a state of systemic low-grade inflammation, which is marked by elevated levels of circulating TNF-α (36). In a model of diet-induced obesity, IL-13Rα2 mRNA expression levels in adipose tissue from mice fed an HFD for 13 wk were significantly increased compared with levels in adipose tissue from mice fed a normal chow diet during the same time period (Fig. 1A). Expression of IL-13Rα1 and common γ-chain (IL-2Rγ) was also elevated, whereas expression of IL-4Rα was not increased in adipose tissue of obese mice. As has been reported previously (36), TNF-α mRNA expression was also elevated in adipose tissue from obese mice. Similar increases in adipose tissue IL-13Rα2 expression were seen in leptin-deficient ob/ob mice, a genetic model of obesity (data not shown). Serum concentrations of sIL-13Rα2 were also higher under conditions of obesity (Fig. 1B). In accordance with the elevated IL-13Rα2, which inactivates and depletes IL-13 in tissue (15, 37), adipose tissue levels of IL-13 cytokine were reduced in mice fed an HFD (Fig. 1C).
IL-13Rα2 expression is induced with HFD, associated with depletion of tissue IL-13. C57BL/6 mice were fed normal chow or a 60% HFD for 13 wk. (A) Gene expression analysis in scWAT, normalized to housekeeping genes. (B) sIL-13Rα2 in mouse serum. (C) IL-13 concentration in adipose tissue extract. (D) IL-13Rα2 protein concentration in extracts of eWAT or scWAT tissue of mice fed chow or HFD. scWAT and eWAT were collected from the same animals. The dashed line represents the limit of assay detection. (E) il13rα2 transcripts in human adipose tissue. VAT denotes the omental adipose depot, corresponding to visceral adipose tissue. Data represent three independent experiments for (A–C) (n = 6–8 mice per group), two independent experiments for (D) (n = 11 mice per group), and one experiment for (E) (n = 7 and 12 subjects for the VAT and scWAT groups, respectively). *p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0005.
We compared IL-13Rα2 expression in epididymal WAT (eWAT), a depot of VAT associated with insulin resistance and dyslipidemia, and in scWAT, a storage depot associated with adipokine secretion, enhanced lipid oxidation, and insulin action (38). In mouse adipose tissue, IL-13Rα2 protein concentration was low in both eWAT and scWAT of mice on a chow diet, and was significantly elevated in eWAT but not scWAT of mice fed an HFD (Fig. 1D). Human adipose tissue was also examined for il13ra2 gene expression. Subjects providing omental fat and scWAT for this analysis had average ages of 31.9 ± 3.5 and 48.6 ± 8.4 y, respectively, and all were female. Average BMI of the subjects providing omental fat and those providing scWAT was 41.9 ± 11.0 (n = 6) and 33.3 ± 3.6 (n = 7), respectively. The difference in BMI was not statistically significant, and with an average BMI >30, each group was considered obese. Expression of il13rα2 was found in both omental WAT (VAT) and in scWAT (Fig. 1E). Given that the subjects providing VAT differed in age and trended toward a higher BMI than those providing scWAT, and that gene expression values could not be directly compared between VAT and scWAT, these findings demonstrate expression of il13rα2 in human adipose tissue, but do not allow firm conclusions to be drawn concerning the relative expression of il13rα2 in VAT compared with scWAT.
IL-13 activity gene signature is enhanced in adipose tissue of il13rα2−/− mice
As obesity-induced IL-13Rα2 expression was associated with reduced adipose tissue IL-13 concentrations, we asked whether IL-13Rα2 deficiency would enhance tissue IL-13 levels. Wt and il13rα2−/− mice were subjected to HFD feeding for 13 wk. At the end of this time, concentrations of IL-13 in scWAT did not differ between wt and il13rα2−/− mice (data not shown). Weight gain (Fig. 2A) and total fat mass (Fig. 2B) were similar between il13rα2−/− mice and littermate controls. There were no significant differences in glucose tolerance or insulin sensitivity either on a chow diet (data not shown) or HFD (Fig. 2C, 2D). A small panel of genes was differentially expressed in adipose tissue of il13rα2−/− mice compared with wt controls (Fig. 2E). Genes with elevated expression in il13rα2−/− mice were known to be IL-13 responsive, including the chitinase family members Ym1 and Ym2 (39), the macrophage lectin Mgl2 (40), the eosinophil chemokine eotaxin 1 (Ccl11) (41), and the repair marker, amphiregulin (42).
HFD-induced metabolic changes do not differ between wt and il13rα2−/− mice, despite adipose tissue IL-13 response profile. C57BL/6 mice were fed a 60% HFD for 13 wk. (A) Body weight; (B) % fat mass, determined by nuclear magnetic resonance; (C) glucose tolerance was assessed in fasted mice after injection of 1.5 g/kg glucose i.p.; (D) insulin tolerance was assessed in fasted mice after i.p. injection of 1 U/kg human insulin; (E) relative mRNA expression of representative genes in eWAT of il13rα2−/− mice compared with wt, normalized to housekeeping genes. Graphs represent median, high, and low values from three independent experiments (n = 10–16 mice per group). *p < 0.05, **p < 0.01, ****p < 0.0001 by t test.
IL-33 promotes Th2 inflammation in chow-fed mice
HFD feeding was not sufficient to differentiate systemic metabolic responses in wt and il13rα2−/− mice, as reflected in the body weight, percentage of fat mass, glucose tolerance, and insulin tolerance findings (Fig. 2A–D). With limiting IL-13 concentrations under resting conditions, lack of IL-13Rα2 may have little impact on metabolic endpoints. IL-33 is known to be a potent inducer of IL-13, IL-5, and other cytokines, both in vitro and in vivo (29). Therefore, we administered IL-33 to mice to elevate systemic IL-13 concentrations. Recombinant IL-33 was given by i.p. injection to wt and il13rα2−/− mice, every other day for seven administrations, at doses of 0.0125 or 0.125 mg/kg. We observed dose-dependent modulation of a range of physiological and metabolic endpoints (Supplemental Fig. 1). Based on these observations, a dose of 0.125 mg/kg was chosen to induce responses of sufficient magnitude to allow interrogation of the role of IL-13Rα2.
IL-33 induced marked splenomegaly in both wt and il13rα2−/− mice on chow or HFD (Fig. 3). After IL-33 administration on chow diet, IL-13 concentration was greatly elevated in serum and in tissues, including liver, eWAT, and scWAT (Fig. 3A). Compared to wt, il13rα2−/− mice on chow diet had reduced concentrations of IL-13 in serum, but higher concentrations in liver, eWAT, and scWAT (Fig. 3A). These observations are consistent with findings that sIL-13Rα2 acts as a reservoir for the cytokine, such that mice lacking sIL-13Rα2 have a reduced IL-13 concentration in the blood (15). The cell-surface form of IL-13Rα2 acts to internalize and deplete IL-13, such that mice lacking IL-13Rα2 have elevated IL-13 concentration in tissues (15). IL-13 administration under comparable conditions failed to elevate tissue IL-13 levels (data not shown), suggesting that endogenous IL-13 production was required to sustain the responses.
IL-33 promotes tissue IL-13 elevation, splenomegaly, and adipose tissue inflammation in: (A) chow-fed; or (B) HFD-fed wt and il13rα2−/− mice. Mice were administered PBS (□) or IL-33 (▪) i.p. at 0.125 mg/kg every other day for 2 wk. Serum and tissues were harvested 24 h after the last injection. Panels depict IL-13 concentration in serum or the indicated tissues, spleen weight, % CD45+ cells in the stromal vascular fraction (SVF) of scWAT or eWAT, % eosinophils (CD45+,CD11b+, SiglecF+) in scWAT or eWAT, % ILC2s (CD45+, Lin−, CD25+, ST2+) in the SVF of eWAT. Data represent mean ± SEM of n = 5–6 mice per group from two to three independent experiments. scWAT and eWAT were collected from the same animals. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, by t test, for comparisons between IL-33–treatment groups.
In mice on an HFD, the magnitude of IL-13 induction in response to IL-33 in eWAT and scWAT was reduced compared with that of mice on a chow diet (Fig. 3). In eWAT, the IL-13 concentration was significantly higher in IL-33–treated il13ra2−/− mice compared with wt mice on chow diet, but significantly lower in il13ra2−/− mice compared with wt mice on an HFD. Trends toward similar findings were seen in scWAT (Fig. 3), showing a strong effect of diet on adipose tissue IL-13 regulation in response to IL-33. In mice on a chow diet, IL-33 also increased adipose tissue eosinophils, ILC2s, and total CD45+ leukocytes in both wt and il13rα2−/− mice (Fig. 3A). On an HFD, IL-33 induction of leukocytes and eosinophils was maintained in wt mice, but attenuated in il13rα2−/− mice (Fig. 3B), emphasizing that effects of IL-13Rα2 deficiency became apparent under the stress of an HFD.
IL-33–induced metabolic changes are IL-13 dependent
Administration of IL-33 at 0.125 mg/kg every other day for 2 wk did not produce appreciable weight change in mice on a chow diet, but caused weight loss in obese mice (Fig. 4A), consistent with previous reports (29, 43). Il13rα2−/− mice showed a greater degree of weight loss than wt mice (Fig. 4A). There was no effect of IL-33 on body weight in il13−/− mice on an HFD, confirming IL-13 dependence of the effect (Fig. 4A). Similarly, fasting blood glucose concentrations were reduced in HFD-fed wt mice administered IL-33, and reduced to a greater extent in il13rα2−/− mice, but were not reduced in il13−/− mice (Fig. 4B). Under conditions of HFD, IL-33 also significantly decreased fat mass (Fig. 4C) and adipocyte size (data not shown). In the course of these studies, we noted that the wt and il13rα2−/− mice administered IL-33 on chow diet exhibited mild perianal irritation and diarrhea, which was relatively more pronounced in the il13rα2−/− animals (Fig. 4E). The perianal irritation and diarrhea were both increased in wt mice fed an HFD compared with chow diet, and heightened further in HFD-fed il13rα2−/− animals (Fig. 4E, 4F). This may have contributed to the markedly reduced food consumption observed in the IL-33–treated animals on an HFD (Fig. 4D). Mice that were deficient in IL-13 were protected from the perianal irritation and diarrhea (Fig. 4E), and maintained food consumption (Fig. 4D) and body weight (Fig. 4A) on HFD.
Differential effects of IL-33 administration in chow versus HFD for il13rα2−/− (α2KO), il13−/− (13KO), and wt mice. Mice on chow diet were administered IL-33 i.p. at 0.125 mg/kg every other day for 2 wk. (A) Body weight; (B) fasting blood glucose; (C) % fat mass; (D) food consumption; and (E) perianal irritation/diarrhea score were assessed 24 h after the final injection of IL-33; (F) perianal irritation in IL-33–treated il13rα2−/− mice fed an HFD. Data represent the mean ± SEM of three independent experiments, with 6–12 mice per group for the chow groups, and 10–26 mice per group for the HFD groups. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 for the indicated comparisons, by t test.
Elevations in serum concentrations of liver enzymes alanine aminotransferase and aspartate aminotransferase, but not alkaline phosphatase, were seen in mice fed an HFD as compared with those on a chow diet (Fig. 5A–C). Concentrations of these markers were reduced to variable degrees in mice administered IL-33 (Fig. 5A–C). Levels of alkaline phosphatase and aspartate aminotransferase were lower in the il13−/− mice compared with il13rα2−/− animals, suggesting an influence of IL-13 on liver function. Serum levels of insulin, low-density lipoprotein, HDL, triglycerides, and cholesterol were all significantly reduced in HFD-fed mice administered IL-33 (Fig. 5D–H). There was a trend toward greater reduction in serum lipids in il13rα2−/− animals, and attenuated reduction in il13−/− mice, producing a significant difference between lipid levels in il13rα2−/− and il13−/− animals (Fig. 5E–H). Taken together, these data implicate IL-13 in mediating IL-33-induced protection from HFD-induced changes in liver enzymes, insulin, and serum lipids.
Effects of IL-33 administration on serum metabolic markers in HFD-fed il13rα2−/− (α2KO), il13−/− (13KO), and wt mice after 13 wk of HFD. (A) Alanine aminotransferase (ALT; units per milliliter); (B) alkaline phosphatase (ALP; units per liter); (C) aspartate aminotransferase (AST; units per liter); (D) insulin (picograms per milliliter); (E) low density lipoprotein (LDL; milligrams per deciliter); (F) HDL (milligrams per deciliter); (G) TG (triglyceride; milligrams per deciliter); and (H) cholesterol (milligrams per deciliter). The dashed line indicates concentration in untreated wt mice on chow diet. Data represent the mean ± SEM of three independent experiments, with 5–19 mice per group for the HFD groups. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 for the indicated comparisons, by t test.
IL-33–induced gene expression changes are modulated by IL-13Rα2
Following IL-33 administration, increased expression of numerous genes was noted in adipose tissue, including those encoding cytokines, cytokine receptors, and metabolic markers (Fig. 6, Supplemental Fig. 2). Whether these changes signified altered expression per cell, or reflected altered cellular composition of the adipose tissue, could not be determined. In eWAT, the expression of genes encoding Th2 cytokines IL-13, IL-4, and IL-5 was increased 30–50-fold following administration of IL-33 to wt mice on an HFD, and was reduced in eWAT of il13rα2−/− mice (Fig. 6A), consistent with the relative concentrations of IL-13 cytokine in eWAT (Fig. 3B). In contrast to the effect of IL-13Rα2 deficiency, expression of IL-5 was elevated in eWAT of il13−/− mice, suggesting that IL-13 normally functions to limit IL-5 elevation (Fig. 6A). Among other Th2-associated cytokines, IL-25 expression was not induced in eWAT. IL-33 expression was mildly increased following IL-33 administration, but the response did not differ between wt, il13rα2−/−, and il13−/− mice (Fig. 6A).
Gene expression changes in eWAT of il13rα2−/− or il13−/− mice with matched wt controls, administered 0.125 mg/kg IL-33 every other day for 2 wk. Expression of each gene was determined relative to housekeeping controls. The dashed line represents the average expression level in all PBS-treated groups (il13rα2−/− mice and wt controls, il13−/− mice and wt controls), normalized to a value of one. Genes were grouped into those encoding: (A) cytokines; (B) cytokine receptors; (C) cell differentiation markers; (D) IL-13 response markers. Graphs represent median, high, and low values from two to three independent experiments, with 5–10 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 compared with matched wt, by t test.
Among cytokine receptor genes, expression of il13rα2 was greatly reduced in eWAT of il13−/− mice, below the baseline level seen in the absence of IL-33 (dashed line, Fig. 6B), suggesting that an active inhibitory pathway exists to block IL-13Rα2 expression in mice lacking IL-13. The IL-13Rα1 and IL-4Rα signaling chains were not regulated by IL-33. The IL-33–associated expression of both IL-2Rγ and IL-1RL1 (ST2) was reduced in il13rα2−/− mice, but elevated in il13−/− mice (Fig. 6B). Among markers of cell differentiation, il13rα2−/− mice also displayed reduced expression of genes encoding the Th2 transcription factor, Gata3, the costimulatory molecule, PD1, the ILC2 transcription factor, Rorα, the Treg marker, Foxp3, and the Th17 transcription factor, Rorc. For all but Rorc, the transcripts were elevated in il13−/− mice (Fig. 6C). Collectively, these gene expression changes in mice lacking IL-13Rα2 displayed an inverse profile to those seen in mice lacking IL-13. A similar gene expression profile was seen in scWAT (Supplemental Fig. 2). Overall, the reduced expression of cellular differentiation genes in adipose tissue of il13rα2−/− mice reflected the reduced level of CD45+ leukocyte infiltration in the stromal vascular fraction of IL-33–treated il13rα2−/− mice compared with wt on an HFD (Fig. 3B). In contrast to the elevated expression of lymphocyte differentiation markers in eWAT of il13−/− mice given IL-33, expression of most IL-13 response markers was reduced in il13−/− animals (Fig. 6D, Supplemental Fig. 2), and did not differ between wt and il13rα2−/− mice.
The expression of a panel of metabolism-related genes was examined in the adipose tissue of mice administered IL-33 compared with PBS-treated control animals (Supplemental Fig. 3A). In eWAT, expression of most genes in the panel was reduced following IL-33 treatment, in both wt and il13rα2−/− mice. In all cases, expression was further reduced in mice lacking IL-13, suggesting that IL-33–induced IL-13 normally functions in eWAT to regulate expression of metabolic response genes (Supplemental Fig. 3A). Similarly, in scWAT, expression of most genes in the panel was reduced in response to IL-33 administration, and was not significantly affected by IL-13Rα2 deficiency. Unlike in eWAT, however, IL-13 deficiency did not exacerbate the IL-33–associated gene expression changes in scWAT (Supplemental Fig. 3B). These observations suggest that although IL-13 plays a key regulatory role in IL-33–induced metabolic changes in eWAT, additional regulatory pathways exist in scWAT, abrogating their IL-13 dependence.
Finally, we @examined expression of a panel of genes in the GI tract of wt and il13rα2−/− mice to identify correlates of the metabolic changes seen with HFD feeding and IL-33 administration. The mice expressed IL-13Rα2 throughout the GI tract, progressively increasing from duodenum to jejunum to ileum to colon (Supplemental Fig. 4A). As has been reported previously (44), IL-13Rα2 expression in the gut was increased following IL-33 administration, and this was seen in the duodenum, jejunum, and ileum (Supplemental Fig. 4A). Because the greatest difference between expression of IL-13Rα2 in IL-33–treated mice compared with PBS-treated mice was seen in the jejunum, we focused on this region for analysis of type 2 markers that could differ in wt versus il13rα2−/− mice. Neither IL-13Rα1 nor IL-4Rα expression was induced in the jejunum following IL-33 dosing (Supplemental Fig. 4B). Expression of IL-1RL1 (ST2) was increased with IL-33 administration, but was not modulated in the absence of IL-13Rα2 (Supplemental Fig. 4B). Several genes were induced with IL-33 in wt mice, and displayed significantly lower expression in mice lacking IL-13Rα2, including those encoding the M2/ILC2 marker Arg1, the Th2 transcription factor, Gata3, the costimulatory molecule, PD1, the Treg marker, Foxp3, the eosinophil chemokine, Ccl24, and the inflammatory marker, IL-12Rβ1 (Supplemental Fig. 4C). Additional markers were elevated throughout the GI tract with IL-33, including IL-5, IL-13, and Ym1, but the expression level did not differ between wt versus il13rα2−/−mice (data not shown).
IL-33–induced extramedullary hematopoiesis and inflammation are modulated by IL-13Rα2
As previously reported (30, 45), IL-33 administration induced infiltration of inflammatory cells in multiple tissues, and resulted in splenomegaly (Fig. 7A, 7B), associated with microscopic evidence of disrupted splenic architecture and increased extramedullary hematopoiesis (EMH) (Fig. 7C, data not shown). Increased cellularity was also observed in bone marrow in IL-33–administered mice compared with PBS-treated controls, consistent with the splenic EMH (data not shown). Marginally elevated eosinophil numbers were found in the circulation, spleen, and peritoneal lavage of wt mice in response to IL-33 (Fig. 7D–F). In contrast, the peritoneal lavage of IL-33–treated wt and il13rα2−/− mice was characterized by large numbers of vacuolated macrophages, comprising up to 70% of total leukocytes (Fig. 7G–I). Because they were found in il13rα2−/− mice that displayed reduced body weight on an HFD, the macrophages were examined for lipid in the vacuoles by Oil Red O staining, and found to be negative (data not shown).
Exacerbated splenomegaly, EMH, and eosinophilia in HFD-fed il13−/− mice treated with IL-33. (A) Spleen size; (B) spleen weight; (C) EMH score; (D) % circulating eosinophils; (E) % eosinophils in spleen; (F) % eosinophils in peritoneal lavage; (G) % macrophages in peritoneal lavage; Wright’s Giemsa staining of peritoneal cells from IL-33–treated: (H) wt; (I) il13rα2−/−; (J) il13−/− mice. Data are representative of three independent experiments. (B–G) Data plotted as mean ± SEM of five to six mice per group. ****p < 0.0001 compared with matched wt, by t test.
In IL-13–deficient mice, eosinophilic inflammation was dramatically increased, with elevations of up to 40–50% total leukocytes in the circulation, spleen, peritoneal lavage (Fig. 7D–F, 7J), and bone marrow (data not shown). The increased eosinophils drove marked splenomegaly (Fig. 7A, 7B). Increased numbers of granulocyte precursors were observed in the bone marrow (data not shown), and were accompanied by a reduction in the relative percentage of vacuolated macrophages in the peritoneal lavage (Fig. 7G, 7J). The striking increase in eosinophil number seen in mice lacking IL-13 suggests that IL-13 acts to limit eosinophil development and/or infiltration in response to IL-33 administration.
IL-13–dependent GI toxicity develops in wt mice following IL-33 administration and is exacerbated in il13rα2−/− mice under HFD conditions
As described above, diarrhea and perianal irritation were observed in mice administered IL-33, and were more severe in il13rα2−/− mice compared with wt littermates (Fig. 4E, 4F). IL-13–deficient mice were protected from these GI effects, suggesting IL-13 dependence. Microscopic examination showed that IL-33 administration increased eosinophilic and mononuclear cell infiltrates in the jejunum and colorectum (Fig. 8A–D), with increased goblet cell size and increased size and secretory material within Paneth cells in the jejunum (Fig. 8E, 8K, 8L). In the pancreatic ducts, IL-33 induced epithelial cell hypertrophy with accumulation of eosinophilic material within epithelial cells (hyalinosis), periductal infiltration of inflammatory cells, and decreased zymogen content in the pancreas, which may have been related to the decreased food intake (Fig. 8F–J). Accumulation of similar eosinophilic material was also present within epithelial cells in the esophagus and stomach (data not shown). Although il13rα2−/− mice had more severe diarrhea and perianal irritation (Fig. 4E, 4F), the microscopic changes throughout the GI tract were similar in severity and distribution between wt and il13rα2−/− mice. Taken together, these findings show that IL-33 induced diarrhea and perianal irritation in the context of HFD, which was exacerbated in IL-13Rα2–deficient mice, associated with reduced food intake, but was not associated with an identifiable gene expression pattern or signature in the gut.
IL-33 increases mononuclear and eosinophilic infiltrates in the GI tract and pancreas, with perianal irritation. HFD-fed il13rα2−/− and wt control mice were administered PBS or 0.125 mg/kg IL-33 every other day for 2 wk. (A) Mononuclear infiltrates in jejunum; (B) eosinophilic infiltrates in jejunum; (C) mononuclear infiltrates in colorectum; (D) eosinophilic infiltrates in colorectum; (E) goblet cells in jejunum; (F) epithelial hyalinosis/hypertrophy in pancreatic duct; (G) pancreatic periductal inflammatory infiltrates; (H) reduction in pancreatic zymogen content. For (A)–(H), data plotted are median of two to five mice per group. For (F), only one mouse was scored in the PBS-treated il13rα2−/− group, representative of no detectable pathology in this group. Pancreatic ductal epithelium from: (I) wt mice administered PBS; or (J) il13rα2−/− mice administered IL-33, showing increased eosinophilic material in ductal epithelial cells (*), decreased zymogen granules within acinar cells (+), and periductal inflammatory cell infiltrates (↑); jejunum from: (K) wt mice administered PBS; or (L) il13rα2−/− mice administered IL-33 showing increased goblet cell numbers (*), increased inflammatory cell infiltrates in lamina propria (+), and increased size/secretory material in Paneth cells (↑).
Discussion
The infiltration of metabolic tissues by macrophages, eosinophils, and lymphocytes establishes a cytokine milieu that influences glucose regulation and insulin responsiveness. A type 1 environment of classically activated macrophages, reactive oxygen species, and proinflammatory cytokines contributes to insulin resistance, whereas a type 2 environment of alternatively activated macrophages and eosinophils promotes glucose homeostasis in adipose tissue (46). Shifting this balance by cytokine manipulation could alleviate insulin resistance and re-establish glucose homeostasis under conditions of metabolic stress. IL-13 may play a role, as IL-13–deficient mice have been reported to have hyperglycemia on HFD (22), and IL-13 overexpression protects against HFD-induced obesity (5). IL-13Rα2 modulates endogenous IL-13 concentration and activity by regulating IL-13 neutralization and depletion (10). IL-13Rα2 is inducibly expressed on fibroblasts, smooth muscle cells, and other cell types, but its expression has not been previously described in adipose tissue. The finding that IL-13Rα2 was induced in adipose tissue under conditions of HFD suggests that blockade or deficiency of IL-13Rα2 could elevate local IL-13 concentrations, which in turn could drive anti-inflammatory macrophage polarization, and help to maintain metabolic homeostasis.
We found that IL-13Rα2 deficiency ameliorated the metabolic consequences of HFD, but only when a type 2 environment was established by IL-33 administration. In the absence of IL-33 administration, a small number of IL-13 response genes was elevated, but metabolic parameters were unchanged in il13rα2−/− mice. These observations suggest that any elevations in IL-13 concentration resulting from IL-13Rα2 deficiency alone were insufficient to affect metabolic responses. To further elevate tissue IL-13 concentrations, we administered IL-33 to the mice. The metabolic protective effects of IL-33 have been well described. Administration of exogenous IL-33 protects mice against obesity and type 2 diabetes (29). In obese mice, IL-33 induces Th2 cytokines, polarizes adipose tissue macrophages toward an anti-inflammatory phenotype, and reduces adipose tissue mass and fasting blood glucose levels (29). IL-33–responsive ILC2 have emerged as an important regulatory cell type in adipose tissue (47, 48), influencing obesity, and browning and beiging of fat (34, 49, 50). Through release of IL-13, IL-5, and other mediators, ILC2 orchestrate macrophage phenotype, eosinophilia, and Treg in adipose tissue (51, 52). In accordance with this, we observed increased ILC2 in adipose tissue of both wt and il13rα2−/− mice administered IL-33.
IL-33 administration to mice on HFD elevated adipose tissue expression of genes encoding IL-13–response markers Ym1, Ym2, Mgl2, Ccl11 (eotaxin 1), and Ccl2/Mcp1, to a similar extent in il13rα2−/− and wt mice. In contrast, markers associated with lymphocyte subsets, including the Th2 differentiation factor Gata3, the costimulatory marker, PD1, the ILC2/M2 differentiation factor, Rorα, the Th17 differentiation factor, Rorc, and the Treg marker, Foxp3, were reduced in il13rα2−/− mice compared with wt under these conditions. Overall, the expression of genes associated with leukocyte infiltration was proportional to the percentage of CD45+ cells detected in the adipose tissue. Because the gene expression analysis was carried out with whole tissue extracts, however, the current findings do not allow us to distinguish expression level per cell compared with changes in infiltrating cell types, or whether the genes of interest were expressed in the immune infiltrate or in the adipose tissue itself. For example, Rorα can be expressed in adipocytes under conditions of inflammation and obesity, leading to endoplasmic reticulum stress and supporting macrophage infiltration (53). IL-13 deficiency produced the opposite profile, resulting in increased expression of markers of immune infiltration, along with reduced expression of IL-13 response genes. These findings support a regulatory role for IL-13 in limiting tissue inflammation in response to IL-33.
Along with the gene expression changes, IL-33 administration reduced body weight on HFD, and reduced circulating levels of insulin, cholesterol, and fasting blood glucose. In both eWAT and scWAT, IL-33 administration reduced expression of genes encoding the adipokine adiponectin, the insulin-sensitive glucose transporter, Glut4 (Slc2A4), and the metabolic hormone leptin. It also reduced expression of the beige adipocyte-associated markers UCP1 and Cidea. Beige adipocytes regulate caloric expenditure, and their induction by IL-33–activated ILC2 is thought to promote metabolic homeostasis (34, 47). Although IL-33 administration induces beiging of WAT under conditions of a low-fat or chow diet, this effect is abrogated under conditions of HFD (47). In response to HFD, UCP1 can be transiently induced in brown adipose tissue, but not WAT (47, 54). Thus, we did not observe increased expression of UCP1 in WAT of mice on HFD in response to IL-33. Although the current study was focused on changes induced by obesity, the effects of IL-33 administration on beiging in il13rα2−/− mice under conditions of caloric restriction or temperature modulation will be of importance to examine.
Whereas IL-33 administration reduced expression of these metabolic markers in adipose tissue, IL-13Rα2 deficiency did not have any further effect. In eWAT, IL-13 deficiency produced a strong additional reduction in expression of these markers, whereas in scWAT, neither IL-13Rα2 deficiency nor IL-13 deficiency affected expression. Both adipose tissues responded to IL-33 administration by skewing toward a type 2 profile, with increased expression of tissue IL-5, IL-13, Gata3, and Ym1. The observation that IL-33–induced gene expression changes were altered in eWAT but not in scWAT of il13−/− mice suggests that IL-13 regulation of the IL-33 response is more prominent in eWAT, consistent with the higher expression of IL-13Rα2 found in that tissue. Compared to scWAT, the VAT depot eWAT displays increased inflammation (55), and is associated with insulin resistance and dyslipidemia (38). In contrast, scWAT is associated with reduced metabolic risk, improved glucose tolerance, and insulin sensitivity (38, 56). Lymphocytes homing to eWAT and scWAT may differ in their responsiveness to IL-33. Recent studies have defined the critical role of IL-33 in expansion and maintenance of tissue-resident Foxp3+ Treg in VAT, which prevent obesity-associated inflammation and preserve insulin sensitivity and glucose tolerance (57–59). In VAT of obese mice, Treg numbers are reduced, contributing to inflammation and insulin resistance (60), and this deficit could be corrected by administration of exogenous IL-33 (43). Although the current studies did not enumerate Foxp3+ cells or demonstrate regulatory function, IL-33–induced Foxp3 expression was significantly increased in eWAT (9.6-fold; p < 0.0005) but not in scWAT (2-fold; NS) of il13−/− mice compared with wt, and reduced in eWAT, scWAT, and gut tissue of IL-33–treated il13rα2−/− mice. Further studies are required to uncover the mechanism by which IL-13 and IL-13Rα2 could influence Foxp3 expression in adipose tissue.
As an epithelial alarmin, IL-33 is stored in the nucleus and released upon cell damage, when it acts to initiate repair mechanisms (24). In the intestine, IL-33 is also stored in a population of pericryptal fibroblasts, in which it functions to promote antimicrobial defense (61). Because IL-33 can be strongly proinflammatory, its activity is tightly regulated by soluble ST2, and by rapid oxidation leading to loss of activity (62). Mice expressing a form of IL-33 lacking the nuclear localization domain exhibit high concentrations of circulating IL-33, with widespread eosinophil-dominated inflammation, culminating in lethality (63). In mouse studies, IL-33 administration leads to multiorgan inflammation, including GI irritation, with release of type 2 cytokines, alternative macrophage activation, macrophage infiltration, smooth muscle hypercontractility/hypertrophy, epithelial hyposecretion, and increased mucosal permeability (25, 30, 64, 65). We observed that HFD-fed wt mice developed mild diarrhea and perianal irritation in response to IL-33 administration. This was exacerbated in the absence of IL-13Rα2, and may have driven the reduced food intake and enhanced weight loss observed in il13rα2−/− mice administered IL-33. Epithelial alteration in the proximal GI tract and pancreatic ducts, and increased eosinophilic and mononuclear infiltrates, were observed in the intestine and pancreas of IL-33–treated mice. However, these findings were comparable in wt and il13rα2−/− mice, and did not appear to account for the increased diarrhea and perianal irritation observed in the il13rα2−/− animals. It is possible that altered GI function (such as abnormal contractility, maldigestion, or malabsorption) drove worsened diarrhea and perianal irritation in il13rα2−/− mice, but further studies are required to understand the basis for the exacerbated irritation in the absence of IL-13Rα2.
In summary, these studies confirm a key role for IL-13 in mediating IL-33–driven metabolic homeostasis on HFD, and implicate involvement of IL-13Rα2 in regulation of this response. Along with protective effects, however, IL-33 administration induced diarrhea and perianal irritation, particularly with HFD. IL-13 appears to be a key driver of these GI-related clinical signs, given that il13−/− mice were protected, whereas irritation was exacerbated in il13rα2−/− mice. These findings suggest that in the presence of IL-33, IL-13Rα2 modulation can elevate adipose tissue IL-13 concentrations to levels sufficient to antagonize metabolic effects of HFD, but at the potential cost of GI toxicity. In the absence of IL-33 administration, however, the lack of IL-13Rα2 was not a sufficient driver to alter the metabolic effects of HFD. Although the strong type 2 skewing modeled in this study would not normally be seen under physiological conditions, these studies help to assess the potential impact of IL-13 modulation by IL-13Rα2.
Disclosures
All authors are employees of Pfizer.
Acknowledgments
We thank Diane Mathis and Thomas A. Wynn for critical reading of the manuscript, and Javier Cote-Sierra, Ann-Marie Richard, and Mylene Perreault for helpful discussion. We thank Laura Danner and Lisa Brideau for animal handling and technical expertise, Mary Bauchmann for management of genetically modified animals, and Kim Lewis and Deborah Carraher for clinical chemistry.
Footnotes
The online version of this article contains supplemental material.
Abbreviations used in this article:
- BMI
- body mass index
- EMH
- extramedullary hematopoiesis
- eWAT
- epididymal white adipose tissue
- GI
- gastrointestinal
- HDL
- high-density lipoprotein
- HFD
- high fat diet
- ILC2
- type 2 innate lymphoid cell
- scWAT
- s.c. WAT
- sIL-13Rα2
- soluble form of IL-13Rα2
- Treg
- T regulatory cell
- VAT
- visceral adipose tissue
- wt
- wild type.
- Received August 31, 2017.
- Accepted December 11, 2017.
- Copyright © 2018 by The American Association of Immunologists, Inc.