Mice with CCR2 deficiency are protected from insulin resistance but only after long periods of high-fat diet (HFD) feeding, despite the virtual absence of circulating inflammatory monocytes. We performed a time course study in mice with hematopoietic and global CCR2 deficiency to determine adipose tissue–specific mechanisms for the delayed impact of CCR2 deficiency on insulin resistance.
Mice with global or hematopoietic CCR2 deficiency (CCR2−/− and BM-CCR2−/−, respectively) and wild-type controls (CCR2+/+ and BM-CCR2+/+, respectively) were placed on an HFD for 6, 12, and 20 weeks. Adipose tissue myeloid populations, degree of inflammation, glucose tolerance, and insulin sensitivity were assessed.
Flow cytometry analysis showed that two different populations of F4/80+ myeloid cells (CD11bloF4/80lo and CD11bhiF4/80hi) accumulated in the adipose tissue of CCR2−/− and BM-CCR2−/− mice after 6 and 12 weeks of HFD feeding, whereas only the CD11bhiF4/80hi population was detected in the CCR2+/+ and BM-CCR2+/+ controls. After 20 weeks of HFD feeding, the CD11bloF4/80lo cells were no longer present in the adipose tissue of CCR2−/− mice, and only then were improvements in adipose tissue inflammation detected. Gene expression and histological analysis of the CD11bloF4/80lo cells indicated that they are a unique undifferentiated monocytic inflammatory population. The CD11bloF4/80lo cells are transiently found in wild-type mice, but CCR2 deficiency leads to the aberrant accumulation of these cells in adipose tissue.
The discovery of this novel adipose tissue monocytic cell population provides advances toward understanding the pleiotropic role of CCR2 in monocyte/macrophage accumulation and regulation of adipose tissue inflammation.
Obesity is an independent risk factor for type 2 diabetes, cardiovascular disease, fatty liver disease, atherosclerosis, and several cancers. Chronic adipose tissue inflammation is a cardinal feature of obesity that leads to other health complications such as insulin resistance (1–3). Recruitment of macrophages is an important factor in adipose tissue inflammation (4) and has been shown to be temporally associated with insulin resistance (5,6).
CCR2 regulates monocyte chemotaxis through direct interactions with its ligands, monocyte chemoattractant protein (MCP)-1 and -3 (7). CCR2−/− mice have an immune deficiency in Th1 responses characterized by low levels of interferon-γ (IFN-γ) production and delayed macrophage recruitment to sites of inflammation (8). A study by Tsou et al. (9) found that CCR2−/− mice have a defect in the egress of inflammatory monocytes from the bone marrow, resulting in a dramatic reduction of these cells in the circulation (9,10). Thus, CCR2−/− mice have a decreased pool of circulating inflammatory monocytes and reduced numbers of differentiated myeloid cells recruited to sites of inflammation.
The role of CCR2 and MCP-1 in macrophage recruitment to adipose tissue and the liver and their contribution to insulin resistance have been evaluated by various groups (11–16). One study showed that obese CCR2−/− mice have a modest decrease in macrophage marker expression and macrophage number in adipose tissue after 24 weeks on a 60% fat diet compared with wild-type controls (11). Furthermore, a bone marrow transplant (BMT) study showed that hematopoietic CCR2 deficiency leads to decreased F4/80 expression in the adipose tissue of obese ob/ob mice (13). These experiments indicate that substantial obesity (i.e., long periods of high-fat diet [HFD] feeding and/or a genetically morbidly obese model) is required before differences in macrophage recruitment and improvements in insulin sensitivity are observed in CCR2−/− mice, despite their striking reduction in circulating inflammatory monocytes. We set out to investigate the mechanism responsible for the delayed protection in adipose tissue inflammation and macrophage recruitment in global and hematopoietic models of CCR2 deficiency (CCR2−/− and BM-CCR2−/−, respectively).
In this study, we analyzed the changes in myeloid populations that take place in the adipose tissue of CCR2−/− and BM-CCR2−/− mice during HFD-induced obesity and contrasted them with changes in CCR2+/+ and BM-CCR2+/+ controls. Our data showed that global and hematopoietic CCR2 deficiency leads to the accumulation of CD11bloF4/80lo myeloid cells during early periods of HFD feeding. Further analysis of the CD11bloF4/80lo cells showed that they have a monocytic and inflammatory gene expression profile characterized by significantly elevated expression of Itgax, Il8rb, Ccl5, Nos2, and Csf1. In addition, no differences in inflammation or glucose tolerance were observed between BM-CCR2+/+ and BM-CCR2−/− mice until after substantial obesity was reached during extended periods of HFD feeding—a condition that corresponds temporally with the absence of CD11bloF4/80lo cells.
RESEARCH DESIGN AND METHODS
All animal procedures were performed with approval from the Institutional Animal Care and Usage Committee of Vanderbilt University. The CCR2−/− and CCR2+/+ mice used for the studies were originally purchased from The Jackson Laboratory, are on a C57BL/6 background, and have been propagated in our animal facility. All studies were performed with male 8-week-old mice. Mice were fed ad libitum and given free access to water. In the BMT study, mice were kept on antibiotic water from 1 week prior to BMT through 4 weeks post-BMT. Mice were then placed on a 60% HFD from Research Diets (D12492).
Bone marrow transplantation.
Bone marrow cells were collected from CCR2+/+ or CCR2−/− donor mice and were injected into the retro-orbital venous plexus of lethally irradiated recipient CCR2+/+ mice as previously described (17). The efficiency of bone marrow reconstitution was confirmed by PCR using DNA isolated from the spleens of the recipient mice (data not shown).
Body weight, body composition, and food intake.
Body weight and food intake were measured weekly for the duration of the study. Total lean, fat, and free fluid mass were measured by nuclear magnetic resonance using a Bruker Minispec (Woodlands, TX) in the Vanderbilt Mouse Metabolic Phenotyping Center at baseline and 6, 12, and 20 weeks after placement on an HFD.
Glucose and insulin tolerance tests.
Mice were fasted for 5 h and baseline glucose levels measured using a LifeScan One Touch Ultra glucometer (Johnson & Johnson, Northridge, CA) via the tail vein. Mice were then injected intraperitoneally with glucose at 2 g/kg lean body mass for the 6-week mice and 1 g/kg lean body mass for the 12- and 20-week mice (glucose tolerance tests [GTTs]) or insulin at 0.4 units/kg lean body mass for the 6-week mice and 0.5 units/kg lean body mass for the 12- and 20-week mice (insulin tolerance tests [ITTs]). (13). Glucose levels were assessed at 15, 30, 45, 60, and 90 min after injection.
Plasma collection and measurements.
Mice were fasted for 5 h before all blood collections and at sacrifice. Mice were bled from the retro-orbital venous plexus using heparinized collection tubes. Plasma was separated by centrifugation and stored at −80°C. Insulin levels were measured using an ELISA kit from Millipore (Billerica, MA).
Real-time RT-PCR.
RNA extraction and cDNA Synthesis were performed as described previously (18). The reactions were carried out using IQ Supermix (BioRad) and FAM-conjugated Assay-on-Demand (Applied Biosystems, Foster City, CA) primer/probe sets normalized to glyceraldehyde-3-phosphate dehydrogenase. Catalog numbers will be provided upon request. Data were analyzed using the Pfaffl method (19) and presented as relative expression.
Adipose tissue lysates.
The epididymal fat pads were collected at the end of the study, flash-frozen, and kept at −80°C until used. Approximately 50 mg adipose tissue was homogenized in 250 μL lysis buffer. The protein concentration in the lysate was quantified using the Lowry method with samples diluted in 0.1 N NaOH, 0.1% SDS.
IFN-γ and interleukin-4 enzyme-linked immunosorbent assay.
IFN-γ and interleukin-4 (IL-4) enzyme-linked immunosorbent assays (ELISAs) were performed on adipose tissue lysates using BD Biosciences OptEIA ELISA kits according to the manufacturer’s instructions; 50 μg total protein per well was used.
Histology.
Toluidine Blue O (TBO) staining was performed as previously described (18). Crown-like structures (CLS) from eight different fields per mouse were quantified. CD11bloF4/80lo- and CD11bhiF4/80hi-sorted cells were cytospun onto slides and stained with hematoxylin-eosin (H-E).
Stromal vascular fraction separation.
After being killed, mice were perfused with 10 mL PBS. Adipose tissue pads were then minced in 3 mL of 0.5% BSA in PBS and placed on ice. Subsequently, 3 mL of 2 mg/mL collagenase was added to the minced fat and incubated at 37°C for 20 min with shaking. The cell suspension was filtered through a 100-μm filter and then spun at 300g for 10 min to separate floating adipocytes from the stromal vascular fraction (SVF) pellet. The SVF pellet was then resuspended in 3 mL ACK lysis buffer and incubated at room temperature for 5 min for erythrocyte lysis. The cells were then washed twice with PBS.
Peritoneal cell collection.
CCR2+/+ and CCR2−/− mice were injected intraperitoneally with 3 mL of 2% thioglycollate solution. The mice were killed 72 h later, and the cells were collected by flushing the peritoneal cavity with 10 mL of 0.5% BSA in PBS. Cells were washed twice with fresh PBS before staining for flow cytometry.
Flow cytometry.
Stromal vascular cells (SVCs) and blood cells were first incubated with Fc block for 5 min at room temperature, followed by incubation for 30 min at 4°C with the fluorophore-conjugated antibodies indicated as follows: F4/80-APC (eBioscience), Ly6C-FITC (BD Biosciences), CD11b-APC-Cy7 (BD Bioscience), CD11c-PeCy7 (BD Bioscience), and Ly6G-PE (BD Bioscience). DAPI (0.2 μg/mL) was added to each sample as a viability stain 10 min before flow cytometry. Samples were processed on either a MACSQuant analyzer (Miltenyi Biotech) or a 5 Laser LSRII machine in the Vanderbilt Flow Cytometry Core and data analyzed using FlowJo software. During the analysis, all DAPI-negative cells (live) were gated (Supplementary Fig. 9) and used for further analysis with the specific fluorophores.
Cell sorting.
Cells were collected and stained as described above for the flow cytometry section. The cells were sorted using a FACSAria I sorter at the Vanderbilt flow cytometry core. The CD11bloF4/80lo and CD11bhiF4/80hi populations were isolated for RNA expression and histological analysis.
Statistics.
GraphPad Prism 4.0 software was used for all statistical analyses. Data were analyzed using two-tailed unpaired t tests to determine differences between two groups with one variable, one-way ANOVA when more than two measurements were compared, and two-way ANOVA to compare measurements with two different variables (i.e., GTTs and ITTs). Outliers were excluded from the data for each individual parameter if outside the range of the mean ± 2 SD, and P ≤ 0.05 was considered significant.
RESULTS
Weight and adiposity.
Four weeks after BMT, BM-CCR2−/− and BM-CCR2+/+ mice were placed on an HFD for 6, 12, or 20 weeks. The recipient groups had similar weight gain, adiposity, and epididymal fat pad weight throughout the study (Supplementary Fig.1). In addition, there were no differences in food intake between groups (data not shown).
Immunophenotype of circulating monocytes.
BM-CCR2−/− mice fed an HFD for 12 weeks had a sevenfold reduction (P < 0.001) in the percentage of circulating Ly6Chi monocytes (Supplementary Fig. 2A–C) compared with the BM-CCR2+/+ mice. In addition, there was a significant increase in the levels of Ly6Clo monocytes in the BM-CCR2−/− mice compared with the controls (P < 0.01) (Supplementary Fig. 2D). CCR2−/− mice fed an HFD for 12 weeks presented a similar reduction in Ly6Chi monocytes compared with CCR2+/+ controls (Supplementary Fig. 3A and E). We also analyzed the expression of CD11c, F4/80, and Ly6G in blood monocytes of CCR2+/+ and CCR2−/− mice fed either an HFD or chow diet for 12 weeks (Supplementary Figs. 3 and 4). Interestingly, the Ly6Clo population from HFD-fed CCR2−/− mice had increased CD11c expression (rightward shift of blue line in Supplementary Fig. 3F) compared with the Ly6Clo cells from HFD-fed CCR2+/+ mice (Supplementary Fig. 3B). In the general interest of assessing the effects of HFD feeding on the immunophenotype of circulating monocytes, we investigated monocytes isolated from HFD- and chow-fed CCR2+/+ mice. When comparing the Ly6Chi and Ly6Clo populations in HFD-fed CCR2+/+ mice, we found that the Ly6Chi cells have a higher expression of CD11c (Supplementary Fig. 3B). In addition, as previously shown (20,21), a comparison of HFD- and chow-fed CCR2+/+ mice showed increased expression of CD11c in Ly6Chi monocytes, whereas the expression of CD11c in Ly6Clo monocytes remained unaffected during HFD feeding (Supplementary Fig. 4B and F).
Glucose tolerance and insulin sensitivity.
To examine the systemic metabolic effects of CCR2 deficiency, we performed GTTs and ITTs. We found no differences in glucose or insulin tolerance between groups at the 6- or 12-week time points (Fig. 1A–D). However, a modest improvement in glucose tolerance was found in the BM-CCR2−/− mice (P < 0.05), 15 min post-glucose injection, after 20 weeks of HFD feeding (Fig. 1E). Fasting glucose and insulin levels were not different between groups at any time (Table 1). Similar results were obtained in the global CCR2−/− mice, in which differences in glucose tolerance were not observed until after 20 weeks of HFD feeding (Supplementary Fig. 5A–C). The discrepancy in glucose tolerance between the hematopoietic and global CCR2-deficient models can be attributed to the reduced ability of mice undergoing BMT to gain weight (22). On average, BM-CCR2−/− mice weighed <40 g after 20 weeks of HFD feeding, whereas the CCR2−/− weighed >48 g during the same HFD feeding period.
. | BM-CCR2+/+ . | BM-CCR2−/− . |
---|---|---|
Insulin (ng/mL) | ||
6 weeks | 0.54 ± 0.24 | 0.60 ± 0.19 |
12 weeks | 0.98 ± 0.76 | 1.18 ± 1.27 |
20 weeks | 1.50 ± 0.33 | 2.27 ± 0.67 |
Glucose (mg/dL) | ||
6 weeks | 156 ± 29 | 178 ± 26 |
12 weeks | 164 ± 20 | 180 ± 34 |
20 weeks | 282 ± 51 | 257 ± 30 |
. | BM-CCR2+/+ . | BM-CCR2−/− . |
---|---|---|
Insulin (ng/mL) | ||
6 weeks | 0.54 ± 0.24 | 0.60 ± 0.19 |
12 weeks | 0.98 ± 0.76 | 1.18 ± 1.27 |
20 weeks | 1.50 ± 0.33 | 2.27 ± 0.67 |
Glucose (mg/dL) | ||
6 weeks | 156 ± 29 | 178 ± 26 |
12 weeks | 164 ± 20 | 180 ± 34 |
20 weeks | 282 ± 51 | 257 ± 30 |
Data are means ± SEM. Fasting glucose levels were assessed after a 5-h morning fast. Fasting insulin was measured by ELISA (n = 7–10 mice per group).
Adipose tissue inflammatory state.
Several reports indicate that CCR2−/− mice have deficient Th1 responses characterized by a low number of antigen-presenting cells, resulting in reduced levels of the Th1 cytokine IFN-γ (8,23) or elevated IL-4 (24). Assessment of IFN-γ and IL-4 in adipose tissue lysates from BM-CCR2−/− and BM-CCR2+/+ mice showed no differences in these cytokines levels between groups after any period of HFD feeding (Supplementary Fig. 6A and B). Gene expression of the inflammatory genes Tnf and Nos2 was not different between groups at the 6- or 12-week time points (Supplementary Fig. 6C and D); however, Tnf expression was significantly lower (P < 0.05) in the BM-CCR2−/− mice after 20 weeks of HFD feeding (Supplementary Fig. 6E). Gene expression of the chemokines MCP-1 (Ccl2) and macrophage inflammatory protein-1α (Ccl3) was significantly increased after 6 (P < 0.01; P < 0.001) and 12 weeks (P < 0.05; P < 0.01) in the BM-CCR2−/− mice; no differences in the expression of these chemokines were observed in the mice fed an HFD for 20 weeks.
The BM-CCR2−/− mice presented a threefold increase in the expression of the anti-inflammatory marker arginase 1 (Arg1) at 6 and 12 weeks (P < 0.01); expression of this gene was not different between groups after 20 weeks of HFD feeding (Supplementary Fig. 6C–E). Expression of Cd163, mgl1, mgl2, and Il10 was assessed, and no consistent changes were observed in these “M2” or anti-inflammatory genes between groups after any period of HFD feeding.
Immune cell infiltration into adipose tissue.
CLS, which are an accumulation of immune cells around presumably dead adipocytes, were used to quantify immune cell infiltration in TBO-stained sections of adipose tissue (Fig. 2A–F). There were no differences in numbers of CLS between BM-CCR2−/− and BM-CCR2+/+ mice after 6 or 12 weeks of HFD feeding (Fig. 2G). However, there was a significant decrease (P < 0.05) in CLS in the BM-CCR2−/− mice after 20 weeks of HFD feeding (Fig. 2E–F). Further analysis of TBO-stained adipose tissue from mice fed an HFD for 20 weeks showed that the number of CLS is dependent on the body weight of the mouse. There were no differences between weight-matched BM-CCR2−/− and BM-CCR2+/+ mice when they weighed <40 g; conversely, there was a decrease in CLS in the BM-CCR2−/− mice compared with weight-matched BM-CCR2+/+ controls at >45 g body weight (Supplementary Fig. 7). In accordance with our histological data, there were no differences in the expression of CD68 (Cd68), CD11b (Itgam), or CD11c (Itgax) in adipose tissue of BM-CCR2−/− compared with BM-CCR2+/+ mice after 6 or 12 weeks of HFD feeding (Fig. 2H and I); however, the expression of these genes was significantly lower (P < 0.05) for the BM-CCR2−/− mice by the 20-week time point compared with the control group (Fig. 2J). The expression of F4/80 (Emr1) in adipose tissue of BM-CCR2−/− mice was significantly (P < 0.05) increased after 6 weeks and unchanged after 12 weeks, and there was a trend (P < 0.07) toward a decrease after 20 weeks of an HFD in comparison with the BM-CCR2+/+ controls (Fig. 2H–J).
CCR2 deficiency leads to the aberrant accumulation of F4/80lo cells in adipose tissue.
Based on our observation that there were no changes in immune infiltration in adipose tissue after 6 and 12 weeks of HFD feeding despite the dramatic reduction in circulating inflammatory cells, we sought to further characterize the macrophages in adipose tissue by flow cytometry. To this end, we performed careful evaluation of the F4/80+ cells within adipose tissue. Based on our gating strategy, the BM-CCR2+/+ mice contained the expected F4/80− and F4/80hi populations (Fig. 3A). Interestingly, the BM-CCR2−/− mice accumulated two discrete myeloid populations in their adipose tissue: F4/80lo and F4/80hi (Fig. 3B). The discrete F4/80lo population was observed in the SVF of the BM-CCR2−/− at the 6-week (Fig. 3B and C) and 12-week (Fig. 3C and Supplementary Fig. 8B) time points but was no longer detected after 20 weeks of HFD feeding (Fig. 3C and Supplementary Fig. 8D).
Real-time RT-PCR analysis of RNA isolated from whole adipose tissue of BM-CCR2−/− mice showed that the presence of the F4/80lo population at 6 and 12 weeks was associated with increases in gene expression of myeloperoxidase (Mpo) (P < 0.05), CXCR2 (IL8rb) (P < 0.01), Cxcl1 (P < 0.001), and Cxcl2 (P < 0.05) (Fig. 3D and E). No differences in the expression of these genes were observed after 20 weeks of an HFD, when the CD11bloF4/80lo cells were no longer detected (Fig. 3F).
Similar to the analysis of the BM-CCR2−/− model, analysis of the SVCs of the CCR2−/− mice showed that CD11bloF4/80lo cells aberrantly accumulate in the adipose tissue of global CCR2−/− mice with similar kinetics. The cells accumulated in the adipose tissue of CCR2−/− mice after 6 or 12 weeks of HFD feeding (Fig. 4D and E), whereas no discrete population was found in the CCR2+/+ controls (Fig. 4A and B). Additionally, after 20 weeks of HFD feeding, the CD11bloF4/80lo cells were no longer found in the adipose tissue of CCR2−/− mice (Fig. 4F). Note that there is downward shift of the double-positive populations in both the CCR2+/+ and CCR2−/− mice after 20 weeks of HFD feeding (Fig. 4C and F). Identical flow conditions were used as in the earlier time points. Therefore, we hypothesize that the difference may be due to the excess lipid interference with signal detection; nonetheless, only one population of CD11b+F4/80+ cells was found in both genetic models at this time point. Global CCR2−/− mice were used for subsequent characterization of the CD11bloF4/80lo cells.
CD11bloF4/80lo cell characterization.
CD11bhiF4/80hi and CD11bloF4/80lo cells were isolated from the adipose tissue of CCR2−/− mice 10 weeks post–HFD feeding (Fig. 5A). Gene expression analysis was performed for both populations and is presented as expression relative to the CD11bhiF4/80hi cells (Fig. 5B). Expression of Emr1 (P < 0.001) was significantly lower for CD11bloF4/80lo cells, as expected based on our gating strategy. The expression of Arg1 (P < 0.05), Cd163 (P < 0.001), Il10 (P < 0.01), Il6 (P < 0.01), Ccr5 (P < 0.05), Ccl2 (P < 0.01), and Csfr1 (P < 0.001) was significantly lower in the CD11bloF4/80lo cells in comparison with the CD11bhiF4/80hi macrophages. Expression of Itgax (P < 0.01), Nos2 (P < 0.05), Ccl5 (P < 0.01), Il8rb (P < 0.05), and Csf1 (P < 0.05) was significantly higher for the CD11bloF4/80lo cells. H-E staining of the sorted CD11bloF4/80lo cells showed that these cells have a monocytic morphology and are smaller in size than the CD11bhiF4/80hi macrophages (Fig. 5C). Despite several reports of neutrophilia at sites of inflammation in CCR2−/− mice (25,26), flow cytometry analysis of the CD11bloF4/80lo population in the global CCR2−/− mice showed that these cells are Ly6G negative and therefore unlikely to be neutrophils (Supplementary Fig. 9).
CD11bloF4/80lo cells are transiently found in CCR2+/+ mice.
To determine the physiological relevance of the CD11bloF4/80lo cells in wild-type conditions, we isolated CD11bloF4/80lo and CD11bhiF4/80hi macrophages from CCR2+/+ mice fed an HFD for 7 weeks. Despite the absence of two discrete populations in the CCR2+/+ mice, we were able to isolate CD11bloF4/80lo cells and found that they have gene expression (Fig. 6B) and morphology (Fig. 6C) similar with the CD11bloF4/80lo cells isolated from the adipose tissue of CCR2−/− mice. In addition, we assessed whether HFD feeding was necessary for the accumulation of CD11bloF4/80lo cells. The expression of several key genes in the CD11bloF4/80lo cells isolated from chow-fed CCR2−/− and CCR2+/+ mice (Supplementary Fig. 10) was similar to that of CD11bloF4/80lo cells isolated from the HFD mice (Figs. 5B and 6B).
CD11bloF4/80lo aberrantly accumulate in the peritoneum of thioglycollate-injected CCR2−/− mice.
To determine whether CD11bloF4/80lo cells accumulated in other metabolic relevant tissues during obesity or in an acute model of inflammation, we assessed the accumulation of these cells in the livers of HFD-fed and thioglycollate-injected CCR2−/− mice, respectively. We did not observe the accumulation of these cells in livers of HFD-fed mice, but we found a decrease in the number of recruited inflammatory macrophages as recently shown by Obstfeld et al. (27) (data not shown). In contrast, we observed a marked aberrant accumulation of CD11bloF4/80lo and a decrease in CD11bhiF4/80hi cells in the peritoneum of CCR2−/− mice compared with CCR2+/+ mice at 72 h after thioglycollate injection (Supplementary Fig. 11).
DISCUSSION
CCR2−/− mice have been shown to have a modest reduction in macrophage recruitment to adipose tissue after extended periods of HFD feeding (11,13). Given that the pool of circulating inflammatory monocytes is strikingly reduced in these mice (9) (Supplementary Figs. 2B and C and 3E), inflammatory macrophage recruitment to adipose tissue should be correspondingly decreased. However, inflammation and inflammatory macrophage recruitment in CCR2−/− mice is only partially decreased and only after substantial obesity is attained (11,13). Our current studies explored mice with global and hematopoietic CCR2 deficiency at different periods of HFD feeding to investigate the mechanism involved in the delayed improvement in inflammation and insulin sensitivity.
To understand processes that take place at earlier stages of obesity that could prevent improvements in inflammation and insulin resistance, we analyzed myeloid populations in adipose tissue of BM-CCR2−/− and CCR2−/− mice and control mice after 6, 12, and 20 weeks of an HFD by flow cytometry. Our data provide evidence for the accumulation of a unique CD11bloF4/80lo myeloid population in the adipose tissue of BM-CCR2−/− (Fig. 3B and C and Supplementary Fig. 8B) and CCR2−/− (Fig. 4D and E) mice after 6 and 12 weeks of HFD feeding. These CD11bloF4/80lo cells are no longer detected in the adipose tissue of the BM-CCR2−/− or CCR2−/− mice after extended periods (>20 weeks) of HFD feeding (Figs. 3C and 4C and F and Supplementary Fig. 8C and D). Our data suggest that the presence of the CD11bloF4/80lo cells is associated with the mechanism(s) for the delay in improvements in inflammation and insulin resistance observed in these mice and that substantial obesity and disappearance of the CD11bloF4/80lo cells are required before metabolic improvements are attained.
To determine the identity and inflammatory phenotype of the novel CD11bloF4/80lo cells, we isolated RNA from these cells for gene expression characterization. Altogether, the expression data suggest that the CD11bloF4/80lo cells do not exhibit classical M1 or M2 expression profiles but, instead, have a unique inflammatory expression in contrast with that of CD11bhiF4/80hi macrophages. Chomarat et al. (28) showed that during monocyte to macrophage differentiation there is an upregulation of CD115 (Csf1r) coupled by a downregulation of its ligand, M-CSF (Csf1). We measured the expression of both Csfr1 and Csf1 and found that Csfr1 was downregulated while Csf1 was upregulated in the CD11bloF4/80lo cells, suggesting that they are in a monocytic transition stage. To test this, we found that CD11bloF4/80lo cells have a monocytic morphology—in contrast with the CD11bhiF4/80hi cells, which have a morphology characteristic of macrophages (Fig. 5C). The expression profile and morphological analysis indicate that the CD11bloF4/80lo cells are inflammatory monocytes that accumulate in adipose tissue in the absence of CCR2.
We also sought to investigate the presence of the CD11bloF4/80lo cells in wild-type mice independent of CCR2 deficiency. We isolated CD11bloF4/80lo and CD11bhiF4/80hi macrophages from CCR2+/+ mice using identical gates as previously used for the CCR2−/− mice (Fig. 6A). Despite the absence of a discrete population under the CD11bloF4/80lo (P2) gate, we were able to isolate a relatively high number of cells. Gene expression analysis of key markers of CD11bloF4/80lo cells previously described showed that CD11bloF4/80lo cells isolated from CCR2+/+ mice have gene expression patterns identical with those of the CCR2−/− mice (Fig. 6B). Additionally, H&E staining of isolated cells showed a similar monocytic morphology between the CCR2−/− and CCR2+/+ CD11bloF4/80lo cells (Figs. 5C and 6C). Moreover, we found that HFD feeding was not necessary for the accumulation of these cells in adipose tissue; CD11bloF4/80lo cells were found in chow-fed CCR2+/+ and CCR2−/− mice (Supplementary Fig. 10). Next, we assessed whether these cells were found in the liver and the peritoneal cavity after thioglycollate injection. We did not observe their accumulation in the liver; however, we did observe the decrease in recruited macrophages without effects on resident Kupffer cells, as recently described by Obstfeld et al. (27) (data not shown). On the other hand, we found a marked accumulation of CD11bloF4/80lo cells in the peritoneum of thioglycollate-injected CCR2−/− mice in comparison with CCR2+/+ controls (Supplementary Fig. 11A and B). In addition, we were able to clearly observe the accumulation of the CD11bloF4/80lo cell population in the CCR2+/+ mice, albeit at lower levels than in the CCR2−/− mice (Supplementary Fig. 11B). These findings indicate that although CCR2 deficiency is not necessary for the recruitment of CD11bloF4/80lo cells to adipose tissue and the peritoneum, it leads to the aberrant accumulation of these cells, which would otherwise only transiently be found at these inflammatory sites.
The mechanism for the accumulation and eventual disappearance of the CD11bloF4/80lo cells is unknown. A recent study using antibody treatment against colony stimulating factor 1 receptor (CSFR1) resulted in the accumulation of inflammatory F4/80lo myeloid cells and a decrease in the differentiated F4/80hi macrophages in lymphoid tissues (29). CSFR1 is involved in the differentiation of monocytes to macrophages (28); thus, inhibition of this receptor led to the accumulation of undifferentiated monocytes (29). Similarly, CCR2 and its ligand CCL2 have been shown to play a role in monocyte to macrophage differentiation (30). Thus, it is possible that just as in CSFR1 antagonism, CCR2 deficiency leads to the aberrant accumulation of CD11bloF4/80lo myeloid cells in adipose tissue. Accordingly, a possible mechanism for their disappearance is the eventual differentiation of the CD11bloF4/80lo myeloid cells into CD11bhiF4/80hi macrophages in adipose tissue during advanced stages of obesity, when no active recruitment is taking place. Future studies will address the origin and mechanism(s) of recruitment and disappearance of the CD11bloF4/80lo cells.
In conclusion, our data provide evidence for the aberrant accumulation of the novel CD11bloF4/80lo myeloid cells in the adipose tissue of CCR2−/− mice, which is associated with delayed improvements in inflammation and glucose tolerance in this mouse model during obesity. We found that the CD11bloF4/80lo cells are also found in the adipose tissue of CCR2+/+ mice (albeit in lower numbers) and that an HFD is not required for their recruitment into adipose tissue. The gene expression and histological analysis of the CD11bloF4/80lo cells indicate that these cells have a monocytic nature. In addition, their transient accumulation in the adipose tissue of CCR2+/+ mice suggests that the CD11bloF4/80lo cells are in a transition stage and that CCR2 is necessary for their differentiation into CD11bhiF4/80hi macrophages. Our study has provided in vivo evidence for the role of CCR2 in macrophage recruitment and differentiation in adipose tissue and has highlighted the metabolic consequences of the aberrant accumulation of undifferentiated CD11bloF4/80lo cells in this tissue. The discovery of this potentially highly inflammatory myeloid population in the adipose tissue of CCR2−/− mice also has implications for the use of CCR2 antagonism as a therapy for chronic inflammatory diseases because it could worsen inflammation during early stages of the inflammatory process (31,32). Future and ongoing studies will investigate the role of these cells in different inflammatory settings.
ACKNOWLEDGMENTS
This work was supported by National Institutes of Health (NIH) Grant HL-089466 and a supplement to this grant to D.A.G. D.A.G. also has received support from Molecular Endocrinology Training Grant NIH DK-07563-23. A.H.H. has received the American Diabetes Association Career Development Award 1-07-CD-10. Flow cytometry experiments were performed in the Vanderbilt Flow Cytometry Shared Resource, supported by the Vanderbilt Digestive Disease Research Center (DK-058404).
No potential conflicts of interest relevant to this article were reported.
D.A.G. collected and analyzed the data and wrote the manuscript. A.K., J.S.O., E.K.A., and C.D.W. assisted with data collection and reviewed the manuscript. W.K.G. assisted with data collection. A.H.H. assisted with data analysis, provided important intellectual contributions, and wrote and edited the manuscript.
The authors thank the Major Laboratory (Vanderbilt) for their assistance with flow cytometry and ELISA experiments, the Southard-Smith Laboratory (Vanderbilt) for their assistance with RNA isolation from sorted cells, and Laura Price from the VUMC Cytopathology Laboratory for her assistance with cytospins.