Depletion of CD40 on CD11c+ cells worsens the metabolic syndrome and ameliorates hepatic inflammation during NASH

The co-stimulatory CD40-CD40L dyad plays a central role in fine-tuning immune reactions, including obesity-induced inflammation. Genetic ablation of CD40L reduced adipose tissue inflammation, while absence of CD40 resulted in aggravated metabolic dysfunction in mice. During obesity, CD40 expressing CD11c+ dendritic cells (DC) and macrophages accumulate in adipose tissue and liver. We investigated the role of CD40+CD11c+ cells in the metabolic syndrome and nonalcoholic steatohepatitis (NASH). DC-CD40-ko mice (CD40fl/flCD11ccre) mice were subjected to obesity or NASH. Obesity and insulin resistance were induced by feeding mice a 54% high fat diet (HFD). NASH was induced by feeding mice a diet containing 40% fat, 20% fructose and 2% cholesterol. CD40fl/flCD11ccre mice fed a HFD displayed increased weight gain, increased adipocyte size, and worsened insulin resistance. Moreover, CD40fl/flCD11ccre mice had higher plasma and hepatic cholesterol levels and developed profound liver steatosis. Overall, regulatory T cell numbers were decreased in these mice. In NASH, absence of CD40 on CD11c+ cells slightly decreased liver inflammation but did not affect liver lipid accumulation. Our experiments suggest that CD40 expressing CD11c+ cells can act as a double-edged sword: CD40 expressing CD11c+ cells contribute to liver inflammation during NASH but are protective against the metabolic syndrome via induction of regulatory T cells.

The integrin CD11c is a DC surface marker, but is also present on B, T and NK cells and subsets of monocytes and macrophages 16 . Increased numbers of CD11c + cells are found in the liver of mice under obese conditions 17 , and liver inflammation contributes to the progression from NAFLD (defined as the presence of ≥5% of hepatic steatosis) to NASH 18 . In humans, adipose tissue CD11c expression is discriminative for crown-like-structure macrophages and the presence of CD11c + macrophages correlates with markers of insulin resistance 19 . CD40 expression was detected on CD11c + adipose tissue cells in obese mice while lowly expressed in lean adipose tissue 20 .
We here investigate the contribution of CD40 on CD11c + cells in the regulation of diet induced obesity and NASH using CD40 fl/fl CD11c cre mice. We found that CD40 expressing CD11c + cells contribute to diet-induced-obesity (DIO) and NASH in opposing ways. In diet induced obesity CD40 expressing CD11c + cells play a crucial role in protection against obesity-induced ectopic lipid storage and metabolic dysfunction, most likely via induction of regulatory T cells. During NASH, CD40 on CD11c + cells contributes to liver inflammation.

Results
CD40 fl/fl CD11c cre mice have increased body weight gain and excess lipid deposition compared to WT mice after high fat diet feeding. During the course of DIO, CD40 fl/fl CD11c cre mice gained more weight than their wild type littermates (two-way ANOVA, p = 0.0012 at week 18, Fig. 1A). This difference was specific for HFD fed mice and was not observed in the mice fed SFD (Fig. 1A). EpAT weight did not differ between CD40 fl/fl CD11c cre and WT mice (Fig. 1B). However, adipocyte size of HFD CD40 fl/fl CD11c cre mice was increased by 20.9% compared to WT mice (unpaired t-test, p = 0.0082, Fig. 1C) indicating excess lipid deposition. In line with these results, total cholesterol levels in plasma were significantly higher in CD40 fl/fl CD11c cre mice compared to WT mice on HFD (two-way ANOVA, p = 0.0476, Fig. 1D). Plasma VLDL, HDL and LDL levels did not differ significantly, but the increase in HDL and LDL levels explain the increase in total cholesterol in the HFD CD40 fl/fl CD11c cre mice (Fig. 1D). Plasma triglyceride levels did not differ (Fig. 1E). Leptin concentrations in the Data is represented as mean ± SEM. *P < 0.05; **P < 0.01 for comparison between WT and CD40 fl/fl CD11c cre mice fed the same diet. n = 7/group for SFD, and n = 8/group for HFD. Adipose tissue inflammation is not affected in CD40 fl/fl CD11c cre mice. FACS analysis of the adipose tissue revealed that leukocyte infiltration of the AT in CD40 fl/fl CD11c cre mice did not significantly differ from WT mice when fed a similar diet (Supplemental Fig. 1A). Histological stainings of the EpAT with CD68 for macrophages and CD3 for T cells confirm these findings (Supplemental Fig. 1B). Further analysis of myeloid and lymphoid subsets also showed no significant differences (Supplemental Fig. 1C,D). This was confirmed by the mRNA transcription profile of inflammatory mediators in EpAT (Supplemental Fig. 2A,B), revealing that absence of CD40 on CD11c + cells does not affect AT inflammation.
Loss of CD11c + CD40 causes the development of liver steatosis during diet-induced obesity. As decreased storage capacity of the adipose tissue leads to ectopic lipid deposition, we next assessed the severity of obesity-induced hepatic steatosis. In CD40 fl/fl CD11c cre mice, the percentage liver weight compared to body weight was significantly increased (6.4 ± 2.3% of BW) compared to WT mice (4.2 ± 0.5% of BW) (unpaired t-test, p = 0.0162, Fig. 3A). Liver total cholesterol levels were highly increased in CD40 fl/fl CD11c cre mice compared to WT mice on the HFD (unpaired t-test, p = 0.0118, Fig. 3B) while triglyceride content was not affected (Fig. 3B). As expected from the high liver cholesterol levels, the grade of steatosis was significantly higher in CD40 fl/fl CD11c cre mice on a HFD compared to WT mice (median steatosis score of grade 3 in CD40 fl/fl CD11c cre mice compared to grade 1 in WT mice, Mann-Whitney test, p = 0.0177, Fig. 3C). The increased steatosis was confirmed by ORO staining showing significantly increased lipid accumulation in the liver (Fig. 3D). Quantification of the ORO staining indicated that 7.8 ± 7.7% of the liver of WT mice is steatotic vs 23.3 ± 6.1% of the liver of CD40 fl/fl CD11c cre mice (mean ± SD), unpaired t-test, p = 0.0005, Fig. 3D). Excess free cholesterol in the liver can accumulate in the hepatocytes, Kupffer cells, and hepatic stellate cells where it can induce cellular toxicity or proinflammatory and profibrotic effects 23 . In order to examine whether these mice suffered from liver damage due to excessive storage of fat in the liver, we measured the presence of liver enzymes and damage related proteins in the plasma that could indicate hepatocellular injury 24 . Significantly increased levels of plasma ALT were found in CD40 fl/fl CD11c cre mice (Mann-Whitney test, p = 0.0207, Fig. 3E), while AST and albumin levels did not differ from levels found in WT mice on HFD (Fig. 3E). Liver diseases that result from hepatocyte necrosis, and not from inflammation, result in elevated serum GDH levels 25 . GDH activity in CD40 fl/fl CD11c cre mice was higher compared to the WT mice, although not significant (unpaired t-test, p = 0.1967, Fig. 3E). Sirius Red staining of the liver shows that there is no fibrosis present after 18 weeks of HFD (data not shown), but the increased levels of ALT and GDH in the plasma of CD40 fl/fl CD11c cre mice indicate that the liver steatosis had caused damage to the hepatocytes of these mice.
Regulatory T cells numbers are decreased in the liver of CD40 fl/fl CD11c cre mice with DIO. Liver inflammation was investigated by histopathological quantification of the CD45 staining and showed no differences in number of leukocytes present between the groups that received a similar diet (Fig. 4A). Liver leukocyte www.nature.com/scientificreports www.nature.com/scientificreports/ subsets were further analyzed by flow cytometry (Fig. 4B,C). The percentage of CD3 + T cells in the liver was increased in both the HFD groups compared to SFD mice (Fig. 4B). Interestingly, the percentage of CD4 + FoxP3 + regulatory T cells (Tregs) was significantly lower in the CD40 fl/fl CD11c cre mice on a HFD compared to WT mice on the same diet (unpaired t-test, p = 0.0067, Fig. 4C). This decrease in Tregs was also observed in blood, spleen and LN of CD40 fl/fl CD11c cre mice (Fig. 4D). This difference in Treg population was only observed after HFD feeding and was not found in the SFD mice (Fig. 4D). Livers of CD40 fl/fl CD11c cre mice on a HFD showed a slightly increased expression of RORγt (unpaired t-test, p = 0.0725, Fig. 4E) and IL17 compared to WT mice mRNA (  www.nature.com/scientificreports www.nature.com/scientificreports/ p = 0.0923, Fig. 4F)), and increased expression in genes related to cholesterol uptake and biosynthesis (LDLr (unpaired t-test, p = 0.0600), and SREBP2 (unpaired t-test, p = 0.0195), Fig. 4F) in HFD fed CD40 fl/fl CD11c cre mice compared to WT mice suggesting and imbalance in liver cholesterol metabolism in these mice.
The degree of hepatosteatosis is not different in CD40 fl/fl CD11c cre mice compared to WT mice on a NASH inducing diet. As the degree of hepatic inflammation in DIO is rather low, we decided to further investigate the (anti)inflammatory potential of CD11c + CD40 + cells in a more inflammatory model of hepatosteatosis, and we induced NASH in both genotypes. The NASH inducing diet did not affect body weight compared to the control diet and there were no differences between CD40 fl/fl CD11c cre mice and WT mice that were fed a similar diet (Fig. 5A). Increased liver weights were observed after feeding the NASH diet, but there was no difference between CD40 fl/fl CD11c cre mice and WT mice (Fig. 5B). HOMA-IR was slightly increased in CD40 fl/fl CD11c cre mice fed a NASH diet (score 1.18 ± 0.42 in WT vs 1.58 ± 0.71 in CD40 fl/fl CD11c cre mice, unpaired t-test, p = 0.0914, Fig. 5C). Furthermore, the degree of steatosis was similar in CD40 fl/fl CD11c cre mice and WT mice on the NASH diet (Fig. 5D,E, representative HE pictures). Liver Cholesterol and TG levels were similar in the different groups on the same diet (Fig. 5F). Fasted plasma glucose measured after 20 weeks of diet was not different between the groups (Supplemental Fig. 3A). Fasted plasma insulin levels were slightly higher in the CD40 fl/fl CD11c cre mice on the NASH diet compared to WT mice (0.23 ug/mL vs 0.31 ug/mL, unpaired t-test, p = 0.0667, Supplemental Fig. 3B). Glucose and insulin sensitivity were measured after 12 weeks of diet and were not affected in the CD40 fl/fl CD11c cre mice and WT mice on the NASH diet as shown by no differences between the groups during the GTT and ITT (Supplemental Fig. 3C,D). Plasma cholesterol and TG did not significantly differ between the groups on the same diet (Supplemental Fig. 3E,F).
CD40 fl/fl CD11c cre mice have reduced liver inflammation during NASH compared to WT mice. Although no differences could be observed in hepatosteastosis after the NASH diet, the degree of hepatic inflammation had decreased in CD40 fl/fl CD11c cre mice compared to WT mice (Mann-Whitney test, p = 0.0301, Fig. 6A). Quantification of liver Mac-3 staining and CD3 staining shows that the number of macrophages and T cells in the liver of CD40 fl/fl CD11c cre mice are not significant different between the groups, but are both slightly decreased compared to WT mice on a NASH diet, accounting for the total decrease in inflammation (Fig. 6B,C). FACS data of the liver shows no difference in leukocyte composition (Fig. 6D). ALT levels www.nature.com/scientificreports www.nature.com/scientificreports/ had increased after feeding the NASH inducing diet but did not differ between WT and CD40 fl/fl CD11c cre mice (Fig. 6E), whereas AST levels were lower in CD40 fl/fl CD11c cre mice (19.02 ng/mL vs 16.57 ng/mL, unpaired t-test, p = 0.0879, Fig. 6F). Furthermore, systemic inflammation was not different between the groups on the same diet (blood, spleen and lymph node leukocyte subsets, Supplemental Fig. 4A-C).

Discussion
In line with the worsened metabolic phenotype observed in CD40 −/− mice on a high fat diet 10-13 , we here show that HFD fed CD40 fl/fl CD11c cre mice become more obese than WT mice, develop hyperinsulinemia and have an increased lipid uptake in adipose tissue and liver, resulting in severe hepatic steatosis. In contrast to the phenotype observed in total body CD40-deficient mice, CD40-deficiency in CD11c + cells did not result in a profound increase in pro-inflammatory mediators in lymphoid organs, adipose tissue or liver. However, upon high fat diet feeding CD40 fl/fl CD11c cre mice did show a reduction in the number of CD4 + FoxP3 + regulatory T cells (Tregs) in blood, adipose tissue, liver and lymphoid organs.
Tregs function as a control mechanism that can affect the behavior of other T cell populations and suppress overactive immune responses 26 . CD40-deficient mice are known to have a reduced amount of Tregs in peripheral blood, spleen and the thymus 27,28 . Especially B cell CD40 is involved in survival of Tregs in the thymus 29 , whereas deficiency of CD11c-CD40 resulted in only a minor decrease in natural Tregs in the thymus 30 . In our experiment, we could not observe decreases in Treg levels in blood, lymphoid organs and tissues in CD40 fl/fl CD11c cre upon SFD, but we did see a HFD induced reduction of Tregs in lymphoid organs, and especially in the liver. This suggests that CD40 + CD11c + cells are particularly involved in the reactive generation of Tregs during chronic inflammatory conditions 31 , and can be generated through homeostatic proliferation of CD4 + CD25 − T cells 32 .
Although we did not observe signs of enhanced inflammation, the decrease in Tregs was surprisingly accompanied by an increase in systemic and hepatic cholesterol levels in CD40 fl/fl CD11c cre mice fed a HFD. Consistent with our findings, depleting FoxP3 + Tregs in a mouse model for atherosclerosis resulted in increased plasma cholesterol levels but did not induce inflammation in the liver or the atherosclerotic plaques 33 . In this study Treg depletion reduced protein expression of the VLDL binding protein sortilin-1 in the liver increasing lipoprotein www.nature.com/scientificreports www.nature.com/scientificreports/ catabolism activity in the plasma, which results in accumulation of VLDL and CM lipoproteins in the circulation 33 . In our experiment we observed upregulated expression of the LDLr, SREBP2 and FAS gene suggesting that the liver takes up more cholesterol from the blood and has increased cholesterol biosynthesis 34 , while reduced gene expression of LXRα (and Idol) and its target genes indicate decreased excretion of cholesterol from hepatocytes 35 . Together this could result in the accumulation of cholesterol in the liver. Another contributor to the increased hepatosteatosis in our DIO model could be the increased leptin levels in the CD40 fl/fl CD11c cre mice. The adipose tissue hormone leptin negatively affects Treg proliferation 36,37 . Increased leptin levels contribute to the progression from NAFLD to NASH via activation of macrophage and Kupffer cells and increased oxidative stress 38 .
Interestingly, mice deficient in the co-stimulatory molecule CD80/86 (B7.1/B7.7) showed increased liver steatosis when fed a HFD, a similar phenotype as observed in our CD40 fl/fl CD11c cre mice 39 . CD80/CD86 is required for proper Treg development and the phenotype observed in these mice was also attributed to the reduction of Treg numbers in B7.1/B7.7 double knock out mice 39 . Together with our results these findings underline the important role for Tregs in obesity associated liver disease and indicate that manipulation of co-stimulatory molecule signaling to increase Treg numbers is therapeutically interesting for the prevention of hepatosteatosis and eventually NASH.
Tregs have been described to play an important role in hepatic steatosis related diseases, particularly in NAFLD and NASH. NAFLD and NASH are important health problems related to obesity and NAFLD affects 25% of the global adult population, of which 59% of the patients exhibits NASH 18 . NAFLD is defined by the presence of ≥5% of hepatic steatosis, and when hepatocyte damage develops, disease progresses into NASH, characterized by hepatic fibrosis 18 . Hepatic steatosis and NASH are associated with reduction of hepatic Treg numbers in both human and mouse 40,41 . Depletion of Tregs is caused by increased oxidative stress in fatty livers inducing Treg apoptosis and impairs suppression of inflammatory responses 40 . Activation of inflammatory signaling through the TNF-α pathway contributes to the progression from simple steatosis to steatohepatitis 40 . Interestingly, pharmacological upregulation of Treg numbers, using 3, 3′-diindolylmethane (DIM), can reduce hepatic inflammation 42 .
However, when NASH was induced in our CD40 fl/fl CD11c cre mouse model, we could not observe a reduction in Tregs in liver or lymphoid organs. Moreover, absence of CD11c + CD40 + cells did not affect the degree of hepatosteatosis, which was massive (~40%) in the NASH model. These data indicate that the reduced Treg numbers observed in the DIO study were probably caused by the adipose tissue driven ectopic lipid accumulation, rather than the intrinsic effects of the liver fat during NASH. Interestingly, absence of CD11c + CD40 + cells did reduce hepatic inflammation during NASH, indicating that CD11c + CD40 + cells have a strong (lipid driven) immune-modulatory function.
In conclusion, we here show that CD40 + CD11c + cells play a crucial role in protection against obesity-induced liver steatosis via activation of regulatory T cells, thereby inducing a tolerogenic immune response that prevents hepatosteatosis.
Mice had ad libitum access to food and water and were maintained under a 12-hour light-dark cycle. Body weight was monitored weekly. At the end of the dietary exposure period mice were euthanized, and blood samples were collected after intracardiac puncture using EDTA coated syringes. Liver, epidydimal adipose tissue (EpAT), muscle, spleen, and lymph nodes were removed after sacrifice and used for subsequent analyses. All the experimental procedures were approved by the Ethical Committee for Animal Experiments of the Academic Medical Centre, Amsterdam (AMC). All experiments were performed in accordance with relevant guidelines and regulations.
Glucose and insulin tolerance test. After  www.nature.com/scientificreports www.nature.com/scientificreports/ Plasma lipids, insulin, adipokine and liver enzyme measurements. Fasting plasma insulin levels were measured in samples from 6 hour fasted mice by using an insulin ELISA kit (Mercodia, Uppsala, Sweden) following manufacturers' protocol. Plasma leptin and adiponectin levels of the DIO mice were measured using a mouse leptin ELISA Kit (ChrystalChem) and a mouse adiponectin ELISA Kit (AssayPro). Liver enzymes were measured in plasma samples using an alanine aminotransferase (ALT) activity kit (DIO mice, Sigma-Aldrich) or ELISA Kit (NASH mice, Biomatik), a mouse aspartate aminotransferase (AST) ELISA Kit (Biomatik), a BCG (Bromocresol Green) albumin assay kit (Sigma-Aldrich), a glutamate dehydrogenase (GDH) activity assay kit (Sigma-Aldrich), and a bilirubin assay kit (Sigma-Aldrich).
Cholesterol and triglyceride measurements. Total cholesterol and triglyceride (TG) concentrations in plasma and liver were measured by standard enzymatic methods (CHOD-PAP and GPO-PAP; Roche Diagnostics). Frozen liver tissues were homogenized in 1 ml of SET buffer (250 mM sucrose, 2 mM EDTA, and 10 mM Tris), and the samples then underwent multiple freeze-thaw cycles for cell-destruction. Individual plasma lipoprotein levels were measured by fast-performance liquid chromatography as described before 45 . Flow cytometry. White blood cells from blood, spleen, lymph nodes (LN), liver and stromal vascular fraction (SVF) from EpAT were analyzed by flow cytometry. EpAT was minced into small pieces and digested with liberase (0.25 mg/mL, Roche) for 45 minutes at 37 °C. The digested samples were passed through a 70-µm nylon mesh (BD Biosciences, Breda, the Netherlands). The SVF was obtained from the resulting pellet and resuspended in FACS buffer. Erythrocytes in blood and spleen were removed by incubation with hypotonic lysis buffer (8.4 g of NH4Cl and 0.84 g of NaHCO3 per liter of distilled water). To prevent non-specific binding of antibodies to the Fc receptor, all cell suspensions were incubated with CD16/32 antibody (eBioscience, San Diego, CA, USA) in FACS buffer (0.5% BSA, 0.01% NaN 3 in PBS) before the antibodies were incubated with the indicated tissues. Fluorescence was measured by flow cytometry (FACSCanto II, BD Biosciences, Breda, The Netherlands) and analyzed with FlowJo software version 7.6.5. (Tree star). The antibodies used are listed in Supplemental Table 1.
Histology. EpAT and liver tissues were collected, fixed in 4% paraformaldehyde and embedded in paraffin.
Liver steatosis was graded on 4 μm thick haematoxylin-eosin (H&E) stained sections. Grades were scored in a range from 0-5 were 0 = no steatosis, 1 = minimal steatosis, 2 = mild steatosis, 3 = moderate steatosis, 4 = severe steatosis, and 5 = marked steatosis. Immunohistochemistry (IHC) on liver and EpAT was performed for antibodies indicated in the graphs and listed in Supplemental Table 1. IHC staining was quantified using Image J. Adipocyte size from DIO mice was measured on EpAT H&E stained sections using image J (NIH, Bethesda, Maryland, USA). Additionally, livers from DIO mice were frozen, embedded in OCT and frozen at −80 degrees Celsius. To measure liver lipid content, 5 μm thick cryosections of the liver were stained with Oil red O (Sigma-Aldrich, Zwijndrecht, the Netherlands). Analyses were performed by an observer who was blinded for the experimental conditions. Gene expression analysis. Total RNA of EpAT and liver was extracted using TRIzol (Invitrogen, Carlsbad, CA, USA), while total RNA from muscle was extracted using the GeneJET RNA Purification Kit (Thermo Scientific, Massachusetts, USA). RNA was reverse transcribed with an iScript cDNA synthesis kit (Bio-Rad, Veenendaal, the Netherlands). qPCR was performed using a SYBR green PCR kit (Applied Biosystems, Leusden, the Netherlands) on a ViiA7 real-time PCR system (Applied Biosystems). Primer sequences can be found in Supplemental Table 2.
Statistical analysis. The experiments were performed with n = 8 mice (HFD), n = 7 mice (SFD), n = 15 mice (NASH), and n = 6 mice (CD), and results are presented as means ± SEM. Data were analyzed by an unpaired t-test, a Mann-Whitney test when appropriate, or a two-way ANOVA when comparing multiple groups using GraphPad Prism 7.0 software (GraphPad Software, Inc., La Jolla, CA, USA). P-values < 0.05 were considered significant.

Data Availability
All data generated or analyzed during this study are included in this published article (and its Supplementary  Information Files).