Deficiency in adipocyte chemokine receptor CXCR4 exacerbates obesity and compromises thermoregulatory responses of brown adipose tissue in a mouse model of diet-induced obesity

Abstract

The chemokine receptor CXCR4 is expressed on adipocytes and macrophages in adipose tissue, but its role in this tissue remains unknown. We evaluated whether deficiency in either adipocyte or myeloid leukocyte CXCR4 affects body weight (BW) and adiposity in a mouse model of high-fat-diet (HFD)-induced obesity. We found that ablation of adipocyte, but not myeloid leukocyte, CXCR4 exacerbated obesity. The HFD-fed adipocyte-specific CXCR4-knockout (AdCXCR4ko) mice, compared to wild-type C57BL/6 control mice, had increased BW (average: 52.0 g vs. 35.5 g), adiposity (average: 49.3 vs. 21.0% of total BW), and inflammatory leukocyte content in white adipose tissue (WAT), despite comparable food intake. As previously reported, HFD feeding increased uncoupling protein 1 (UCP1) expression (fold increase: 3.5) in brown adipose tissue (BAT) of the C57BL/6 control mice. However, no HFD-induced increase in UCP1 expression was observed in the AdCXCR4ko mice, which were cold sensitive. Thus, our study suggests that adipocyte CXCR4 limits development of obesity by preventing excessive inflammatory cell recruitment into WAT and by supporting thermogenic activity of BAT. Since CXCR4 is conserved between mouse and human, the newfound role of CXCR4 in mouse adipose tissue may parallel the role of this chemokine receptor in human adipose tissue.—Yao, L., Heuser-Baker, J., Herlea-Pana, O., Zhang, N., Szweda, L. I., Griffin, T. M., Barlic-Dicen, J. Deficiency in adipocyte chemokine receptor CXCR4 exacerbates obesity and compromises thermoregulatory responses of brown adipose tissue in a mouse model of diet-induced obesity.

Obesity is an independent risk factor for insulin resistance, type 2 diabetes, and cardiovascular disease. The root cause of obesity is a prolonged imbalance between caloric intake and energy expenditure that leads to increased lipid storage and adipose tissue expansion. Adipose tissue occurs in white and brown forms that serve different functions (1).

The main role of brown adipose tissue (BAT) is adaptive thermogenesis, the process of regulated heat production in response to the environmental temperature or diet. There are 3 subcategories of adaptive thermogenesis. Cold exposure induces shivering thermogenesis in skeletal muscle and nonshivering thermogenesis in brown adipocytes. Overfeeding triggers diet-induced or metabolic thermogenesis, which allows for excess energy received in the form of food to dissipate as heat, and thereby this form of adaptive thermogenesis prevents obesity (2).

White adipose tissue (WAT) provides an important energy depot in the form of stored lipids (1), and it serves as an endocrine organ that produces adipokines, which have multiple effects at both local and systemic levels. Excessive fat uptake results in overproduction and secretion of signals that recruit inflammatory cells into WAT, triggering low-grade chronic inflammation that is mediated by the cells of innate and adaptive immune systems (3, 4).

In lean adipose tissue, the main resident immune cell subtypes are the alternatively activated adipose tissue macrophages (M2 ATMs), regulatory T (T_reg) cells, and T helper 2 (T_h2) cells. These anti-inflammatory leukocytes produce interleukin 10 (IL-10) and transforming growth factor β (TGF-β), which maintain adipose tissue homeostasis. Aberrant adipose tissue expansion triggers an influx of classically activated ATMs (M1 ATMs), neutrophils, and natural killer cells. In addition to innate immune cells, adaptive response immune cells, including CD8 effector memory and B cells, are also increased in the adipose tissue during the course of obesity (4).

The chemokine system, composed of chemokines and chemokine receptors, directs inflammatory leukocytes into obese adipose tissue and is therefore thought to be the promoter of obesity-induced adipose tissue inflammation. Chemokines support cell recruitment by interacting with their cognate receptors expressed on leukocytes (5). Chemokines, including CCL2, -5, -7, -8, -11, and -13 and CXCL5, -8, and -10, are up-regulated in different depots of adipose tissue. Serum levels of these chemokines are dramatically increased in obese vs. lean individuals. Expression of the chemokine receptors CCR1, -2, -3, and -5 is elevated on inflammatory cells in omental and subcutaneous adipose tissues of obese patients (6). In mice, targeted deletion of Ccr2 decreases ATM content and adipose tissue inflammation and inhibits insulin resistance (7). Furthermore, Cxcr2^−/− bone marrow chimeras show decreased obesity-induced inflammation and are partially protected from disorders of glucose metabolism (8). Moreover, CCR5-mediated signaling in the adipose tissue is thought to maintain obesity-induced inflammation and insulin resistance (9). The chemokine receptor CXCR4 in adipose tissue is expressed on adipocytes and ATMs (10); however, its role in fat tissue remains unknown.

CXCR4 signals upon ligation of its cognate chemokine CXCL12. This chemokine receptor is unique because it is expressed in a wide variety of cell types, not only on leukocytes, as is the case with most other chemokine receptors. CXCR4 coordinates the cell trafficking and homing that are essential during embryonic cardiac and cerebellar development and for homeostasis and function of the immune and stem cell systems (11). The adipose tissue CXCR4 expression pattern suggests that CXCR4 controls obesity-induced leukocyte recruitment, adipose tissue inflammation, homeostasis, and functional responses of adipocytes. Thus, we inactivated CXCR4 in adipocytes and myeloid leukocytes, which are ATM precursors, and examined how CXCR4 deficiency in either cell type affects body weight (BW) and adiposity in a mouse model of high-fat-diet (HFD)-induced obesity.

We present evidence that ablation of adipocyte, not myeloid leukocyte, CXCR4 exacerbated HFD-induced obesity, which was not a result of hyperphagia but was associated with increased adiposity and WAT and BAT hypertrophy. We also show that the obese phenotype in mice lacking CXCR4 specifically in adipocytes [adipocyte-specific CXCR4-knockout (AdCXCR4ko) mice] resulted in increased proinflammatory leukocyte content in WAT. In contrast, no significant difference in the number and immunophenotype of ATMs and lymphocytes was observed in WAT of obesogenic diet-fed mice lacking CXCR4 in myeloid leukocytes [myeloid leukocyte-specific CXCR4 knockout (MyeCXCR4ko) mice] and wild-type (WT) C57BL/6 controls. Interestingly, HFD-fed AdCXCR4ko mice had significantly lower metabolic rates and strikingly different BAT. Notably, inactivation of adipocyte CXCR4 prevented the HFD-induced increase in expression of uncoupling protein 1 (UCP1), which acts as the main regulator of cold- and diet-induced (metabolic) adaptive nonshivering thermogenesis that, during exposure to cold or chronic overeating, increases energy expenditure to maintain body temperature or protect from obesity (12–16). The lack of UCP1 up-regulation in BAT of HFD-fed AdCXCR4ko mice was indeed associated with the inability of these mice to maintain body temperature when exposed to cold. Thus, our data provide important information on a unique and previously unknown role of CXCR4 in adipose tissue. Our study suggests that contrary to other chemokine receptors, which promote obesity by supporting adipose tissue inflammation, CXCR4 limits it by preventing excessive inflammatory leukocyte influx into WAT and by supporting an increase in the thermogenic response of BAT.

MATERIALS AND METHODS

Experimental model

C57BL/6 mice mimic development of obesity in humans; when fed an HFD (i.e., 60% kcal HFD), these mice gain BW and develop hyperinsulinemia and hyperglycemia. However, when the strain is fed a control diet (CD; 10% kcal), they remain lean without metabolic abnormalities. A 60% kcal HFD is usually fed to expedite development of obesity. If C57BL/6 mice are fed this diet, the increase in BW relative to controls fed the 10% kcal CD becomes noticeable after 4–8 wk of HFD feeding, a significant BW gain occurs after 8–12 wk on an HFD, and after 16–22 wk of HFD feeding, mice typically exhibit a 20–30% increase in BW compared with the weight increase in the CD-fed animals. If a 20–30% increase in BW is accompanied by adipocyte hypertrophy, expansion of WAT, and fat deposition in the abdominal cavity, the mice are considered obese (17).

Animals

LoxP-floxed Cxcr4 exon 2 (Cxcr4^f/f) strain (C57BL/6 background); fatty acid-binding protein 4–Cre recombinase (Fabp4-cre) transgenic mice expressing Cre recombinase under the control of mouse Fabp4 adipocyte promoter (C57BL/6 background); lysozyme M–Cre recombinase (LysM^Cre) mice (C57BL/6 background) that express Cre recombinase under the control of the M lysozyme locus in myeloid cells; and WT C57BL/6 controls were all obtained from The Jackson Laboratory (Bar Harbor, ME, USA).

Cre recombinase is detected in BAT and WAT in Fabp4-cre mice. Therefore, crossing of Fabp4-cre mice with a strain containing a loxP site–flanked sequence of interest results in deletion of the floxed allele in adipose tissues (18). We crossed the Cxcr4^f/f mice with the Fabp4-cre mice to obtain AdCXCR4ko mice in the F2 generation (Supplemental Fig. S1). The Cxcr4^f/f mice were also crossed with the LysM^Cre strain, which supported deletion of the floxed allele in myeloid leukocytes, including granulocytes, monocytes, and mature macrophages (19), to obtain MyeCXCR4ko mice in the F2 generation (Supplemental Fig. S1).

To determine how deficiency in adipocyte or myeloid leukocyte CXCR4 affects development of obesity, AdCXCR4ko, MyeCXCR4ko, and C57BL/6 mice of both genders were fed, starting at 4 wk of age, either the 10% kcal CD (Harlan Teklad, Indianapolis, IN, USA; TD.06416) or the 60% kcal HFD (Harlan Teklad; TD.06414) for up to 24 wk, and BW, adiposity, food consumption, and metabolic rates were evaluated. All animal care and experimental protocols were approved by the Institutional Animal Care and Use Committee of the Oklahoma Medical Research Foundation.

Immunohistochemistry and immunofluorescence

WAT, including subcutaneous and visceral fat pads (mesenteric, retroperitoneal, and epididymal/parametrial gonadal), and BAT from the AdCXCR4ko and MyeCXCR4ko mice and the C57BL/6 controls fed CD or HFD for up to 24 wk were excised, fixed with 4% paraformaldehyde, embedded in paraffin, and serially sectioned. The sections were stained with hematoxylin and eosin. CXCR4 expression in adipose tissue was evaluated on tissue sections stained with primary rabbit anti-CXCR4 antibody (Santa Cruz Biotechnology, Dallas, TX, USA) or isotype control (IC) IgG (Santa Cruz Biotechnology) and secondary biotinylated goat anti-rabbit IgG (Santa Cruz Biotechnology), followed by incubation in streptavidin/horseradish peroxidase (HRP) and diaminobenzidine (Life Technologies, Grand Island, NY, USA). The tissue sections were examined under a light microscope, and images were obtained with an AxioCam MRC 12-bit color digital camera (Zeiss, Thornwood, NY, USA).

Paraformaldehyde-fixed WAT and BAT were also loaded onto 20% sucrose and embedded into optimal cutting temperature compound, frozen at −80°C, serially sectioned, and costained with rabbit polyclonal anti-CXCR4 (Santa Cruz Biotechnology), polyclonal goat anti-mouse Fabp4 (R&D Systems, Minneapolis, MN, USA), or monoclonal rat anti-mouse CD68 (AbD Serotec, Raleigh, NC, USA) antibodies, followed by incubation with the respective secondary donkey anti-rabbit or anti-goat Alexa Fluor 488– or donkey anti-rat Alexa Fluor 568–conjugated antibodies (Life Technologies). CXCL12 expression in BAT was detected by using primary rabbit polyclonal anti-CXCL12 antibody (Santa Cruz Biotechnology) and the secondary donkey anti-rabbit Alexa Fluor 488 antibody (Life Technologies).

UCP1 expression in BAT from AdCXCR4ko and C57BL/6 controls fed the CD or the HFD was determined on tissue sections stained with rabbit polyclonal anti-UCP1 antibody (Abcam, Cambridge, MA, USA) and secondary goat anti-rabbit IgG Alexa Fluor 568 antibody (Life Technologies). Images were collected with a C1 confocal system on a TE2000U microscope (Nikon, Belmont, CA, USA), with computer-controlled lasers.

Analysis of leukocyte counts in peripheral blood

Peripheral blood was obtained from age- and gender-matched AdCXCR4ko, MyeCXCR4ko, and C57BL/6 mice, and the absolute numbers of white blood cells, lymphocytes, monocytes, neutrophils, platelets and red blood cells were evaluated with a Hemavet (Drew Scientific, Dallas, TX, USA).

Flow cytometry

Visceral mesenteric, retroperitoneal, and epididymal/parametrial gonadal WAT from AdCXCR4ko, MyeCXCR4ko, and WT C57BL/6 mice fed the CD or the HFD for 24 wk were excised and digested with collagenase type I (Worthington Biochemical, Lakewood, NJ, USA) at 37°C for 1 h, and stromovascular cells were separated from mature adipocytes by centrifugation at 200 g for 10 min. CXCR4 expression on ATMs was detected by costaining mononuclear cells with monoclonal antibodies directed against mouse macrophage marker F4/80 (BD Biosciences, San Jose, CA, USA) and CXCR4 (eBioscience, San Diego, CA, USA). M1 and M2 ATM contents were determined by staining stromovascular cells with monoclonal anti-mouse F4/80, CD11b (eBioscience), CD206 (AbD Serotec), and CD11c (BD Biosciences) antibodies. Dead cells were excluded by propidium iodide (PI; Life Technologies). The gate was set on F4/80⁺CD11b⁺ cells (see Fig. 5), and the percentage and absolute number of F4/80⁺CD11b⁺CD206⁺CD11c⁻ and F4/80⁺CD11b⁺CD206⁻CD11c⁺ cells indicating M2 and M1 ATMs, respectively, were determined by flow cytometry. To evaluate the presence of CD8⁺ and CD4⁺ T lymphocytes, viable stromovascular cells were costained with CD3ε (BioLegend, San Diego, CA, USA) and anti-mouse CD8 (eBioscience) or CD4 (eBioscience) monoclonal antibodies. Cells in stromovascular fraction were also costained with CD19-directed (BD Biosciences) and B220-directed (BioLegend) antibodies, to determine B lymphocytes in WAT. Flow cytometry was performed on a BD LSRII system (BD Biosciences), correcting for nonspecific staining with isotype antibody controls. FlowJo software (Tree Star, Ashland, OR, USA) was used for data analysis.

Figure 5.

Adipocyte CXCR4 deficiency alters ATM and lymphocyte contents in WAT. WT C57BL/6 control (n=15), AdCXCR4ko (n=15), and MyeCXCR4ko (n=15) mice were fed an HFD for 24 wk and then euthanized. Blood was collected and plasma separated. Visceral mesenteric, retroperitoneal, and epididymal/parametrial gonadal WAT pads were excised and digested with collagenase type I. Stromovascular cells were separated from mature adipocytes and stained for surface marker CD11b; macrophage-specific marker F4/80; M1 marker CD11c; and M2 markers CD206, CD3, CD4, CD8, CD19, and B220. C, F, H, J) Cells were analyzed by flow cytometry; representative results are shown. Numbers in quadrants indicate percentage of cells with the indicated immunophenotype. Dead cells were excluded by PI stain (top center panels). FSC, forward scatter; SSC, side scatter; PE, phycoerythrin. A, B) Plasma levels of CCL2 (A) and CCL5 (B) were measured in all 3 strains by ELISA. C) Quantification of ATM phenotypes in WAT. F4/80⁺CD11b⁺ double-positive cells (top right panel) were evaluated for expression of CD11c or CD206 in WT (middle left panel), AdCXCR4ko (middle center panel), and MyeCXCR4ko (middle right panel) mice. Bottom panels: IC IgG stains. D, E) Absolute numbers of F4/80⁺CD11b⁺CD206⁻CD11c⁺ M1 ATMs (D) and F4/80⁺CD11b⁺CD206⁺CD11c⁻ M2 ATMs (E) per gram of visceral WAT from WT, AdCXCR4ko, and MyeCXCR4ko mice fed the HFD. F) Evaluation of CD3⁺CD8⁺ T lymphocytes in WAT. Stromovascular fraction was costained for CD3e and CD8; the percentage of CD3⁺CD8⁺ cells is shown for WT (bottom left panel), AdCXCR4ko (bottom center panel), and MyeCXCR4ko (bottom right panel) mice. Top right panel: IC IgG stains. G) Absolute numbers of CD3⁺CD8⁺ T lymphocytes per gram of visceral WAT from WT, AdCXCR4ko, and MyeCXCR4ko mice fed the HFD. H) Determination of CD3⁺CD4⁺ T lymphocytes in WAT. Stromovascular fraction was costained for CD3e and CD4, and percentage of CD3⁺CD4⁺ cells is shown for WT (bottom left panel), AdCXCR4ko (bottom center panel), and MyeCXCR4ko (bottom right panel) mice. Top right panel: IC IgG stains. I) Absolute numbers of CD3⁺CD4⁺ T lymphocytes per gram of visceral WAT from WT, AdCXCR4ko, and MyeCXCR4ko mice fed the HFD. J) Quantification of B lymphocytes in WAT. Stromovascular fraction was costained for CD19 and B220, and percentage of CD19⁺B220⁺ cells is shown for WT (bottom left panel), AdCXCR4ko (bottom center panel), and MyeCXCR4ko (bottom right panel) mice. Top right panel: IC IgG stains. K) Absolute numbers of CD19⁺B220⁺ B lymphocytes per gram of visceral WAT from WT, AdCXCR4ko, and MyeCXCR4ko mice fed the HFD.

Glucose tolerance test (GTT) and insulin tolerance test (ITT)

For the GTT, mice were denied access to food overnight, then injected intraperitoneally with glucose (1 g/kg BW), and tail blood glucose was measured (Glucometer Contour; Bayer, Mishawaka, IN, USA) at 0, 30, 60, 90, and 120 min after injection. For the ITT, the mice were denied access to food for 5 h and then injected intraperitoneally with insulin (0.75 U/kg BW; Humulin; Eli Lilly, Indianapolis, IN, USA). Tail blood glucose concentrations were measured at 0, 30, 60, and 120 min after injection.

Cold tolerance test

Mice were housed singly and unrestrained, and had free access to food and water. Body temperature readings were conducted on conscious mice with a MicroTherma thermometer with a rectal probe (Braintree Scientific, Braintree, MA, USA). Body temperature was measured at ambient temperature of 25°C and at 30, 60, 90, 120, 150, and 180 min after the mice were moved to a 4°C cold room.

RNA quantification

RNA was extracted by RNeasy (Qiagen, Valencia, CA, USA) and reverse transcribed with RETROscript (Life Technologies). Relative target quantification was calculated with the 2^−ΔΔCT method and normalized to β-actin. Primer sequences were as follows: Cxcr4, forward 5′-CTGCCCACCATCTACTTCATC-3′ and reverse 5′-CGTCATGCTCCTTAGCTTCTT-3′; nuclear respiratory factor 1 (Nfr1), forward 5′-GGTGAAATAAGCCTCCCGATAG-3′ and reverse 5′-TGAGGCAGTTTAGACAGAATGG-3′; cytochrome c oxidase 4 (Cox4), forward 5′-AGTTGTACCGCATCCAGTTT-3′ and reverse 5′-GGCCATACACATAGCTCTTCTC-3′; ATP synthase 5β (ATP5b), forward 5′-CTCAGAGGTGTCTGCCTTATTG-3′ and reverse: 5′-TTGGTGGTGGTGATCCTTTC-3′; carnitine palmitoyltransferase 1β (CPT1b), forward 5′-TCCAAACGTCACTGCCTAAG-3′ and reverse 5′-CCAATGTCTCCATGCGGTAATA-3′; transcription factor A, mitochondrial (Tfam), forward 5′-CTGAAGTTGGACGAAGTGATCT-3′ and reverse 5′-GGGCCTAATCCCAATGACAA-3′; UCP1, forward: 5-CACCTTCCCGCTGGACACT-3′ and reverse 5′-CCTGGCCTTCACCTTGGAT-3′; and β-actin, forward 5′-ACCAGTTCGCCATGGATGAC-3′ and reverse 5′-TGCCGGAGCCGTTGTC-3′.

Protein analysis

BAT from AdCXCR4ko mice and C57BL/6 controls fed CD or HFD was lysed by sonication in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (Bio-Rad, Hercules, CA, USA). Samples were loaded onto a 12% polyacrylamide gel, and the proteins were separated by using a Hoefer SE 600 electrophoresis unit (GE Healthcare Life Sciences, Pittsburgh, PA, USA) and transferred onto polyvinylidene difluoride (PVDF) membranes (Bio-Rad). The membranes were blocked overnight in tris(hydroxymethyl)-aminomethane-buffered saline (TBS) containing 0.05% Tween 20 (TTBS) and 5% skim milk. Antibody-detecting mouse UCP1 was diluted 1:1000 and incubated with the membranes for 1 h. HRP-linked secondary antibody (Thermo Fisher Scientific, Rockford, IL, USA) was incubated with membranes for an additional hour, and the protein bands were visualized by chemiluminescence (GE Healthcare Life Sciences). Equal protein loading was confirmed by reprobing the membranes with monoclonal anti-GAPDH antibody (Abcam).

Dual-energy X-ray absorptiometry (DEXA)

The AdCXCR4ko mice and the C57BL/6 controls fed the CD or the HFD for 22–24 wk were weighed immediately before the scan, and a DEXA system was used to measure total body fat content with the mice under anesthesia (Lunar PIXImus2; GE Lunar Corp., Madison, WI, USA). This system uses software that has been reported to require additional modification to improve the accuracy of fat content quantification (20), which was included in the current analysis. The percentage of body fat was measured as body fat content, excluding the head, divided by total body mass.

Indirect calorimetry

Energy expenditure was measured by indirect calorimetry in AdCXCR4ko mice and C57BL/6 controls fed either the CD or HFD for 24–26 wk. The animals were acclimated to the testing cages and metabolic cabinet for 48 h before data collection with ad libitum access to food and water. After acclimation, oxygen consumption and carbon dioxide production were measured with a multiple animal respirometry system (MARS; Sable Systems, Las Vegas, NV, USA). Ten-minute averages were collected hourly over a continuous 24-h period. The respiratory exchange ratio was unaltered by genotype. Thus, data were reported as average light-or dark-phase rates of oxygen consumption normalized to total mass or lean body mass.

Statistical analysis

Data are presented as means ± sem and were analyzed with Graph Pad Prism (San Diego, CA, USA). Comparisons between 2 groups were performed with an unpaired, 2-tailed Student's t test and 1-way analysis of variance (ANOVA). Data derived from the same animal at several time points (indirect calorimetry) were analyzed with 2-way ANOVA for repeated measurements, to evaluate the differences between experimental groups. Values of P < 0.05 were considered statistically significant.

RESULTS

Chemokine receptor CXCR4 is expressed on adipocytes and ATMs in adipose tissue

During the course of adipose tissue inflammation, the expression of chemokine receptors that support obesity development is limited to leukocytes invading the tissue (5, 6). Interestingly, CXCR4 was detected on adipocytes and ATMs in human adipose tissue (10). However, how CXCR4 functions in this tissue remains unclear. To learn about the roles of CXCR4 in adipose tissue, we first examined whether its expression in mouse adipose tissue reflects that of humans.

C57BL/6 mice fed either the 10% kcal CD or maintained for 24 wk on the obesity-inducing 60% kcal HFD were euthanized, adipose tissues were removed, and CXCR4 expression was evaluated in BAT and in subcutaneous and visceral mesenteric, retroperitoneal, and epididymal/parametrial WAT.

As in human adipose tissue, CXCR4 was detected in mouse adipose tissue (Fig. 1A) on adipocytes, as well as on macrophage marker CD68⁺ mononuclear cells (Fig. 1B). This expression pattern suggests that adipose tissue CXCR4 controls a multitude of functions, including inflammatory cell influx and efflux into and out of fat tissue, fat uptake, storage, and mobilization, and adipocyte functional responses, homeostasis, and survival.

Figure 1.

In adipose tissue, CXCR4 is expressed on adipocytes and adipose tissue macrophages. WT C57BL/6 control mice fed either a CD (n=10) or an HFD (n=10) for 24 wk were euthanized, and WAT, including subcutaneous and visceral fat pads (mesenteric, retroperitoneal and epididymal/parametrial gonadal) and BAT pads, were excised, fixed, frozen, sectioned, and stained. A) Visceral epididymal WAT sections were stained with primary rabbit polyclonal anti-CXCR4 (CXCR4) or respective IC antibody, followed by biotin-conjugated secondary antibody, and treated with streptavidin/HRP and diaminobenzidine, and the sections were evaluated under a light microscope. B) Visceral epididymal WAT sections were costained with primary rabbit polyclonal anti-CXCR4 and rat anti-mouse CD68 antibodies, followed by donkey anti-rabbit Alexa Fluor 488– or donkey anti-rat Alexa Fluor 568–conjugated secondary antibodies, and mounted in DAPI-containing medium. The sections were examined by fluorescence microscopy. Green, CXCR4; red, CD68; yellow, red + green emission overlap; yellow arrows indicate CD68⁺CXCR4⁺ adipose tissue macrophages; green arrows indicate CXCR4⁺ adipocytes.

Ablation of adipocyte and not myeloid leukocyte CXCR4 increases susceptibility to HFD-induced obesity

Since CXCR4 in adipose tissue is expressed on adipocytes and ATMs (Fig. 1B), we evaluated how deficiency of CXCR4 in either cell type affects development of diet-induced obesity in mice fed the HFD for 24 wk.

The AdCXCR4ko mice were viable, fertile, and normal in size. Under normal conditions, animals appeared healthy despite neutrophilia and lymphocytosis (Table 1). In these mice, CXCR4 expression was greatly reduced in BAT and subcutaneous, mesenteric, and epididymal/parametrial gonadal WAT. Low amounts of Cxcr4 mRNA, probably due to normal expression of CXCR4 in stromovascular cells, including leukocytes and endothelial cells, were detected in BAT and WAT, whereas expression of this chemokine receptor in control organs, including heart, spleen, and hypothalamus, remained unaltered (Fig. 2A). In AdCXCR4ko mice, CXCR4 protein expression was not detected in Fabp4⁺ white and brown adipocytes (Fig. 2B), but was observed on F4/80⁺ ATMs from WAT at levels comparable to those in C57BL/6 controls (Fig. 2C). Expression of CXCR4 on cells of myeloid lineage in AdCXCR4ko mice was also confirmed by costaining WAT (Supplemental Fig. S2A) and BAT (Supplemental Fig. S2B) with anti-CD68 and anti-CXCR4 antibodies.

Table 1.

Peripheral blood cell counts

Genotype	WBC (×103/μl)	NE (×103/μl)	LY (×103/μl)	MO (×103/μl)	RBC (×106/μl)	PLT (×103/μl)
WT	4.9 ± 0.6	0.6 ± 0.1	3.7 ± 0.7	0.2 ± 0.04	7.6 ± 0.5	934 ± 40
AdCXCR4ko	11.8 ± 1.5	3.7 ± 1.2	6.9 ± 1.4	0.7 ± 0.1	8.9 ± 0.5	936 ± 45
MyeCXCR4ko	19.3 ± 1.4	6.9 ± 1.3	10.2 ± 0.1	1.9 ± 0.2	8.7 ± 0.1	895 ± 17.8

WBC, white blood cells; NE, neutrophils; LY, lymphocytes; MO, monocytes; RBC, red blood cells; PLT, platelets.

Figure 2.

Ablation of adipocyte CXCR4 exacerbates HFD-induced obesity. WT C57BL/6 control (n=40) and AdCXCR4ko (n=40) mice were fed either a CD or an HFD. During this time, the animals were evaluated for BW and euthanized after 24 wk of CD or HFD feeding. A) Detection of CXCR4 mRNA was performed in heart, spleen, and hypothalamus (Hypothal.); and BAT and subcutaneous (Subcut.), visceral mesenteric (Mesent.), and epididymal/parametrial gonadal (Gon.) WAT from AdCXCR4ko and WT mice. B) WAT and BAT tissue sections were also costained with primary monoclonal anti-mouse Fabp4 and polyclonal anti-CXCR4 (CXCR4) antibodies, followed by donkey anti-rat Alexa Fluor 488– or donkey anti-rabbit Alexa Fluor 568–conjugated secondary antibodies. Sections were mounted in DAPI-containing medium and evaluated by fluorescence microscope. Staining of visceral epididymal WAT and BAT sections is shown as representative. Green, Fabp4; red, CXCR4; blue, DAPI; yellow, red + green emission overlap. C) Visceral mesenteric, retroperitoneal, and gonadal WAT was removed and digested with collagenase type I, and the stromovascular fraction containing leukocytes was separated from the adipocytes by centrifugation. Cells were costained with anti-F4/80- and anti-CXCR4-directed antibodies or their respective IgG ICs (top right panel) and analyzed by flow cytometry. Dead cells were excluded by PI. Numbers in top right corners indicate the percentage of cells with the indicated immunophenotype. FSC, forward scatter; SSC, side scatter; PE, phycoerythrin. D) Weight over time of mice fed a CD. F) Representative AdCXCR4ko and WT control mice fed an HFD. D) Weight over time of mice fed an HFD. *P < 0.01.

AdCXCR4ko mice fed CD did not differ in their BWs from C57BL/6 controls (Fig. 2D). The HFD triggered BW increases in both the AdCXCR4ko and the C57BL/6 mice. The AdCXCR4ko mice became significantly heavier than the C57BL/6 controls after 8 wk on the HFD, and they continued to gain weight at an accelerated rate. After 24 wk on the HFD, the AdCXCR4ko mice were considered obese (Fig. 2E), weighing ∼52 g, or ∼34% more than the C57BL/6 controls (Fig. 2F).

MyeCXCR4ko mice, like the AdCXCR4ko strain, are also viable, fertile, and normal in size. As previously reported (21), the MyeCXCR4ko mice developed neutrophilia, monocytosis, and lymphocytosis (Table 1). The mice expressed minimal amounts of CXCR4 on F4/80⁺ leukocytes in WAT (Fig. 3A), whereas the expression of this chemokine receptor on Fabp4⁺ adipocytes (Fig. 3B) remained high. Loss of CXCR4 expression on ATMs in MyeCXCR4ko mice was observed after costaining of WAT (Supplemental Fig. S2A) and BAT (Supplemental Fig. S2B) with anti-CD68 and -CXCR4 antibodies.

Figure 3.

HFD feeding does not potentiate obesity in MyeCXCR4ko mice. WT C57BL/6 control (n=40) and MyeCXCR4ko (n=40) mice were fed a CD for 18 wk or an HFD for 24 wk. Animals were weighed 1×/wk and euthanized at the end of the feeding regimen. A) Visceral mesenteric, retroperitoneal, and epididymal/parametrial gonadal WAT was removed and digested with collagenase type I, and the stromovascular fraction containing leukocytes was separated from the adipocytes by centrifugation. Stromovascular cells were costained with anti-F4/80- and anti-CXCR4-directed antibodies or their respective isotype IgG controls (top right panel) and analyzed by flow cytometry. Dead cells were excluded by PI stain. Numbers in top right corners indicate the percentage of cells with the indicated immunophenotype. FSC, forward scatter; SSC, side scatter; PE, phycoerythrin. B) WAT and BAT tissue sections were also costained with primary monoclonal anti-mouse Fabp4 and polyclonal anti-CXCR4 (CXCR4) antibodies followed by donkey anti-rat Alexa Fluor 488– or donkey anti-rabbit Alexa Fluor 568–conjugated secondary antibodies. Sections were mounted in DAPI-containing medium and evaluated by fluorescence microscopy. Staining of a visceral epididymal WAT section is shown as representative. Green, Fabp4; red, CXCR4; blue, DAPI; yellow, red + green emission overlap. C, D). Weight of CD-fed (C) and HFD-fed (D) mice.

The MyeCXCR4ko mice fed the CD gained BW at the same rate as the C57BL/6 controls (Fig. 3C). Although the HFD triggered a BW increase in both the MyeCXCR4ko and the C57BL/mice, no significant difference was recorded between the strains (Fig. 3D). Thus, ablation of adipocyte, not myeloid leukocyte, CXCR4 increases susceptibility to HFD-induced obesity.

Exacerbated obesity in AdCXCR4ko mice is not a result of hyperphagia, but positively correlates with increased mass, adiposity, and hypertrophy of WAT and BAT

In both animals and humans, hyperphagia is the most frequent cause of obesity (22). To determine whether obesity in AdCXCR4ko mice is a result of altered feeding behavior, weekly food consumption for age-matched AdCXCR4ko, MyeCXCR4ko, and C57BL/6 mice was assessed over 24 wk of HFD feeding.

The AdCXCR4ko mice did not have increased weekly caloric intake compared to that of the C57BL/6 controls (Fig. 4A). This finding suggests that there were no differences in feeding patterns between the 2 groups. Moreover, food consumption in the AdCXCR4ko mice was not greater than in the MyeCXCR4ko group (Fig. 4A, B), indicating that obesity in the AdCXCR4ko mice was not a result of hyperphagia.

Figure 4.

Obesity in AdCXCR4ko mice is not a result of hyperphagia but positively correlates with increased adiposity, mass, and hypertrophy of BAT and WAT. WT C57BL/6 control (n=15), AdCXCR4ko (n=15), and MyeCXCR4ko (n=15) mice were fed a CD or an HFD for 24 wk. A, B) During 24 wk of HFD feeding, food consumption of AdCXCR4ko (A) and MyeCXCR4ko (B) mice was recorded 1×/wk. C, D) AdCXCR4ko mice (n=5/group) were evaluated for fat tissue mass (C) and lean tissue mass (D) by DEXA. E, F) Subcutaneous and visceral WAT (mesenteric, retroperitoneal, and epididymal/parametrial gonadal; E) and BAT (F) from WT and AdCXCR4ko mice were removed and weighed. G) WAT (epididymal) and BAT sections from WT and AdCXCR4ko mice were stained with hematoxylin and eosin. H, I) Subcutaneous and visceral WAT (mesenteric, retroperitoneal, and epididymal/parametrial gonadal; H) and BAT (I) from WT and MyeCXCR4ko mice were removed and weighed. J) WAT (epididymal) and BAT sections from WT and MyeCXCR4ko mice were stained with hematoxylin and eosin.

Adipose tissue responds rapidly and dynamically to nutrient excess through adipocyte hypertrophy (increased size), or hyperplasia (increased number), or both. It is well established that adipocyte hypertrophy is the main mechanism supporting adipose tissue expansion in obesity (23). We therefore evaluated whether obesity in AdCXCR4ko mice is due to increases in mass and adiposity of WAT and BAT and adipose tissue hypertrophy.

Measurement of adipose tissue volume using DEXA (24) clearly demonstrated a 2-fold increase in adiposity in the HFD-fed AdCXCR4ko mice relative to that in the controls (Fig. 4C). No difference between the AdCXCR4ko mice and the C57BL/6 group was observed if the animals were fed CD (Fig. 4C). Notably, lean body mass remained similar in the AdCXCR4ko mice and the C57BL/6 controls fed either an HFD or a CD, indicating that ablation of adipocyte CXCR4 had no effect on the mass of nonadipose tissues, such as muscle (Fig. 4D). Furthermore, obesity in the AdCXCR4ko mice positively correlated with 1.8- and 1.6-fold increases in subcutaneous and visceral (mesenteric, retroperitoneal, and gonadal) WAT mass (Fig. 4E), respectively, as well as with a ∼2-fold BAT mass increase (Fig. 4F). Obesity in the HFD-fed AdCXCR4ko mice (Fig. 2E, F) also correlated with a profound WAT and BAT hypertrophy (Fig. 4G).

In contrast to the AdCXCR4ko mice, no significant differences in WAT (Fig. 4H) or BAT (Fig. 4I) masses were recorded between the MyeCXCR4ko mice and the C57BL/6 controls. Although the MyeCXCR4ko mice were fed the HFD for 24 wk and had higher BWs than their CD-fed counterparts (Fig. 3D), no obvious hypertrophy was observed in WAT or BAT obtained from these mice (Fig. 4J). Together these results suggest that diet-induced weight gain in AdCXCR4ko mice is due to aberrant, likely hypertrophic, adipose tissue expansion.

Adipocyte CXCR4 deficiency alters ATMs and lymphocyte contents in WAT

The defining feature of adipose tissue inflammation in obesity is an increase in the total number of ATMs in predominantly WAT that is supported by the influx of proinflammatory CD11b⁺ monocytes that mature to classic CD40⁺ and CD11c⁺ M1 ATMs. This increase in CD40⁺ and CD11c⁺ M1 ATMs is accompanied by a decrease in the number of alternative CD163⁺ and CD206⁺ M2 ATMs. Cells of the adaptive immunity, especially CD8⁺ and interferon-γ-producing CD4⁺ T lymphocytes and B cells, were also demonstrated to mediate adipose tissue inflammation (25, 26). Recruitment of proinflammatory leukocytes, which is thought to contribute to metabolic syndrome and systemic insulin resistance (3, 4), is considered to be chemokine driven in part (6).

The chemokines CCL2 and CCL5, acting through their cognate chemokine receptors CCR2 and CCR5 were shown to recruit inflammatory leukocytes, especially monocytes, to obese WAT, and their plasma levels were increased in obese individuals and mice, relative to lean control subjects (7, 9). Thus, we examined how inactivation of adipocyte or myeloid leukocyte CXCR4 affects plasma levels of these proinflammatory chemokines. As shown in Fig. 5A, B, CCL5 and CCL2 were increased in plasma of AdCXCR4ko, MyeCXCR4ko, and control C57BL/6 mice in response to HFD feeding; however, the strongest increases in CCL2 and CCL5 were recorded in the HFD-fed AdCXCR4ko mice.

We also determined how ablation of CXCR4 in adipocytes or myeloid leukocytes affects obesity-induced proinflammatory monocyte influx and ATM immunophenotype and lymphocyte populations in WAT. We found that WAT of CD-fed AdCXCR4ko, MyeCXCR4ko, and C57BL/6 mice had a similar number of F4/80⁺CD11b⁺CD206⁻CD11c⁺ M1 or F4/80⁺CD11b⁺CD206⁺CD11c⁻ M2 ATMs; the M1 population, however, was smaller than the M2 population (data not shown). HFD induced a profound influx of proinflammatory monocytes into WAT of AdCXCR4ko mice; F4/80⁺CD11b⁺CD206⁻CD11c⁺ M1 ATMs increased 3.2-fold relative to M1 ATMs detected in WAT of C57BL/6 controls (Fig. 5C, D). Whereas M1 ATMs increased in WAT of HFD-fed AdCXCR4ko mice, the number of F4/80⁺CD11b⁺CD206⁺CD11c⁻ M2 ATMs decreased 2.3-fold (Fig. 5C, E). In addition, CXCR4 ablation resulted in a 2.6-fold increase in CD8⁺ T lymphocytes in WAT of HFD-fed mice (Fig. 5F, G). It is interesting that deletion of adipocyte CXCR4 was associated with 2.4-fold decrease in CD4⁺ T-lymphocyte populations (Fig. 5H, I) and a 40% reduction in B lymphocytes (Fig. 5J, K). Contrary to changes in the number of monocytes and lymphocytes detected in WAT of the HFD-fed AdCXCR4ko mice, no significant differences in CD206⁺ M2 and CD11c⁺ M1 ATM, CD8⁺ and CD4⁺ T-lymphocyte, and B-lymphocyte counts were detected between the MyeCXCR4ko mice and C57BL/6 controls, indicating that inactivation of myeloid leukocyte CXCR4 had no effect on leukocyte content or immunophenotype (Fig. 5C–K) in WAT. This observation suggests that adipocyte and not myeloid leukocyte CXCR4 controls development of obesity by preventing excessive inflammatory immune cell influx into WAT.

Ablation of adipocyte CXCR4 does not alter glucose tolerance or insulin sensitivity

Obesity-induced adipose tissue inflammation is thought to trigger insulin resistance, type 2 diabetes, and metabolic syndrome (3, 4). The chemokine system was reported to link obesity to insulin resistance by triggering the onset of localized inflammation caused by crosstalk between inflamed obese adipose tissue and other organs (i.e., pancreas and liver). Evidence shows that obesity-induced chemokine receptor–mediated inflammatory monocyte recruitment and the presence of M1 macrophages in insulin target organs other than adipose tissue trigger metabolic complications including insulin resistance and glucose intolerance (27).

C57BL/6 mice fed the HFD become mildly to moderately hyperglycemic. The increase in blood glucose is accompanied with mild hyperinsulinemia. However, the C57BL/6 mice fed the HFD fail to develop overt diabetes that, in humans, is associated with islet atrophy and nephropathy, and thus, the development of diabetes in diet-induced mouse obesity models remains controversial (17). Deficiency in adipocyte CXCR4 exacerbated diet-induced obesity (Fig. 2E, 2F) and resulted in increased M1 content in WAT (Fig. 5C, D). Therefore, we determined how lack of adipocyte CXCR4 affects glucose and insulin tolerance.

No significant difference in glucose sensitivity was recorded between the AdCXCR4ko and C57BL/6 mice (Fig. 6A). We also observed that the AdCXCR4ko and C57BL/6 mice had comparable glucose clearance levels, indicating similar insulin sensitivity (Fig. 6B). These data suggest that ablation of adipocyte CXCR4 does not alter glucose tolerance and insulin sensitivity.

Figure 6.

Ablation of adipocyte CXCR4 does not alter glucose tolerance or insulin sensitivity. WT C57BL/6 (n=8) and AdCXCR4ko (n=8) mice were fed the HFD for 24 wk and were denied access to food either overnight or for 5 h before they were injected with glucose (1 g/kg BW; A) or insulin (0.75 U/kg BW; B). In both assays, blood glucose levels were measured 30, 60, 90, and 120 min after injection.

Deficiency in adipocyte CXCR4 results in lower metabolic rates

The HFD exacerbated obesity in AdCXCR4ko mice. The striking feature of these mice was the markedly increased mass (Fig. 4F) and hypertrophy of BAT (Fig. 4G) that often indicate dysregulated brown adipocyte–mediated energy consumption (12, 16, 28). We tested whether ineffective energy expenditure is a reason for compromised BAT homeostasis by determining metabolic rates in AdCXCR4ko mice and C57BL/6 controls fed the CD or HFD, by using indirect calorimetry. This method allows for assessment of BAT function in vivo because it provides an accurate estimate of energy consumption from measures of carbon dioxide production or oxygen consumption during the dark and light phases of the 24-h cycle (29).

As shown in Fig. 7, the CD-fed AdCXCR4ko mice and C57BL/6 controls had similar rates of oxygen consumption in light and dark phases, regardless of whether oxygen consumption was normalized to total body mass or to lean mass. The HFD-fed AdCXCR4ko mice had, compared to the control group, significantly lower oxygen consumption, indicating lower metabolic rates during the light (Fig. 7A, P<0.0001) as well as the dark (Fig. 7B, P<0.0001) phase, if oxygen consumption was normalized to total body mass. When oxygen consumption was normalized to lean body mass, no significant difference in metabolic rates between the obesogenic diet-fed AdCXCR4ko mice and WT controls was recorded (Fig. 7C, D). This suggests that adipose and not lean tissue contributes to lower total body mass energy consumption in HFD-fed AdCXCR4ko mice.

Figure 7.

Deficiency in adipocyte CXCR4 results in lower metabolic rate. Rate of O₂ consumption was measured by indirect calorimetry during 12 h light (A, C) and 12 h dark (B, D) phases of the daily cycle in WT C57BL/6 control (n=5) and AdCXCR4ko mice (n=5) fed either a CD or an HFD for 24–26 wk. Rate of O₂ consumption was normalized to total body mass (A, B) or to lean body mass (C, D).

AdCXCR4ko mice display impaired adaptive thermogenesis

Adipocyte CXCR4 could be involved in regulation of energy metabolism in BAT, since lower energy expenditure was recorded in HFD-fed AdCXCR4ko mice (Fig. 7). Thus, we hypothesized that HFD increases expression of in CXCR4 and CXCR12 in BAT. We determined expression levels of CXCR4 and CXCR12 in CD- or HFD-fed AdCXCR4ko mice and C57BL/6 controls. We found that CD-fed C57BL/6 and AdCXCR4ko mice expressed comparable amounts of CXCL12 in BAT. HFD feeding increased expression of this chemokine in brown adipocytes of C57BL/6 controls but not in AdCXCR4ko mice (Fig. 8A). CXCR4 was moderately up-regulated in response to the HFD in white adipocytes of the C57BL/6 and MyeCXCR4ko mice (Fig. 8B), and its expression was strongly increased in brown adipocytes of the HFD-fed C57BL/6 controls and MyeCXCR4ko mice. CXCR4 was not detected in Fabp4⁺ cells in CD- or HFD-fed AdCXCR4ko mice (Fig. 8B, C). Thus, HFD triggers an increase in CXCR4 and CXCR12 expression especially in BAT, which is associated with increased energy consumption (Fig. 7). This suggests that CXCR4 and CXCR12 are critical determinants of BAT energy metabolism.

Figure 8.

AdCXCR4ko mice display impaired adaptive thermogenesis. WT C57BL/6 control (n=5), AdCXCR4ko (n=5), and MyeCXCR4ko mice were fed either a CD or an HFD for 24 wk. Body temperature was measured at 25°C or at the indicated times (D) at 4°C. Mice were euthanized; BAT and visceral mesenteric, retroperitoneal, and epididymal/parametrial gonadal WAT pads were removed and processed; and expression of CXCL12, CXCR4, and UCP1 was evaluated. A) Expression of CXCL12 (red) in Fabp4⁺ (green) brown adipocytes from AdCXCR4ko and WT mice fed a CD or an HFD. B) Expression of CXCR4 (red) in Fabp4⁺ (green) adipocytes in WAT from AdCXCR4ko, MyeCXCR4ko, and WT mice fed a CD or an HFD. C) Expression of CXCR4 (red) in Fabp4⁺ (green) adipocytes in BAT from AdCXCR4ko, MyeCXCR4ko, and WT mice fed a CD or an HFD. D) Body temperature of HFD-fed AdCXCR4ko and WT mice at 25°C (time 0) and at different times after exposure to cold. *P < 0.05. E) Western blot analysis of UCP1 expression using commercially available rabbit polyclonal anti-mouse UCP1. Equal protein loading was confirmed by reprobing membranes with monoclonal anti-GAPDH antibody. F) Densitometry Western blot analysis. G) UCP1 expression in BAT from WT and AdCXCR4ko mice was also assessed by immunofluorescence using rabbit anti-mouse UCP1 primary antibody and secondary goat anti-rabbit IgG Alexa Fluor 568 antibody.

BAT has the inherent capacity to support cellular energy expenditure (bioenergetics) through oxidative phosphorylation and adaptive thermogenesis. During mitochondrial oxidative phosphorylation, energy from food is turned into adenosine-5′-triphosphate (ATP), which is used for many cellular processes, including biosynthetic reactions, motility, and cell division (30). Notably, in homeotherms, some energy from food is stored as triglycerides, mostly within WAT depots throughout the body. This energy may be mobilized during nonshivering adaptive thermogenesis for the purpose of producing the heat needed to maintain core body temperature, thereby ensuring that cellular functions and physiological processes continue in cold environments. However, nonshivering adaptive thermogenesis is also activated by chronic overfeeding. Metabolic or diet-induced adaptive thermogenesis can turn excess caloric intake into heat, which prevents the excessive triglyceride deposition and adipose tissue expansion that lead to obesity. Metabolic and cold-induced, nonshivering adaptive thermogenesis take place in the mitochondria of brown adipocytes and are regulated by UCP1 (12–16). Thus, if CXCR4 is a molecule regulating bioenergetics responses of BAT, then adipocyte CXCR4 deficiency should affect thermogenic and possibly oxidative responses of BAT in AdCXCR4ko mice.

We first assessed how adipocyte CXCR4 affects activity of BAT in thermal stress conditions. Body temperature in the AdCXCR4ko mice and C57BL/6 controls was measured in an ambient temperature of 25°C and at different times after acute exposure to cold (4°C). We found that AdCXCR4ko mice and controls had similar body temperature when kept at 25°C; however, the AdCXCR4ko mice were unable to maintain constant body temperature and displayed significant cold sensitivity as early as 2 h after being exposed to 4°C (Fig. 8D), suggesting that adipocyte CXCR4 is an important regulator of cold-induced thermogenic responses in BAT.

To determine whether adipocyte CXCR4 is involved in regulation of metabolic thermogenesis, we assessed expression of UCP1 in CD- and HFD-fed animals, as studies have demonstrated that excess caloric intake increases UCP1 expression in brown adipocytes (31). Interestingly, we found that AdCXCR4ko and control mice express similar amounts of BAT UCP1 when fed CD. HFD induced a 3.5-fold increase in UCP1 expression in C57BL/6 mice (Fig. 8E, F). However, no increase in expression of UCP1 was detected in BAT from obesogenic HFD-fed AdCXCR4ko mice (Fig. 8E–G) suggesting that deficiency in adipocyte CXCR4 also compromises metabolic thermogenesis by preventing a high calorie diet–induced increase in UCP1 expression.

Last, we investigated whether adipocyte CXCR4 affects oxidative capacity of BAT, as this function supplies energy for thermogenesis (32). Thus, the expression of genes involved in mitochondrial biogenesis and oxidative function including Nfr1, Cox4, ATP5b, CPT1b, and Tfam (33) was evaluated in BAT from C57BL/6 controls and AdCXCR4ko mice when housed at 25°C and after a 3 h exposure to 4°C. We observed that at 25°C, expression of Nrf1, Cox4, ATP5b, CPT1b, and Tfam in BAT of the AdCXCR4ko mice was significantly reduced relative to expression in the controls (Fig. 9A), suggesting that mitochondrial metabolism and oxidative capacity of BAT are compromised in this strain. Interestingly, exposure to cold further down-regulated the examined mitochondrial genes including UCP1 (Fig. 9B), indicating that disturbed oxidative responses together with inability to increase UCP1 expression impair adaptive thermogenesis in BAT of AdCXCR4ko mice, resulting in cold sensitivity and increased susceptibility to HFD-induced obesity.

Figure 9.

Expression of mitochondrial genes involved in mitochondrial biogenesis and oxidative function is reduced in AdCXCR4ko mice. WT C57BL/6 control (n=10) and AdCXCR4ko (n=10) mice were fed an HFD for 24 wk. Some animals were maintained at 25°C, and others were moved into a cold room (4°C), maintained for up to 180 min, and euthanized. BAT was removed; total RNA was isolated and reverse transcribed; and gene expression for Nrf1, Cox4, ATP5b, CPT1b, Tfam and UCP1 was determined by quantitative PCR with gene-specific primers. A) Expression of Nrf1, Cox4, ATP5b, CPT1b, and Tfam in BAT from WT and AdCXCR4ko mice maintained at 25°C. B) Expression of Nrf1, Cox4, ATP5b, CPT1b, Tfam, and UCP1 in BAT from WT and AdCXCR4ko mice maintained at 4°C for 180 min. *P < 0.01, **P < 0.05.

DISCUSSION

Sedentary living and the consumption of calorie-dense food, combined with the cornerstone methods for managing obesity (i.e., dieting), which are proven to be largely ineffective, have dramatically increased the prevalence of obesity in the Western countries (34). The result has been a profound increase in the occurrence of metabolic syndrome and cardiovascular disease (3, 4). Given the financial burden that obesity-linked comorbidities impose on society, it is imperative to understand molecules and signals that are fundamental supporters of adipose tissue homeostasis, and it is necessary to identify mechanisms that lead to impaired and aberrant adipocyte functions and obesity. In this regard, our study has identified the chemokine receptor CXCR4, whose role in adipose tissue was unknown, as a critical regulator of adipose tissue homeostasis. More precisely, we show that adipocyte CXCR4 expression is required in BAT for this adipose tissue depot to maintain and, when necessary, increase its thermogenic capacity, thereby increasing overall energy expenditure and decreasing susceptibility to diet-induced obesity. We also demonstrated that CXCR4 prevents excessive inflammatory leukocyte influx into obese WAT, an event that exacerbates development of obesity and obesity-related comorbidities.

The thermogenic capacity of BAT is conferred by UCP1, which is up-regulated in conditions of chronic overfeeding or cold and, in such conditions, acts as the main switch that uncouples oxidative phosphorylation from ATP synthesis. In this way, it is ensured that surplus caloric intake dissipates during metabolic thermogenesis as heat and that resistance to cold is maintained in homeotherms (12–16).

The finding that CXCR4 controls thermogenic activities of BAT is of great importance because the presence of BAT was, until recently, thought to be relevant only in small mammals and infant humans, with negligible physiological relevance in adult humans. However, recent studies have reported the existence of variable but significant amounts of BAT in human adults. It is maintained that human BAT depots contain a mixture of brown adipocytes interspersed within a greater volume of white adipocytes, which have a much lower metabolic activity. As in small mammals and infants, BAT in human adults serves as a thermogenic organ that burns food energy to generate the heat necessary for either thermal homeostasis or to prevent surplus energy storage in adipose tissues. As such, BAT has a major impact on metabolic rate, and because of this, alterations in BAT activity affect BW (12, 35–40). Therefore, clear understanding of signals that control thermogenic capacity of BAT may ultimately generate a platform to develop novel strategies for targeting BAT activity to combat obesity in humans. This possibility is especially true of CXCR4, because this chemokine receptor is conserved across species (11), and learning how CXCR4 functions in mouse BAT may explain how this chemokine receptor affects activity of human BAT.

CXCR4 is constitutively expressed on a variety of different cell types. During embryogenesis, CXCR4 regulates development of hematopoietic, cerebellar, and endothelial tissues by controlling tissue progenitor cell migration, homing, and survival. In adult life, this chemokine receptor serves as the key factor for physiological stem and immune cell trafficking (11). Because of its pivotal role in embryo/organogenesis and tissue/organ regeneration, CXCR4 is often described as the homeostatic chemokine receptor. Our finding that adipocyte CXCR4 controls UCP1-mediated energy expenditure mechanisms, which maintain adipose tissues homeostasis, profoundly increases the homeostatic value of this chemokine receptor, especially because of the detrimental effects that the obesity epidemic imposes on quality of life for affected individuals and broadly on healthcare systems around the globe.

Our study also shows that CXCR4 maintains adipose tissue homeostasis by more than 1 mechanism, and with that we reveal a previously unrecognized role of the chemokine system in development of obesity. Chronic low-grade inflammation in adipose tissues is a hallmark of obesity that is mediated in part by the chemokine receptor–mediated influx of inflammatory leukocytes into WAT (5, 6). In our study, however ablation of myeloid leukocyte CXCR4 did not facilitate inflammatory monocyte recruitment into WAT, which indicates that this chemokine receptor, although expressed by the leukocyte subtype critical in the exacerbation of adipose tissue inflammation, does not contribute to it. Furthermore, we demonstrated that deficiency in adipocyte CXCR4 was associated with an increased number of inflammatory M1 ATMs in WAT, suggesting that CXCR4 does not promote but rather prevents excessive inflammatory cell recruitment into obese adipose tissue. This conclusion is further supported by our data demonstrating that absence of adipocyte CXCR4 also allows for excessive recruitment of CD8⁺ T lymphocytes into WAT. This lymphocyte subset was demonstrated to have an essential role in the initiation and propagation of adipose tissue inflammation (25).

Interestingly, decreased influx of CD4⁺ T lymphocytes into obese WAT has been demonstrated to ameliorate metabolic complications (26), whereas reduced recruitment of regulatory B cells has been shown to be associated with progression of adipose tissue inflammation in obesity and insulin resistance (41). We observed decreased influxes of B cells and CD4⁺ T cells into WAT of AdCXCR4ko mice. Furthermore, we detected similar glucose tolerance and insulin sensitivity in AdCXCR4ko mice and controls, which suggests that potentiated adipose tissue inflammation due to reduced B-cell counts has been negatively affected by a decrease in CD4⁺ T-cell recruitment, preventing development of obvious metabolic complications in HFD-fed AdCXCR4ko mice. Thus, CXCR4 supports adipose tissue homeostasis not only by allowing BAT to adapt to the necessary increase in thermogenic responses induced by the nutrient excess or cold, but also by controlling HFD-induced adipose tissue inflammation.

With the growing worldwide epidemic of obesity, it is clear that new and effective antiobesity therapies are desperately needed. Compelling data suggest that targeting cellular bioenergetics and mediators of adipose tissue inflammation would provide a new therapeutic approach for treatment or prevention of obesity (12, 37). However, signals and molecules controlling energy expenditure and adipose tissue inflammation remain incompletely understood. Our finding that CXCR4 is an important regulator of bioenergetic and inflammatory responses in adipose tissues identifies this chemokine receptor as a potential therapeutic target in combating obesity. However, it remains to be investigated how CXCR4, which is located in the membrane of brown adipocytes, signals to increase UCP1 expression to support adaptation of BAT to the higher energy expenditure imposed by chronic overfeeding or by cold. Furthermore, we observed that the expression of several genes controlling mitochondrial biogenesis, oxidative function, and thermogenesis was greatly reduced in BAT of obese AdCXCR4ko mice. Thus, it has to be clarified whether the observed decrease is due to a reduced number of mitochondria or is a result of post-translational modifications that repress expression of these genes. Despite these questions that remain to be investigated, our study suggests that by stimulating CXCR4 expression and signaling in adipocytes, we may be able to increase energy consumption and reduce adipose tissue inflammation, and by so doing, we may be able to oppose the unnecessary BW gain that supports the development of obesity.