Cellular source of hypothalamic macrophage accumulation in diet-induced obesity

C57BL/6 (C57) male mice were purchased from Orient Bio (Seongnam, Korea). Lysozyme M (LysM)-cre mice were cross-mated with mice harboring an enhanced green fluorescent protein (GFP) reporter allele with an upstream loxP-flanked STOP cassette (both from Jackson Laboratory, Bar Harbor, MA) to generate mice with myeloid cell-specific GFP expression (LysM^GFP mice). The animals were fed either a standard chow diet (CD; Samyang, Seoul, Korea) or a high-fat diet (HFD; 58% fat, Research Diet Inc., New Brunswick, NJ) and maintained under a controlled temperature (22 ± 1 °C) and a 12-h light-dark cycle (lights on 8 AM) with free access to food and water.

Mice were anesthetized with 40 mg/kg Zoletil® (zolazepam and tiletamine) and 5 mg/kg Rompun® and then perfused with 50 ml saline followed by 50 ml 4% paraformaldehyde (PFA) via the left ventricle. Whole brains were collected, fixed with 4% PFA for 24 h, and dehydrated in 30% sucrose solution until the brains sank to the bottom of the container. Coronal brain sections (20 μm thick) including the hypothalamus were then obtained using a cryostat (Leica, Wetzlar, Germany). Sections were stored at − 70 °C until immunostaining. Hypothalamic slices were permeabilized in 0.5% phosphate-buffered saline with Tween 20 for 5 min and blocked with 5% normal donkey serum at room temperature (RT) for 1 h and then incubated with primary antibodies against GFP (Aves Labs, #GFP-1010, 1:1000), the microglia marker Iba1 (Abcam, #Ab5076, 1:500), or vascular marker PECAM1 (BD Pharmingen, #550274, 1:200) at 4 °C for 16 h and then at RT for 1 h. After washing, slides were incubated with the appropriate Alexa-Flour 488- or 633-conjugated secondary antibodies (Invitrogen, 1:1000) at RT for 1 h. To stain the fenestrated vasculature, fresh-frozen hypothalamic slices were fixed in 4% PFA for 30 min, blocked with 5% normal donkey serum, and then incubated with anti-MECA32 antibody (BD Pharmingen, #550563, 1:200). The incubation protocol for primary and secondary antibodies was performed as described above. For nuclear staining, slides were treated with 4′,6-diamidino-2-phenylindole (DAPI, Sigma, #D9564, 1:10,000) for 10 min before mounting. Immunofluorescence was detected using confocal microscopy (Carl Zeiss 780, Germany). Quantitation of fluorescence intensity of Evans blue and MECA32 was performed throughout the entire rostro-caudal axis of the ARC using ImageJ.

To test hypothalamic vascular permeability, 3% Evans blue diluted in saline (Sigma, #E2129, 10 ml/kg) was administered via a tail vein to mice fed with a CD or an HFD for the indicated periods (n = 10). Whole brains were collected at 20 min post-injection and frozen with pre-chilled isopentane. Hypothalamic slices (20 μm thick) were sectioned using a cryostat, washed three times with phosphate-buffered saline, and incubated with DAPI prior to mounting. Evans blue fluorescence (excitation at 620 nm, emission at 680 nm) was imaged using a confocal microscope. Alternatively, the mediobasal hypothalamus (MBH) was harvested 20 min after the intravenous injection of Evans blue, weighed, and incubated in 300 μl formamide (AMRESCO, #0606-500 ml) at 70 °C for 24 h. Samples were centrifuged at 16,000×g for 45 min, and supernatants were collected to measure the optical density at 630 nm using a spectrophotometer (Eppendorf, Germany). Optical density readings were normalized to tissue weight. Hypothalamic vascular permeability was also tested using Alexa-Fluoro 680-conjugated albumin (Molecular Probes, #A34787, 10 mg/kg diluted in saline) (n = 3). Fluorescent-labeled albumin was administered via a tail vein. Whole brains were collected 20 min after injection, immediately frozen, sliced, and immunostained for PECAM1 as described above to confirm the extravasation of fluorescence-conjugated albumin. Immunofluorescence images were captured using a confocal microscope. All vascular permeability tests were performed during the early light period under free-feeding conditions.

Parabiont pairs of 13-week-old, weight-matched LysM^GFP mice and C57 male mice were surgically conjoined (6 pairs). Under anesthesia with Zoletil® and Rompun®, body skin from the lateral side of the parabionts was opened from the elbow and knee and sutured with a parabiont pair. Parabiosis success was confirmed by detecting the presence of LysM^GFP monocytes in the peripheral blood of C57 mice using flow cytometry (BD FACS Canto™). These blood samples were collected from the orbital vein of parabionts at 1 week post-surgery. At 6 weeks after parabiosis, animals were separated and perfused with 4% PFA. Brains were collected to observe LysM^GFP cells in the hypothalamus and choroid plexus of parabionts using GFP immunohistochemistry. Mice were fed either a CD or an HFD for 12 weeks (6 weeks before surgery and 6 weeks after surgery) before being sacrificed.

To generate bone marrow chimeric mice, 7-week-old C57 male mice were irradiated (6.5 Gy twice with 4-h intervals) with a head shield under anesthesia (n = 5). Twenty-four hours after irradiation, they were injected via the tail vein with bone marrow cells (3 × 10⁶ cells) harvested from the femur of LysM^GFP mice. At 6 weeks post-transplantation, we confirmed successful transplantation by detecting the presence of LysM^GFP monocytes in the peripheral blood using flow cytometry. Animals showing successful BMT were fed with a CD or an HFD for 12 or 30 weeks before being sacrificed. Brain, liver, and leg skeletal muscle (quadriceps femoris) were collected for GFP and DAPI immunostaining after cardiac perfusion with 4% PFA.

Whole brains of embryos and neonates born to CD-fed LysM^GFP mothers were collected at the indicated ages (n = 2~5, from 2~3 different litters each time point). Mice younger than 7 days old were sacrificed without cardiac perfusion. Brains were fixed in 4% PFA for 24 h, dehydrated, sliced, and then subjected to GFP and Iba1 immunostaining as described above.

We hypothesized that hypothalamic macrophage accumulation in DIO mice may result from the enhanced recruitment of circulating monocytes. Most brain regions including the hypothalamus are immune-privileged as they are separated from the blood circulation by the blood-brain barrier (BBB) [19]. Thus, BBB disruption may be a prerequisite for the recruitment of circulating immune cells to the sites of neuroinflammation.

To test the vascular permeability of the hypothalamic ARC in DIO mice, we first injected a vessel-impermeable dye, Evans blue, into the tail veins of lean and obese mice fed on a CD or an HFD for the indicated periods. In the lean mice, Evans blue fluorescence was observed inside the hypothalamic blood vessels, although it leaked into the parenchyma in the circumventricular organ ME (Fig. 1a). In contrast, increased Evans blue leakage was observed in the ARC of 10-week HFD-fed mice. A time-course quantification study also demonstrated an increased Evans blue content in the MBH of obese mice on an HFD for 4 and 10 weeks (Fig. 1b). Consistently, increased extravasation of fluorescence-conjugated albumin was observed in the ARC of obese 20-week HFD-fed mice whereas no leakage was found in the lean controls (Fig. 1c). These data indicated that the BBB may become more permeable in the ARC of DIO mice. However, it is also possible that fluorescence-labeled albumin may infiltrate from permeable capillaries in the ME, which is in close proximity to the ARC.

We also stained hypothalamic slices with the fenestrated vessel marker MECA32 [10]. CD-fed lean mice showed a restricted distribution of MECA32-expressing permeable vessels in the external zone of the ME and no MECA32 expression in the ARC (Fig. 1d). However, ARC MECA32 expression significantly increased after 2 weeks of HFD feeding (Fig. 1d), indicating that the ARC microvasculature becomes permeable upon persistent HFD consumption.

In the hypothalamus, macrophages are mostly located in the perivascular space under normal conditions [14]. ARC perivascular macrophages (PVM) adopt a tubular shape without cellular processes [14]. However, upon activation by chronic exposure to HFD, they undergo morphological changes that include an enlarged body and branched processes [14]. MECA32 and GFP double immunostaining revealed activated LysM^GFP macrophages around the MECA32⁺ ARC vessels in 20-week HFD-fed LysM^GFP mice (Fig. 1e). Taken together, these data demonstrated increased vascular permeability and activated PVMs around permeable vessels in the ARC of DIO mice.

To directly demonstrate enhanced recruitment of circulating monocytes to the ARC via permeable vessels, we performed a parabiosis experiment in which the circulation of LysM^GFP mice was connected to that of C57 mice by conjoining the frank skin (Fig. 2a). LysM^GFP partner-driven LysM^GFP cells were detected by FACS analysis in the blood of C57 mice at 6 weeks post-surgery (Fig. 2b), confirming that the parabiosis had been successful. Following a 12-week HFD, the parabionts were sacrificed and their brains were collected to determine whether LysM^GFP cells had infiltrated the hypothalamus of the C57 mice. In LysM^GFP mice, GFP-expressing cells were frequently found in the choroid plexus and in the hypothalamic ARC and ME (Fig. 2c). Of note, LysM^GFP cells were also detected in the choroid plexus of C57 parabionts fed on an HFD for 12 weeks (Fig. 2c). However, we could not detect any LysM^GFP cells in the hypothalamus of the C57 mice. These results suggested that LysM^GFP monocytes infiltrate the choroid plexus but may not enter the hypothalamus of DIO mice despite its vascular hyperpermeability.

We next conducted a BMT experiment to validate our parabiosis results. We established a BM chimera model by transplanting the BM from LysM^GFP mice to C57 mice. One week prior to this BMT, C57 mice received whole body irradiation with head shielding to prevent an artificial BBB breakdown by X-ray irradiation (Fig. 3a). BMT success was confirmed by the presence of LysM^GFP cells in the blood of BMT recipients at 6 weeks post-operation (Fig. 3b). Mice were fed with a HFD for either 12 weeks or 30 weeks before being sacrificed to observe LysM^GFP cells in the brain, liver, and skeletal muscle of the recipients. LysM^GFP cells were frequently found in the liver, muscle, and choroid plexus of the recipients that were fed with a HFD for 12 weeks or 30 weeks (Fig. 3c). Under both HFD feeding conditions, a few LysM^GFP cells were also found along the meningeal lining covering the ME (Fig. 3c). However, no LysM^GFP cells could be detected inside the hypothalamus of these animals, including the ARC. These findings strongly supported the notion that circulating monocytes do not significantly contribute to hypothalamic macrophage pool expansion under chronic HFD conditions.

Given the observed lack of any definite recruitment of LysM^GFP circulating monocytes to the hypothalamus, we next investigated whether these cells populate the hypothalamus during the developmental period. The brains of LysM^GFP mice were collected on embryonic days (E) 16.5 and 18.5 and on postnatal days (P) 1, 7, 14, and 28 for double immunostaining of GFP and the microglial marker Iba1. At E16.5, LysM^GFP cells were barely detectable in the hypothalamic area, although a small number could be observed along the meningeal lining (Fig. 4a, b). In contrast, many Iba1-expressing microglia were evident in the hypothalamic area at E16.5, which were thought to be yolk sac derived (Fig. 4a, c). Of note, a hypothalamic pool of LysM^GFP cells started to increase from E18.5 and dramatically expanded immediately after birth (Fig. 4a, b). During the mid-lactation period (P14), a huge expansion of LysM^GFP cells was observed in the hypothalamus, with an activated microglia-like morphology (Fig. 4a). These results indicated that these cells may be highly activated even when nourished by CD-fed dams. At P28, the hypothalamic LysM^GFP pool was profoundly reduced and had adopted the adult pattern of distribution (Fig. 4a, b), as reported previously [14]. In sharp contrast to the dynamic changes in the hypothalamic LysM^GFP cellular pool, the Iba1-alone expressing microglia pool maintained relatively consistent numbers during early postnatal life (Fig. 4a, c). Interestingly, a significant proportion of the meningeal and ME LysM^GFP cells coexpressed Iba1, whereas most of the LysM^GFP cells inside the hypothalamus did not (Fig. 4a).