https://orcid.org/0000-0001-9490-3726
Abstract
Liver X receptor (LXR) exerts anti-inflammatory effects in macrophages. The aim of this study was to explore the expression and function of LXR in the colonic epithelium under inflammatory conditions.
The expression of LXR was explored by Western blot and immunohistochemistry in colonic biopsies from patients diagnosed with inflammatory bowel disease (IBD) and control patients. In addition, LXR and its target gene expression were analyzed in the colon from interleukin (IL)-10-deficient (IL-10-/-) and wild-type mice. Caco-2 cells were pretreated with the synthetic LXR agonist GW3965 and further challenged with IL-1β, the expression of IL-8 and chemokine (C-C motif) ligand (CCL)-28 chemokines, the activation of mitogen-activated protein (MAP) kinases, and the nuclear translocation of the p65 subunit of nuclear factor kappa B was evaluated. Glibenclamide was used as an ABCA1 antagonist.
We found that LXR expression was downregulated in colonic samples from patients with IBD and IL-10-/- mice. The nuclear positivity of LXR inversely correlated with ulcerative colitis histologic activity. Colonic IL-1β mRNA levels negatively correlated with both LXRα and LXRβ in the colon of IL-10-/- mice, where a decreased mRNA expression of the LXR target genes ABCA1 and FAS was shown. In addition, IL-1β decreased the expression of the LXR target gene ABCA1 in cultured intestinal epithelial cells. The synthetic LXR agonist GW3965 led to a decreased nuclear positivity of the p65 subunit of nuclear factor kappa B, a phosphorylation ratio of the p44-42 MAP kinase, and the expression of CCL-28 and IL-8 in IL-1β-stimulated Caco-2 cells. The pharmacological inhibition of ABCA1 increased the phosphorylation of p44-42 after GW3965 treatment and IL-1β stimulation.
The LXR-ABCA1 pathway exerts anti-inflammatory effects in intestinal epithelial cells and is impaired in the colonic mucosa of patients with IBD and IL-10-/- mice.
INTRODUCTION
A dysregulated immune response is a hallmark of inflammatory bowel disease (IBD) pathogenesis. In fact, Crohn disease (CD) and ulcerative colitis (UC) are characterized by the development of a chronic T-cell-mediated intestinal inflammatory process triggered by microbial and environmental factors.1 Intestinal epithelial cells (IECs), as the main sensors of luminal microbial and alimentary antigens and the orchestrators of an effective mucosal immune response, may play a key role in this phenomenon; in this sense, it has been shown that IECs from patients with IBD produce and secrete high amounts of proinflammatory cytokines and chemokines such as interleukin (IL)-8,2 IL-17,3 and IL-33.4 This phenomenon may be related to decreased anti-inflammatory peroxisome proliferator-activated receptor gamma5 and an increased toll-like receptor (TLR)-4 expression in the apical domain of the IECs from patients with IBD.6 In addition, IECs constitute the main cellular element of an effective and selective barrier between the gastrointestinal mucosal immune system and the virtually infinite microbial and alimentary antigens on the mucosal surface; intestinal epithelial barrier dysfunction is a pathophysiological characteristic of IBD that is related, at least in part, to defective intercellular junction molecule expression and assembly7 as a primary, genetically determined phenomenon and/or as a consequence of subepithelial inflammation.
Liver X receptor (LXR) is a transcriptional regulator that functions as a heterodimer with retinoic X receptor, modulating the transcription of several target genes including ABCA1 and FAS.8 Two isoforms have been identified, α and β (LXRα and LXRβ), and its nuclear localization has been related to transcription activity.9 Activation of LXR depends on ligand interaction by its endogenous agonist oxysterols, oxidized cholesterol metabolites, and by synthetic agonists.10-12
Several studies have identified LXR’s anti-inflammatory properties and its capability to regulate innate and adaptive immune responses, apoptosis, and phagocytosis.13 The activation of LXR has led to the inhibition of such inflammatory genes as iNOS, COX-2, IL-6, and IL-1β in macrophages when they were challenged with lipopolysaccharides (LPS) and cytokines because of interference with the action of nuclear factor kappa B (NF-κB) on the promoter of proinflammatory genes.14 Furthermore, LXR activation in macrophages inhibits the TLR-2, -4, and -9 signaling pathway through ABCA1-dependent changes.15, 16 The analysis of LXR in tumors has also exhibited interesting results: LXR and ABCA1 expression was reduced in human colorectal cancer and metastases, and LXR activation showed an inhibition of proliferation in colorectal cancer cells, xenograft, and colon carcinogenesis models through the blockage at the first phase of the cell cycle and the activation of an apoptosis program.17
The activation of LXR leads to improvement in several animal models of inflammatory and immune-mediated diseases18-20; this effect has also been studied in chemical models of IBD induced by dextran sodium sulfate and 2,4,6-trinitrobenzene sulfonic acid21 but not in spontaneous models such as IL-10-deficient mice. The LXR-deficient mice had less weight, more rectal bleeding, and diarrhea when exposed to dextran sodium sulfate or 2,4,6-trinitrobenzene sulfonic acid, and these effects were more intense in both LXRβ-and LXRαβ-deficient mice; in addition, treatment with the synthetic agonist GW3965 led to a faster weight gain and a less inflammatory response in the colon of wild-type (WT) but not LXR-deficient mice.
Regarding LXR target genes that mediate the anti-inflammatory effects of this receptor in the gut, it has been shown that FAS expression in the intestine modulates homeostasis by regulating Mucin 2, and its blockade consequently triggers mucus diminution, raised intestinal permeability, colitis, systemic inflammation, and dysbiosis.22 Nevertheless, ABCA1-dependent anti-inflammatory actions have not been investigated in the intestinal epithelium. In fact, LXR and pregnane X receptor polymorphisms implied a higher susceptibility to UC in a Danish case-control study.23
In the present study, we aimed to explore the role of LXR in the colon under inflammatory conditions and to show its anti-inflammatory effects in this organ and the responsible target gene. Herein, we show a decreased expression of LXR in the IECs from patients with IBD and mice with spontaneous colitis because of IL-10 deficiency. We also show that LXR activation has anti-inflammatory effects in cultured IECs challenged with IL-1β, indicating that this effect is mediated by the inhibition of NF-κB and mitogen-activated protein (MAP) kinases via ABCA1.
PATIENTS, MATERIALS, AND METHODS
Ethical Considerations
Written informed consent was given by every patient included in the study. The ethics committee of our institution approved the study (reference number 204/11). Mouse studies were also approved by the animal research ethics committee of our institution, and it was performed in compliance with animal welfare protocols (Guide for the Care and Use of Laboratory Animals; National Institutes of Health publication 85-23, 1985).
Patients
Biopsy samples from the inflamed and preserved colonic mucosa of patients with IBD (UC, CD, and indeterminate colitis [IC]) and healthy control patients were collected during the colonoscopies that were performed according to clinical management. Patients who did not present both inflamed and preserved colonic mucosa were excluded. Grading of mucosal inflammation was standardized using scores in humans and mice. In patients with UC, we used the Mayo index24 for clinical and endoscopic evaluation; for histologic evaluation, we used scores described by Sandborn et al25 and by Riley et al.26 In patients with CD, we used the Harvey-Bradshaw Index27 and the Simple Endoscopic Score for Crohn’s Disease28; for CD histologic scoring, we used the Baars index.29 Asymptomatic patients who were included in the colorectal cancer screening program from our institution and without evidence of inflammation in the colonoscopy were recruited as healthy control patients.
Mice
We used the same colony of mice detailed in previous work from our laboratory.7 IL-10-/- mice (Jackson Laboratory, Bar Harbor, ME) and WT C57BL/6J mice were stored in standard conditions and fed a chow diet ad libitum. If rectal prolapse occurred then the animal was killed and excluded from the study. At ages 18 to 20 weeks, the mice were killed using CO2 overdose and cervical dislocation, and the colon was harvested.
Cells
Caco-2 cells (HTB-37, ATCC strain) were cultivated with 20% fetal bovine serum supplemented-Eagle’s Minimum Essential Medium (EMEM) at 37ºC in a 5% CO2 incubator. Cells were used between passages 10 to 20.
Histopathology
For histologic procedures, the colon from the murine model was fixed in Methacarn (60% methanol, 30% chloroform, 10% acetic acid), dehydrated in grown concentrations of ethanol, and finally embedded in paraffin. We mounted 5 μm slices from the colon specimens or colon biopsies in a microscope slide, and hematoxylin and eosin staining was performed according to the standard protocols. In a subset of patients with IBD and control patients, immunohistochemistry (IHC) was also performed following the standard protocols using antibodies against LXR (Imgenex for mice slices, 1:50 concentration, and Santa Cruz Biotechnology for human slices, 1:10) and ABCA1 (Santa Cruz Biotechnology, 1:25). Positive epithelial and nonepithelial nuclei and total nuclear counting was performed using Image J.
Quantitative Polymerase Chain Reaction
We obtained cDNA was obtained from the cellular extract or colon specimen. First, the sample was homogenized using a sonicator or a homogenizer in addition to 1 mL Tri-pure isolation reagent (Invitrogen, Darmstadt, Germany) according to manufacturer protocol. Second, DNase was added to avoid genomic DNA contamination. Reverse transcription was carried out with 500 ng of total RNA using a Supercript II kit (Bio-Rad Laboratories, Hercules, CA). Quantitative polymerase chain reaction was performed using Sybergreen Master Mix (Bio-Rad) and thermocycler IQ5 (Multicolor Real-time PCR Detection System, Bio-Rad). The primers used here are detailed as follows. 5´→3´ sequence: β-Actin: (F) CGGTTCCGATGCCCTGAGGCTCTT, (R) CGTCACACTTCATGATGGAATTGA. LXRα: (F) CCTTCCTCAAGGACTTCAGTTAGTTACAA, (R) CATGG CTCTGGAGAACTCAAAGAT. LXRβ: (F) CATTGCG ACTCCAGGACAAGA, (R) CCCAGATCTCGGACAG CAAG. IL-1β:(F) TCACAGCAGCACATCAACAA, (R) TGTCCTCATCCTCGAAGGTC. ABCA1: (F) GCCCTA CTGTCGGTTGACAT, (R) ATGCCACACCCTGAACTGAG. FAS: (F) CAAGTGTCCACCAACAAGCG, (R) GGAGCGCA GGATAGACTCAC. Villin: (F) GTGGAGTAACACCAAA TCCT, (R) ATCTGCCACAGGACTAAAAC. All primers were designed using the NCBI/Primer-BLAST designing tool. β-Actin and villin were used as housekeeping genes in the mouse and cell culture mRNA evaluations, respectively. A triplicated sample measurement was performed in all the experiments. We used the relative ddCt method to calculate the results.
Protein Isolation and Western Blot
In another subset of patients with IBD and control patients, we performed a Western blot (WB) of the colonic samples. Radioimmunoprecipitation assay buffer was used for protein isolation from the cellular extract, colon specimens, and biopsies. After standardization, electrophoresis was performed using 10% acrylamide precast minigels (Mini-Protean TGX Stain-Free Gels, Bio-Rad Laboratories). We used the primary antibody against LXRαβ (Santa Cruz Biotechnology, 1:100 concentration), ABCA1 (NOVUS Biologicals, 1:1000), β-actin (Sigma-Aldrich, 1:20 000), p44-42 MAP kinases (Cell Signaling Technology, 1:1000), and p44-42 MAP kinases phosphorylated form (Cell Signaling Technology, 1:1000) for an 8-hour incubation at 4ºC, and after secondary antibody incubation we developed the images of the membrane using the Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare) and the Alliance 4.7 developer. Densitometry analysis was performed using Image J. Glyceraldehyde 3-phosphate dehydrogenase was used as the reference protein (Cell Signaling Technology, 1:1000) for the human colonic biopsy WB, β-actin was used as the reference protein in the ABCA1 WB in Caco-2 cells, and total p44-42 MAP kinase was used for the phosphorylated MAP kinase WB.
Indirect Immunofluorescence Staining and Confocal Microscopy
Gelatin-embedded crystals preheated at 37ºC were used as a support for cell growth in p24 devices, where the experiment took place in the conditions described in the Results section. When the experiment was finished, the crystals were blocked, permeabilized, and incubated with the primary antibody against p65 (Santa Cruz Biotechnology, 1:200 concentration) for 8 hours, with the secondary fluorescent antibody and the fluorescein isothiocyanate (FITC)-conjugated phalloidin (1:40) and 4′,6-diamidino-2-phenylindole (DAPI) (1:500). After mounting, we used the confocal inverted microscope AOBS/SP2 (Leica Microsystems, Wetzlar, Germany) and Image J software to stratify the cells into low fluorescence (<75 arbitrary units), medium fluorescence (75-125 units), and high fluorescence (>125 units).
Enzyme-Linked Immunosorbent Assay
The IL-8 protein levels in culture media were evaluated using the human enzyme-linked immunosorbent assay kit according to the instructions provided by the manufacturer (eBioscience).
Drugs and Chemicals
We used GW3965 (Sigma-Aldrich) as an LXR agonist. Glibenclamide was used as a pharmacologic antagonist of ABCA1 (Sigma-Aldrich).
Statistical Analysis
Data are expressed as mean and standard deviation. Comparison of means was performed using a 2-tailed t test or 1-way analysis of variance for multiple comparisons. The Tukey test was used as posttest comparison after the analysis of variance. The Mann-Whitney U test or Kruskal-Wallis test were used to compare nonparametric data. The χ 2 test was used to compare 2 proportions. Correlations were tested using the Pearson or Spearman proof. Paired comparisons were used when the samples from the same patient were analyzed. We used IBM SPSS Statistics for Windows, version 21.0 (IBM Corp, Armonk, NY) and GraphPad Prism 7 (GraphPad Software). We considered P < 0.05 as statistically significant.
RESULTS
Patient Characteristics
Seventeen patients and 5 healthy control patients were included in the study. Patient characteristics are shown in Table 1. The mean age was 45 years, 47% were male, and 11.8% were active smokers. Twelve patients (70.6%) were previously diagnosed with UC, 4 (23.5%) with CD, and 1 (5.9%) with IC. At the time of colonoscopy, the mean partial Mayo score for patients with UC was 1.67, the mean Harvey-Bradshaw Index score for patients with CD was 2.2, the mean hemoglobin was 14 g/dL, the mean leukocyte count was 8023/mL, and the mean C-reactive protein was 0.5 mg/dL. The Montreal classification of the patients with IBD is shown in Supplementary Table 1. We found that 23% of the patients were on anti-tumor necrosis factor (TNF)-α treatment at the time of inclusion, 17.6% were on azathioprine, 11.8% were on low-bioavailability oral corticosteroids (budesonide or beclometasone dipropionate), and 76.5% were on mesalazine. No one was on systemic corticosteroid treatment. Index colonoscopy showed a Mayo endoscopic score for patients with UC of 1 (50%), 2 (41.7%), and 3 (8.3%), but none of the patients had 0 points. The mean Simple Endoscopic Score for Crohn’s Disease was 5.8. During the following year, 5 (29.4%) patients with IBD of the patients with IBD required treatment escalation, 2 (11.8%) required hospital admission, and one patient (5.9%) required surgery.
Patient Characteristics
| IBD (n = 17) | Control Patients (n = 5) | P | |
|---|---|---|---|
| Age, y, mean (SD) | 45.24 (14) | 64.6 (13) | P = 0.016 |
| Sex, % (male) | 47.1% | 40% | P = 0.793 |
| Smoking habit (%) | |||
| No | 64.7 | ||
| Yes | 11.8 | ||
| Ex smoker | 5.9 | ||
| Unknown | 17.6 | ||
| IBD type (%) | |||
| UC | 70.6 | ||
| CD | 23.5 | ||
| IC | 5.9 | ||
| Baseline Mayo partial score, mean (SD) | 1.67 (1.67) | ||
| Baseline Harvey-Bradshaw Index, mean (SD) | 2.20 (1.3) | ||
| Hemoglobin, g/dL, mean (SD) | 14.05 (1.47) | ||
| Leukocytes/μL, mean (SD) | 8023 (2359) | ||
| C-reactive protein, mg/dL, mean (SD) | 0.51 (0.51) |
| IBD (n = 17) | Control Patients (n = 5) | P | |
|---|---|---|---|
| Age, y, mean (SD) | 45.24 (14) | 64.6 (13) | P = 0.016 |
| Sex, % (male) | 47.1% | 40% | P = 0.793 |
| Smoking habit (%) | |||
| No | 64.7 | ||
| Yes | 11.8 | ||
| Ex smoker | 5.9 | ||
| Unknown | 17.6 | ||
| IBD type (%) | |||
| UC | 70.6 | ||
| CD | 23.5 | ||
| IC | 5.9 | ||
| Baseline Mayo partial score, mean (SD) | 1.67 (1.67) | ||
| Baseline Harvey-Bradshaw Index, mean (SD) | 2.20 (1.3) | ||
| Hemoglobin, g/dL, mean (SD) | 14.05 (1.47) | ||
| Leukocytes/μL, mean (SD) | 8023 (2359) | ||
| C-reactive protein, mg/dL, mean (SD) | 0.51 (0.51) |
Patient Characteristics
| IBD (n = 17) | Control Patients (n = 5) | P | |
|---|---|---|---|
| Age, y, mean (SD) | 45.24 (14) | 64.6 (13) | P = 0.016 |
| Sex, % (male) | 47.1% | 40% | P = 0.793 |
| Smoking habit (%) | |||
| No | 64.7 | ||
| Yes | 11.8 | ||
| Ex smoker | 5.9 | ||
| Unknown | 17.6 | ||
| IBD type (%) | |||
| UC | 70.6 | ||
| CD | 23.5 | ||
| IC | 5.9 | ||
| Baseline Mayo partial score, mean (SD) | 1.67 (1.67) | ||
| Baseline Harvey-Bradshaw Index, mean (SD) | 2.20 (1.3) | ||
| Hemoglobin, g/dL, mean (SD) | 14.05 (1.47) | ||
| Leukocytes/μL, mean (SD) | 8023 (2359) | ||
| C-reactive protein, mg/dL, mean (SD) | 0.51 (0.51) |
| IBD (n = 17) | Control Patients (n = 5) | P | |
|---|---|---|---|
| Age, y, mean (SD) | 45.24 (14) | 64.6 (13) | P = 0.016 |
| Sex, % (male) | 47.1% | 40% | P = 0.793 |
| Smoking habit (%) | |||
| No | 64.7 | ||
| Yes | 11.8 | ||
| Ex smoker | 5.9 | ||
| Unknown | 17.6 | ||
| IBD type (%) | |||
| UC | 70.6 | ||
| CD | 23.5 | ||
| IC | 5.9 | ||
| Baseline Mayo partial score, mean (SD) | 1.67 (1.67) | ||
| Baseline Harvey-Bradshaw Index, mean (SD) | 2.20 (1.3) | ||
| Hemoglobin, g/dL, mean (SD) | 14.05 (1.47) | ||
| Leukocytes/μL, mean (SD) | 8023 (2359) | ||
| C-reactive protein, mg/dL, mean (SD) | 0.51 (0.51) |
LXR Protein Expression Decreased in Colonic Biopsies From Patients With IBD Compared With Control Patients
The semiquantitative protein expression of LXRαβ in colonic biopsies was analyzed in 11 patients with IBD (6 diagnosed with UC, 4 with CD, and 1 with IC) and 5 control patients. As shown in Fig. 1A, LXRαβ expression was higher in the control patients compared with both the inflamed and undamaged segments from patients with IBD without differences between both segments. No significant correlation between LXRαβ expression—assessed by WB—in both inflamed and healthy colonic mucosa and leukocyte count, C-reactive protein, endoscopic activity scores, and histologic scores were found (Supplementary Table 2).
(A) WB representative membrane and densitometry quantification of LXRαβ (1:100) and glyceraldehyde 3-phosphate dehydrogenase (1:1000) in endoscopic biopsies of healthy and inflamed colonic mucosa of patients with IBD (n = 11) and control patients (n = 5). (B) Representative images of LXRαβ immunohistochemistry (1:10) in endoscopic biopsies of healthy and inflamed colonic mucosa of patients with IBD and control patients along with the isotype control patients. (C) LXRαβ nuclei quantification in the epithelial compartment, given positive nuclei every 100 IEC, in endoscopic biopsies of healthy and inflamed colonic mucosa of patients with IBD (n = 7) and control patients (n = 4). The Mann-Whitney U test with paired analysis was used in the comparison of the biopsies of the same patient.
We also explored the expression of LXRαβ by IHC in colonic biopsies from patients with IBD (n = 7, 6 of whom were diagnosed with UC) and control patients (n = 4). The colonic samples from 1 patient with CD were evaluated using both IHC and WB. The positivity for LXRαβ staining in the epithelium was expressed as positive nuclei every 100 IECs. As shown in Figs. 1B and C, colonic epithelial nuclear positivity was highest in biopsies from the control patients compared with biopsies from the inflamed mucosa of patients with IBD; similarly, healthy mucosal segments from patients with IBD showed higher LXRαβ nuclear expression compared with inflamed segments, where LXRαβ staining was predominantly cytoplasmic (Supplementary Fig. 1). In addition, we analyzed the correlation between LXRαβ nuclear positivity in both segments and several clinical, endoscopic, and histologic variables (Supplementary Table 3), showing that LXRαβ nuclear positivity in endoscopically inflamed mucosa negatively correlated with the Riley histologic score in patients with UC (rho, –0.829; P = 0.042).
Decrease in LXR mRNA, Protein Expression, and Function in the Colon of IL-10-Deficient Mice vs WT Mice
Nineteen IL-10-deficient and 12 WT (C57BL/6) mice were killed at ages 18 to 20 weeks and the colon was harvested. Data regarding body weight, colon weight and length, and colonic expression of the proinflammatory cytokines TNF-α, IL-1β, and CCL28; iNOS; and the adhesion molecules E-cadherin and ZO-1 (previously reported by our group7) have been replicated for the present work (data not shown). As shown in Figs. 2A and B, we concluded that colonic mRNA levels of both LXRα and LXRβ isoforms were significantly lower in IL-10-/- mice compared with WT mice. Colonic IL-1β mRNA levels negatively correlated with both LXRα and LXRβ (Figs. 2C and D); a significant negative correlation between colonic TNFα mRNA levels and the expression of LXRα isoform was also shown (Fig. 2E), but no significant correlation was found between LXRβ and TNFα (Fig. 2F).
LXRα (A) and b) LXRβ RNA (B) expression levels related to β-actin expression in the colon of WT (n = 12) and IL-10-/- mice (n = 16). Correlation of IL-1β expression levels with (C) LXRα and (D) LXRβ, and correlation of TNF-α levels with (E) LXRα and (F) LXRβ in the colon of WT and IL-10-/- mice (n = 27). The Spearman test was used.
Colonic LXRαβ, LXRα, and LXRβ protein expression was subsequently assessed by immunostaining (Figs. 3A and B). We found that LXRαβ nuclear staining was higher in the colon of WT mice compared with that of IL-10-/- mice, at the expense of IEC expression (Fig. 3C); no significant differences were shown when nonepithelial cell nuclei were counted. Furthermore, we separately evaluated the α and β isoforms of LXR. No significant differences in the colonic epithelial expression of LXRα were found in IL-10-/- mice compared with WT mice (Fig. 3D), whereas LXRβ showed the same expression pattern as the total LXR, ie, a decreased nuclear staining in the epithelial cell compartment of IL-10-/- mice (Fig. 3E).
Representative images of (A) LXRαβ (1:50) and (B) LXRα (1:50) and LXRβ (1:50) IHC in the colon of WT mice and IL-10-/- mice. (C) LXRαβ, (D) LXRα, and (E) LXRβ positive nuclear staining quantification in epithelial and nonepithelial cell compartment IHC in the colon of WT mice (n = 3) and IL-10-/- mice (n = 6 for LXRαβ, n = 4 for LXRα and LXRβ).
In addition, we evaluated the expression of 2 LXR target genes—FAS and ABCA1—in the colon of WT and IL-10-/- mice as an estimation of LXR function; compared with the colon of WT mice, the colon of IL-10-/- mice showed decreased mRNA expression of both genes (Figs. 4A, B). The analysis of ABCA1 protein expression by immunostaining confirmed those results (Fig. 4C).
(A) FAS and (B) ABCA1 RNA expression levels in the colon of WT (n = 10) and IL-10-/- mice (n = 16). β-actin was used as the housekeeping gene. (C) Representative images of ABCA1 (1:25) IHC in the colon of WT mice and IL-10-/- mice.
Anti-Inflammatory Effects of LXR Activation When Caco-2 Cells Were Challenged With IL-1β
The incubation of Caco-2 cells with the proinflammatory cytokine IL-1β (25 ng/mL) for 6 hours significantly decreased the expression of the LXR target gene ABCA1 (Fig. 5A); no significant effect on LXR isoforms expression was found (Figs. 5B, C). On the other hand, LXR activation by the synthetic agonist GW3965 increased the expression of ABCA1 mRNA in Caco-2 cells in a dose-dependent manner (Fig. 5D). We also confirmed the increased protein expression of ABCA1 after GW3965 stimulation (Fig. 5E). Although not significant, 2 mM GW3965 seemed to increase the expression of LXRα without affecting the expression of the LXRβ isoform (Figs. 5F, G); GW3965 treatment (2 μM) led to a decreased expression of IL-8 mRNA in IL-1β-stimulated Caco-2 cells and to lower but not significant levels of chemokine (C-C motif) ligand (CCL)-28 (Figs. 5H, I, respectively). We also confirmed a decreased IL-8 expression at the protein level in the supernatant of cultured media (Fig. 5J).
RNA expression levels of (A) ABCA1, (B) LXRα, and (C) LXRβ in IL-1β-challenged (25 ng/mL) Caco-2 cells (n = 3) during 3, 6, and 24 hours. RNA expression levels of (D) ABCA1 in Caco-2 cells (n = 4) treated during 18 hours with the pharmacological solvent alone (vehicle) dimethylsulfoxide (DMSO) or GW3965 at 0.5, 1, and 2 μM concentrations. (E) WB membrane and densitometry quantification of ABCA1 in cultured Caco-2 cells (n = 4) treated during 18 hours with vehicle (DMSO) or GW3965 (2 μM). RNA expression levels of (F) LXRα and (G) LXRβ in Caco-2 cells (n = 4) treated during 18 hours with vehicle (DMSO) or GW3965 at 0.5, 1,and 2 μM concentrations. RNA expression levels of (H) IL-8 and (I) CCL-28 in IL-1β challenged (6 hours, 25 ng/mL) Caco-2 cells (n = 8) treated during 18 hours with vehicle (DMSO) or GW3965 (2 μM). Villin was used as the housekeeping gene. (J) Supernatant IL-8 concentration (measured by enzyme-linked immunosorbent assay) in IL-1β challenged (24 hours, 25 ng/mL) Caco-2 cells (n = 6) treated during 18 hours with vehicle (DMSO) or GW3965 (2 μM). *P < 0.05, in comparison with respect to vehicle. #P < 0.05, comparison with vehicle treated and IL-1β-challenged cells.
Mediation of Anti-Inflammatory Effect of LXR in Caco-2 Cells by Inhibition of NF-κB and MAP Kinases via ABCA1
To investigate in depth the mechanisms of this anti-inflammatory effect, we tested 2 of the main pathways of IL-1β receptor activation: NF-κB and MAP kinases. The nuclear translocation of the p65 subunit of NF-κB was evaluated by IFI in Caco-2 cells after GW3965 treatment and IL-1β stimulation. The GW3965 treatment decreased the high-positive p65 nuclei after IL-1β stimulation (Figs. 6A, B). We next tested the phosphorylation of p44-42 MAP kinase by Western blot under the same experimental conditions. The GW3965 treatment significantly reduced the phosphorylation ratio of p44-42 after IL-1β stimulation (Fig. 6C).
![(A) Representative images of indirect immunofluorescence staining of p65 subunit of NF-κB (1:200), F-actin (1:40), and both dyes superposed with DAPI (1:500) in IL-1β challenged (30 minutes, 25 ng/mL) Caco-2 cells (n = 6) treated during 18 hours with vehicle (dimethylsulfoxide [DMSO]) or GW3965 (2 μM). (B) Quantification of the percentage of high- (>125 arbitrary fluorescence units), medium- (125-75 units), and low- (<75 units) fluorescence nuclei in the previous mentioned experiment. (C) WB membrane and densitometry quantification of the phosphorylated p44-42 MAP kinase in IL-1β-challenged (10 minutes, 25 ng/mL) Caco-2 cells (n = 11) treated during 18 hours with vehicle (DMSO) or GW3965 (2 μM). The total p44-42 MAP kinase form was used as the reference protein. The Tukey comparison test was used.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/ibdjournal/27/10/10.1093_ibd_izab034/2/m_izab034_fig6.jpeg?Expires=1747877857&Signature=bOyeuezVl5PA~-6EK19EogYj85qzE0c33WrPQcizxiMP0-rPe9zsSp5vNwHOgaaPIicTecjt5mF8pgMSnIyK3avhoSQIABjMM6~4hDv-51fvBboWF80~s00RuTiCDhDAT17lX6eOnwV5o86ItJVBbutc81NKqlbal4b7cJnFoC~8LEdHN1EETfzUnF8-S1WNXNbJnTqt~U1juWPjDqnVk3HgECndy~TNWK5E60cEb9Bg6ePSpLke1gGFR5qTth7Tt-7iZ2FENEyB4Fcu4wkQ4~hH8HmujzAQew~LtGqtERytp~9yvn0WFABQFZJ6xB6Y89EzeI08-KSn-vWnToi86A__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
(A) Representative images of indirect immunofluorescence staining of p65 subunit of NF-κB (1:200), F-actin (1:40), and both dyes superposed with DAPI (1:500) in IL-1β challenged (30 minutes, 25 ng/mL) Caco-2 cells (n = 6) treated during 18 hours with vehicle (dimethylsulfoxide [DMSO]) or GW3965 (2 μM). (B) Quantification of the percentage of high- (>125 arbitrary fluorescence units), medium- (125-75 units), and low- (<75 units) fluorescence nuclei in the previous mentioned experiment. (C) WB membrane and densitometry quantification of the phosphorylated p44-42 MAP kinase in IL-1β-challenged (10 minutes, 25 ng/mL) Caco-2 cells (n = 11) treated during 18 hours with vehicle (DMSO) or GW3965 (2 μM). The total p44-42 MAP kinase form was used as the reference protein. The Tukey comparison test was used.
Research has shown that ABCA1 has been related to anti-inflammatory effects itself in the macrophage,15, 30 and these effects have been related to NF-κB and the MAP kinase pathway.31 We used glibenclamide as a pharmacological antagonist of ABCA1 in cultured Caco-2 cells. As shown in Fig. 7, glibenclamide treatment increased the phosphorylation of p44-42 after GW3965 treatment and IL-1β stimulation.
![WB membrane and densitometry quantification of the phosphorylated MAP kinase in IL-1β-challenged (10 minutes, 25 ng/mL) Caco-2 cells (n = 8) treated during 18 hours with vehicle (dimethylsulfoxide [DMSO]) or GW3965 (2 μM) and in the presence or absence of glybenclamide 0.5 mM for 6 hours. The total p44-42 MAP kinase form was used as the reference protein. The Student t test was used.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/ibdjournal/27/10/10.1093_ibd_izab034/2/m_izab034_fig7.jpeg?Expires=1747877857&Signature=sxJz2Emme-vcDPCeN3eRe~81qg2vm75FihsvdAQficSZPlbNxq1XmJ3jwoc1di1NtOzli4ebL6vMF0AjMr5c7l0NTWBM9kyzJ0lOlHHWaJJFJ7TD~zncKU87aoSS09wPF0j7cPlbrRl4~ktyQXCMQhXM8U54GvfeXW~tAy~ORHu-vswPQgk0AZ1~MuO4DS6NKVQueZmcjM0am54JCzxcyiKtVJbj2nrPYzZU3M7hgedpLPYcmlcT2WARaJlJHKZ9yoi0GltLBiuiipnV2TSD25zBQZm52aZnK12wZyh5zBHGPmgVTR-hMewM-a2~R6a~A6I4uuObDrjVat4Kjshzhw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
WB membrane and densitometry quantification of the phosphorylated MAP kinase in IL-1β-challenged (10 minutes, 25 ng/mL) Caco-2 cells (n = 8) treated during 18 hours with vehicle (dimethylsulfoxide [DMSO]) or GW3965 (2 μM) and in the presence or absence of glybenclamide 0.5 mM for 6 hours. The total p44-42 MAP kinase form was used as the reference protein. The Student t test was used.
DISCUSSION
Studies have shown that LXR is a ubiquitous transcriptional factor of great relevance to lipid metabolism, as it regulates reverse cholesterol transport in both the macrophage and the intestine.32-34 It is also crucial in cell proliferation and death35, 36 and has shown anti-inflammatory effects in macrophages.37 In the present work, we aimed to explore the relevance of epithelial LXR in intestinal inflammation, focusing on its expression and function in the colon of patients with IBD, in a mouse model of spontaneous genetically driven colitis resembling human IBD, and in Caco-2 cells.
As previously reported,21 we confirmed that LXR expression in colonic biopsies of patients with IBD included in the present study was significantly lower compared with those of control patients. Our results indicate that LXR nuclear positivity quantification—reflecting not only protein expression but also, presumably, its function because the nuclear localization of LXR has been linked with the activated form9—in the inflamed mucosa from the subgroup of patients diagnosed with UC negatively correlated with histologic activity. The downregulation of LXR expression in this scenario could be secondary to mucosal inflammation. Nevertheless, and although TLR and IL-1 receptor activation abrogate LXR transcription in macrophages,37, 38 the incubation of cultured IECs with IL-1β had no significant effect on LXR isoform expression, whereas it decreased the expression of the LXR target gene ABCA1. Moreover, LXR downregulation could also represent a consequence of microbiota disturbances. Butyrate, a short-chain fatty acid produced by colonic flora metabolism with known anti-inflammatory effects in colonic mucosal inflammation,39, 40 is a key regulator of peroxisome proliferator-activated receptor–driven nuclear receptor expression.41 A diminished butyrate production may downregulate LXR expression42 and may trigger gut inflammation. On the other hand, we have shown decreased LXR protein expression, not only in endoscopically inflamed segments from patients with IBD but also in the apparently healthy mucosa from the same individuals. In this sense, a previous work from our group showed that increased iNOS expression and activity in the macroscopically healthy mucosa from patients with left-sided UC predicted the proximal extent of the disease in follow-up,43 and it is already known that LXR activation inhibits iNOS expression in macrophages.14 The possibility of a primary defect of LXR expression and function in the colonic epithelium of patients with IBD and the significance of its decreased expression in healthy mucosa from patients with IBD deserves further investigation.
We also show for the first time decreased LXR expression in the epithelial cells compartment from the colon of IL-10-/- mice; moreover, colonic IL-1β and TNF-α mRNA levels were negatively correlated with epithelial LXR expression. With regard to the latter, note the differential relationship between each LXR isophorm α and β and the expression of these cytokines. Both isoforms are present in the colon, and LXRβ has shown anti-inflammatory effects.21, 44 However, LXRα may also have this beneficial role, resulting at least from its regulation of the cannabinoid receptors and their known protective effect in the colon.45, 46
The IL-10-/- mice develop spontaneous colitis and constitute a well-characterized tool for the study of IBD pathophysiology. The relevance of IL-10 in the human IBD setting was outlined by the fact that carrying mutations in both IL-10 and IL-10 receptor genes is associated with early-onset pediatric CD. Research has shown that IL-10 derived from FoxP3+ regulatory T cells exerts a pivotal role in the maintenance of mucosal tolerance.47 Furthermore, the ability of IL-10 to increase the expression of LXRα in macrophages has been previously shown,48 and IL-10 has cross-talk with LXRα in such cell types, increasing the expression of ABCA1 and reverse cholesterol transport.49 Our results suggest that LXR function is also disturbed in the gut of IL-10-/- mice because the expression of 2 LXR target genes—FAS and ABCA1—was significantly diminished. The altered expression of such molecules, crucial in gut homeostasis, may result in an uncontrolled inflammatory response. In this sense, FAS has been found in lower amounts in the colon of patients with IBD,50 and this fact may lead to lower phosphatidil colin in the rectal mucus of patients with UC 51 and intestinal barrier dysfunction.52 Interestingly, an intestine-specific LXR ligand named GW6340 has been developed.32 The relationship between IL-10 and LXR in IECs in vivo or in vitro has not been previously explored, and the mechanisms involved in intestinal LXR downregulation in the absence of IL-10 remain to be elucidated.
Our results indicate that LXR activation downregulates the expression of the chemokines IL-8 and CCL-28 in Caco-2 cells, and this anti-inflammatory effect of LXR is mediated by the inhibition of NF-κB and MAP kinases. The production of IL-8 after IL-1β challenge has been previously published in epithelial cell lines.53, 54 We found a remarkable increase in this chemokine in the Caco-2 cells measured by enzyme-linked immunosorbent assay, higher than previously observed with IL-1β; the long period of incubation (24 hours) and high IL-1β concentration (25 ng/mL) used in our experiments may explain this magnitude of increase because time and dose response seem to influence IL-8 production after cytokine challenge.55 This process is dependent, at least in part, on ABCA1, because the inhibition of this molecule with glibenclamide partially inhibits the downregulation of MAP kinase phosphorylation observed after LXR activation with GW3965. Although the anti-inflammatory role of ABCA1 in the intestinal epithelium has been proven in the macrophage, our investigation is the first one to outline it. Studies have shown that ABCA1-deficient mice have higher plasmatic levels of IL-1β and TNF-α under basal conditions and after intraperitoneal LPS.31 It has been suggested that the anti-inflammatory effect of LXR activation in the macrophage is mediated by ABCA1, and TLR and their effector pathway including MAP kinases are essential in triggering the production of inflammatory mediators.15
Interestingly, ABCA1 represents a link between lipid metabolism and inflammation. Lipid rafts, cholesterol-enriched membrane microdomains that participate in the transmembrane traffic and immunologic processes, recruit TLR depending on cholesterol saturation and consequently activate proinflammatory pathways. Higher cholesterol levels in the lipid rafts entail TLR-4 recruitment and proinflammatory pathway activation in alveolar macrophages.56 Studies have shown that LXR agonists reduced cholesterol levels in the lipid rafts and cytokine expression in macrophages and reduced the recruitment of MyD88 and TRAF6 in LPS-stimulated macrophages, and this effect disappeared in ABCA1-deficient macrophages.15, 57 Finally, ABCA1-deficient macrophages had higher cholesterol levels and higher TLR-4 recruitment in the lipid rafts.57
Our work has several limitations. First, our human study population was not big enough to show a correlation of LXR with some clinically important follow-up variables such as the need for surgery, colonic dysplasia, and cancer. Second, adenocarcinoma-derived Caco-2 cells may not represent an optimal model for in vitro studies because metabolic requirements and behaviors differ from primary IECs. In addition, we illustrate the importance of LXR in colonic inflammation but cannot assure its cause-consequence status. Finally, although statistically significant, the differences of LXR expression in the colon between inflamed and healthy mucosa, in both humans and mice, are subtle and should be confirmed in a larger series. Further, the biological significance of LXR expression should be explored in depth; in this sense, we have shown that colonic LXR expression in mice runs in parallel with its function because its target genes, FAS and ABCA1, are decreased in IL-10-/- mice compared with those in WT mice; nevertheless, this point deserves further investigation in human IBD.
CONCLUSIONS
We found that LXR exerts anti-inflammatory effects in Caco-2 cells that are mediated by ABCA1, and its expression is diminished in the colonic epithelium of patients diagnosed with IBD and IL-10-/- mice. One may consider LXR a key molecule in the development of gut inflammation in human and experimental colonic inflammation in these scenarios and, consequently, a promising therapeutic target.
Supported by: Work for this study was funded by grants from the Spanish Ministry of Health (Instituto de Salud Carlos III -ISCIII- FIS 13/02555—19/01746 and CIBEREHD) to LM.
ACKNOWLEDGMENTS
We thank Paola Lastra, Mileidis Acosta, Carmen Martínez, Celia de Gracia, Esperanza Martos, Rodrigo Borobia, María López, Helena Martínez, and Jon de la Maza for helping us with the recruitment of patients in the Endoscopy Unit.
REFERENCES
Author notes
Both authors contributed equally.
