Resistin-like molecule α decreases glucose tolerance during intestinal inflammation

Abstract

Resistin-like molecule α (Relm-α), is a secreted cysteine-rich protein belonging to a newly defined family of proteins including resistin, Relm-β and Relm-γ. Resistin was initially defined based on its insulin resistance activity, but the family members are highly upregulated in various inflammatory states, especially those involving intestinal inflammation. Herein, we report the role of Relm-α at baseline and following an experimental model of colitis. Relm-α was readily detected in the serum at baseline (4−5 ng/ml) and its level was regulated by energy uptake. Retnla^−/− mice had decreased baseline circulating leptin levels but displayed normal glucose, glucose clearance and insulin levels. Following exposure to the oral innate trigger dextran sodium sulfate (DSS), a non-redundant pro-inflammatory role for Relm-α was uncovered as Retnla^−/− mice were markedly protected from DSS-induced disease activity and histopathological features. Relm-α regulated eosinophil-directed cytokines (e.g. IL-5, CCL11/eotaxin-1 and CCL5/RANTES) ex vivo. Consistently, DSS-treated Retnla^−/− mice displayed substantially decreased eosinophil accumulation and decreased phosphorylation of NFκB, ERK1/2 and p38 in macrophages and eosinophils. Following DSS exposure, serum level of Relm-α was upregulated and DSS-treated Retnla^−/− mice were markedly protected from hyperglycemia induced by glucose injection independent of changes in insulin levels. Retnla^−/− mice were protected from increases in gut hormone serum levels of gastric inhibitory polypeptide and peptide YY that were induced following DSS-treatment. These findings demonstrate a central pro-inflammatory role for Relm-α in the regulation of colonic inflammation and a novel link between colonic injury, glucose tolerance and energy intake.

Introduction

Resistin is a hormone that was originally identified in the adipose tissue with physiological roles in promoting insulin resistance and linked to obesity with insulin resistance (1, 2). Recently, a new family of Resistin-like molecules including Relm-α, Relm-β and Relm-γ have been identified (3). This group of Relm proteins share sequence homology with resistin and contain highly conserved C-terminal cysteine residues, which support the assembly of disulfide-dependent multimeric units (4). Interestingly, and despite opposing physiological effects on insulin resistance, the multimeric assembly of the resistin family is similar to that of adiponectin and suggests a functional role for the Relm proteins in the regulation of glucose (4). Indeed, it is assumed that resistin and the Relm protein family may have a role in the metabolism and energy balance (5).

Relm-α, formerly known as found in inflammatory zone 1 (FIZZ1), has been implicated in various inflammatory conditions including asthma and helminth infections (6, 7). Following Th2 stimuli Relm-α is highly upregulated in the lung and gastrointestinal tract and is thought to promote fibrosis via direct stimulation of fibroblasts (7, 8). Although Relm-α has not been identified in the human genome, the expression pattern of human resistin is more similar to that of Relm-α than to murine resistin (6). Thus, Relm-α may share functional roles similar to those of resistin.

Notably, various metabolic hormones including resistin and Relm-β have been attributed pro-inflammatory roles in inflammatory bowel disease (IBD). Patients with IBD, display increased serum levels of resistin, leptin, and adiponectin (9, 10). In addition, gut hormones such as gastric inhibitory polypeptide (GIP) and peptide YY (PYY) are elevated in patients with Crohn's disease (11, 12). Importantly, studies with Retnlb^−/− mice revealed a significant role for Relm-β in experimental models of intestinal inflammation (13, 14). Taken together, these data implicate that Relm family members may have a role in intestinal inflammation and metabolism. Nevertheless, the role of Relm-α in intestinal inflammation is not determined and its contribution to glucose metabolism or energy uptake is currently unknown. Herein, we further characterize recently generated Retnla-deficient mice (15) and elucidate a non-redundant role for Relm-α in regulating innate colitis and inflammation-associated glucose tolerance.

Materials and Methods

Mice

Male and female, 8- to 12-week-old Retnla^−/− mice (backcrossed to c57BL/6 or BALB/c background at least 7 and 10 generations, respectively) were generated using the Velocigene™ technology as described (15). Il6^−/− (c57BL/6 background) mice were obtained from the Jackson Laboratory (Bar Harbor, ME). For all experiments, 4−5 week old wild type mice were obtained from Taconic Laboratories (Hudson, NY) and environmentally matched with the Retnla^−/− mice for 2−3 weeks. All mice were housed under specific pathogen-free conditions and treated according to institutional guidelines.

High fat diet

In some experiments, mice were fed a high-fat diet consisted of 58% fat, 25.6% carbohydrate, and 16.4% protein (total 23.4 kJ/g), whereas the normal diet contained 11.4% fat, 62.8% carbohydrate, and 25.8% protein (total 12.6 kJ/g).

DSS-induced of colonic injury

DSS (ICN Biomedical Inc., USA, average molecular weight of 41kDA) was supplied in the drinking water as a 2.5% (w/v, for c57BL/6 mice) and 5% (w/v, for BALB/c mice) solution for up to 8 days. The appearance of diarrhea was defined as mucus-fecal material adherent to anal fur. The presence or absence of diarrhea was scored as either 1 or 0, respectively. The presence or absence of diarrhea was confirmed by means of examination of the colon after completion of the experiment. Mice were killed, and the colon was excised from the animal. Diarrhea was defined by the absence of fecal pellet formation in the colon and the presence of continuous fluid fecal material in the colon. The appearance of rectal bleeding was defined as diarrhea containing visible blood, mucus, or both or gross rectal bleeding and scored as described for diarrhea. A change in body weight was calculated by the percent change (gain/loss) from the initial weight. The disease activity index (DAI) was derived by scoring 3 major clinical signs (weight loss, diarrhea, and rectal bleeding) (13).

Intestinal histopathologic examination

Animals were killed on day 7, and the colon was excised. Tissue specimens were then fixed in 4% paraformaldehyde and stained with hematoxylin and eosin using standard histologic techniques. The histological score was determined by calculating the percentage of colon length with mucosal ulceration, edema, lymphoid aggregates and epithelial cell loss, by performing a blinded morphometric analysis of the colon with the ImageProPlus 4.5 software package (Media Cybernetics, Inc, Silver Spring, Md) (13, 16).

Immunofluorescence

Fixed frozen sections were treated with 100% acetone and blocked with 3% goat serum in PBS. Slides were incubated with isotype controls (Rat IgG1 and Rabbit IgG; Vector, Burlingame, CA) anti-F4/80 (BD Pharmingen, San Jose, CA), anti-MBP (Kindly provided by James Lee, Mayo Clinic), anti-phospho NFκB, anti-phospho-ERK1/2, anti-phospho p38 (Cell Signaling, Danvers, MA), anti-Relm-α (kind gift of Peprotech, Rocky Hill, NJ) (overnight, 4°C) followed by goat anti-rabbit alexa 488; donkey anti-rat Alexa 594 (Invitrogen) and counterstained with 4′,6-diamidino-2-phenylindole dihydrochloride DAPI/Supermount G solution (Fluoromount-G) Images were captured using a Zeiss microscope and Axioviewer image analysis software (Deutsland; Carl Zeiss Corp., Jena, Germany). Quantification of eosinophil numbers in the tissue was performed by counting the number of immunoreactive cells (magnification ×40) from at least three random sections/mouse. Values were expressed as eosinophils per high power field.

Macrophage activation

Bone marrow-derived macrophages (2×10⁵ cells) were incubated for 24 hrs with recombinant Relm-α (kind gift of Peprotech, Rocky Hill, NJ). Thereafter, the supernatant was collected and stored at −20°C until assessed for chemokines.

Punch Biopsies

The colons of control or DSS-treated mice was flushed with PBS and open along a longitudinal axis. Thereafter, 3 mm² punch biopsies were obtained and incubated for 24hrs in RPMI supplemented with 10% FCS and antibiotics. Supernatants were collected and kept in −20°C until assessed for chemokines.

ELISA and multiplex chemokine assays

Insulin and leptin levels were examined in the serum using a commercial EIA kit (Crystal Chem, Downers Grove, IL) according to the manufacturer's instructions. Lower detection limits for insulin and leptin were 0.2 and 0.1 ng/ml respectively. For detection of serum Relm-α, purified biotinylated anti-Relm-α and anti-Relm-α (Peprotech, Rocky Hill, NJ) were used according to a protocol provided by the manufacturer. Lower detection limit for Relm-α was 125.62 pg/ml.

Chemokines and gut hormone levels were determined by a mouse multiplex kit (Millipore, Billerica, MA) according to the manufacturer's instructions. Lower detection limits for chemokines were 6.4 pg/ml. Lower detection limits for gut hormones were 41.5 pg/m (Amylin and gherlin), 8.7 pg/ml (GLP-1 and PYY) and 2.7 pg/ml (GIP), respectively.

Glucose tolerance and measurements

For glucose tolerance tests, D-glucose (2 mg/g of body weight) was intraperitoneally injected into overnight fasted mice, and glucose levels were monitored at 0, 15, 30, and 60 min after injection by retro orbital bleeding using an Accu-Chek glucometer (Roche Diagnostics Corp., Indianapolis, Indiana, USA).

Statistical analysis

Data were analyzed by either by ANOVA followed by Tukey post hoc test or by unpaired, two-tailed t-test using GraphPad Prism 4 (San Diego, CA). Data are presented as mean ± SD; values of p < 0.05 were considered statistically significant.

Results

Baseline serum Relm-α expression

Since Relm-α is a secreted protein, we examined the baseline circulating levels of Relm-α in wild type BALB/c and c57BL/6 mice. Relm-α was detected at high levels at baseline in the serum and no significant differences were observed between both mice strains (Figure 1) and between male and female mice (data not shown). Interestingly, following overnight fasting both BALB/c and c57BL/6 mice displayed a significant reduction in Relm-α expression (Figure 1). To control for nonspecific binding of the anti-Relm-α antibody, serum from Retnla^−/− was subjected to the ELISA and displayed no immunoreactivity (data not shown).

Figure 1. Serum expression of Relm-α.

The expression of Relm-α in the serum of BALB/c or c57BL/6 mice was assessed at baseline or following an overnight fasting period , n=22−27 mice, *-p<0.05, **-p<0.01.

Regulation of leptin and weight gain by Relm-α

Next, we were interested to examine whether Relm-α may regulate metabolic features and/or affect the expression of other adipokine expression (17, 18). Interestingly, Retnla^−/− mice displayed significantly lower levels of leptin at baseline whereas no alterations in insulin levels were detected (Figure 2A-B); No baseline difference was observed in serum levels of TNF-α and IL-6. Furthermore, Retnla^−/− mice exhibited similar weight to wild type mice following normal food (data not shown) and gained weight similarly under high-fat diet conditions (data not shown).

Figure 2. The effects of Relm-α on leptin and baseline glucose metabolism.

Serum of 24 hours starved-wild type and Retnla^−/− mice, kept on a normal diet, was analyzed for leptin and insulin levels (A-B respectively) n=14; **-p<0.01, ns-non significant. Following normal (C, E) or high fat diet (D, F) wild type and Retnla^−/− mice were starved for 24 hrs and blood glucose levels assessed (C-D). In addition, the mice were subjected to a glucose tolerance test (E-F). Glucose clearance was measured at the indicated time points; n=3 (4−8 mice per experimental group), ns-non significant.

Baseline glucose metabolism in Retnla^−/− mice

Given the association between insulin resistin and glucose metabolism (2), we aimed to examine the role of Relm-α in glucose metabolism and tolerance. Thus, we examined glucose levels in Retnla^−/− mice at baseline and following normal or high fat diet. Retnla^−/− mice had comparable glucose levels to wild type mice at baseline (114.3 ± 4.5 and 102.5 ± 13.3 mg/dL in wild type and Retnla^−/− mice, respectively) (Figure 2C). In addition, following a high fat diet, serum glucose levels were comparable between Retnla^−/− and wild type mice (147.3 ± 1.8 and 183.4 ± 28.57 mg/dL in wild type and Retnla^−/− mice, respectively) (Figure 2D).

Resistin has been shown to regulate blood glucose levels in association with increased weight gain (2). Therefore, we examined whether Relm-α regulates glucose clearance when fed with normal or high fat diet. These sets of experiments revealed that Retnla^−/− mice cleared glucose normally under regular diet, and displayed similar kinetics to wild type mice (Figure 2E). Furthermore, intraperitoneal glucose challenge following a high fat diet, revealed no significant difference in glucose clearance between wild type and Retnla^−/− mice (Figure 2F).

Retnla^−/− mice are protected from DSS-induced colitis

Following DSS-treatment wild type BALB/c and c57BL/6 mice display increased levels of circulating Relm-α (Figure 3A). For example, in BALB/c mice Relm-α was elevated in the serum after DSS-treatment from 5.4 ± 3.2 (baseline) to 13.8 ± 1.7 ng/ml (DSS-treated, p<0.05) (Figure 3A); the ng/ml level of Relm-α in the serum is notably high. The increase in Relm-α levels was independent of IL-6, as Il6^−/− mice, which have been previously shown to be protected from DSS-induced colitis (19), increased Relm-α similar to control (c57BL/6) mice (from 4.1 ± 4.3 at baseline to 14.1 ± 3.9 ng/ml following DSS-treatment). To examine the role of Relm-α in experimental colitis Retnla^−/− mice were subjected to DSS in their drinking water and assessed for disease progression. Retnla^−/− mice were protected from the major clinical features of DSS-induced colitis and displayed reduced rectal bleeding, diarrhea and weight loss that was reflected by reduced disease activity index (Figure 3B-C). Importantly, the protection from DSS-induced damage was observed in both c57BL/6 and BALB/c mouse strains (Figure 3B-C). Furthermore, histological examination of colons obtained from Retnla^−/− mice showed significant reduction in histopathological findings (both in c57BL/6 and BALB/c mice). Upon DSS-treatment wild type mice displayed increased inflammation whereas Retnla^−/− displayed decreased edema formation, epithelial cell damage and leukocyte infiltration (Figure 3D-E). Quantitation of the histological findings showed marked protection in Retnla^−/− mice that was observed both in c57BL/6 and BALB/c mice (Figure 3F-G).

Figure 3. Relm-α expression and function in DSS-induced colonic injury.

Wild type BALB/c and c57BL/6 mice as well as Il6^−/− mice were subjected to DSS (5% for BALB/c and 2.5% for c57BL/6 and Il-6^−/− mice) in their drinking water. Five days after disease induction, serum samples of untreated or DSS-treated mice were collected and Relm-α expression was examined using an ELISA (A). *-p<0.05, **-p<0.01, n=8−12 mice. Wild type and Retnla^−/− mice were subjected to the aforementioned experimental regime and analyzed for the clinical disease features including diarrhea, rectal bleeding, weight loss and scored for the overall disease activity index (B-C) and hostological score (F-G), n=4 (6−8 mice per experimental group); ***-p<0.001. Seven days after DSS treatment the colons were excised, fixed and stained with H&E. Representative photomicrographs of low power magnification (D and E left panels [×10] respectively) and high power (D and E right panels [×40] respectively) magnification of control treated mice (Ctrl, D) and DSS-treated mice (E) is shown. Digital morphometric analysis of the histological score of c57BL/6 and BALB/c mice is shown (F-G). Data represent n=4 (6−8 mice per experimental group); **-p<0.01, ***-p<0.001.

Regulation of eosinophil-active cytokines by Relm-α

We have previously shown that Relm-α is capable of inducing cellular infiltration including eosinophil accumulation into the peritoneal cavity (15). This suggested that the proinflammatory effect of Relm-α may involve regulation of chemokine expression. To determine whether Retnla^−/− mice display an altered chemokine expression pattern following DSS-treatment, punch biopsies from DSS-treated Retnla^−/− were analyzed for chemokine production ex-vivo. Consistent with our previous findings, DSS-treated Retnla^−/− mice displayed an altered eosinophil-related chemokine response; CCL11/eotaxin-1 and CCL5/RANTES levels were substantially reduced ex-vivo (Figure 4A-B). In addition, IL-5 levels were significantly reduced as well (Figure 4C). These results were not attributed to a general inhibition since the levels of G-CSF, JE/CCL2 and IP-10/CXCL10 were not significantly altered (Figure 4D and data not shown). Notably, the DSS-induced CXCL1/KC was also Relm-α dependent (Figure 4E). Consequently, we hypothesized that eosinophil recruitment into the colon, previously shown to be mediated by CCL11/eotaxin-1 ((20), would be altered in DSS-treated Retnla^−/− mice. Indeed, eosinophil accumulation in the colon of DSS-treated Retnla^−/− mice was significantly attenuated (Figure 4G-H). Of note, Retnla^−/− mice also displayed a significant reduction in IL-17 production (Figure 4F). Importantly, the aforementioned altered chemokine response is likely not due to the effects of Relm-α on macrophages as Relm-α was not capable to induce or potentiate chemokine release from bone marrow macrophages directly or in combination with LPS, respectively (data not shown).

Figure 4. The effects of Relm-α on colon cytokine production and eosinophil levels.

Punch biopsies were obtained from control and DSS-treated wild type (WT) or Retnla^−/− mice six days after the beginning of DSS-treatment. Following a 24-hour incubation period, the supernatants were obtained and assessed for chemokine and cytokine expression (A-F), n=4, *-p<0.05, **-p<0.01. Six days following DSS-treatment, the colons were excised, frozen sections prepared and slides were stained with anti-MBP (Green) and DAPI (Blue). A representative photomicrograph (magnification ×10) (G) and quantitative analysis (H) is shown, HPF-high power field, **-p<0.01.

Activation of pro-inflammatory signaling cascades in vivo

Given our previous findings on the proinflammatrory effects of Relm-α on macrophage activation (15) and the aforementioned data on eosinophil accumulation into the colon, we hypothesized that both cell types will display decreased activation in vivo in DSS-treated Retnla^−/− mice. To explore this hypothesis, frozen sections of control and DSS-treated wild type and Retnla^−/− mice were stained with for the F4/80 antigen and analyzed for activation of NFκB, ERK1/2 and p38. In agreement with the pro-inflammatory role for Relm-α, DSS-treated Retnla^−/− mice displayed substantially decreased NFκB-, ERK1/2-and p38-phosphorylation. These alterations were observed both in F4/80⁺ cells corresponding with macrophage morphology (Figure 5A-C white arrows and 5D) and in F4/80^low/negative cells that corresponded with eosinophil morphology (Figure 5A-C yellow arrows and Figure 5E). Importantly, we specifically analyzed micro-inflammatory foci where the distribution of F4/80⁺ cells was homogeneous. This analysis revealed that the altered phosphorylation in Retnla^−/− mice was likely due to decreased phosphorylation and not due to overall less cellular recruitment (Figure 5A-C).

Figure 5. The effects of Relm-α on pro-inflammatory signaling cascades in vivo.

Wild type (WT) and Retnla^−/− BALB/c mice were exposed to 5% DSS for seven days. Thereafter, colon was excised, frozen sections prepared and slides were stained with F4/80 (Green), anti-phospho ERK1/2 (pERK1/2, A), anti-phosphop38 (pp38, B), anti-phospho-NFκB (pNFκB, C) and DAPI (Blue). Representative photomicrographs are shown. White and yellow arrows indicate F4/80⁺ and F4/80⁻ phosho-antibody⁺ (pAb) immunoreactive cells respectively. A high power magnification of F4/80⁺/pAb⁺ and F4/80⁻/pAb⁺ cells is shown (D-E).

Glucose tolerance and insulin assessment following colonic inflammation

Given the substantial role for Relm-α during colonic inflammation (Figures 3-5) and the role of resistin in glucose metabolism, we hypothesized that Relm-α may have a role in glucose metabolism specifically under inflammatory conditions. Assessment of glucose levels six days following DSS-treatment demonstrated normal serum glucose levels in Retnla^−/− mice (Figure 6A). Since the levels of circulating Relm-α were highly induced following the DSS-experimental regime, we hypothesized that under colonic inflammatory conditions, Relm-α may regulate glucose clearance. Although baseline glucose levels were unaltered in Retnla^−/− mice following DSS-treatment (Figure 6A), DSS-treated Retnla^−/− mice were significantly protected from hyperglycemia induced by glucose challenge while wild type mice displayed markedly elevated levels of serum glucose (Figure 6B). For example, while the levels of glucose in wild type mice increased after 15 minutes to 278 ± 84 mg/dL, glucose levels in Retnla^−/− mice hardly increased (p<0.001). Even more striking was the difference observed at 30 minutes where glucose levels increased up to 362 ± 48 mg/dL in wild type mice, whereas in Retnla^−/− mice it was increased only up to 223 ± 74 mg/dL (p<0.001).

Figure 6. The effects of Relm-α on glucose metabolism and serum levels of gut-derived hormones during DSS-indcued colon injury.

Wild type (WT) and Retnla^−/− BALB/c mice were exposed to 5% DSS in their drinking water. Mice were starved overnight seven days after disease induction and serum glucose was assessed (A). Glucose clearance following intraperitoneal glucose injection was measured at the indicated time points (B). ***-p<0.05, ns-non significant, n=15 mice. Serum samples of untreated or DSS-treated mice were collected and insulin (C), GIP (D) and PYY (E) levels were examined using ELISA or a multiplex array kit, ns-non significant, *-p<0.05, n=12. Punch biopsy samples were analyzed for PYY expression (F), ns-non significant, *-p<0.05, n=6.

In order to determine whether the changes in glucose clearance may be due to a DSS-induced change in insulin levels in the Retnla^−/− mice, serum insulin levels were assessed. Importantly, the metabolic effects of Relm-α were independent of changes in insulin, as insulin levels were similar at baseline and following DSS-administration between wild type and Retnla^−/− mice (Figure 6C).

Next, we hypothesized that gut hormone levels that have been linked to glucose metabolism and energy uptake (such as gherlin, amylin, GIP, glucagon-like peptide-1/GLP-1 and PYY) might be altered in response to DSS and modulated by Relm-α. Thus, we assessed the levels of active gherlin, active amylin, total GLP-1, GIP and PYY. Following DSS-treatment, GIP and PYY levels were significantly increased in the serum of wild type BALB/c mice; whereas, Retnla^−/− mice did not display enhanced gut hormone levels (Figure 6D-E). Gherlin and amylin were not detected (data not shown). Although readily detected, no changes were observed in GLP-1 following DSS-treatment (data not shown). To further elucidate whether the changes in PYY directly correlated with the decreased disease phenotype that was observed in Retnla^−/− mice, we examined PYY levels in colon punch biopsies obtained from DSS-treated wild type and Retnla^−/− mice. Although PYY levels were significantly upregulated in DSS-treated punch biopsies, no difference was observed in PYY levels between wild type and Retnla^−/− mice (Figure 6F).

Discussion

Immune-related diseases such as IBD, diabetes, obesity and asthma have become some of the fastest growing and persistent public health problems in the western world, and are currently on the rise (21-23). These diseases share a component of inflammation that is involved in disease pathogenesis and complications (21-25). Thus, defining molecular pathways that may be shared between several immune-related diseases such as asthma, obesity and IBD that often co-occur is of great interest and importance (21-24).

The Relm family of proteins draws much attention, as these proteins share sequence and structural homology to resistin and are highly upregulated in various inflammatory states including asthma and IBD. Nevertheless, the role of Relm-α is still unclear (6, 26). In this study we demonstrate several key findings. First, we demonstrate that Relm-α is consistently detectable in the serum and its expression levels are regulated by food intake and colonic inflammation. Second, Retnla^−/− mice are protected from DSS-induced colitis and under these conditions, Relm-α has a role in hyperglycemia induced by glucose injection and in regulating gut-derived hormones such as GIP and PYY. Third, we provide substantial evidence that Relm-α regulates pro-inflammatory eosinophil directed cytokines in vivo (e.g. CCL11/eotaxin-1 and IL-5) and activates intracellular pro-inflammatory signaling cascades. Our data support a model wherein Relm-α contributes to glucose metabolism when it is induced during the setting of specific intestinal inflammatory conditions and the host is exposed to increased pro-inflammatory cytokines (e.g. IL-6 and TNF-α) and high glucose intake.

Several molecules that are involved in energy intake were found to be dysregulated in Retnla^−/− mice either at baseline (e.g leptin) or following DSS-treatment (e.g GIP and PYY). In fact, GIP stimulates glucose dependent insulin secretion and PYY regulates satiety via the hypothalamus (12, 27). Moreover, the findings that the levels of Relm-α are regulated by food intake strongly suggests that Relm-α has a metabolic role. Notably, leptin, an important protein regulating energy intake and expenditure including appetite metabolism (28, 29), is up-regulated in IBD patients and has a pro-inflammatory role in experimental colitis (17, 18, 30). Furthermore, PYY and GIP are upregulated in the serum of patients with Crohn's disease (11, 12). This suggests that under colonic-inflammatory conditions where the body is in energy deficit, multiple pathways (either gut-derived or from other endocrine sources) act in concert to increase food ingestion. These latter findings are of specific interest since resistin was initially described as a factor linking obesity and insulin resistance (2). In particular, PYY has been found to be upregulated in the serum of human and mice with diet-induced obesity (31, 32). Hence, the ability of Relm-α to regulate glucose metabolism in DSS-induced colitis may be due to the overall energy deficit state and the alteration in energy-related hormonal status; these findings argue against the protection from hyperglycemia simply due to the protection from inflammation. Consistent with this hypothesis, we show that GIP and PYY are upregulated in the serum (and PYY in the colon) following DSS treatment in wild type but not Retnla^−/− mice. Interestingly, while circulating PYY levels were altered in the Retnla^−/− mice, colonic generation of PYY was not dysregulated, indicating that a central pathway may be regulated by Relm-α during colonic inflammatory conditions. Furthermore, no difference is observed in glucose tolerance between control- and DSS-treated wild type mice (data not shown) indicating that hyperglycemia involves cooperativity between DSS-treatment and Relm-α.

While the inflammatory state alone is not likely to be the main factor leading to glucose tolerance in Retnla^−/− mice (given all the metabolic alterations), our findings cannot exclude this possibility. In fact, several observations link increased inflammation and glucose metabolism. For example, adipokines (leptin, resistin and adiponectin) have been all shown to have important roles in inflammation and are elevated in the serum of IBD patients (9, 10, 30). Of note, and similar to Relm-α, the serum levels of leptin and resistin are also detected in the ng/ml range (9). Furthermore, high fat diet induces increased serum endotoxin levels and mice that are chronically perfused with low dose LPS develop hepatic insulin resistance and increased IL-6 and TNF-α (33). In these settings, toll-like receptor (TLR) 4-mediated MyD88 activation has a key role in promoting insulin resistance by diet-induced obesity (34). In addition, recent reports that overweight Crohn's disease patients (body mass index >24) develop more severe disease (as indicated by more frequent anoperineal complications, a marked year-by-year disease activity and require earlier surgical intervention), compared with lean patients (35, 36). In agreement with our data, a recent study by Al-Azzawi et al. demonstrated that prolonged administration of intraperitoneal Relm-α (but not resistin), significantly increased insulin resistance that is associated with decreased gallbladder tension (37). Thus, while Relm-α and resistin share similar structure and expression pattern, they may have distinct roles under different settings.

The ability of Relm-α to regulate leptin levels may also contribute to its overall pro-inflammatory role in vivo. Nevertheless, we have recently shown that Relm-α acts as a co-factor with LPS to induce IL-6 and TNF-α production (15) and we now demonstrate that Relm-α can regulate eosinophil-directed chemokines (e.g CCL11/eotaxin-1) and cytokines (e.g. IL-5). This latter effect is relatively specific since G-CSF and other chemokines, which are significantly induced by DSS-treatment, were not attenuated. These data argue for a specific effect and not a general inhibition of chemokine production due to decreased disease state and further distinguishes the role of Relm-α and leptin. Our findings regarding the pro-inflammatory role of Relm-α suggest that Relm-α is a novel link between the innate and adaptive immune response. It is likely that Relm-α induces its responses via regulating various cell types. Supporting this hypothesis are our findings that Relm-α did not induce or potentiate chemokine release from macrophages. Thus, the effects of Relm-α on chemokine expression is possibly by other cells including epithelial cells and T cells. Of note, Relm-α was found to significantly regulate colonic expression of IL-17, a cytokine that has been shown to be critical in colitis (38). These findings suggest that Relm-α can either directly (via acting on T cells) or indirectly (via regulating macrophage IL-6 production (15)) regulate Th17 cell function. Although the receptor for Relm-α has yet to be identified, our data suggest that Relm-α is capable to induce intracellular MAPK and NFκB activation.

In summary, we demonstrate a novel role for Relm-α in the orchestration of the colonic immune reaction in response to DSS by regulating colon-derived eosinophil directed cytokines. Furthermore, our data establishes a novel link between colonic inflammation, energy uptake and glucose metabolism and provide an important insight into the role of Relm-α in this process. As the health of the modern world is under increasing threat of chronic co-occurring inflammatory diseases, defining the roles of shared components such as Relm-α in the pathophysiology of multiple diseases may provide new targets for future therapeutics.