Plasmacytoid dendritic cells (pDCs) display an increased abundance in visceral adipose tissue (VAT) of humans with obesity. In the current study, we set out to decipher the molecular mechanisms of their recruitment to VAT and the functional relevance of this process. We observed increased pDC numbers in murine blood, liver, spleen, and VAT after feeding a high-fat diet (HFD) for 3 wk when compared with a standard diet. pDCs were enriched in fat-associated lymphoid clusters representing highly specific lymphoid regions within VAT. HFD led to an enlargement of fat-associated lymphoid clusters with an increased density and migratory speed of pDCs as shown by intravital multiphoton microscopy. For their recruitment into VAT, pDCs employed P-selectin with E-selectin and L-selectin being only critical in response to HFD, indicating that the molecular cues underlying pDC trafficking were dependent on the nutritional state. Subsequent recruitment steps required α4β1 and α4β7 integrins and engagement of CCR7. Application of fingolimod (FTY720) abrogated egress of pDCs from VAT, indicating the involvement of sphingosine-1-phosphate in this process. Furthermore, HFD altered pDC functions by promoting their activation and type 1 IFN expression. Blocking pDC infiltration into VAT prevented weight gain and improved glucose tolerance during HFD. In summary, a HFD fundamentally alters pDC biology by promoting their trafficking, retention, and activation in VAT, which in turn seems to regulate metabolism.
This article is featured in Top Reads, p.
Obesity is a major risk factor for metabolic diseases and became an enormous health burden in modern societies in the last 30 years (1). Excessive nutritional energy is stored in the white adipose tissue dispersed s.c. throughout the body and between internal organs. It serves as a major lipid reservoir and additionally as a source of endocrine mediators. Especially, visceral adipose tissue (VAT) is recognized to be highly metabolic active in humans (2). In mice, white adipose tissue attached to the uterus of females or epididymis of males is primarily known as VAT. In recent years it has become evident that excessive nutritional intake does not only affect body weight and disturb metabolism, but it also dramatically influences immunological homeostasis (3, 4). In VAT, classical dendritic cells (DCs) (5), invariant NK T cells (6), and γδ T cells (7) decrease with obesity whereas CD8+ T cells (8), innate lymphoid cells (9), and macrophages change in composition and performance. VAT-resident immune cells have a variety of different functions, including the regulation of adipokine expression, clearance of apoptotic cells, and extracellular matrix remodeling (3, 10). The shift of immune cell composition, numbers, and phenotype initiated with obesity finally results in a switch toward a proinflammatory cytokine milieu (11–13) that leads to a systemic increase of different inflammatory mediators, including free fatty acids, TNF-α, IL-1β, and IL-6 (14, 15), which in turn promote the development of insulin insensitivity and metabolic diseases (16).
Within the white adipose tissue, clusters of hematopoietic cells have been identified and named fat-associated lymphoid clusters (FALCs) (17). These non-classical lymphoid clusters have been described in omental, mesenteric, gonadal, and pericardial fat (17, 18). Composed of B1 and B2 cells (19), CD4+ T cells, and CD11b+ myeloid cells (18), FALCs are highly vascularized (18) and, in contrast to lymph nodes (LNs), they are not encapsulated (17). However, FALCs of the omentum contain high endothelial venules (HEVs) (20), a specialized type of postcapillary venule essential for lymphocyte trafficking to LNs. Interestingly, FALCs increase in number and size in response to acute or chronic inflammation of the peritoneal cavity to initiate adaptive immune responses (17, 18, 21).
Recently, plasmacytoid DCs (pDCs) gained attention, as these rare innate immune cells increase in frequency in VAT and liver of obese mice and humans (22–24). In contrast to classical DCs, pDCs have a round morphology and are characterized by their secretory function (25). They are capable to sense viruses and self-nucleic acids (26) and respond with rapid and robust type 1 IFN (IFN-I) expression and secretion (27). Despite few molecular differences, the function, the overall phenotype, and the core gene expression program is conserved between human and murine pDCs (28), and several pDC-specific surface markers have been established. Murine pDCs express sialic acid binding Ig-like lectin H (SiglecH), plasmacytoid dendritic cell Ag-1 (PDCA-1, CD317), Ly6C, and CD45R. In steady state, pDCs express low levels of MHC class II and costimulatory molecules and their expression increases upon activation (29). The generation of pDCs is mainly restricted to the bone marrow (BM), from where they enter the blood and secondary lymphoid organs to continuously patrol through the body. In the LN, they reach T cell areas mainly through HEVs by employing a specific adhesion cascade (30, 31). Under steady-state conditions, pDCs require L-selectin and during inflammation additionally E-selectin to allow rolling in the LN (32). Furthermore, β1 and β2 integrins as well as multiple chemokine receptors such as CCR7 and CCR9 are essential for pDCs to adhere and transmigrate into the LN, gut, and thymus tissue (33, 34). Similar to T cells, pDCs egress from secondary lymphoid organs back into the blood by sphingosine-1-phosphate (S1P)–mediated signaling (35). S1P is a lipid mediator that is present at high concentrations in plasma and lymph and couples to five different G protein–coupled S1P receptors (S1PRs) (36). The decisive role of S1P in immune cell trafficking was discovered when the immunosuppressive agent fingolimod (FTY720) was found to induce S1PR1, S1PR3, S1PR4, and S1PR5 internalization in T cells to render them unresponsive to the S1P gradient toward the circulation trapping them in secondary lymphatic organs (37, 38).
Within the immune system, pDCs are key players in orchestrating sensitization, activation, and differentiation of innate and adaptive immune cells (39). Interestingly, gene expression analyses during obesity show an upregulation of IFN-I genes in VAT (40), and IFN-I receptor-deficient mice fail to develop obesity and insulin insensitivity (24). Furthermore, recent studies suggest that pDCs, which represent a major source of IFN-I, may play a critical role in promoting obesity (41). In this study, we set out to decipher the impact of a high-fat diet (HFD) on the trafficking of pDCs into VAT and their activation within this key region of obesity. We identified the localization of pDCs in VAT and their multistep adhesion cascade facilitating their homing into VAT. HFD resulted not only in increased pDC infiltration and retention, but it also promoted an activated pDC phenotype compared with a standard fat diet (SFD). Within VAT, pDCs accumulated in FALCs where they increased in density and mobility after HFD application. The blockade of pDC infiltration to adipose tissue prevented weight gain and improved glucose tolerance. In summary, we identified a unique trafficking and activation profile of pDCs within VAT in response to HFD that opens up new avenues for treatment of obesity by targeting pDC homing and activation.
Materials and Methods
C57BL/6 wild-type (WT) mice were purchased from Charles River Laboratories (Wilmington, MA). CCR7−/− mice were bred and housed in the animal facilities of the New Research Building, Harvard Medical School (Boston, MA). The ubiquitin C–tdTomato mice provided by Wolfgang Kastenmüller (MPI Würzburg, Würzburg, Germany) and the SiglecH-GFP reporter mice (42) were bred in the animal facility of the University of Bonn (Bonn, Germany). Homozygous messenger of IFN-β (MOB) mice were obtained from Stefanie Scheu (HHU Düsseldorf, Düsseldorf, Germany) (43). Mice were housed under specific pathogen–free conditions in accordance with National Institutes of Health guidelines. Animal experiments have been approved by the Institutional Review Board and local authorities. Experiments were performed with mice at 6–10 wk of age at Harvard Medical School (Boston, MA), the University of Bonn, the Biomedical Center of the Ludwig-Maximilians-Universität München, and at the Multiphoton Imaging Core Facility at the Walter Brendel Center of Experimental Medicine in Munich, Germany. Normal chow, HFD (Research Diets, open source; D12492), or SFD (Research Diets, open source; D12450J) were applied for 3 wk.
Ab staining and flow cytometry
Unless stated otherwise, all Abs were obtained from BioLegend. The following anti-mouse Abs were used: CD31 (390), SiglecH (551-PE or -FITC), CD45R (RA3-6B2), CD11b (M1/70), CD45 (30-F11), PDCA-1 (927), CD49d (R1-2), CD62L (MEL-14), CXCR3 (CXCR3-176), and CCR7 (4B12). E-selectin and P-selectin binding sites on pDCs were evaluated by staining with E-selectin–human Ig Fc chimera (R&D Systems) and P-selectin–human Ig Fc chimera Ab (R&D Systems). Staining for flow cytometry was performed at 4°C for 20–30 min in FACS buffer (1% FCS in PBS with 2 mM EDTA). Dead cells were excluded by staining with Zombie Aqua (BioLegend, 1:1000 in PBS). Data were acquired by using FACSCanto II (Becton Dickinson) and analyzed with FlowJo software (Becton Dickinson). For in vivo imaging, SiglecH-Alexa 488, CD45-PE, and CD31-Alexa 647 were used.
All data are presented as mean ± SEM. Significance was calculated with GraphPad Prism 8. Statistical analyses were performed using a Student t test or one-way ANOVA as indicated. Significance was defined as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.
In vivo and in vitro expansion of pDCs
Female C57BL/6 mice were injected s.c. with melanoma cell line B16 secreting FLT3L that results in a massive expansion of all DC subsets in vivo (4, 44). Seven days after inoculation, spleen and liver were harvested and pDCs were enriched by gradient centrifugation (Lymphoprep, STEMCELL Technologies) at 800 × g for 20 min at 20°C without a break. pDC content and number were measured by flow cytometry staining before transfer. Cell suspensions were stained with CSFE (Molecular Probes). When indicated, pDCs were generated in vitro by incubating BM with FLT3L for 7 d. Cells were harvested and pDCs were quantified by FACS staining before labeling with a CellTrace Violet cell proliferation kit (Invitrogen).
Pertussis toxin pretreatment
pDCs were treated with pertussis toxin (PTx) (final concentration 200 ng/ml) for 2 h at 37°C and combined with CSFE labeling during the last 20 min before washing with RPMI 1640 containing 2% FCS. Media-treated pDCs were mixed with differentially stained cells and injected (in 200 μl of solution) into the tail veins of mice.
Enriched labeled pDCs were suspended in 2–4 × 107 cells/ml and injected i.v. After 18 h, recipient mice were sacrificed and spleen, inguinal LNs, liver, and VAT were harvested and analyzed by flow cytometry. Homing of pDCs was calculated to the number of injected pDCs. For inhibitory studies, the number of transferred pDCs was measured and pDCs recovered 18 h later in different organs were calculated per 1 million injected pDCs. In competitive homing assays, WT and CCR7−/− cells were either stained with CFSE or with tetramethylrhodamine-5-isothiocyanate (TRITC; Molecular Probes). Switching labeling ensured no effect of toxicities for either cell population. An index for homing was calculated by a ratio of PTx-treated or CCR7−/− versus control cells (100%).
Blocking Abs for inhibition studies
The following mAbs were used: Mel-14 (rat IgG2a, anti-murine L-selectin (45), Bio X Cell] and PS/2 (rat IgG2b, anti-murine α4), which were stored at −70°C in endotoxin-free saline at 1 mg/ml. P-selectin and E-selectin were blocked with RMP-1 and ultra-RME-1/CD62E (BioLegend), respectively. The integrin α4β7 (LPAM) was blocked with DATK32 (BioLegend), CCR9 with blocking Abs (9B1, BioLegend), and CXCR3 with blocking Abs (CXCR3-173, BioLegend). For inhibition studies, 100 μg of Ab was directly injected i.v. before cell transfer (∼1–2 h) or coinjected with pDC suspension. Administration of the S1PR inhibitor FTY720 (1 mg/kg) or PBS was performed as indicated or by daily i.p. injections.
For long-term blockade mice were fed with SFD (D12450J, Research Diets) or HFD (D12492, Research Diets) for 1 wk and injected daily i.p. with 100 µg of mAb against P-selectin (BioLegend) or its isotype control (IgG2a, MOPC-173). Mice were weighed daily and the weight of VAT was analyzed on day 7. Glucose tolerance was tested (i.p. glucose tolerance test [IPGTT]) by injecting 1 mg/kg glucose (Life Technologies) i.p. into 5-h-starved mice. The level of blood glucose was measured at 10-min intervals until 90 min after glucose injection.
LN, spleen, liver, and epididymal VAT were harvested 18 h after pDC transfer. LNs were smashed through a nylon net (pore size 70 µm). Spleen was digested with collagenase D (1 mg/ml) and DNase 1 (1 mg/ml) for 20 min at 37°C in RPMI 1640 media. Subsequently, the tissue was smashed through a nylon net and RBC lysis (Pharm Lyse, BD Biosciences) was performed before FACS staining. Liver was smashed through a nylon net by scraping it with a lid of the petri dish with PBS and 2 mM EDTA. The cell suspension was placed on a Lymphoprep gradient (STEMCELL Technologies) and centrifuged at 800 × g without break for 20 min at room temperature. Cells were harvested from the interphase, washed with PBS, and stained with Abs for flow cytometry. VAT tissue was minced with a scalpel and digested with collagenase 2 (1 mg/ml in RPMI 1640 media) for 25 min at 37°C under continuous rotation. The suspension was centrifuged and the cell pellet (stromal-vascular fraction) was taken up in RBC lysis buffer before staining for flow cytometry.
For the ELISA of IFN-α (PBL IFN Source) and IFN-β (BioLegend) one fat pad of epididymal VAT was lysed in T-PER tissue protein extraction reagent (Thermo Fisher Scientific) with protease inhibitor (Roche). The amount of cytokine was calculated per one total fat pad.
Intravital multiphoton imaging
In female mice VAT is directly attached to the uterus. When establishing the imaging protocol, we observed that the uterus contracted uncontrollably, which led to unwanted movements in the videos. Therefore, we preferred male mice for in vivo imaging. C57BL/6 mice were anesthetized with an i.p injection of MMF-mix anesthesia (5.0 mg/kg midazolam, 0.5 mg/kg medetomidine, and 0.05 mg/kg fentanyl). SiglecH-FITC and CD45-PE Abs were injected i.v. 15 min before imaging. When indicated, isolated SiglecH-GFP cells from BM and liver of SiglecH GFP reporter mice (46) were injected i.p. 6 or 12 h before imaging the rear side of the VAT (facing the intestine).
pDCs from SiglecH-GFP reporter mice were enriched from bone marrow and transferred via i.p. injection into ubiquitin C–tdTomato mice. Eighteen hours later, mice were anesthetized with isoflurane (Baxter; 2.5% for induction, 1–1.5% for maintenance, vaporized in an 80:20 mixture of O2 and air), VAT was exposed, and intravital microscopy was performed. The imaging system was composed of a Zeiss 780 upright microscope equipped with a 20-water immersion lens (numerical aperture 1.0; Zeiss), a Chameleon laser (Coherent) tuned to 820 or 930 nm, and with the ZEN acquisition control software. The microscope was enclosed in an environmental chamber in which anesthetized mice were warmed by heated air and the VAT was kept at 37°C with warmed PBS. Raw imaging data were processed and analyzed with Imaris (Bitplane). In addition, imaging was performed using a LaVision BioTec multiphoton TriM Scope II equipped with a Chameleon pulsed laser with the range from 800 to 1100 nm. The images were acquired by Inspector software and processed with Imaris (Bitplane).
Confocal imaging of VAT
VAT was harvested and fixed using PLP buffer (0.05 M phosphate buffer containing 0.1 M l-lysine [pH 7.4], 2 mg/ml NaIO4, and 10 mg/ml paraformaldehyde) for 6 h, then dehydrated in 30% sucrose and stained. Tissue was permeabilized, blocked, and stained in PBS containing 0.3% Triton X-100 and 1% FCS in a humidified chamber (overnight, 4°C). After washing, the tissue was mounted with GenTeal gel (Novartis) or Dako and imaged as whole mount with the Olympus FluoView BX50WI inverted microscope with 10×/0.4 and 20×/0.5 objectives or with an LSM 880 with Airyscan module (Carl Zeiss), with EC Plan-Neofluar 10×/0.3 and Plan-Apochromat 20×/0.8 objectives. Image analysis was performed using ImageJ software (National Institutes of Health) or with Zen blue software (Carl Zeiss).
Liver, spleen, LNs, and VAT were placed in TRIzol (Ambion) and minced. RNA extraction was performed with a Qiagen RNeasy kit, and cDNA synthesis was performed with an iScript kit (Bio-Rad). RT-PCR was performed on a LightCycler 480 II (Roche) using a SYBR Green QuantiFast kit (Qiagen). Relative gene expression was calculated using the ΔCt method, and the mean expression level of liver from SFD was set as 1. The following primers were designed with Roche probe library: CCL19 forward, 5′-TGTGGCCTGCCTCAGATTAT-3′, reverse, 5′-AGTCTTCCGCATCATTAGCAC-3′; CCL21a forward, 5′-TCCAAGGGCTGCAAGAGA-3′, reverse, 5′-TGAAGTTCGTGGGGGATCT-3′; CCL21b forward, 5′-TCCAAGGGCTGCAAGAGA-3′, reverse, 5′-TGAAGTTCGTGGGGGATCT-3′; CCL25 forward, 5′-GAGTGCCACCCTAGGTCATC-3′, reverse, 5′-CCAGCTGGTGCTTACTCTGA-3′; CD62E (sele) forward, 5′-TCCTCTGGAGAGTGGAGTGC-3′, reverse, 5′-GGTGGGTCAAAGCTTCACAT-3′; CD62P (selp) forward, 5′-TCCAGGAAGCTCTGACGTACTTG-3′, reverse, GCAGCGTTAGTGAAGACTCCGTAT-3′; as the housekeeping gene, 18S rRNA was used, forward, 5′-GCCGCTAGAGGTGAAATTCTT-3′, reverse, 5′-CGTCTTCGAACCTCCGACT-3′.
HFD induces systemic increase of pDCs
In humans, pDC numbers in VAT correlate with an increase in body mass index (24), suggesting that obesity may affect the homeostasis of pDCs. Therefore, we studied whether high fat intake alone was instrumental in provoking systemic alterations of the pDC profile within the body. We fed male mice for 3 wk with either HFD or SFD as control, and subsequently analyzed blood, peripheral tissue, lymphatic organs, and VAT for the presence of pDCs by flow cytometry (Supplemental Fig. 1A, 1B). The HFD mediated a gain in body weight (Fig. 1A) and increased pDC numbers in blood and tissues (Fig. 1B, Supplemental Fig. 1C–E). In LNs, the absolute pDC number was reduced, which coincided with a decreased total number of CD45+ cells in the LNs with a HFD (data not shown and Ref. 47). However, the pDC frequency in the LN revealed no change. The VAT increased 2-fold in weight after 3 wk of a HFD, and it displayed increased pDC numbers and frequency (Fig. 1C, Supplemental Fig. 1C–E). Thus, application of HFD altered pDC homeostasis throughout the body. This observation was largely independent of sex, as female mice presented a similar systemic increase in pDC number with a HFD with the exception of BM (Supplemental Fig. 2A–C). Thus, an obesogenic diet for only 3 wk fundamentally modified the presence of pDCs in VAT.
pDC localization in FALCs
Next, we studied the distribution profile of pDCs in VAT and found that they were present in CD45+ cell clusters (defined as accumulation of more than five CD45+ cells, minimal diameter of 50 µm, and minimal proximity between cells of 10 µm), so-called FALCs, which mainly localized in the peripheral area of the tip part of the epididymal VAT (Fig. 2A, 2B, blue areas). In 3-wk-old mice, only some VAT-pDCs were detectable at CD31+ blood vessels and FALCs were absent (Fig. 2C, left), whereas 8-wk-old mice presented FALCs (Fig. 2C, right), suggesting that FALCs develop in postnatal life. When mice were fed with HFD for 3 wk, FALCs significantly increased in number from ∼22 to 38 per mm3 (Fig. 2D, 2E). A 3-fold increased mean fluorescence intensity (MFI) of CD45 within the FALCs after feeding a HFD further indicated an accumulation of CD45+ cells in VAT resulting in a larger size of FALCs (Fig. 2D, 2F). Thus, a HFD promotes FALC formation and enlargement. Next, we established intravital multiphoton microscopy to uncover the mobility of pDCs in VAT in detail. Isolated SiglecH-GFP+ pDCs were transferred into mice ubiquitously expressing tdTomato to identify the tissue structure, including hematopoietic cells, blood vessels, stromal cells, and adipocytes. After 18 h, pDCs were found in peripheral areas of VAT (Supplemental Fig. 2D). Interestingly, most pDCs were present within clusters of tdTomato+ cells (Supplemental Fig. 2D, Supplemental Video 1). Again, these clusters were localized in the tip area, similar to the CD45+ clusters, the so-called FALCs, indicating the dynamics of pDC trafficking into FALCs (see (Fig. 2A). Intravital imaging confirmed that the increase of FALCs upon HFD (Fig. 2G) was accompanied by a substantial rise in VAT-resident pDCs (Fig. 2H, left), which preferentially localized within the FALCs. In this study, the abundance of pDCs in the FALCs increased from ∼18 cells per mm2 with SFD to 63 cells per mm2 with HFD (Fig. 2H, middle). The migratory speed of endogenous pDCs in VAT changed from resident-like (3 µm/min) to migratory (10 µm/min) behavior (Fig. 2H, right). For a control, we compared the migratory pattern of endogenous and transferred pDCs. There was no difference in migratory speed within these groups. This was true for application of SFD or HFD, respectively (Supplemental Fig. 2E, 2F). In addition, transferred pDCs increased in FALCs similar to endogenous pDCs when HFD was applied (Supplemental Fig. 2G, 2H). This supports the notion that pDC trafficking and accumulation within FALCs are altered by HFD.
Multistep adhesion cascade for pDC homing to the VAT requires P-selectin
Next, the recruitment of pDCs to VAT, secondary lymphoid organs, and non-lymphoid, yet metabolic active tissue, here the liver, was evaluated in mice with a SFD. pDCs were expanded, harvested, enriched, and labeled before transfer (Supplemental Fig. 3A, 3B). As selectin-mediated tethering and rolling are the first steps of the multistep adhesion cascade in postcapillary venules (48), we inhibited selectin function in short-term adoptive transfer experiments using specific function-blocking mAbs (Fig. 3A). PSGL-1, the carbohydrate ligand of P-selectin, was detectable on pDCs (Supplemental Fig. 3C) (49), and blocking P-selectin reduced immigration of pDCs to liver and LNs to some extent (Fig. 3B). The pDC immigration to VAT was completely abolished upon blocking P-selectin, indicating an absolute dependency on P-selectin for pDCs homing to VAT under steady-state conditions (Fig. 3B). In contrast, splenic pDCs did not require P-selectin for immigration (Fig. 3B). As pDCs express and employ L-selectin to home to peripheral LNs (Supplemental Fig. 3C) (50, 51), we blocked L-selectin via MEL-14 mAb treatment. Indeed, the immigration to LNs was significantly reduced (Fig. 3C), but trafficking of pDCs to all other tissues such as liver, spleen, and VAT was not affected (Fig. 3C). Moreover, we did not find any relevance of E-selectin for pDC homing in mice fed with a SFD (Fig. 3D).
HFD treatment changes the rolling activity of pDCs in VAT
To elucidate whether HFD modified the homing process, we studied pDC homing in mice with a SFD or HFD (Fig. 4A). Upon adoptive transfer of pDCs, substantial pDC homing to liver, spleen, and LNs was observed within 18 h (Fig. 4B). Strikingly, VAT of HFD mice showed more immigrated pDCs compared with SFD mice (Fig. 4B). The application of blocking Abs against P-selectin during a HFD reduced homing of pDCs to VAT, albeit in contrast to a SFD, this blockade was not complete (Fig. 4C). Interestingly, after HFD, L-selectin blockade significantly impaired homing of pDCs to VAT similar to the LNs (Fig. 4D). Moreover, E-selectin, which also did not have any relevance for pDC homing with a SFD (see (Fig. 3D), contributed to pDC immigration to VAT, as blocking of E-selectin diminished pDC homing whereas homing to spleen, liver, or LNs was unaffected (Fig. 4E). In line with these findings, the mRNA expression of CD62E specifically increased 50-fold in VAT but not in liver, spleen, and LNs upon HFD compared with a SFD, indicating that high caloric intake alters the molecular signature of VAT (Supplemental Fig. 4A).
pDC homing to VAT requires integrins and Gαi protein signaling
The next critical step in the multistep adhesion cascade is the transition from rolling to stable arrest, which is mediated by adhesion receptors of the integrin family (52). As pDCs expressed the integrins α4β1 and α4β7 (51, 53) (Supplemental Fig. 3C), we analyzed whether they mediate pDC immigration to VAT. Blocking α4β7 integrins revealed that they are required for pDC homing to VAT but not to liver, spleen, or LNs (Fig. 5A). The blockade of all α4 integrins showed that pDC homing to liver and spleen did not require α4 integrin activity (Fig. 5B) but, as expected (51), the immigration to LNs was reduced by ∼50% (Fig. 5B). pDC immigration to VAT was strictly dependent on α4 integrins, as its blockade completely abolished pDC infiltration into VAT (Fig. 5B), indicating that pDC homing to LNs and VAT required α4β1 integrins, but VAT immigration additionally relied on α4β7 integrins. Integrin adhesion requires G protein–coupled receptor (GPCR) signaling, which is sensitive to PTx (54). To decipher the involvement of GPCRs in pDC homing to VAT, competitive homing assays were performed where 50% of pDCs were treated with PTx and coinjected with 50% control pDCs (Fig. 5C, left). Indeed, inhibiting GPCR signaling blocked pDC homing to LNs and, interestingly, also to VAT (Fig. 5C, right), whereas homing to spleen and liver was not affected (Fig. 5C, right). As CCR7 is expressed by pDCs (Supplemental Fig. 3C) and represents a key factor for LN entry via HEVs (34), we performed a competitive homing assay with WT and CCR7−/− pDCs (Fig. 5D). While pDC homing to liver and spleen did not require CCR7, the infiltration to peripheral LNs was significantly reduced as expected (34). Strikingly, the immigration of pDCs to VAT was also inhibited, indicating a necessity of CCR7 signaling for pDC immigration to VAT already with SFD (Fig. 5D). Further treatment with a HFD did not change this requirement (Fig. 5D). Accordingly, qPCR analysis revealed that different CCR7 ligands, namely CCL19, CCL21a, and CCL21b, were expressed in VAT (Supplemental Fig. 4A), pointing toward a role of the CCR7/CCR7 ligand axis. The application of a HFD increased the expression of CCL21a in spleen but not significantly in VAT (Supplemental Fig. 4A). Other chemokine receptors expressed by pDCs such as CXCR3 (Supplemental Fig. 3C) and CCR9 were not essential for pDC homing to VAT, and CCL25 expression was not detected in any organ analyzed (Supplemental Fig. 4B–D). Thus, in addition to α4 integrins, homing of pDCs to VAT involved CCR7 signaling.
pDC turnover in VAT changes with HFD
In LNs, pDCs egress to the blood via efferent lymphatics in a S1P-mediated fashion (55). To analyze the number of pDCs that entered VAT, remained there, and potentially egressed to the bloodstream, we measured the number of pDCs in VAT under homeostatic conditions (Fig. 6A, c1) and after daily application of FTY720 for 7 d, resulting in S1PR internalization (Fig. 6A, c2, 6B). FTY720 led to a significant increase in number of pDCs in VAT with a SFD (Fig. 6B), which became 3-fold higher when applied with a HFD (Fig. 6B), indicating that pDC dynamics and turnover in VAT changed with a HFD. Calculating the magnitude from the cell number gained in condition c1 and c2 from SFD and HFD showed that the pDC infiltration was 3-fold higher when mice received additionally a HFD (Fig. 6D, left). To gain further insight into this process, we blocked the infiltration of pDCs to VAT by inhibiting α4β7 integrins (Fig. 6A, R, c3), which led to a decrease in pDC number. The additional treatment with FTY720 (Fig. 6A, c4) revealed an equilibrium of infiltrating and emigrating pDCs under SFD conditions (Fig. 6C, black bars), as it rescued the pDC number to the level seen with isotype control. Upon HFD this changed as the inhibition of pDC infiltration could not decrease the pDC number within the VAT as efficient as with a SFD. Furthermore, the additional blockade of the egress revealed that there was no equilibrium of incoming and emigrating cells anymore (Fig. 6C, red bars). According to the scheme in (Fig. 6A, we calculated that the pDC infiltration increased 3-fold within 1 wk of a HFD (Fig. 6D, left), although the egress did not change (Fig. 6D, right) and retention was prolonged (Fig. 6D, middle). Thus, pDC turnover changed dramatically in VAT with a high caloric diet.
HFD induces activation of pDCs in VAT
As a HFD rapidly altered recruitment and turnover of pDCs in VAT as seen in LNs during viral infection (42, 56), we analyzed whether a HFD mediated pDC activation accordingly. Indeed, pDCs in VAT expressed significantly higher levels of activation markers with a HFD compared with a SFD, namely PDC-TREM, CD69, CD86, and CD107, as measured by MFI. The tetraspanin molecule CD9 expression did not change (Fig. 7A, left), suggesting that a HFD did not simply changed pDC composition (57, 58). In contrast, the expression of activation markers of splenic pDCs showed no difference with a HFD versus SFD (Fig. 7A, right), indicating that the change of pDC phenotype upon an obesogenic diet was specific for VAT. As activation of pDCs is usually accompanied by robust IFN-I expression, we measured the expression of IFN-I by pDCs in VAT using MOB mice known to mark IFN-β production by YFP expression (43). In MOB mice fed with SFD, ∼20% of all pDCs expressed YFP (Fig. 7B), which increased to >40% with a HFD (Fig. 7B). Furthermore, the intensity of YFP increased with a HFD, indicating an increase of IFN-β expression per pDC (Fig. 7B, right). Accordingly, the absolute amount of IFN-β protein in the VAT was significantly upregulated (Fig. 7C, left). Furthermore, IFN-α was 7-fold upregulated in VAT when a HFD was fed (Fig. 7C, right). In summary, a high-caloric diet specifically activated pDCs in VAT resulting in increased expression of IFN-I. To study the functional impact of pDC trafficking and activation during HFD, we blocked P-selectin by daily injection of a function-blocking Ab (Fig. 7D). Mice receiving isotype control Ab in parallel to a HFD significantly gained body weight starting at day 4 of HFD. In contrast, mice that received a HFD and the P-selectin blocking Ab did not show increased body weight and remained at the level of SFD-treated mice (Fig. 7E). An IPGTT showed that mice treated with HFD in combination with P-selectin blocking Abs had an improved capability to systemically use glucose when compared with isotype-treated HFD-fed mice (Fig. 7F). Furthermore, whereas the weight of VAT in HFD- and isotype-treated mice increased, the treatment with P-selectin blocking Ab during HFD led to reduced weight gain of adipose tissue (Fig. 7G), suggesting that blocking P-selectin can prevent an obesogenic switch at least to some extent upon HFD in our model.
The immune system components within VAT are important and diverse regulators of metabolic processes. They can drive the onset of the metabolic syndrome but may also contribute to its prevention (6, 8, 19). Upon activation, pDCs provide a rapid IFN-I response characterizing the early reaction of the host to disturbances in homeostasis, as described for infections (59). pDCs also respond to metabolic cues as has been shown by increased pDC abundance in VAT of humans with obesity (24). Our comprehensive analysis of pDC distribution in different organs of lean and obese mice revealed a systemic pDC response, with increased numbers in blood, liver, BM, spleen, and VAT upon HFD, with the exception of the LN, as reported earlier (47). In 3-wk-old mice, VAT-resident pDCs were located at the abluminal side of small capillaries even before FALCs were detectable. Accordingly, FALCs start to develop only postnatally (18), whereas secondary lymphoid organs develop during embryogenesis (60, 61). In 8-wk-old mice, however, pDCs were preferentially detectable in FALCs where they showed a similar migratory speed as in LNs under steady-state conditions (56). As FALCs represent lymphoid tissues that orchestrate adaptive immune responses that in turn require pDCs, our results imply a prominent role of these cells in FALC and VAT biology (17, 21, 62). During a HFD, the number of FALCs increased and their volume was enlarged. Furthermore, pDCs within FALCs showed a higher density and an increased migratory speed. To investigate the biology of pDCs, we expanded the pDC population using the B16-FLT3L system, as the endogenous pDC number was too low to perform homing experiments. This approach has been widely used in earlier studies to analyze the function and homing of pDCs (33, 34). Moreover, we showed that the expression profile of adhesion molecules of FLT3L-expanded pDCs (Supplemental Fig. 3C) was similar to the molecular signature of adhesion molecules expressed by endogenous pDCs, demonstrating their suitability for our studies. The homing of pDCs to VAT required P-selectin already under normal conditions, indicating that the VAT milieu differs from other peripheral organs (63). In contrast, E-selectin was not employed by pDCs under normal conditions but with HFD. The involvement of E-selectin in pDC recruitment to VAT but not to LNs suggests that 3 wk of a HFD were sufficient to cause an inflammatory state in VAT, whereas the peripheral LN was spared. This was confirmed by the fact that we found a substantial increase of E-selectin mRNA expression specifically in VAT upon a HFD compared with a SFD but not in other organs, including liver, spleen, and LNs. Whereas P-selectin can be prestored in Weibel–Palade bodies of endothelial cells, E-selectin is expressed de novo upon stimulation by inflammatory mediators (49, 64). In addition, L-selectin was expressed by pDCs but it was not required for pDCs homing to VAT under normal conditions. However, L-selectin became important with a HFD when VAT is remodeled (65). Interestingly, the expression profile of homing molecules on pDCs, including PSGL-1, CD62L, CD49d, α4β7, and CCR7, was largely unaffected by HFD (data not shown), suggesting that rather a switch of VAT toward an inflammatory milieu than a pDC-intrinsic mechanism mediates increased pDC homing to and retention in VAT (20, 66).
Two of the most relevant adhesion molecules for leukocyte homing are α4β1 and α4β7 integrins, and we confirmed their expression by pDCs (51, 67). Whereas homing of pDCs to LN involved α4β1 integrins only, the VAT infiltration additionally relied on α4β7 integrins, usually employed for pDC homing to the gut (68). Integrin activation requires Gαi-mediated PTx-sensitive signaling and, indeed, the treatment with PTx prevented pDC infiltration to VAT and LNs, indicating an essential Gαi signaling for this recruitment process.
pDCs expressed CCR7, which signals via Gαi, and identical to LN homing, pDCs required CCR7 signaling to infiltrate VAT during SFD and HFD. Gene expression analyses revealed expression of CCR7 ligands within spleen, LNs, and VAT, suggesting a functional role of this chemokine receptor axis under both conditions. Although pDCs express CCR9 (33, 69), mRNA expression of its ligand CCL25 could not be identified. However, chemokines do not require local gene expression, as they can be transported and presented via glycosaminoglycans, but CCR9 engagement for pDC homing to VAT, spleen, liver, and LNs as shown for thymus and gut (33, 70) could not be observed. Similarly, CXCR3 signaling was not involved in pDC homing to VAT, but may be required for inner tissue migration as shown for intranodal positioning within the LN (71).
Studying pDC dynamics in VAT revealed that a HFD led to a rise in pDC number in VAT, and blocking the egress with FTY720 further resulted in a 3-fold increase, indicating that pDCs actively left the VAT via a S1P-dependent mechanism. Moreover, our data indicated an equilibrium of pDC influx and efflux under homeostatic conditions with SFD. A HFD increased the number of infiltrating pDCs and prolonged their retention, resulting in a rise of total pDC number in VAT. In addition, pDCs showed a higher expression of the early activation marker CD69 with a HFD. For activated T cells in the LN, it has been demonstrated that CD69 physically interacts with S1PR, resulting in downregulation of S1PR surface expression and thereby blocking S1P-mediated T cell egress (72). Although obese humans present elevated S1P levels in the blood (73), we did not find an increased egress of pDCs from VAT, which may be due to their upregulated CD69 expression with HFD. In LNs, higher retention times of pDCs coincide with a change to an activated phenotype (55). Indeed, VAT pDCs displayed an activated phenotype with a HFD. Moreover, 20% of all pDCs expressed IFN-I in the steady state, which was similar to pDCs in other peripheral tissues (43), whereas obesogenic diet increased IFN-I expression by pDCs up to 40% in VAT. IFN-I in turn contributes to the polarization of M2 macrophages to M1 macrophages, thereby promoting proinflammatory cytokine release and fueling meta-inflammation, leading to systemic insulin resistance (24).
Thus, pDCs reside preferentially in FALCs of VAT in adult mice under steady-state conditions and display increased homing to and retention in VAT after 3 wk of HFD, resulting in an accumulation of pDCs caused by a shift of VAT toward an inflammatory state as outlined by dramatically increased E-selectin expression. In vivo imaging showed an increase of migratory speed after high caloric intake, indicating a transition from a rather stationary to scanning mode within the FALCs. The dramatic disturbances of trafficking with a HFD coincide with an activation of pDCs and increased IFN-I expression in VAT, suggesting a potential mechanistic link by which IFN-I may render VAT into a chronic inflammatory tissue (41).
This hitherto unknown modulation of pDC homeostasis by a HFD may shed new light on the impact of our daily diet on the complex interplay between the immune system and adipose tissue, a topic that is of great clinical importance. Interestingly, the genetic abrogation of IFN-I signaling prevents the development of obesity (22), identifying pDCs as a potentially powerful target to treat obesity. Recently, IFN-I from pDCs in VAT has been demonstrated to fuel the loss of VAT-resident regulatory T cells that results in an obesogenic phenotype (41). Furthermore, the blockade of pDC infiltration into VAT by P-selectin blockade in combination with a HFD prevented gain of body weight and improved glucose tolerance at least during the first week of high caloric intake. Thus, pDC trafficking may represent an interesting target for therapeutic intervention in obesity by preventing weight gain and the metabolic syndrome.
We thank Michael Flynn for assistance performing animal experiments, Wolfgang Kastenmüller for providing reagents and the ubiquitin C–tdTomato mice, and Stefanie Scheu for providing mice. We acknowledge the Core Facility Flow Cytometry at the Biomedical Center, Ludwig-Maximilians-Universität München for providing FACS equipment and the Multiphoton Imaging Core Facility at the Walter Brendel Center for Experimental Medicine.
This work was supported by Deutsche Forschungsgemeinschaft Grants STU 528/1-1 (to S.S.), SFB 1054 B03 (to T.B.), SFB 914 A06 (to S.S. and T.B.), SFB 914 A02 (to B.W. and D.M.-B.), SFB 914 B10 (to L.W.), SFB 914 Z01 (to H.I.-A.), and SFB 914 A10 (to C.S.), Friedrich Baur-Stiftung (to S.S.), and by National Institute of Allergy and Infectious Diseases Grant R01 AI155865 (to U.v.A.).
S.S. designed the study, analyzed the data, and wrote the paper. S.S., H.I.-A., S.N., C.G., D.v.d.H., D.S., L.L., and S.E. performed experiments. T.B., M.C., and C.S. provided reagents and mice; L.W. provided reagents, and D.M.-B. and B.W. contributed in paper writing and provided reagents. U.v.A. designed the study, provided reagents and mice, and was involved in writing the paper. All authors discussed the results and commented on the manuscript.
The online version of this article contains supplemental material.
Abbreviations used in this article:
fat-associated lymphoid cluster
G protein–coupled receptor
high endothelial venule
type I IFN
i.p. glucose tolerance test
mean fluorescence intensity
messenger of IFN-β
standard fat diet
sialic acid binding Ig-like lectin H
visceral adipose tissue
The authors have no financial conflicts of interest.