Deficiency in plasmacytoid dendritic cells and type I interferon signalling prevents diet-induced obesity and insulin resistance in mice

C57BL/6 (B6) male mice were purchased from Taconic (Ejby, Denmark). IFNAR^−/− mice were kindly provided by B. Johansson-Lindbom, Lund University (Lund, Sweden). B6.E2-2 ^fl/fl mice were generated as previously detailed [23] and B6.Cg-Tg(Itgax-cre)1-1Reiz/J mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). B6.E2-2 ^fl/fl and B6.Cg-Tg(Itgax-cre)1-1Reiz/J mice were paired; the offspring were screened for flox-sites and cre-recombinase and were bred to produce the B6.E2-2 ^fl/fl Itgax.cre⁺ and B6.E2-2 ^fl/fl Itgax.cre⁻ mice used in this study.

B6.E2-2 ^fl/fl Itgax.cre mice were bred in a specific pathogen-free animal facility at Lund University and kept under standard conditions with ad libitum access to water and food during experiments. The experiments were conducted in compliance with the National Institutes of Health guidelines, and all animal experimental procedures were approved by the ethical committee of Lund University Animal Care and Use.

The mice were maintained on either a high-fat diet (HFD; 21.9 kJ/g [5.24 kcal/g], 34.9% (wt/wt) fat, 26.2% (wt/wt) protein, 26.2% (wt/wt) carbohydrate; D12492, Research Diets, New Brunswick, NJ, USA) or a normal diet (ND; 12.6 kJ/g [3 kcal/g], 4% (wt/wt) fat, 18.5% (wt/wt) protein, 55.7% (wt/wt) carbohydrate; R36, Lactamin AB, Stockholm, Sweden) for 18 weeks, starting at 5 weeks of age.

Body weight was recorded weekly. For endpoint analysis, the animals were fasted for 6 h before body composition scanning using a GE-Lunar PIXImus 2 scanner (GE Healthcare, Wauwatosa, WI, USA) or an OGTT, where blood glucose was measured at various time points before and after oral glucose administration (2 g/kg). Cheek blood samples were collected before and after glucose administration for hormone analysis. The mice were anaesthetised with Hypnorm/Dormicum (Vetapharma, Leeds, UK/Roche, Basel, Switzerland) and transcardially perfused with 15 ml PBS, followed by dissection of subcutaneous white adipose tissue (SAT), visceral white adipose tissue (VAT), brown adipose tissue (BAT), the spleen and the liver for flow cytometry.

Single cell suspensions from the spleen were obtained by disrupting the tissue through a 70 μm cell strainer. The livers were mechanically minced into small >1 mm pieces, washed twice in PBS, treated with collagenase II (Invitrogen, Carlsbad, CA, USA) for 40 min at 37°C, and single cells were suspended through a 70 μm cell strainer and separated by 50/30 Percoll (GE Healthcare) gradient centrifugation. The white adipose tissue samples were mechanically minced and collagenase II digestion was performed for 40 min at 37°C to fractionate the AT into adipocytes and a stromal vascular fraction (SVF). The SVF was resuspended in HEPES-buffered RPMI medium (Gibco, Waltham, MA, USA) and single cells were suspended through a 70 μm cell strainer. All samples were resuspended in FACS buffer (PBS, 3% [vol./vol.] FCS, 2 mmol/l EDTA) prior to surface staining, preincubated with anti-CD16/32 (Fc ‘blocking’ antibody, clone 2.4G2) (BD Biosciences, San Jose, CA, USA) for 15 min at 4°C and subsequently stained with fluorescent-labelled primary antibodies for 25 min at 4°C. The following antibodies were used: anti-TCRβ (clone H57-597) and anti-SiglecF (clone E50-2440) from BD Biosciences; anti-CD317 (clone 120G8.04) from Dendritics (Lyon, France); anti-Ly6G (clone 1A8), anti-Ly6C (clone HK1.4), anti-NK1.1 (clone PK136), anti-CD45 (clone 30-F11), anti-I-A/I-E (clone M5/114.15.2), anti-CD11b (clone M1/70) and anti-CD206 (clone C068C2) from Biolegend (San Diego, CA, USA); and anti-CD4 (clone GK1.5), anti-F4/80 (clone BM8), anti-CD45R(B220) (clone RA3-6B2), anti-CD8 (clone 53-6.7), anti-CD11c (clone N418), anti-SiglecH (clone 440c), anti-CD19 (clone 1D3) and fixable viability dye (FVD) from eBioscience (Santa Clara, CA, USA). The cells were fixed using the Foxp3 Staining Buffer Set (eBioscience). Samples containing biotin-coupled antibodies were incubated for 25 min at 4°C with secondary Streptavidin-PECF594 antibodies (BD Biosciences). Cells were washed twice in FACS buffer and data acquisition was performed on a BD LSR II flow cytometer with FACSDiva software (BD Biosciences). A minimum of 50,000 CD45⁺ events were analysed for each sample. The analysis was performed using the FlowJo software (FlowJo LLC, Ashland, OR, USA).

To gain insight into the potential role of type I IFN in obesity and obesity-related metabolic abnormalities, we used the HFD mouse model of type 2 diabetes. From 5 weeks of age, male C57BL/6 (B6) mice were given ad libitum access to either an HFD or ND for 18 weeks. As shown before [24], mice fed an HFD showed a significant weight increase of 64% compared with ND-fed littermates (Fig. 1a). At the 18 week endpoint, the obese mice had developed glucose intolerance and insulin resistance (Fig. 1b, c). The detrimental stage of metabolic disease was evident by the increase in the sizes of the fat pads and the liver, which were significantly enlarged in obese mice (HFD, 2.35 ± 0.29 g; ND, 1.10 ± 0.16 g) (Fig. 1d). Using an x-ray densitometer, lean and fat mass was compared between the two groups, by which the difference in weight gain was attributed to a significant increase in fat mass (Fig. 1e, f).

To directly test the hypothesis that type I IFN plays a significant role in the development of obesity and type 2 diabetes-related metabolic abnormalities, we exposed mice lacking the receptor for IFN-α (IFNAR^−/−) to an HFD. In support of the hypothesis, we found that the HFD-fed IFNAR^−/− mice displayed a relative resistance to diet-induced weight gain compared with the B6 control mice on an HFD (Fig. 1g). This difference was significant after 6 weeks following the initiation of HFD feeding. At the endpoint after 18 weeks of HFD, the IFNAR^−/− mice had an 11% lower body weight compared with the control B6 mice (IFNAR^−/−HFD 44.7 g vs B6 HFD 49.9 g). This was associated with significantly smaller subcutaneous fat depots and a lower total fat mass in the HFD-fed IFNAR^−/− mice compared with B6 mice (Fig. 1f, h).

Obese mice developed a low-grade inflammation in the visceral fat deposits, with an increased number of leucocytes and an altered immune cell repertoire characterised by elevated levels of CD11c⁺ proinflammatory macrophages (CD11c⁺CD11b⁺F4/80⁺) (Fig. 2a–d).

When examining the infiltrating cell population in the inflamed AT, we found an expectedly large proportion of the CD11c⁺ cells to be proinflammatory macrophages (F4/80⁺CD11b⁺CD11c⁺) (Fig. 2b–d). In the liver the number of proinflammatory macrophages was also found to be significantly increased (Fig. 2c). When applying a subsequent analysis of the total CD11c⁺ population in the liver, we found significantly higher numbers in the obese state but also noted that a large proportion of these cells could not be accounted for by the macrophage population (Fig. 2c).

To determine whether the metabolic effects observed in the absence of type I IFN signalling were linked to a decreased inflammation in VAT and the liver we exposed the IFNAR^−/− mice and B6 control mice to an 18 week period of HFD feeding and compared them to groups fed an ND (Fig. 2e–g). We observed that the accumulation of proinflammatory macrophages in VAT of B6 mice was reduced in the absence of IFNAR1 (Fig. 2g). This would be in line with a previously suggested role of IFN-I in proinflammatory polarisation of VAT-recruited macrophages [22].

Because it has been previously reported that pDCs increase in both the liver and AT of HFD-treated animals [4], we hypothesised that this DC subset could account for the observed increase in CD11c⁺ cells in the liver. We therefore identified the pDCs among the CD11c⁺ population by gating away B cells (CD19⁺), natural killer cells (NK1.1⁺) and macrophages (F4/80⁺ and CD11b⁺) and focusing on the B220⁺120G8⁺CD11c^(int) subset (Fig. 2h–j). While the expression of the pDC marker Siglec-H was high in the splenic pDC population, the same subset in the liver and fat depots demonstrated variable expression profiles (Fig. 2k); the marker itself was also recently found to be less specific than initially proposed [25] and is partly sensitive to collagenase treatment [26]. The 120G8 antibody specifically targets BST-2 and constitutes a suitable marker for pDCs, as illustrated by its ability to eliminate pDCs through antibody depletion without significantly affecting other cell populations [27]. Using this gating strategy, we found that the pDC population was significantly increased in both the liver and VAT after HFD feeding (a 2.7-fold increase in the liver and 4.3-fold increase in VAT), and was most abundant in the liver (Fig. 2j). In the liver, pDCs accounted for almost 50% of the CD11c⁺ population in the obese group and approximately 30% in the lean group (Fig. 2h–j).

The observed accumulation of pDCs in both the liver and AT, as well as the protection from obesity development in the IFNAR knockout mice, suggest that this cellular subset could play a role in the pathology observed in the DIO model. To directly address this issue, we used the B6.E2-2 ^fl/fl .Itgax-cre mouse strain, in which the conditional knockout of E2-2 results in a specific block of the development of pDCs [23] (Fig. 3). A pronounced decrease in this subset is evident in both the bone marrow [23] and peripheral organs, including the spleen, liver and AT (Fig. 3a, b). In contrast, no significant alteration was seen in other haematopoietic subsets, such as macrophages and DCs (Fig. 3c–e).

Next we exposed E2-2 conditional knockout (E2-2.cre⁺) mice and control littermates (E2-2.cre⁻) to an 18 week period of HFD feeding and compared them to groups fed an ND. As with the IFNAR^−/− mice, we observed a significant difference in weight gain between the E2-2.cre⁺ and the E2-2.cre⁻ groups, with a significantly lower weight gain in the E2-2.cre⁺ mice (45.6 g for E2-2.cre⁺ and 54.4 g for E2-2.cre⁻ mice) that was evident from as early as week 4 on an HFD (Fig. 4a). The reduced weight gain was accompanied by increased insulin sensitivity (Fig. 4b). Thus, while the control group at the endpoint had reached the level of insulin resistance, the E2-2.cre⁺ mice were protected from developing insulin resistance (Fig. 4b). An improved, albeit not significant, GTT was also noted (Fig. 4c).

To investigate if the observed protection from weight gain among the IFNAR^−/− mice was the result of a decrease in energy intake or energy expenditure, we registered the food intake throughout the experiment. As illustrated in Fig. 5a, no significant difference between the IFNAR^−/− and B6 mice was observed at any time point, suggesting that the observed difference in weight gain was due to different energy expenditure. Similarly, when the metabolic differences between E2-2.cre⁺ and E2-2.cre⁻ mice were assessed through their food intake, no significant difference was observed (Fig. 5b). Additionally, when assessing the mice, which were kept in groups of 4–6 individuals, no differences in behavioural patterns were seen between the groups. These results suggest that the ablation of type I IFN signalling, as well as pDC deficiency, prevented the mice from developing DIO and insulin resistance by affecting energy metabolism.