- Email: [email protected]
Obesity-Induced Lymphocyte Hyperresponsiveness to Chemokines: A New Mechanism of Fatty Liver Inflammation in Obese Mice
GASTROENTEROLOGY 2008;134:1459 –1469
BASIC–LIVER, PANCREAS, AND BILIARY TRACT Obesity-Induced Lymphocyte Hyperresponsiveness to Chemokines: A New Mechanism of Fatty Liver Inflammation in Obese Mice
AMÉLIE E. BIGORGNE,*,‡ LAURENCE BOUCHET–DELBOS,*,‡ SYLVIE NAVEAU,*,‡,§ IBRAHIM DAGHER,储 SOPHIE PRÉVOT,¶ INGRID DURAND–GASSELIN,* JACQUES COUDERC,* PHILIPPE VALET,#,** DOMINIQUE EMILIE,*,‡,‡‡ and GABRIEL PERLEMUTER*,‡,§ *INSERM, U764, Clamart; ‡Univ Paris-Sud, Faculté de Médecine Paris-Sud, Institut Fédératif de Recherche 13, Clamart; §AP-HP, Hôpital Antoine Béclère, Service d’Hépato-Gastroentérologie, Clamart; 储AP-HP, Hôpital Antoine Béclère, Service de Chirurgie, Clamart; ¶AP-HP, Hôpital Antoine Béclère, Service d’AnatomoPathologie, Clamart; #INSERM, U858, Toulouse; **Université de Toulouse, UPS, Institut de Médecine Moléculaire de Rangueil, Equipe no. 3, Toulouse Cedex 4; and ‡‡AP-HP, Hôpital Antoine Béclère, Service de Microbiologie-Immunologie Biologique, Clamart, France
N
onalcoholic fatty liver disease (NAFLD) is an increasingly recognized condition that may lead to end-stage liver disease. NAFLD corresponds to a wide spectrum of pathologic lesions, ranging from pure steatosis to steatohepatitis, fibrosis, cirrhosis, and hepatocellular carcinoma. Nonalcoholic steatohepatitis (NASH) is merely a stage in NAFLD: steatosis—the retention of lipids in hepatocytes—predisposes the liver to the development of more severe lesions.1 However, steatosis per se may not be a severe disease because many patients do not progress to necroinflammation or fibrosis.2 Steatohepatitis may be seen as a “double-hit” lesion,3 with steatosis (the first hit) rendering the liver vulnerable to further injury. A second hit, such as lipid peroxidation or cytokine secretion,4,5 then triggers the recruitment of inflammatory cells by the liver.6 Several lines of evidence suggest a role for endotoxin in fatty liver inflammation. Lipopolysaccharide (LPS) is a membrane component of gram-negative commensal bacteria present in the digestive tract. A role for endotoxin in the pathogenesis of NASH is suggested by the high incidence of NASH and cirrhosis following jejuno-ileal bypass surgery for obesity. This situation can induce bacterial overgrowth in the defunctionalized small intestine, resulting in LPS absorption.7 The hepatotoxicity of LPS results from the release of cytokines, such as interferon ␥, which increases the sensitivity of hepatocytes to tumor necrosis factor ␣.8 Abbreviations used in this paper: HFD, high-fat diet; LPS, lipopolysaccharide; mAbs, monoclonal antibodies; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NK, natural killer. © 2008 by the AGA Institute 0016-5085/08/$34.00 doi:10.1053/j.gastro.2008.02.055
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Background & Aims: Hepatic lipid retention (steatosis) predisposes hepatitis. We investigated the mechanisms of lymphocyte homing to fatty liver and the role of lipopolysaccharide (LPS) in the onset of inflammation in ob/ob mice. Methods: We decreased intestinal bacterial compounds by oral antibiotic treatment to test the role of endogenous LPS in liver inflammation. Adoptive transfer of lymphocytes was used to study the respective contributions of steatosis and lymphocytes to liver inflammation. We tested lymphocyte response to chemokines by in vitro chemotaxis assays in ob/ob, their lean controls, and “nonobese ob/ob” mice, generated by controlling caloric intake to distinguish between the effects of obesity and leptin deficiency. Results: Antibiotic treatment decreased liver infiltration with CD4ⴙ T, CD8ⴙ T, natural killer (NK)T, B, and NK cells. Adoptive transfer of lymphocytes from ob/ob or control mice showed that (1) steatosis increased lymphocyte recruitment to the liver; (2) CD4ⴙ T, CD8ⴙ T, and B cells from ob/ob mice had a greater propensity to migrate specifically to the liver. This migration was enhanced by LPS. These results were also observed in a model of high-fat diet-induced obesity. CD4ⴙ T and B cells were hyperresponsive to CXCL12 and CXCL13, respectively. Weight normalization in “non-obese ob/ob” mice decreased liver inflammation, lymphocyte response to chemokines, and homing to the liver. Conclusions: Our study provides the first evidence that liver inflammation in mice with genetic or diet-induced obesity results from both steatosis and lymphocyte hyperresponsiveness to chemokines expressed in the liver. These abnormalities are reversible with weight normalization.
1460
BIGORGNE ET AL
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
C57BL/6 ob/ob mice have a mutation that prevents leptin synthesis, resulting in a phenotype of obesity, hepatomegaly, severe steatosis, and high serum levels of aminotransferase.9 Ob/ob mice are much more sensitive to LPS than their lean littermates, rapidly developing liver inflammation after exposure to LPS.8,10,11 The mechanism underlying the role of LPS in the onset of lymphocyte homing to fatty liver remains unknown. CXC chemokine expression has been shown to be induced in endotoxemic animals.12 CXCL12 (stromal cellderived factor 1␣ [SDF-1␣]) is expressed in a wide variety of tissues, including bile duct epithelial cells.13 CXCL13 (B cell chemoattractant chemokine, BLC/BCA-1) selectively attracts B cells and is expressed in the liver.14,15 Several studies have suggested that the homing of circulating lymphocytes to the liver may increase during inflammation, but no study has yet formally confirmed this hypothesis. We explored the pathogenesis of NASH further by investigating the role of digestive LPS in the onset of liver inflammation and the mechanisms of lymphocyte homing to fatty liver in ob/ob mice. We first demonstrate that digestive bacterial compounds are involved in liver inflammation in ob/ob mice. We then show that CD4⫹ T, CD8⫹ T, and B cells from ob/ob mice have an intrinsic propensity to migrate specifically to the liver, which is enhanced by LPS challenge. Moreover, we obtain similar results in a model of high-fat diet (HFD)-induced obesity. Lymphocytes from ob/ob mice are hyperresponsive to at least 2 CXC chemokines expressed in the liver. We demonstrate that, in the ob/ob model, obesity itself—rather than leptin deficiency—is directly involved in liver inflammation and lymphocyte sensitivity to chemokines. Our study provides the first evidence that liver inflammation in obesity is due to both steatosis and lymphocyte hyperresponsiveness to chemotactic agents, which are reversible with weight normalization.
Materials and Methods Mice Five-week-old male C57BL/6 ob/ob mice and controls (lean littermates) were purchased from Janvier (Le Genest St. Isle, France). Mice were treated in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996). They were fed a diet consisting of 22% protein, 4.3% lipid, and 70% carbohydrate supplied ad libitum. “Non-obese ob/ob” mice were supplied daily with the same caloric intake as lean controls, from the age of 5 weeks, and experiments were performed on 12-week-old mice. C57BL/6 mice fed a HFD were purchased at the age of 3 weeks and fed a diet consisting of 15.5% protein, 46% lipid, and 38.5% carbohydrate (Safe, Augy, France) supplied ad libitum, and experiments were performed after 18 weeks on the HFD.
GASTROENTEROLOGY Vol. 134, No. 5
Reduction of the Level of Endogenous Digestive Bacterial Compounds Ciprofloxacin (Bayer Pharma, Germany) and ornidazol (Roche, Switzerland) were given in drinking water at a daily dose of 10 g/g and 15 g/g body weight, respectively, for 10 days, whereas the control group received water only.
Metabolic Status of Mice Blood was collected by retroorbital vein puncture. The serum was stored at ⫺20°C until use for alanine and aspartate aminotransferase (ALT, AST, respectively), triglyceride, glucose, total cholesterol, and high-density lipoprotein-cholesterol determinations (OLYMPUS AU400). Serum insulin was determined by enzyme-linked immunosorbent assay (ELISA) (R&D Systems, Minneapolis, MN). Insulin resistance was determined by the product: fasting serum glucose (mmol/L) ⫻ insulin (mU/L).
Quantification and Phenotyping of Lymphocytes Infiltrating the Liver Liver samples of 6 mice from each group were fixed in 36% formaldehyde and embedded in paraffin. Sections (5 m) were cut and stained with hematoxylin, eosin, and saffron. Slides were read blindly. Lobular inflammation was defined according to Kleiner’s score16: 0, no focus; 1, ⬍2 foci; 2, 2– 4 foci; 3, ⬎4 foci per 200⫻ field. For quantification and phenotyping of lymphocytes infiltrating the liver, lymphocytes were isolated from mouse liver by a 2-step perfusion procedure. In the first step, circulating cells were eliminated by perfusing the liver in situ through the portal vein with Ca2⫹- and Mg2⫹-free HEPES (pH 7.65) supplemented with 5 mmol/L EDTA at 37°C, at a flow rate of 4 mL/min. The liver was then perfused with 0.05% collagenase IV (Sigma–Aldrich, St. Louis, MO) buffered with 0.1 mol/L HEPES at 37°C for 5 minutes, excised, and homogenized. The suspension was filtered through a 70-m filter, and the filtrate was centrifuged at 50g at room temperature for 2 minutes to remove liver parenchymal cells. The lymphocyte fraction was stained with monoclonal antibodies (mAbs) for flow cytometry analysis.
Adoptive Transfer for Studying the Contribution of Steatosis to Lymphocyte Homing Source. Lean donor mice were challenged intraperitoneally (IP) with LPS (0.5 g/g body weight) (Alexis, Switzerland) diluted in sterile saline and killed 6 hours later. A lymphocyte suspension was obtained by tearing the spleen with fine forceps in 1X phosphate-buffered saline (PBS) in sterile conditions. Lymphocytes were purified by centrifuging this suspension through a sterile Lympholyte density gradient (Cedarlane, Canada) at 800g for 20 minutes at room temperature. Labeling. Lymphocytes were labeled with fluorescent cell tracers for living cells (CellTracker Orange
CMTMR; Molecular Probes, Eugene, OR). A suspension of 106 cells/mL was incubated for 30 minutes at 37°C with Orange CMTMR (1:250). Lymphocyte transfer to recipient mice. We transferred 6 ⫻ 106 lymphocytes intravenously (IV) to ob/ob and control recipient mice 6 hours after IP challenge with LPS (0.5 g/g body weight). Lymphocyte harvest. Recipient mice were killed 18 hours after lymphocyte transfer. Circulating cells were eliminated by the first perfusion step, as described previously. A fraction of the liver was then excised and weighed before digestion in a Petri dish with 0.05% collagenase IV buffered with 0.1 mol/L HEPES at 37°C for 10 minutes. The lymphocyte fraction was stained with mAbs. Analysis. Lymphocyte migration was quantified by flow cytometry and expressed as the number of fluorescent lymphocytes recruited per gram of liver.
Adoptive Transfer for Studying the Contribution of Lymphocytes to Homing to the Liver Source. For each experiment, 2 ob/ob and 2 control donor mice were challenged IP with LPS and killed 6 hours later. Lymphocytes were purified as described previously. Labeling. A suspension of 106 cells/mL was incubated for 30 minutes at 37°C with either Calcein AM (1:1000) (Molecular Probes) or CellTracker Orange CMTMR (1:250), according to their source. Lymphocyte transfer to recipient mice. We pooled 6 ⫻ 106 cells labeled with each cell tracer and transferred them IV into lean recipient mice 6 hours after LPS challenge. Lymphocyte harvest. Recipient mice were killed 18 hours after lymphocyte transfer. Lymphocytes were isolated from the spleen, lymph node, and thymus by homogenizing and seeping through a 40-m filter in 1X PBS and from the liver by a 2-step perfusion procedure. The lymphocyte fraction was stained with mAbs. Analysis. Lymphocyte migration was quantified by flow cytometry, according to lymphocyte source and phenotype. Results are expressed as the number of fluorescent cells recruited to the liver per 105 transferred cells.
In Vitro Chemotaxis Assays Lymphocyte chemotaxis was evaluated with the Transwell system (Corning Costar, MA). Lymphocytes were isolated from the spleen of ob/ob or control mice as described above 6 hours after LPS challenge and stained with fluorescent mAbs. We added 1.5 ⫻ 106 cells in 150 L RPMI supplemented with 20 mmol/L HEPES and 1% calf fetal serum (Perbio Sciences, Belgium) to the upper chamber and 600 L of the same medium with (1 g/ mL) or without murine CXCL12, CXCL13, CCL19, and CCL21 (R&D Systems) to the lower chamber. A sample of 1.5 ⫻ 106 cells was used as input control. Cells were
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1461
incubated at 37°C for 1 hour (for CXCL13 and CCL19) or 2 hours (for CXCL12 and CCL21). Results are expressed as the percentage of cells added to the upper chamber that migrated to the lower chamber for each phenotype.
Monoclonal Antibodies for Flow Cytometry Lymphocytes were stained with allophycocyaninconjugated rat anti-CD4, anti-CD8, and anti-CD19 mAbs; mouse anti-NK1.1; and PerCP-Cy5.5-conjugated hamster anti-CD3 mAbs (Pharmingen, CA). Lymphocyte subpopulations were analyzed by 4-color flow cytometry, using a FACScalibur machine (Becton Dickinson Immunocytometry Systems, NJ).
Immunodetection Immunofluorescence was performed on frozen liver sections for CXCL12, CXCL13, and CD31 (endothelial cells). Immunohistochemistry was performed on paraffin-embedded sections for CK19 (bile duct epithelial cells) and CXCL12 (see Supplementary Material online at www.gastrojournal.org for technical details).
Statistical Analysis Data are expressed as means ⫾ SEM. Quantitative data were compared using nonparametric (Mann–Whitney) tests and Kruskall–Wallis variance analysis. Multiple comparisons were performed, using Fisher PLSD test. A paired, 2-group Wilcoxon signed-rank test was used to compare the contributions of lymphocytes from obese and control mice to homing to the liver. P values less than .05 were considered significant.
Results Reducing Digestive Bacterial Compounds by Oral Antibiotic Treatment Decreases Liver Inflammation in ob/ob Mice We investigated the possible role of digestive bacterial compounds in the occurrence of inflammation in fatty liver by administering oral antibiotics against aerobic and anaerobic bacteria to ob/ob and control mice. Antibiotic treatment did not modify body weight (Table 1). Basal liver weight, expressed as a percentage of total body weight, was higher in ob/ob than in control mice. Liver weight significantly decreased in ob/ob mice after a 10-day course of antibiotic treatment (from 6.7% ⫾ 0.8% to 5.8% ⫾ 0.8%, respectively, P ⬍ .05) (Table 1). We assessed the impact of this treatment on serum aminotransferase levels. Antibiotic treatment strongly decreased ALT and AST levels in ob/ob mice but had no significant effect in control mice (Table 1). We then assessed the impact of digestive decontamination on liver inflammation. Fewer foci were observed in antibiotic-treated ob/ob mice (Kleiner’s score, 1.1 ⫾ 0.7) than in untreated ob/ob mice (Kleiner’s score, 1.9 ⫾ 0.5) (P ⬍ .001) (Table 1, Figure 1A and B). The number of CD4⫹ T (CD3⫹CD4⫹), CD8⫹ T (CD3⫹CD8⫹), natural killer
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
May 2008
1462
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Table 1. Decreased Liver Inflammation in ob/ob Mice After Antibiotic Treatment Control mice
Body weight (g) Liver weight (%) ALT (IU/L) AST (IU/L) Inflammation score (Kleiner’s score)
Ob/ob mice
Untreated
Antibiotics
Untreated
27.6 ⫾ 0.6 4.8 ⫾ 0.5 58 ⫾ 12 175 ⫾ 34 0.5 ⫾ 0.5
27.8 ⫾ 0.5 5.3 ⫾ 0.6 66 ⫾ 16 148 ⫾ 21 0.1 ⫾ 0.4
48.7 ⫾ 1.3 6.7 ⫾ 0.8 419 ⫾ 42 400 ⫾ 30 1.9 ⫾ 0.5
Antibiotics 47.3 ⫾ 0.7 5.8 ⫾ 0.8a 158 ⫾ 23b 224 ⫾ 29c 1.1 ⫾ 0.7b
NOTE. Table 1 shows means ⫾ SEM of 6 mice per group. Statistically significant differences between antibiotic-treated and untreated mice are indicated (nonparametric variance analysis [Mann–Whitney test], aP ⬍ .05, bP ⬍ .001, cP ⬍ .01).
(NK)T (CD3⫹NK1.1⫹), B (CD3⫺CD19⫹), and NK (CD3⫺NK1.1⫹) cells in the liver of ob/ob mice was significantly decreased by antibiotic treatment (Figure 1C). Overall, liver weight, biochemistry, pathology, and flow cytometry results demonstrated that reducing the level of endogenous digestive bacterial products decreased liver inflammation.
Steatosis Increases Lymphocyte Recruitment by the Liver
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
The low-grade liver inflammation induced by digestive bacterial products was amplified by challenging mice with LPS to study lymphocyte homing to fatty liver. We developed an adoptive transfer system for evaluating the contribution of liver steatosis to lymphocyte recruitment. Lymphocytes were isolated from LPS-challenged lean donor mice, labeled with a fluorescent cell tracer, and transferred to ob/ob or control LPS-challenged recipient mice (Figure 2A). The level of recruitment for CD4⫹ T, CD8⫹ T, NKT, B, and NK fluorescent cells to fatty liver, normalized for liver weight, was higher than the level of recruitment to control liver (Figure 3). Fatty livers therefore have a higher capacity to recruit B and T cells after LPS challenge than control livers.
Lymphocytes From Obese Mice Have a Stronger Propensity to Migrate to the Liver We then investigated whether inflammatory cells isolated from ob/ob mice displayed an intrinsic capacity to migrate to the liver. We first isolated and phenotyped lymphocytes from the spleens of ob/ob and control mice. There was no significant difference in the number of T and B cells between ob/ob and control mice. LPS challenge did not affect the composition of lymphocyte subsets (data not shown). Lymphocytes isolated from the spleens of LPS-challenged ob/ob and control donors were used for adoptive transfer. They were labeled to identify their source (ie, ob/ob or control) and transferred to recipient mice (Figure 2B). We found that the level of CD4⫹ T- and CD8⫹ T-cell migration to the liver was higher for ob/ob than for control lymphocytes: 46.3 ⫾ 9.2 vs 30.7 ⫾ 7.9 (P ⬍ .05) and 44.0 ⫾ 10.5 vs 25.7 ⫾ 7.0 cells recruited to the liver
for 105 transferred cells (P ⬍ .01), respectively. The level of B-cell migration into the liver was also higher for ob/ob than for control B cells: 17.1 ⫾ 2.6 vs 10.1 ⫾ 2.0 cells, respectively, recruited to the liver for 105 transferred cells (P ⬍ .001) (Figure 4, left column). No significant difference was observed for NK cells (data not shown). Lymphocyte migration into the spleen, thymus, and lymph nodes did not differ significantly between ob/ob and control cells (data not shown). Thus, the higher level of ob/ob CD4⫹ T-, CD8⫹ T-, and B-cell migration was specific to the liver. We validated the results obtained for ob/ob mice in another model of obesity by feeding C57BL/6 mice a HFD. These mice had a higher body weight than C57BL/6 control mice (42.8 ⫾ 1.0 vs 30.8 ⫾ 0.7 g, respectively; P ⬍ .05). Serum glucose, triglyceride, total cholesterol, high-density lipoprotein-cholesterol, and insulin ⫻ glucose product were significantly higher in HFD mice than in controls (see Supplementary Table online at www.gastrojournal.org). HFD mice showed steatosis and mild lobular inflammation (Kleiner’s score: 1.4 ⫾ 0.2 vs 0.75 ⫾ 0.16, respectively; P ⬍ .05; see Supplementary Figure 1 online at www.gastrojournal.org). Using the adoptive transfer procedure described above, we showed that the level of CD4⫹ T-, CD8⫹ T-, and B-cell migration to the liver was significantly higher for HFD lymphocytes than for controls (see Supplementary Figure 2 online at www.gastrojournal.org). Thus, obesity, whether genetic or diet induced, leads to an increase in the homing of circulating lymphocytes to the liver.
The Propensity of ob/ob Lymphocytes to Migrate to the Liver Persists in the Absence of LPS Challenge We investigated whether the activation of ob/ob lymphocytes by LPS challenge contributed to increases in lymphocyte migration to the liver by transferring lymphocytes isolated from unchallenged donors. The level of ob/ob lymphocyte migration was significantly lower than that for LPS-challenged donors for CD4⫹ T (Figure 4A and 4D, solid columns), CD8⫹ T (Figure 4B and 4E, solid columns), and B cells (Figure 4C and 4F, solid columns). Significantly lower levels of control lymphocyte migration were also observed in the absence of LPS challenge to
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1463
donors (Figure 4, open columns). Recruitment to lymphoid organs was also decreased by the absence of LPS challenge of donors (data not shown). Thus, LPS enhances lymphocyte migration to several organs, including the liver. However, despite the low level of lymphocyte migration in the absence of LPS challenge, ob/ob lymphocytes still displayed higher levels of migration to the liver than
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 1. Oral antibiotic treatment decreases liver inflammation in ob/ob mice. Oral ciprofloxacin and ornidazol treatment. Liver pathology of (A) untreated and (B) antibiotic-treated ob/ob mice (original magnification, ⫻40 and ⫻100, respectively). Solid arrows point to inflammatory cell foci, which are enlarged in the lower panels. Representative figures of 6 mice per group. (C) Flow cytometry analysis of lymphocytes infiltrating the liver. Graphs show means ⫾ SEM of 5 mice per group. Statistically significant differences (P ⬍ .05) between antibiotic-treated and untreated ob/ob mice are indicated by an asterisk (nonparametric variance analysis [Mann–Whitney test]).
control cells. This was found for CD4⫹ T, CD8⫹ T, and B cells (Figure 4, right columns). Migration was similar for ob/ob and control lymphocytes in the spleen, thymus, and lymph nodes in the absence of LPS challenge (data not shown). Our results show that ob/ob CD4⫹ T, CD8⫹ T, and B cells display higher levels of migration to the liver than control cells, whether or not they are exposed to LPS.
1464
BIGORGNE ET AL
Figure 2. Schematic representation of the adoptive transfer procedure. (A) Study of the contribution of steatosis to lymphocyte homing. Lymphocytes, isolated from the spleen of LPS-challenged lean donor mice, were labeled with a fluorescent cell tracer and transferred to LPS-challenged ob/ob or control mice. Liver lymphocytes of recipient mice were harvested, and recruitment was analyzed by flow cytometry. (B) Study of the contribution of lymphocytes to homing to the liver. Lymphocytes, isolated from the spleen of LPS-challenged ob/ob or control donor mice, were labeled with fluorescent cell tracers, pooled, and transferred to LPS-challenged lean mice. Liver lymphocytes of recipient mice were harvested, and recruitment was analyzed by flow cytometry.
CD4ⴙ T and B Cells From ob/ob Mice Are Hyperresponsive to CXCL12 and CXCL13 We next explored the mechanisms underlying the preferential homing of ob/ob lymphocytes to the liver. In Transwell experiments (Figure 5A), CXCL12 induced the chemotaxis of both B and T cells, and CXCL13 had only a mild chemotactic effect on T cells but induced B-cell
GASTROENTEROLOGY Vol. 134, No. 5
chemotaxis. CCL19 and CCL21 preferentially attracted CD4⫹ T and CD8⫹ T cells, but no difference was observed for the migration of ob/ob and control lymphocytes. In contrast, ob/ob CD4⫹ T cells responded more strongly to CXCL12 than controls, and ob/ob B cells responded more strongly to CXCL13 than controls. Thus, ob/ob lymphocytes were more sensitive to both CXCL12 and CXCL13 than control cells, with statistically significant differences observed for the CXCL12/CD4⫹ T-cell and CXCL13/B-cell combinations. We analyzed by immunochemistry and immunofluorescence the source of chemokines in the liver. CXCL12 and CXCL13 were colocalized (Figure 5B) both in control and fatty livers. The chemokine-producing cells were CD31 negative (Figure 5B) and had the same morphology than CK19-positive cells, demonstrating their bile duct epithelial origin (Figure 5C).
Obesity Is Directly Responsible for the Stronger Response of ob/ob Lymphocytes to CXCL12 and CXCL13 We generated “non-obese ob/ob” mice to distinguish between the contributions of obesity itself and leptin deficiency in the lymphocyte response to CXCL12 and CXCL13. These mice recovered mobility, with no sign of suffering or lethality, and had a significantly lower body weight than ob/ob mice (29.22 ⫾ 0.41 vs 56.07 ⫾ 0.94 g, respectively; P ⬍ .01), similar to that of lean control mice (Figure 6A). Serum glucose, triglyceride, total cholesterol, high-density lipoprotein-cholesterol, and insulin ⫻ glucose product were lower in these mice, as compared with ob/ob mice (see Supplementary Table online at www.gastrojournal. org). Liver weight, serum aminotransferase, and the number of inflammatory foci returned to control levels (Figure 6B–E). The quantification of total liver lymphocytes
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 3. The contribution of steatosis to lymphocyte homing. Recruitment of (A) CD4⫹ T cells, (B) CD8⫹ T cells, (C) NKT cells, (D) B cells, and (E) NK cells. Results are expressed as the number of fluorescent lymphocytes recruited per gram of liver. Graphs show means ⫾ SEM of 3 experiments. Statistically significant differences (P ⬍ .05) between ob/ob and control livers are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]).
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1465
also showed significantly lower levels of liver inflammation in “non-obese ob/ob” mice (Figure 6F). Overall, these results demonstrate that liver inflammation in ob/ob mice is related to obesity itself rather than to leptin deficiency. Because “non-obese ob/ob” mice are leptin deficient, comparisons of this model with “conventional ob/ob” mice can be used for studies of the role of obesity itself in the response of lymphocytes to CXCL12 and CXCL13. CD4⫹ T and B cells from “non-obese ob/ob” mice migrated significantly less in response to CXCL12 and CXCL13 than did the equivalent cells from ob/ob mice (Figure 7A and 7B). This migration level was in a similar range to that for lean controls. Our results demonstrate that the restoration of normal weight in ob/ob mice
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 4. The contribution of lymphocytes to homing to the liver. Lymphocytes were isolated from ob/ob and control mice, challenged (left column) or not (right column) with LPS, and used for adoptive transfer experiments. Migration into the liver of (A and D) CD4⫹ T cells, (B and E) CD8⫹ T cells, and (C and F) B cells. Results are expressed as the number of fluorescent lymphocytes recruited to the liver per 105 transferred lymphocytes. Graphs show means ⫾ SEM of 6 experiments. Statistically significant differences between ob/ob and control cells are indicated (paired 2-group Wilcoxon signed-rank test, *P ⬍ .05, **P ⬍ .01, ***P ⬍ .001).
abolishes lymphocyte hyperresponsiveness to chemokines.
Obesity Is Directly Responsible for the Higher Propensity of ob/ob Lymphocytes to Migrate to the Liver We then investigated whether a decrease in weight in ob/ob mice prevented lymphocyte migration to the liver. Ob/ob and “non-obese ob/ob” lymphocytes were used for adoptive transfer. The level of CD4⫹ T-, CD8⫹ T-, and B-cell migration to the liver was lower for “non-obese ob/ob” lymphocytes than for ob/ob lymphocytes (Figure 7C–E). In conclusion, weight normalization abolishes both obesity-induced in vitro hyper-
1466
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Figure 5. In vitro response to chemokines and liver expression of CXCL12 and CXCL13. (A) Lymphocytes of LPS-challenged ob/ob or control mice were isolated from the spleen. The number of ob/ob CD4⫹ T, CD8⫹ T, and B cells migrating in response to CXCL12, CXCL13, CCL19, and CCL21 was compared with that of control lymphocytes. Results are expressed as the percentage of cells that migrated in response to chemokines. Graphs show means ⫾ SEM of 5 experiments. Statistically significant differences (P ⬍ .05) between ob/ob and control cells are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]). (B) Double immunofluorescent detection of CXCL12/CXCL13, CD31 (endothelial cells)/ CXCL12, and CD31/CXCL13 in control and ob/ob liver (original magnification, ⫻20). (C) Immunochemistry of CK19 (bile duct epithelial cells) and CXCL12 in control and ob/ob liver (original magnification, ⫻100).
sensitivity to CXCL12 and CXCL13 and in vivo migration to the liver.
Discussion
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
This report demonstrates a new pathophysiologic mechanism of inflammation in fatty liver. We show that liver inflammation in ob/ob mice results not only from the higher potency of the fatty liver to attract circulating lymphocytes but also from the intrinsic tendency of ob/ob lymphocytes to migrate specifically to the liver. We demonstrate that LPS increases lymphocyte migration to the liver by directly acting on lymphocytes. T and B cells from ob/ob mice are hyperresponsive to CXCL12 and CXCL13 and are prone to migrate to the liver. These properties are directly due to obesity and not to leptin deficiency and are reversible with weight normalization. Overall, our study provides the first evidence that obesity leads to a lymphocyte dysfunction characterized by abnormal sensitization to chemotactic agents. The liver, which is closely linked to the gastrointestinal tract, must act as a bivalent immune organ, developing an inflammatory response against pathogens and tolerance to nonpathogen digestive bacterial compounds.17–19 We show that liver inflammation decreased after antibiotic treatment in ob/ob mice. Gut-derived endotoxins may trigger inflammation in the fatty liver, both in murine models of NASH and in humans.20,21 The mechanism underlying the LPS sensitization of ob/ob mice has been connected to impaired macrophage function8,22 and the depletion of hepatic CD4⫹ NKT cells.23 In our experiments, ob/ob lymphocytes showed higher levels of migration to the liver than control lymphocytes,
even in the absence of LPS challenge. LPS challenge merely magnifies this spontaneous property. This suggests that ob/ob lymphocytes are activated in basal conditions, which may be more relevant to human NASH than the deliberate priming of lymphocytes with LPS injection. Eliminating bacterial compounds from the intestine decreased the numbers of CD4⫹ T, CD8⫹ T, NKT, B, and NK cells infiltrating the liver. NKT cells levels are low in the livers of ob/ob24 and diet-induced25 obese mice. The adoptive transfer of regulatory NKT lymphocytes improves NASH in ob/ob mice,26 suggesting that NKT cells protect against obesity-related liver damage. However, we observed smaller numbers of NKT cells in the liver after antibiotic treatment, together with an improvement in liver function. These findings may be accounted for by observations that bacterial glycolipid antigens induce the expansion and proliferation of NKT cells in culture.27 We identified 2 concomitant mechanisms involved in lymphocyte homing to fatty liver. We first showed that steatosis increases lymphocyte recruitment by the liver. We subsequently demonstrated that fatty liver inflammation is also due to abnormal lymphocyte behavior. The CD4⫹ T, CD8⫹ T, and B cells of ob/ob mice displayed a higher propensity to migrate to the liver. Splenocyte subpopulations were identical in ob/ob and control mice, showing that the differential migration was related to specific properties of ob/ob lymphocytes, rather than to the enrichment of ob/ob spleen in certain lymphocyte subsets. The fatty liver recruited significantly more B cells than control liver. Intrahepatic B cells play a major role in liver
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1467
fibrosis, by producing cytokines and/or by cell contact.28 However the mechanism of B-lymphocyte accumulation in the liver remains unknown. Our results show that obesity itself directly triggers B-cell recruitment within the liver—a critical step in the onset of fatty liver inflammation. Interference with CXC chemokine function protects against septic liver damage and may constitute a possible
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 6. Decrease in liver weight, aminotransferase level, and inflammation in “non-obese ob/ob” mice. (A) Body weight. (B) Liver weight, expressed as a percentage of body weight. (C) Serum ALT and (D) serum AST levels. (E) Liver inflammation (Kleiner’s score). (F) Quantification of liver lymphocytes by flow cytometry, expressed as the total number of liver lymphocytes. Graphs show means ⫾ SEM of 8 mice per group. Statistically significant differences are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]).
treatment for controlling pathologic inflammation in endotoxemic animals.12 CXCL12 acts as a chemoattractant for both B and T cells, via the recognition of its receptors, CXCR4 and CXCR7.29 CXCL13 selectively attracts B lymphocytes via CXCR5 recognition.14 We showed that CXCL12 and CXCL13 induced higher levels of migration of ob/ob CD4⫹ T and B cells, respectively, than of control cells. Lymphocytes from ob/ob mice dis-
1468
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Figure 7. Normalization of B- and T-cell responses to CXCL12 and CXCL13 and decreased lymphocyte migration to the liver in “non-obese ob/ob” mice. (A) CXCL12-induced chemotaxis of CD4⫹ T cells and (B) CXCL13-induced chemotaxis of B cells. Results are expressed as the percentage of cells that migrated in response to chemokines. Graphs show means ⫾ SEM of 3 experiments. Statistically significant (P ⬍ .05) and nonsignificant (ns) results are indicated (Kruskall–Wallis and Fisher PLSD tests). (C–E) Adoptive transfer of lymphocytes from ob/ob and “non-obese ob/ob” donors. Results are expressed as the number of fluorescent lymphocytes recruited to the liver per 105 transferred lymphocytes. Migration into the liver of (C) CD4⫹ T cells, (D) CD8⫹ T cells, and (E) B cells. Graphs show means ⫾ SEM of 9 experiments. Statistically significant differences are indicated (paired 2-group Wilcoxon signed-rank test, *P ⬍ .05, **P ⬍ .01).
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
play hypersensitivity to certain chemokines (CXCL12 and CXCL13) but not to others (CCL19 and CCL21). The precise mechanism underlying lymphocyte sensitivity to chemokines remains unclear, but our results point to several nonmutually exclusive hypotheses. The signaling pathway triggered in response to the activation of specific chemokine receptors, such as CXCR4 and CXCR5, may be abnormal in ob/ob lymphocytes. If this were the case, higher levels of recruitment of ob/ob lymphocytes to organs other than the liver would have been expected, including the thymus, spleen, and lymph nodes, all of which express CXCL12 and CXCL13.14,30,31 This was not the case. Flow cytometry showed no difference in the expression of CXCR4 and CXCR5 on the surface of T and B cells from ob/ob and control mice (data not shown). These findings do not exclude the possibility that these receptors are differentially activated without modulation of their membrane expression.29 Moreover, the increase in chemokine sensitivity may be restricted to a lymphocyte subpopulation prone to migrate to the liver, for example, because of the expression of particular adhesion molecules. Alternatively, ob/ob lymphocytes may be hyperresponsive to additional chemokines not tested in our study and specifically expressed in the liver. We also cannot rule out the possible involvement of additional mechanisms. Fatty liver has an abnormal he-
patic microvasculature, and a larger number of leukocytes adhere to the sinusoidal lining than in control liver.32 Hepatic microvasculature dysfunction and hypersensitivity to chemotactic agents are not mutually exclusive, and both may play a role in lymphocyte recruitment to the liver. In addition to metabolic abnormalities and liver dysfunction, leptin deficiency can modulate the immune system.33,34 Lymphocyte infiltration into the liver was decreased by the abolition of obesity in leptin-deficient mice, and a higher migration of lymphocytes was observed in another model of obesity—HFD-induced obesity. Obesity itself was therefore responsible for lymphocyte infiltration in fatty liver in ob/ob mice. We also showed that lymphocyte sensitivity to chemokines was significantly lower in “non-obese ob/ob” mice. Our results therefore demonstrate that obesity is directly responsible for the higher level of chemotaxis of CD4⫹ T and B cells in response to CXCL12 and CXCL13. In conclusion, our results demonstrate that liver inflammation in obesity results from both steatosis and lymphocyte hyperresponsiveness to chemotactic agents. A decrease in excess weight abolishes lymphocyte disorders in obesity. Obesity and endotoxin therefore act together to increase lymphocyte hyperresponsiveness to chemokines and to induce liver inflammation.
Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at 10.1053/ j.gastro.2008.02.055. References 1. Perlemuter G, Bigorgne A, Cassard-Doulcier AM, et al. Nonalcoholic fatty liver disease: from pathogenesis to patient care. Nat Clin Pract Endocrinol Metab 2007;3:458 – 469. 2. Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002; 346:1221–1231. 3. Day CP, James OF. Steatohepatitis: a tale of two “hits?” Gastroenterology 1998;114:842– 845. 4. Seki S, Kitada T, Sakaguchi H. Clinicopathological significance of oxidative cellular damage in non-alcoholic fatty liver diseases. Hepatol Res 2005;33:132–134. 5. Tilg H, Diehl AM. Cytokines in alcoholic and nonalcoholic steatohepatitis. N Engl J Med 2000;343:1467–1476. 6. Brunt EM. Nonalcoholic steatohepatitis: definition and pathology. Semin Liver Dis 2001;21:3–16. 7. Lichtman SN, Sartor RB, Keku J, et al. Hepatic inflammation in rats with experimental small intestinal bacterial overgrowth. Gastroenterology 1990;98:414 – 423. 8. Yang SQ, Lin HZ, Lane MD, et al. Obesity increases sensitivity to endotoxin liver injury: implications for the pathogenesis of steatohepatitis. Proc Natl Acad Sci U S A 1997;94:2557–2562. 9. Koteish A, Mae Diehl A. Animal models of steatohepatitis. Best Pract Res Clin Gastroenterol 2002;16:679 – 690. 10. Faggioni R, Fantuzzi G, Gabay C, et al. Leptin deficiency enhances sensitivity to endotoxin-induced lethality. Am J Physiol 1999;276: R136 –R142. 11. Romics L Jr, Mandrekar P, Kodys K, et al. Increased lipopolysaccharide sensitivity in alcoholic fatty livers is independent of leptin deficiency and toll-like receptor 4 (TLR4) or TLR2 mRNA expression. Alcohol Clin Exp Res 2005;29:1018 –1026. 12. Li X, Klintman D, Liu Q, et al. Critical role of CXC chemokines in endotoxemic liver injury in mice. J Leukoc Biol 2004;75:443– 452. 13. Coulomb-L’Hermin A, Amara A, Schiff C, et al. Stromal cellderived factor 1 (SDF-1) and antenatal human B cell lymphopoiesis: expression of SDF-1 by mesothelial cells and biliary ductal plate epithelial cells. Proc Natl Acad Sci U S A 1999;96:8585– 8590. 14. Legler DF, Loetscher M, Roos RS, et al. B cell-attracting chemokine 1, a human CXC chemokine expressed in lymphoid tissues, selectively attracts B lymphocytes via BLR1/CXCR5. J Exp Med 1998;187:655– 660. 15. Meijer J, Zeelenberg IS, Sipos B, et al. The CXCR5 chemokine receptor is expressed by carcinoma cells and promotes growth of colon carcinoma in the liver. Cancer Res 2006;66:9576 –9582. 16. Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005;41:1313–1321. 17. Bowen DG, McCaughan GW, Bertolino P. Intrahepatic immunity: a tale of two sites? Trends Immunol 2005;26:512–517. 18. Crispe IN, Giannandrea M, Klein I, et al. Cellular and molecular mechanisms of liver tolerance. Immunol Rev 2006;213:101– 118. 19. Li Z, Diehl AM. Innate immunity in the liver. Curr Opin Gastroenterol 2003;19:565–571.
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1469
20. Brun P, Castagliuolo I, Leo VD, et al. Increased intestinal permeability in obese mice: new evidence in the pathogenesis of nonalcoholic steatohepatitis. Am J Physiol Gastrointest Liver Physiol 2007;292:G518 –G525. 21. Wigg AJ, Roberts-Thomson IC, Dymock RB, et al. The role of small intestinal bacterial overgrowth, intestinal permeability, endotoxaemia, and tumour necrosis factor ␣ in the pathogenesis of non-alcoholic steatohepatitis. Gut 2001;48:206 –211. 22. Loffreda S, Yang SQ, Lin HZ, et al. Leptin regulates proinflammatory immune responses. FASEB J 1998;12:57– 65. 23. Li Z, Oben JA, Yang S, et al. Norepinephrine regulates hepatic innate immune system in leptin-deficient mice with nonalcoholic steatohepatitis. Hepatology 2004;40:434 – 441. 24. Guebre-Xabier M, Yang S, Lin HZ, et al. Altered hepatic lymphocyte subpopulations in obesity-related murine fatty livers: potential mechanism for sensitization to liver damage. Hepatology 2000;31:633– 640. 25. Li Z, Soloski MJ, Diehl AM. Dietary factors alter hepatic innate immune system in mice with nonalcoholic fatty liver disease. Hepatology 2005;42:880 – 885. 26. Elinav E, Pappo O, Sklair-Levy M, et al. Adoptive transfer of regulatory NKT lymphocytes ameliorates non-alcoholic steatohepatitis and glucose intolerance in ob/ob mice and is associated with intrahepatic CD8 trapping. J Pathol 2006;209:121– 128. 27. Mattner J, Debord KL, Ismail N, et al. Exogenous and endogenous glycolipid antigens activate NKT cells during microbial infections. Nature 2005;434:525–529. 28. Novobrantseva TI, Majeau GR, Amatucci A, et al. Attenuated liver fibrosis in the absence of B cells. J Clin Invest 2005;115:3072– 3082. 29. Balabanian K, Lagane B, Infantino S, et al. The chemokine SDF1/CXCL12 binds to and signals through the orphan receptor RDC1 in T lymphocytes. J Biol Chem 2005;280:35760 –35766. 30. Ebert LM, Schaerli P, Moser B. Chemokine-mediated control of T-cell traffic in lymphoid and peripheral tissues. Mol Immunol 2005;42:799 – 809. 31. Kim CH, Broxmeyer HE. Chemokines: signal lamps for trafficking of T and B cells for development and effector function. J Leukoc Biol 1999;65:6 –15. 32. McCuskey RS, Ito Y, Robertson GR, et al. Hepatic microvascular dysfunction during evolution of dietary steatohepatitis in mice. Hepatology 2004;40:386 –393. 33. De Rosa V, Procaccini C, Cali G, et al. A key role of leptin in the control of regulatory T-cell proliferation. Immunity 2007;26:241– 255. 34. La Cava A, Matarese G. The weight of leptin in immunity. Nat Rev Immunol 2004;4:371–379.
Received March 23, 2007. Accepted January 31, 2008. Address requests for reprints to: Gabriel Perlemuter, MD, Prof, INSERM, U764, Clamart, F-92140, France. e-mail: gabriel.perlemuter@ abc.aphp.fr; fax: (33) 1 40 94 06 56. Supported by INSERM, Conseil régional d’Ile-de-France, and Association pour la recherche contre le cancer (ARC; to A.E.B.); by INSERM, Fondation pour la Recherche Médicale, and AP-HP (to G.P.); and by INSERM Avenir, Univ Paris-Sud, and The European Union FP6 (INNOCHEM, grant number LSHB-CT-2005-518167). The authors thank Chantal Desdouets for technical support for liver perfusion, Didier Robrieux for rearing the mice, and Cédric Dray for HFD mice, insulin and glucose data. Conflicts of interest: No conflicts of interest exist.
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
May 2008
1469.e1
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Supplementary Document Supplementary Methods Immunodetection protocols. Double immunofluorescent detection was performed on frozen sections of the liver. Endogenous biotin activity was quenched (Dakocytomation) and sections were sequentially incubated with a rat anti-CD31 mAb (Pharmingen) and either with the biotinylated anti-CXCL12 antibody or with a goat anti-CXCL13 antibody (R&D Systems). Anti-CD31 antibody was revealed with a donkey Alexa Fluor 594conjugated anti-rat antibody, the biotinylated antiCXCL12 antibody with Alexa Fluor 488-conjugated streptavidin and the CXCL13 antibody with a rabbit biotinylated anti-goat antibody followed by Alexa Fluor 488-conjugated streptavidin. For the colocalization of CXCL12 and CXCL13, CXCL13 was revealed as described
previously and CXCL12 with the Alexa Fluor 594-conjugated streptavidin. Immunohistochemistry was performed on paraffin-embedded sections of the liver. Antigen retrieval was performed by microwaving in citrate buffer pH 6 (DakoCytomation, Glostrup, Denmark). Endogenous peroxidase and biotin activities were quenched with a peroxidase blocking solution followed by the Biotin Blocking System (DakoCytomation). Sections were incubated with a rabbit anti-CK19 antibody (Abcam; Cambridge, UK) or a biotinylated anti-CXCL12 mAb (gift from Dr F. Arenzana-Seisdedos, France). CK19 binding was detected with a biotinylated goat anti-rabbit (Biogenex, San Ramon, CA), followed, as for CXCL12, by peroxidase-conjugated streptavidin (Biogenex). AEC (3amino-9-ethylcarbozole; DakoCytomation) was used as the chromogen and slides were counterstained with hematoxylin.
Supplementary Table 1. Metabolic Status of Control, ob/ob, “Non-obese ob/ob” and High-Fat Diet Mice
ALT (IU/L) AST (IU/L) Glucose (mmol/L) Triglyceride (mmol/L) Cholesterol (mmol/L) HDL-cholesterol (mmol/L) Glucose x insulin
Control (n ⫽ 12)
ob/ob (n ⫽ 12)
Non-obese ob/ob (n ⫽ 12)
High-fat diet (n ⫽ 8)
40 ⫾ 4 131 ⫾ 28 3.2 ⫾ 0.3 0.90 ⫾ 0.08 3.5 ⫾ 0.1 2.6 ⫾ 0.1 408 ⫾ 92
381 ⫾ 37 383 ⫾ 19 5.5 ⫾ 0.8 1.12 ⫾ 0.05 5.2 ⫾ 0.2 3.8 ⫾ 0.1 2251 ⫾ 252
78 ⫾ 10a 183 ⫾ 22a 2.1 ⫾ 0.5a 1.02 ⫾ 0.04 3.9 ⫾ 0.2a 3.0 ⫾ 0.1a 1092 ⫾ 633a
53 ⫾ 12 145 ⫾ 25 7.8 ⫾ 0.5b 1.3 ⫾ 0.1b 4.5 ⫾ 0.2b 3.2 ⫾ 0.1b 4913 ⫾ 1550b
This Supplementary Table shows means ⫾ SEM. aStatistically significant (P ⬍ .05) differences between “non-obese ob/ob” mice and ob/ob mice. bStatistically significant (P ⬍ .05) differences between high-fat diet mice and control mice (nonparametric variance analysis (Mann–Whitney).
Supplementary Figure 1. Liver histology of mice fed a high-fat diet. Liver histology of: (A) Control and (B) high-fat diet mice (original magnification 40⫻). Black arrow points to inflammatory cell foci. Representative figures of 8 mice per group.
May 2008
Supplementary Figure 2. High-fat diet-induced obesity increases lymphocyte migration to the liver. Lymphocytes were isolated from high-fat diet (HFD) or control LPS-challenged donor mice, labeled, pooled, and transferred to LPS-challenged lean recipient mice. After migration, liver lymphocytes were harvested. Results are expressed as the number of fluorescent lymphocytes recruited into the liver per 105 transferred lymphocytes. Migration into the liver of: (A) CD4⫹T, (B) CD8⫹T, and (C) B cells. Graphs show means ⫾ SEM of 5 experiments. Statistically significant differences between control and HFD cells are indicated (paired 2-group Wilcoxon signed rank test, *P ⬍ .05).
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1469.e2
BASIC–LIVER, PANCREAS, AND BILIARY TRACT Obesity-Induced Lymphocyte Hyperresponsiveness to Chemokines: A New Mechanism of Fatty Liver Inflammation in Obese Mice
AMÉLIE E. BIGORGNE,*,‡ LAURENCE BOUCHET–DELBOS,*,‡ SYLVIE NAVEAU,*,‡,§ IBRAHIM DAGHER,储 SOPHIE PRÉVOT,¶ INGRID DURAND–GASSELIN,* JACQUES COUDERC,* PHILIPPE VALET,#,** DOMINIQUE EMILIE,*,‡,‡‡ and GABRIEL PERLEMUTER*,‡,§ *INSERM, U764, Clamart; ‡Univ Paris-Sud, Faculté de Médecine Paris-Sud, Institut Fédératif de Recherche 13, Clamart; §AP-HP, Hôpital Antoine Béclère, Service d’Hépato-Gastroentérologie, Clamart; 储AP-HP, Hôpital Antoine Béclère, Service de Chirurgie, Clamart; ¶AP-HP, Hôpital Antoine Béclère, Service d’AnatomoPathologie, Clamart; #INSERM, U858, Toulouse; **Université de Toulouse, UPS, Institut de Médecine Moléculaire de Rangueil, Equipe no. 3, Toulouse Cedex 4; and ‡‡AP-HP, Hôpital Antoine Béclère, Service de Microbiologie-Immunologie Biologique, Clamart, France
N
onalcoholic fatty liver disease (NAFLD) is an increasingly recognized condition that may lead to end-stage liver disease. NAFLD corresponds to a wide spectrum of pathologic lesions, ranging from pure steatosis to steatohepatitis, fibrosis, cirrhosis, and hepatocellular carcinoma. Nonalcoholic steatohepatitis (NASH) is merely a stage in NAFLD: steatosis—the retention of lipids in hepatocytes—predisposes the liver to the development of more severe lesions.1 However, steatosis per se may not be a severe disease because many patients do not progress to necroinflammation or fibrosis.2 Steatohepatitis may be seen as a “double-hit” lesion,3 with steatosis (the first hit) rendering the liver vulnerable to further injury. A second hit, such as lipid peroxidation or cytokine secretion,4,5 then triggers the recruitment of inflammatory cells by the liver.6 Several lines of evidence suggest a role for endotoxin in fatty liver inflammation. Lipopolysaccharide (LPS) is a membrane component of gram-negative commensal bacteria present in the digestive tract. A role for endotoxin in the pathogenesis of NASH is suggested by the high incidence of NASH and cirrhosis following jejuno-ileal bypass surgery for obesity. This situation can induce bacterial overgrowth in the defunctionalized small intestine, resulting in LPS absorption.7 The hepatotoxicity of LPS results from the release of cytokines, such as interferon ␥, which increases the sensitivity of hepatocytes to tumor necrosis factor ␣.8 Abbreviations used in this paper: HFD, high-fat diet; LPS, lipopolysaccharide; mAbs, monoclonal antibodies; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NK, natural killer. © 2008 by the AGA Institute 0016-5085/08/$34.00 doi:10.1053/j.gastro.2008.02.055
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Background & Aims: Hepatic lipid retention (steatosis) predisposes hepatitis. We investigated the mechanisms of lymphocyte homing to fatty liver and the role of lipopolysaccharide (LPS) in the onset of inflammation in ob/ob mice. Methods: We decreased intestinal bacterial compounds by oral antibiotic treatment to test the role of endogenous LPS in liver inflammation. Adoptive transfer of lymphocytes was used to study the respective contributions of steatosis and lymphocytes to liver inflammation. We tested lymphocyte response to chemokines by in vitro chemotaxis assays in ob/ob, their lean controls, and “nonobese ob/ob” mice, generated by controlling caloric intake to distinguish between the effects of obesity and leptin deficiency. Results: Antibiotic treatment decreased liver infiltration with CD4ⴙ T, CD8ⴙ T, natural killer (NK)T, B, and NK cells. Adoptive transfer of lymphocytes from ob/ob or control mice showed that (1) steatosis increased lymphocyte recruitment to the liver; (2) CD4ⴙ T, CD8ⴙ T, and B cells from ob/ob mice had a greater propensity to migrate specifically to the liver. This migration was enhanced by LPS. These results were also observed in a model of high-fat diet-induced obesity. CD4ⴙ T and B cells were hyperresponsive to CXCL12 and CXCL13, respectively. Weight normalization in “non-obese ob/ob” mice decreased liver inflammation, lymphocyte response to chemokines, and homing to the liver. Conclusions: Our study provides the first evidence that liver inflammation in mice with genetic or diet-induced obesity results from both steatosis and lymphocyte hyperresponsiveness to chemokines expressed in the liver. These abnormalities are reversible with weight normalization.
1460
BIGORGNE ET AL
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
C57BL/6 ob/ob mice have a mutation that prevents leptin synthesis, resulting in a phenotype of obesity, hepatomegaly, severe steatosis, and high serum levels of aminotransferase.9 Ob/ob mice are much more sensitive to LPS than their lean littermates, rapidly developing liver inflammation after exposure to LPS.8,10,11 The mechanism underlying the role of LPS in the onset of lymphocyte homing to fatty liver remains unknown. CXC chemokine expression has been shown to be induced in endotoxemic animals.12 CXCL12 (stromal cellderived factor 1␣ [SDF-1␣]) is expressed in a wide variety of tissues, including bile duct epithelial cells.13 CXCL13 (B cell chemoattractant chemokine, BLC/BCA-1) selectively attracts B cells and is expressed in the liver.14,15 Several studies have suggested that the homing of circulating lymphocytes to the liver may increase during inflammation, but no study has yet formally confirmed this hypothesis. We explored the pathogenesis of NASH further by investigating the role of digestive LPS in the onset of liver inflammation and the mechanisms of lymphocyte homing to fatty liver in ob/ob mice. We first demonstrate that digestive bacterial compounds are involved in liver inflammation in ob/ob mice. We then show that CD4⫹ T, CD8⫹ T, and B cells from ob/ob mice have an intrinsic propensity to migrate specifically to the liver, which is enhanced by LPS challenge. Moreover, we obtain similar results in a model of high-fat diet (HFD)-induced obesity. Lymphocytes from ob/ob mice are hyperresponsive to at least 2 CXC chemokines expressed in the liver. We demonstrate that, in the ob/ob model, obesity itself—rather than leptin deficiency—is directly involved in liver inflammation and lymphocyte sensitivity to chemokines. Our study provides the first evidence that liver inflammation in obesity is due to both steatosis and lymphocyte hyperresponsiveness to chemotactic agents, which are reversible with weight normalization.
Materials and Methods Mice Five-week-old male C57BL/6 ob/ob mice and controls (lean littermates) were purchased from Janvier (Le Genest St. Isle, France). Mice were treated in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996). They were fed a diet consisting of 22% protein, 4.3% lipid, and 70% carbohydrate supplied ad libitum. “Non-obese ob/ob” mice were supplied daily with the same caloric intake as lean controls, from the age of 5 weeks, and experiments were performed on 12-week-old mice. C57BL/6 mice fed a HFD were purchased at the age of 3 weeks and fed a diet consisting of 15.5% protein, 46% lipid, and 38.5% carbohydrate (Safe, Augy, France) supplied ad libitum, and experiments were performed after 18 weeks on the HFD.
GASTROENTEROLOGY Vol. 134, No. 5
Reduction of the Level of Endogenous Digestive Bacterial Compounds Ciprofloxacin (Bayer Pharma, Germany) and ornidazol (Roche, Switzerland) were given in drinking water at a daily dose of 10 g/g and 15 g/g body weight, respectively, for 10 days, whereas the control group received water only.
Metabolic Status of Mice Blood was collected by retroorbital vein puncture. The serum was stored at ⫺20°C until use for alanine and aspartate aminotransferase (ALT, AST, respectively), triglyceride, glucose, total cholesterol, and high-density lipoprotein-cholesterol determinations (OLYMPUS AU400). Serum insulin was determined by enzyme-linked immunosorbent assay (ELISA) (R&D Systems, Minneapolis, MN). Insulin resistance was determined by the product: fasting serum glucose (mmol/L) ⫻ insulin (mU/L).
Quantification and Phenotyping of Lymphocytes Infiltrating the Liver Liver samples of 6 mice from each group were fixed in 36% formaldehyde and embedded in paraffin. Sections (5 m) were cut and stained with hematoxylin, eosin, and saffron. Slides were read blindly. Lobular inflammation was defined according to Kleiner’s score16: 0, no focus; 1, ⬍2 foci; 2, 2– 4 foci; 3, ⬎4 foci per 200⫻ field. For quantification and phenotyping of lymphocytes infiltrating the liver, lymphocytes were isolated from mouse liver by a 2-step perfusion procedure. In the first step, circulating cells were eliminated by perfusing the liver in situ through the portal vein with Ca2⫹- and Mg2⫹-free HEPES (pH 7.65) supplemented with 5 mmol/L EDTA at 37°C, at a flow rate of 4 mL/min. The liver was then perfused with 0.05% collagenase IV (Sigma–Aldrich, St. Louis, MO) buffered with 0.1 mol/L HEPES at 37°C for 5 minutes, excised, and homogenized. The suspension was filtered through a 70-m filter, and the filtrate was centrifuged at 50g at room temperature for 2 minutes to remove liver parenchymal cells. The lymphocyte fraction was stained with monoclonal antibodies (mAbs) for flow cytometry analysis.
Adoptive Transfer for Studying the Contribution of Steatosis to Lymphocyte Homing Source. Lean donor mice were challenged intraperitoneally (IP) with LPS (0.5 g/g body weight) (Alexis, Switzerland) diluted in sterile saline and killed 6 hours later. A lymphocyte suspension was obtained by tearing the spleen with fine forceps in 1X phosphate-buffered saline (PBS) in sterile conditions. Lymphocytes were purified by centrifuging this suspension through a sterile Lympholyte density gradient (Cedarlane, Canada) at 800g for 20 minutes at room temperature. Labeling. Lymphocytes were labeled with fluorescent cell tracers for living cells (CellTracker Orange
CMTMR; Molecular Probes, Eugene, OR). A suspension of 106 cells/mL was incubated for 30 minutes at 37°C with Orange CMTMR (1:250). Lymphocyte transfer to recipient mice. We transferred 6 ⫻ 106 lymphocytes intravenously (IV) to ob/ob and control recipient mice 6 hours after IP challenge with LPS (0.5 g/g body weight). Lymphocyte harvest. Recipient mice were killed 18 hours after lymphocyte transfer. Circulating cells were eliminated by the first perfusion step, as described previously. A fraction of the liver was then excised and weighed before digestion in a Petri dish with 0.05% collagenase IV buffered with 0.1 mol/L HEPES at 37°C for 10 minutes. The lymphocyte fraction was stained with mAbs. Analysis. Lymphocyte migration was quantified by flow cytometry and expressed as the number of fluorescent lymphocytes recruited per gram of liver.
Adoptive Transfer for Studying the Contribution of Lymphocytes to Homing to the Liver Source. For each experiment, 2 ob/ob and 2 control donor mice were challenged IP with LPS and killed 6 hours later. Lymphocytes were purified as described previously. Labeling. A suspension of 106 cells/mL was incubated for 30 minutes at 37°C with either Calcein AM (1:1000) (Molecular Probes) or CellTracker Orange CMTMR (1:250), according to their source. Lymphocyte transfer to recipient mice. We pooled 6 ⫻ 106 cells labeled with each cell tracer and transferred them IV into lean recipient mice 6 hours after LPS challenge. Lymphocyte harvest. Recipient mice were killed 18 hours after lymphocyte transfer. Lymphocytes were isolated from the spleen, lymph node, and thymus by homogenizing and seeping through a 40-m filter in 1X PBS and from the liver by a 2-step perfusion procedure. The lymphocyte fraction was stained with mAbs. Analysis. Lymphocyte migration was quantified by flow cytometry, according to lymphocyte source and phenotype. Results are expressed as the number of fluorescent cells recruited to the liver per 105 transferred cells.
In Vitro Chemotaxis Assays Lymphocyte chemotaxis was evaluated with the Transwell system (Corning Costar, MA). Lymphocytes were isolated from the spleen of ob/ob or control mice as described above 6 hours after LPS challenge and stained with fluorescent mAbs. We added 1.5 ⫻ 106 cells in 150 L RPMI supplemented with 20 mmol/L HEPES and 1% calf fetal serum (Perbio Sciences, Belgium) to the upper chamber and 600 L of the same medium with (1 g/ mL) or without murine CXCL12, CXCL13, CCL19, and CCL21 (R&D Systems) to the lower chamber. A sample of 1.5 ⫻ 106 cells was used as input control. Cells were
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1461
incubated at 37°C for 1 hour (for CXCL13 and CCL19) or 2 hours (for CXCL12 and CCL21). Results are expressed as the percentage of cells added to the upper chamber that migrated to the lower chamber for each phenotype.
Monoclonal Antibodies for Flow Cytometry Lymphocytes were stained with allophycocyaninconjugated rat anti-CD4, anti-CD8, and anti-CD19 mAbs; mouse anti-NK1.1; and PerCP-Cy5.5-conjugated hamster anti-CD3 mAbs (Pharmingen, CA). Lymphocyte subpopulations were analyzed by 4-color flow cytometry, using a FACScalibur machine (Becton Dickinson Immunocytometry Systems, NJ).
Immunodetection Immunofluorescence was performed on frozen liver sections for CXCL12, CXCL13, and CD31 (endothelial cells). Immunohistochemistry was performed on paraffin-embedded sections for CK19 (bile duct epithelial cells) and CXCL12 (see Supplementary Material online at www.gastrojournal.org for technical details).
Statistical Analysis Data are expressed as means ⫾ SEM. Quantitative data were compared using nonparametric (Mann–Whitney) tests and Kruskall–Wallis variance analysis. Multiple comparisons were performed, using Fisher PLSD test. A paired, 2-group Wilcoxon signed-rank test was used to compare the contributions of lymphocytes from obese and control mice to homing to the liver. P values less than .05 were considered significant.
Results Reducing Digestive Bacterial Compounds by Oral Antibiotic Treatment Decreases Liver Inflammation in ob/ob Mice We investigated the possible role of digestive bacterial compounds in the occurrence of inflammation in fatty liver by administering oral antibiotics against aerobic and anaerobic bacteria to ob/ob and control mice. Antibiotic treatment did not modify body weight (Table 1). Basal liver weight, expressed as a percentage of total body weight, was higher in ob/ob than in control mice. Liver weight significantly decreased in ob/ob mice after a 10-day course of antibiotic treatment (from 6.7% ⫾ 0.8% to 5.8% ⫾ 0.8%, respectively, P ⬍ .05) (Table 1). We assessed the impact of this treatment on serum aminotransferase levels. Antibiotic treatment strongly decreased ALT and AST levels in ob/ob mice but had no significant effect in control mice (Table 1). We then assessed the impact of digestive decontamination on liver inflammation. Fewer foci were observed in antibiotic-treated ob/ob mice (Kleiner’s score, 1.1 ⫾ 0.7) than in untreated ob/ob mice (Kleiner’s score, 1.9 ⫾ 0.5) (P ⬍ .001) (Table 1, Figure 1A and B). The number of CD4⫹ T (CD3⫹CD4⫹), CD8⫹ T (CD3⫹CD8⫹), natural killer
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
May 2008
1462
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Table 1. Decreased Liver Inflammation in ob/ob Mice After Antibiotic Treatment Control mice
Body weight (g) Liver weight (%) ALT (IU/L) AST (IU/L) Inflammation score (Kleiner’s score)
Ob/ob mice
Untreated
Antibiotics
Untreated
27.6 ⫾ 0.6 4.8 ⫾ 0.5 58 ⫾ 12 175 ⫾ 34 0.5 ⫾ 0.5
27.8 ⫾ 0.5 5.3 ⫾ 0.6 66 ⫾ 16 148 ⫾ 21 0.1 ⫾ 0.4
48.7 ⫾ 1.3 6.7 ⫾ 0.8 419 ⫾ 42 400 ⫾ 30 1.9 ⫾ 0.5
Antibiotics 47.3 ⫾ 0.7 5.8 ⫾ 0.8a 158 ⫾ 23b 224 ⫾ 29c 1.1 ⫾ 0.7b
NOTE. Table 1 shows means ⫾ SEM of 6 mice per group. Statistically significant differences between antibiotic-treated and untreated mice are indicated (nonparametric variance analysis [Mann–Whitney test], aP ⬍ .05, bP ⬍ .001, cP ⬍ .01).
(NK)T (CD3⫹NK1.1⫹), B (CD3⫺CD19⫹), and NK (CD3⫺NK1.1⫹) cells in the liver of ob/ob mice was significantly decreased by antibiotic treatment (Figure 1C). Overall, liver weight, biochemistry, pathology, and flow cytometry results demonstrated that reducing the level of endogenous digestive bacterial products decreased liver inflammation.
Steatosis Increases Lymphocyte Recruitment by the Liver
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
The low-grade liver inflammation induced by digestive bacterial products was amplified by challenging mice with LPS to study lymphocyte homing to fatty liver. We developed an adoptive transfer system for evaluating the contribution of liver steatosis to lymphocyte recruitment. Lymphocytes were isolated from LPS-challenged lean donor mice, labeled with a fluorescent cell tracer, and transferred to ob/ob or control LPS-challenged recipient mice (Figure 2A). The level of recruitment for CD4⫹ T, CD8⫹ T, NKT, B, and NK fluorescent cells to fatty liver, normalized for liver weight, was higher than the level of recruitment to control liver (Figure 3). Fatty livers therefore have a higher capacity to recruit B and T cells after LPS challenge than control livers.
Lymphocytes From Obese Mice Have a Stronger Propensity to Migrate to the Liver We then investigated whether inflammatory cells isolated from ob/ob mice displayed an intrinsic capacity to migrate to the liver. We first isolated and phenotyped lymphocytes from the spleens of ob/ob and control mice. There was no significant difference in the number of T and B cells between ob/ob and control mice. LPS challenge did not affect the composition of lymphocyte subsets (data not shown). Lymphocytes isolated from the spleens of LPS-challenged ob/ob and control donors were used for adoptive transfer. They were labeled to identify their source (ie, ob/ob or control) and transferred to recipient mice (Figure 2B). We found that the level of CD4⫹ T- and CD8⫹ T-cell migration to the liver was higher for ob/ob than for control lymphocytes: 46.3 ⫾ 9.2 vs 30.7 ⫾ 7.9 (P ⬍ .05) and 44.0 ⫾ 10.5 vs 25.7 ⫾ 7.0 cells recruited to the liver
for 105 transferred cells (P ⬍ .01), respectively. The level of B-cell migration into the liver was also higher for ob/ob than for control B cells: 17.1 ⫾ 2.6 vs 10.1 ⫾ 2.0 cells, respectively, recruited to the liver for 105 transferred cells (P ⬍ .001) (Figure 4, left column). No significant difference was observed for NK cells (data not shown). Lymphocyte migration into the spleen, thymus, and lymph nodes did not differ significantly between ob/ob and control cells (data not shown). Thus, the higher level of ob/ob CD4⫹ T-, CD8⫹ T-, and B-cell migration was specific to the liver. We validated the results obtained for ob/ob mice in another model of obesity by feeding C57BL/6 mice a HFD. These mice had a higher body weight than C57BL/6 control mice (42.8 ⫾ 1.0 vs 30.8 ⫾ 0.7 g, respectively; P ⬍ .05). Serum glucose, triglyceride, total cholesterol, high-density lipoprotein-cholesterol, and insulin ⫻ glucose product were significantly higher in HFD mice than in controls (see Supplementary Table online at www.gastrojournal.org). HFD mice showed steatosis and mild lobular inflammation (Kleiner’s score: 1.4 ⫾ 0.2 vs 0.75 ⫾ 0.16, respectively; P ⬍ .05; see Supplementary Figure 1 online at www.gastrojournal.org). Using the adoptive transfer procedure described above, we showed that the level of CD4⫹ T-, CD8⫹ T-, and B-cell migration to the liver was significantly higher for HFD lymphocytes than for controls (see Supplementary Figure 2 online at www.gastrojournal.org). Thus, obesity, whether genetic or diet induced, leads to an increase in the homing of circulating lymphocytes to the liver.
The Propensity of ob/ob Lymphocytes to Migrate to the Liver Persists in the Absence of LPS Challenge We investigated whether the activation of ob/ob lymphocytes by LPS challenge contributed to increases in lymphocyte migration to the liver by transferring lymphocytes isolated from unchallenged donors. The level of ob/ob lymphocyte migration was significantly lower than that for LPS-challenged donors for CD4⫹ T (Figure 4A and 4D, solid columns), CD8⫹ T (Figure 4B and 4E, solid columns), and B cells (Figure 4C and 4F, solid columns). Significantly lower levels of control lymphocyte migration were also observed in the absence of LPS challenge to
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1463
donors (Figure 4, open columns). Recruitment to lymphoid organs was also decreased by the absence of LPS challenge of donors (data not shown). Thus, LPS enhances lymphocyte migration to several organs, including the liver. However, despite the low level of lymphocyte migration in the absence of LPS challenge, ob/ob lymphocytes still displayed higher levels of migration to the liver than
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 1. Oral antibiotic treatment decreases liver inflammation in ob/ob mice. Oral ciprofloxacin and ornidazol treatment. Liver pathology of (A) untreated and (B) antibiotic-treated ob/ob mice (original magnification, ⫻40 and ⫻100, respectively). Solid arrows point to inflammatory cell foci, which are enlarged in the lower panels. Representative figures of 6 mice per group. (C) Flow cytometry analysis of lymphocytes infiltrating the liver. Graphs show means ⫾ SEM of 5 mice per group. Statistically significant differences (P ⬍ .05) between antibiotic-treated and untreated ob/ob mice are indicated by an asterisk (nonparametric variance analysis [Mann–Whitney test]).
control cells. This was found for CD4⫹ T, CD8⫹ T, and B cells (Figure 4, right columns). Migration was similar for ob/ob and control lymphocytes in the spleen, thymus, and lymph nodes in the absence of LPS challenge (data not shown). Our results show that ob/ob CD4⫹ T, CD8⫹ T, and B cells display higher levels of migration to the liver than control cells, whether or not they are exposed to LPS.
1464
BIGORGNE ET AL
Figure 2. Schematic representation of the adoptive transfer procedure. (A) Study of the contribution of steatosis to lymphocyte homing. Lymphocytes, isolated from the spleen of LPS-challenged lean donor mice, were labeled with a fluorescent cell tracer and transferred to LPS-challenged ob/ob or control mice. Liver lymphocytes of recipient mice were harvested, and recruitment was analyzed by flow cytometry. (B) Study of the contribution of lymphocytes to homing to the liver. Lymphocytes, isolated from the spleen of LPS-challenged ob/ob or control donor mice, were labeled with fluorescent cell tracers, pooled, and transferred to LPS-challenged lean mice. Liver lymphocytes of recipient mice were harvested, and recruitment was analyzed by flow cytometry.
CD4ⴙ T and B Cells From ob/ob Mice Are Hyperresponsive to CXCL12 and CXCL13 We next explored the mechanisms underlying the preferential homing of ob/ob lymphocytes to the liver. In Transwell experiments (Figure 5A), CXCL12 induced the chemotaxis of both B and T cells, and CXCL13 had only a mild chemotactic effect on T cells but induced B-cell
GASTROENTEROLOGY Vol. 134, No. 5
chemotaxis. CCL19 and CCL21 preferentially attracted CD4⫹ T and CD8⫹ T cells, but no difference was observed for the migration of ob/ob and control lymphocytes. In contrast, ob/ob CD4⫹ T cells responded more strongly to CXCL12 than controls, and ob/ob B cells responded more strongly to CXCL13 than controls. Thus, ob/ob lymphocytes were more sensitive to both CXCL12 and CXCL13 than control cells, with statistically significant differences observed for the CXCL12/CD4⫹ T-cell and CXCL13/B-cell combinations. We analyzed by immunochemistry and immunofluorescence the source of chemokines in the liver. CXCL12 and CXCL13 were colocalized (Figure 5B) both in control and fatty livers. The chemokine-producing cells were CD31 negative (Figure 5B) and had the same morphology than CK19-positive cells, demonstrating their bile duct epithelial origin (Figure 5C).
Obesity Is Directly Responsible for the Stronger Response of ob/ob Lymphocytes to CXCL12 and CXCL13 We generated “non-obese ob/ob” mice to distinguish between the contributions of obesity itself and leptin deficiency in the lymphocyte response to CXCL12 and CXCL13. These mice recovered mobility, with no sign of suffering or lethality, and had a significantly lower body weight than ob/ob mice (29.22 ⫾ 0.41 vs 56.07 ⫾ 0.94 g, respectively; P ⬍ .01), similar to that of lean control mice (Figure 6A). Serum glucose, triglyceride, total cholesterol, high-density lipoprotein-cholesterol, and insulin ⫻ glucose product were lower in these mice, as compared with ob/ob mice (see Supplementary Table online at www.gastrojournal. org). Liver weight, serum aminotransferase, and the number of inflammatory foci returned to control levels (Figure 6B–E). The quantification of total liver lymphocytes
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 3. The contribution of steatosis to lymphocyte homing. Recruitment of (A) CD4⫹ T cells, (B) CD8⫹ T cells, (C) NKT cells, (D) B cells, and (E) NK cells. Results are expressed as the number of fluorescent lymphocytes recruited per gram of liver. Graphs show means ⫾ SEM of 3 experiments. Statistically significant differences (P ⬍ .05) between ob/ob and control livers are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]).
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1465
also showed significantly lower levels of liver inflammation in “non-obese ob/ob” mice (Figure 6F). Overall, these results demonstrate that liver inflammation in ob/ob mice is related to obesity itself rather than to leptin deficiency. Because “non-obese ob/ob” mice are leptin deficient, comparisons of this model with “conventional ob/ob” mice can be used for studies of the role of obesity itself in the response of lymphocytes to CXCL12 and CXCL13. CD4⫹ T and B cells from “non-obese ob/ob” mice migrated significantly less in response to CXCL12 and CXCL13 than did the equivalent cells from ob/ob mice (Figure 7A and 7B). This migration level was in a similar range to that for lean controls. Our results demonstrate that the restoration of normal weight in ob/ob mice
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 4. The contribution of lymphocytes to homing to the liver. Lymphocytes were isolated from ob/ob and control mice, challenged (left column) or not (right column) with LPS, and used for adoptive transfer experiments. Migration into the liver of (A and D) CD4⫹ T cells, (B and E) CD8⫹ T cells, and (C and F) B cells. Results are expressed as the number of fluorescent lymphocytes recruited to the liver per 105 transferred lymphocytes. Graphs show means ⫾ SEM of 6 experiments. Statistically significant differences between ob/ob and control cells are indicated (paired 2-group Wilcoxon signed-rank test, *P ⬍ .05, **P ⬍ .01, ***P ⬍ .001).
abolishes lymphocyte hyperresponsiveness to chemokines.
Obesity Is Directly Responsible for the Higher Propensity of ob/ob Lymphocytes to Migrate to the Liver We then investigated whether a decrease in weight in ob/ob mice prevented lymphocyte migration to the liver. Ob/ob and “non-obese ob/ob” lymphocytes were used for adoptive transfer. The level of CD4⫹ T-, CD8⫹ T-, and B-cell migration to the liver was lower for “non-obese ob/ob” lymphocytes than for ob/ob lymphocytes (Figure 7C–E). In conclusion, weight normalization abolishes both obesity-induced in vitro hyper-
1466
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Figure 5. In vitro response to chemokines and liver expression of CXCL12 and CXCL13. (A) Lymphocytes of LPS-challenged ob/ob or control mice were isolated from the spleen. The number of ob/ob CD4⫹ T, CD8⫹ T, and B cells migrating in response to CXCL12, CXCL13, CCL19, and CCL21 was compared with that of control lymphocytes. Results are expressed as the percentage of cells that migrated in response to chemokines. Graphs show means ⫾ SEM of 5 experiments. Statistically significant differences (P ⬍ .05) between ob/ob and control cells are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]). (B) Double immunofluorescent detection of CXCL12/CXCL13, CD31 (endothelial cells)/ CXCL12, and CD31/CXCL13 in control and ob/ob liver (original magnification, ⫻20). (C) Immunochemistry of CK19 (bile duct epithelial cells) and CXCL12 in control and ob/ob liver (original magnification, ⫻100).
sensitivity to CXCL12 and CXCL13 and in vivo migration to the liver.
Discussion
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
This report demonstrates a new pathophysiologic mechanism of inflammation in fatty liver. We show that liver inflammation in ob/ob mice results not only from the higher potency of the fatty liver to attract circulating lymphocytes but also from the intrinsic tendency of ob/ob lymphocytes to migrate specifically to the liver. We demonstrate that LPS increases lymphocyte migration to the liver by directly acting on lymphocytes. T and B cells from ob/ob mice are hyperresponsive to CXCL12 and CXCL13 and are prone to migrate to the liver. These properties are directly due to obesity and not to leptin deficiency and are reversible with weight normalization. Overall, our study provides the first evidence that obesity leads to a lymphocyte dysfunction characterized by abnormal sensitization to chemotactic agents. The liver, which is closely linked to the gastrointestinal tract, must act as a bivalent immune organ, developing an inflammatory response against pathogens and tolerance to nonpathogen digestive bacterial compounds.17–19 We show that liver inflammation decreased after antibiotic treatment in ob/ob mice. Gut-derived endotoxins may trigger inflammation in the fatty liver, both in murine models of NASH and in humans.20,21 The mechanism underlying the LPS sensitization of ob/ob mice has been connected to impaired macrophage function8,22 and the depletion of hepatic CD4⫹ NKT cells.23 In our experiments, ob/ob lymphocytes showed higher levels of migration to the liver than control lymphocytes,
even in the absence of LPS challenge. LPS challenge merely magnifies this spontaneous property. This suggests that ob/ob lymphocytes are activated in basal conditions, which may be more relevant to human NASH than the deliberate priming of lymphocytes with LPS injection. Eliminating bacterial compounds from the intestine decreased the numbers of CD4⫹ T, CD8⫹ T, NKT, B, and NK cells infiltrating the liver. NKT cells levels are low in the livers of ob/ob24 and diet-induced25 obese mice. The adoptive transfer of regulatory NKT lymphocytes improves NASH in ob/ob mice,26 suggesting that NKT cells protect against obesity-related liver damage. However, we observed smaller numbers of NKT cells in the liver after antibiotic treatment, together with an improvement in liver function. These findings may be accounted for by observations that bacterial glycolipid antigens induce the expansion and proliferation of NKT cells in culture.27 We identified 2 concomitant mechanisms involved in lymphocyte homing to fatty liver. We first showed that steatosis increases lymphocyte recruitment by the liver. We subsequently demonstrated that fatty liver inflammation is also due to abnormal lymphocyte behavior. The CD4⫹ T, CD8⫹ T, and B cells of ob/ob mice displayed a higher propensity to migrate to the liver. Splenocyte subpopulations were identical in ob/ob and control mice, showing that the differential migration was related to specific properties of ob/ob lymphocytes, rather than to the enrichment of ob/ob spleen in certain lymphocyte subsets. The fatty liver recruited significantly more B cells than control liver. Intrahepatic B cells play a major role in liver
May 2008
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1467
fibrosis, by producing cytokines and/or by cell contact.28 However the mechanism of B-lymphocyte accumulation in the liver remains unknown. Our results show that obesity itself directly triggers B-cell recruitment within the liver—a critical step in the onset of fatty liver inflammation. Interference with CXC chemokine function protects against septic liver damage and may constitute a possible
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
Figure 6. Decrease in liver weight, aminotransferase level, and inflammation in “non-obese ob/ob” mice. (A) Body weight. (B) Liver weight, expressed as a percentage of body weight. (C) Serum ALT and (D) serum AST levels. (E) Liver inflammation (Kleiner’s score). (F) Quantification of liver lymphocytes by flow cytometry, expressed as the total number of liver lymphocytes. Graphs show means ⫾ SEM of 8 mice per group. Statistically significant differences are indicated by asterisks (nonparametric variance analysis [Mann–Whitney test]).
treatment for controlling pathologic inflammation in endotoxemic animals.12 CXCL12 acts as a chemoattractant for both B and T cells, via the recognition of its receptors, CXCR4 and CXCR7.29 CXCL13 selectively attracts B lymphocytes via CXCR5 recognition.14 We showed that CXCL12 and CXCL13 induced higher levels of migration of ob/ob CD4⫹ T and B cells, respectively, than of control cells. Lymphocytes from ob/ob mice dis-
1468
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Figure 7. Normalization of B- and T-cell responses to CXCL12 and CXCL13 and decreased lymphocyte migration to the liver in “non-obese ob/ob” mice. (A) CXCL12-induced chemotaxis of CD4⫹ T cells and (B) CXCL13-induced chemotaxis of B cells. Results are expressed as the percentage of cells that migrated in response to chemokines. Graphs show means ⫾ SEM of 3 experiments. Statistically significant (P ⬍ .05) and nonsignificant (ns) results are indicated (Kruskall–Wallis and Fisher PLSD tests). (C–E) Adoptive transfer of lymphocytes from ob/ob and “non-obese ob/ob” donors. Results are expressed as the number of fluorescent lymphocytes recruited to the liver per 105 transferred lymphocytes. Migration into the liver of (C) CD4⫹ T cells, (D) CD8⫹ T cells, and (E) B cells. Graphs show means ⫾ SEM of 9 experiments. Statistically significant differences are indicated (paired 2-group Wilcoxon signed-rank test, *P ⬍ .05, **P ⬍ .01).
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
play hypersensitivity to certain chemokines (CXCL12 and CXCL13) but not to others (CCL19 and CCL21). The precise mechanism underlying lymphocyte sensitivity to chemokines remains unclear, but our results point to several nonmutually exclusive hypotheses. The signaling pathway triggered in response to the activation of specific chemokine receptors, such as CXCR4 and CXCR5, may be abnormal in ob/ob lymphocytes. If this were the case, higher levels of recruitment of ob/ob lymphocytes to organs other than the liver would have been expected, including the thymus, spleen, and lymph nodes, all of which express CXCL12 and CXCL13.14,30,31 This was not the case. Flow cytometry showed no difference in the expression of CXCR4 and CXCR5 on the surface of T and B cells from ob/ob and control mice (data not shown). These findings do not exclude the possibility that these receptors are differentially activated without modulation of their membrane expression.29 Moreover, the increase in chemokine sensitivity may be restricted to a lymphocyte subpopulation prone to migrate to the liver, for example, because of the expression of particular adhesion molecules. Alternatively, ob/ob lymphocytes may be hyperresponsive to additional chemokines not tested in our study and specifically expressed in the liver. We also cannot rule out the possible involvement of additional mechanisms. Fatty liver has an abnormal he-
patic microvasculature, and a larger number of leukocytes adhere to the sinusoidal lining than in control liver.32 Hepatic microvasculature dysfunction and hypersensitivity to chemotactic agents are not mutually exclusive, and both may play a role in lymphocyte recruitment to the liver. In addition to metabolic abnormalities and liver dysfunction, leptin deficiency can modulate the immune system.33,34 Lymphocyte infiltration into the liver was decreased by the abolition of obesity in leptin-deficient mice, and a higher migration of lymphocytes was observed in another model of obesity—HFD-induced obesity. Obesity itself was therefore responsible for lymphocyte infiltration in fatty liver in ob/ob mice. We also showed that lymphocyte sensitivity to chemokines was significantly lower in “non-obese ob/ob” mice. Our results therefore demonstrate that obesity is directly responsible for the higher level of chemotaxis of CD4⫹ T and B cells in response to CXCL12 and CXCL13. In conclusion, our results demonstrate that liver inflammation in obesity results from both steatosis and lymphocyte hyperresponsiveness to chemotactic agents. A decrease in excess weight abolishes lymphocyte disorders in obesity. Obesity and endotoxin therefore act together to increase lymphocyte hyperresponsiveness to chemokines and to induce liver inflammation.
Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at 10.1053/ j.gastro.2008.02.055. References 1. Perlemuter G, Bigorgne A, Cassard-Doulcier AM, et al. Nonalcoholic fatty liver disease: from pathogenesis to patient care. Nat Clin Pract Endocrinol Metab 2007;3:458 – 469. 2. Angulo P. Nonalcoholic fatty liver disease. N Engl J Med 2002; 346:1221–1231. 3. Day CP, James OF. Steatohepatitis: a tale of two “hits?” Gastroenterology 1998;114:842– 845. 4. Seki S, Kitada T, Sakaguchi H. Clinicopathological significance of oxidative cellular damage in non-alcoholic fatty liver diseases. Hepatol Res 2005;33:132–134. 5. Tilg H, Diehl AM. Cytokines in alcoholic and nonalcoholic steatohepatitis. N Engl J Med 2000;343:1467–1476. 6. Brunt EM. Nonalcoholic steatohepatitis: definition and pathology. Semin Liver Dis 2001;21:3–16. 7. Lichtman SN, Sartor RB, Keku J, et al. Hepatic inflammation in rats with experimental small intestinal bacterial overgrowth. Gastroenterology 1990;98:414 – 423. 8. Yang SQ, Lin HZ, Lane MD, et al. Obesity increases sensitivity to endotoxin liver injury: implications for the pathogenesis of steatohepatitis. Proc Natl Acad Sci U S A 1997;94:2557–2562. 9. Koteish A, Mae Diehl A. Animal models of steatohepatitis. Best Pract Res Clin Gastroenterol 2002;16:679 – 690. 10. Faggioni R, Fantuzzi G, Gabay C, et al. Leptin deficiency enhances sensitivity to endotoxin-induced lethality. Am J Physiol 1999;276: R136 –R142. 11. Romics L Jr, Mandrekar P, Kodys K, et al. Increased lipopolysaccharide sensitivity in alcoholic fatty livers is independent of leptin deficiency and toll-like receptor 4 (TLR4) or TLR2 mRNA expression. Alcohol Clin Exp Res 2005;29:1018 –1026. 12. Li X, Klintman D, Liu Q, et al. Critical role of CXC chemokines in endotoxemic liver injury in mice. J Leukoc Biol 2004;75:443– 452. 13. Coulomb-L’Hermin A, Amara A, Schiff C, et al. Stromal cellderived factor 1 (SDF-1) and antenatal human B cell lymphopoiesis: expression of SDF-1 by mesothelial cells and biliary ductal plate epithelial cells. Proc Natl Acad Sci U S A 1999;96:8585– 8590. 14. Legler DF, Loetscher M, Roos RS, et al. B cell-attracting chemokine 1, a human CXC chemokine expressed in lymphoid tissues, selectively attracts B lymphocytes via BLR1/CXCR5. J Exp Med 1998;187:655– 660. 15. Meijer J, Zeelenberg IS, Sipos B, et al. The CXCR5 chemokine receptor is expressed by carcinoma cells and promotes growth of colon carcinoma in the liver. Cancer Res 2006;66:9576 –9582. 16. Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology 2005;41:1313–1321. 17. Bowen DG, McCaughan GW, Bertolino P. Intrahepatic immunity: a tale of two sites? Trends Immunol 2005;26:512–517. 18. Crispe IN, Giannandrea M, Klein I, et al. Cellular and molecular mechanisms of liver tolerance. Immunol Rev 2006;213:101– 118. 19. Li Z, Diehl AM. Innate immunity in the liver. Curr Opin Gastroenterol 2003;19:565–571.
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1469
20. Brun P, Castagliuolo I, Leo VD, et al. Increased intestinal permeability in obese mice: new evidence in the pathogenesis of nonalcoholic steatohepatitis. Am J Physiol Gastrointest Liver Physiol 2007;292:G518 –G525. 21. Wigg AJ, Roberts-Thomson IC, Dymock RB, et al. The role of small intestinal bacterial overgrowth, intestinal permeability, endotoxaemia, and tumour necrosis factor ␣ in the pathogenesis of non-alcoholic steatohepatitis. Gut 2001;48:206 –211. 22. Loffreda S, Yang SQ, Lin HZ, et al. Leptin regulates proinflammatory immune responses. FASEB J 1998;12:57– 65. 23. Li Z, Oben JA, Yang S, et al. Norepinephrine regulates hepatic innate immune system in leptin-deficient mice with nonalcoholic steatohepatitis. Hepatology 2004;40:434 – 441. 24. Guebre-Xabier M, Yang S, Lin HZ, et al. Altered hepatic lymphocyte subpopulations in obesity-related murine fatty livers: potential mechanism for sensitization to liver damage. Hepatology 2000;31:633– 640. 25. Li Z, Soloski MJ, Diehl AM. Dietary factors alter hepatic innate immune system in mice with nonalcoholic fatty liver disease. Hepatology 2005;42:880 – 885. 26. Elinav E, Pappo O, Sklair-Levy M, et al. Adoptive transfer of regulatory NKT lymphocytes ameliorates non-alcoholic steatohepatitis and glucose intolerance in ob/ob mice and is associated with intrahepatic CD8 trapping. J Pathol 2006;209:121– 128. 27. Mattner J, Debord KL, Ismail N, et al. Exogenous and endogenous glycolipid antigens activate NKT cells during microbial infections. Nature 2005;434:525–529. 28. Novobrantseva TI, Majeau GR, Amatucci A, et al. Attenuated liver fibrosis in the absence of B cells. J Clin Invest 2005;115:3072– 3082. 29. Balabanian K, Lagane B, Infantino S, et al. The chemokine SDF1/CXCL12 binds to and signals through the orphan receptor RDC1 in T lymphocytes. J Biol Chem 2005;280:35760 –35766. 30. Ebert LM, Schaerli P, Moser B. Chemokine-mediated control of T-cell traffic in lymphoid and peripheral tissues. Mol Immunol 2005;42:799 – 809. 31. Kim CH, Broxmeyer HE. Chemokines: signal lamps for trafficking of T and B cells for development and effector function. J Leukoc Biol 1999;65:6 –15. 32. McCuskey RS, Ito Y, Robertson GR, et al. Hepatic microvascular dysfunction during evolution of dietary steatohepatitis in mice. Hepatology 2004;40:386 –393. 33. De Rosa V, Procaccini C, Cali G, et al. A key role of leptin in the control of regulatory T-cell proliferation. Immunity 2007;26:241– 255. 34. La Cava A, Matarese G. The weight of leptin in immunity. Nat Rev Immunol 2004;4:371–379.
Received March 23, 2007. Accepted January 31, 2008. Address requests for reprints to: Gabriel Perlemuter, MD, Prof, INSERM, U764, Clamart, F-92140, France. e-mail: gabriel.perlemuter@ abc.aphp.fr; fax: (33) 1 40 94 06 56. Supported by INSERM, Conseil régional d’Ile-de-France, and Association pour la recherche contre le cancer (ARC; to A.E.B.); by INSERM, Fondation pour la Recherche Médicale, and AP-HP (to G.P.); and by INSERM Avenir, Univ Paris-Sud, and The European Union FP6 (INNOCHEM, grant number LSHB-CT-2005-518167). The authors thank Chantal Desdouets for technical support for liver perfusion, Didier Robrieux for rearing the mice, and Cédric Dray for HFD mice, insulin and glucose data. Conflicts of interest: No conflicts of interest exist.
BASIC–LIVER, PANCREAS, AND BILIARY TRACT
May 2008
1469.e1
BIGORGNE ET AL
GASTROENTEROLOGY Vol. 134, No. 5
Supplementary Document Supplementary Methods Immunodetection protocols. Double immunofluorescent detection was performed on frozen sections of the liver. Endogenous biotin activity was quenched (Dakocytomation) and sections were sequentially incubated with a rat anti-CD31 mAb (Pharmingen) and either with the biotinylated anti-CXCL12 antibody or with a goat anti-CXCL13 antibody (R&D Systems). Anti-CD31 antibody was revealed with a donkey Alexa Fluor 594conjugated anti-rat antibody, the biotinylated antiCXCL12 antibody with Alexa Fluor 488-conjugated streptavidin and the CXCL13 antibody with a rabbit biotinylated anti-goat antibody followed by Alexa Fluor 488-conjugated streptavidin. For the colocalization of CXCL12 and CXCL13, CXCL13 was revealed as described
previously and CXCL12 with the Alexa Fluor 594-conjugated streptavidin. Immunohistochemistry was performed on paraffin-embedded sections of the liver. Antigen retrieval was performed by microwaving in citrate buffer pH 6 (DakoCytomation, Glostrup, Denmark). Endogenous peroxidase and biotin activities were quenched with a peroxidase blocking solution followed by the Biotin Blocking System (DakoCytomation). Sections were incubated with a rabbit anti-CK19 antibody (Abcam; Cambridge, UK) or a biotinylated anti-CXCL12 mAb (gift from Dr F. Arenzana-Seisdedos, France). CK19 binding was detected with a biotinylated goat anti-rabbit (Biogenex, San Ramon, CA), followed, as for CXCL12, by peroxidase-conjugated streptavidin (Biogenex). AEC (3amino-9-ethylcarbozole; DakoCytomation) was used as the chromogen and slides were counterstained with hematoxylin.
Supplementary Table 1. Metabolic Status of Control, ob/ob, “Non-obese ob/ob” and High-Fat Diet Mice
ALT (IU/L) AST (IU/L) Glucose (mmol/L) Triglyceride (mmol/L) Cholesterol (mmol/L) HDL-cholesterol (mmol/L) Glucose x insulin
Control (n ⫽ 12)
ob/ob (n ⫽ 12)
Non-obese ob/ob (n ⫽ 12)
High-fat diet (n ⫽ 8)
40 ⫾ 4 131 ⫾ 28 3.2 ⫾ 0.3 0.90 ⫾ 0.08 3.5 ⫾ 0.1 2.6 ⫾ 0.1 408 ⫾ 92
381 ⫾ 37 383 ⫾ 19 5.5 ⫾ 0.8 1.12 ⫾ 0.05 5.2 ⫾ 0.2 3.8 ⫾ 0.1 2251 ⫾ 252
78 ⫾ 10a 183 ⫾ 22a 2.1 ⫾ 0.5a 1.02 ⫾ 0.04 3.9 ⫾ 0.2a 3.0 ⫾ 0.1a 1092 ⫾ 633a
53 ⫾ 12 145 ⫾ 25 7.8 ⫾ 0.5b 1.3 ⫾ 0.1b 4.5 ⫾ 0.2b 3.2 ⫾ 0.1b 4913 ⫾ 1550b
This Supplementary Table shows means ⫾ SEM. aStatistically significant (P ⬍ .05) differences between “non-obese ob/ob” mice and ob/ob mice. bStatistically significant (P ⬍ .05) differences between high-fat diet mice and control mice (nonparametric variance analysis (Mann–Whitney).
Supplementary Figure 1. Liver histology of mice fed a high-fat diet. Liver histology of: (A) Control and (B) high-fat diet mice (original magnification 40⫻). Black arrow points to inflammatory cell foci. Representative figures of 8 mice per group.
May 2008
Supplementary Figure 2. High-fat diet-induced obesity increases lymphocyte migration to the liver. Lymphocytes were isolated from high-fat diet (HFD) or control LPS-challenged donor mice, labeled, pooled, and transferred to LPS-challenged lean recipient mice. After migration, liver lymphocytes were harvested. Results are expressed as the number of fluorescent lymphocytes recruited into the liver per 105 transferred lymphocytes. Migration into the liver of: (A) CD4⫹T, (B) CD8⫹T, and (C) B cells. Graphs show means ⫾ SEM of 5 experiments. Statistically significant differences between control and HFD cells are indicated (paired 2-group Wilcoxon signed rank test, *P ⬍ .05).
CHEMOKINES, LPS, AND FATTY LIVER INFLAMMATION
1469.e2