Funding agencies: This work was supported by grants from the Canadian Diabetes Association and the Canadian Institutes of Health Research (to AK). LNF was funded by a postdoctoral fellowships from Novo Nordisk A/S. TEJ was funded by a Danish Research Council postdoctoral fellowship. SRC was supported by a Mitacs fellowship. Human muscle biopsy collection was supported by Deutsche Forschungsgemeinschaft, Clinical Research group “Atherobesity” (KFO152; BL 833/1-1). We thank Dr. Assaf Rudich for his initial stimulus of this study and Zhi Liu for excellent technical assistance.Additional Supporting Information may be found in the online version of this article.
Disclosure: LNF and AS are employed by Novo Nordisk and own stocks. All other authors declared no conflict of interest.
Author Contributions: LNF designed the study, researched and analyzed data and wrote the manuscript; SRC researched and analyzed data and contributed to discussion and editing of manuscript; YSL, TEJ, AO, and MB researched data and contributed to discussion; PJB analyzed data and contributed to discussion and editing of manuscript; JMO and AS contributed to discussion and editing of manuscript; AK designed the study, analyzed data and wrote the manuscript. All authors approved the submitted and published versions.
In obesity, immune cells infiltrate adipose tissue. Skeletal muscle is the major tissue of insulin-dependent glucose disposal, and indices of muscle inflammation arise during obesity, but whether and which immune cells increase in muscle remain unclear.
Immune cell presence in quadriceps muscle of wild type mice fed high-fat diet (HFD) was studied for 3 days to 10 weeks, in CCL2-KO mice fed HFD for 1 week, and in human muscle. Leukocyte presence was assessed by gene expression of lineage markers, cyto/chemokines and receptors; immunohistochemistry; and flow cytometry.
After 1 week HFD, concomitantly with glucose intolerance, muscle gene expression of Ly6b, Emr1 (F4/80), Tnf, Ccl2, and Ccr2 rose, as did pro- and anti-inflammatory markers Itgax (CD11c) and Mgl2. CD11c+ proinflammatory macrophages in muscle increased by 76%. After 10 weeks HFD, macrophages in muscle increased by 47%. Quadriceps from CCL2-KO mice on HFD did not gain macrophages and maintained insulin sensitivity. Muscle of obese, glucose-intolerant humans showed elevated CD68 (macrophage marker) and ITGAX, correlating with poor glucose disposal and adiposity.
Mouse and human skeletal muscles gain a distinct population of inflammatory macrophages upon HFD or obesity, linked to insulin resistance in humans and CCL2 availability in mice.
Inflammation is an important factor in the development of insulin resistance and type 2 diabetes mellitus (T2DM) . Cells of hematopoietic origin (i.e., leukocytes) are strikingly abundant in adipose tissue (AT) of insulin-resistant humans  and mice . Macrophages are the best-characterized cell type in AT of high fat diet-fed (HFD) mice and obese humans, although AT macrophages coexist with multiple leukocyte subsets [4-6]. Importantly, selective phenotypes and subgroups of these leukocytes correlate with peripheral insulin resistance [7, 8]. Accordingly, the presence of conventionally activated macrophages, neutrophils, mast cells, dendritic cells, B cells, noninvariant natural killer T cells and effector CD4+ and CD8+ T cells in AT is considered detrimental to normal insulin action [9, 10]. In contrast, AT regulatory CD4+ T cells, invariant natural killer T cells, eosinophils, and alternatively activated macrophages are believed to promote insulin sensitivity [11-13].
Although the increase in AT immune cells in obesity has been well-documented, there are contradicting reports as to whether or not immune cells infiltrate skeletal muscle in the context of overnutrition and insulin resistance. The discrepancy surrounding the presence of macrophages in skeletal muscle is illustrated in Table 1 and may originate from the low abundance of these cells compared to those detected in AT, from the use of a single method in each case to detect macrophages and/or from biological variability. Given the paramount importance of skeletal muscle for whole-body disposal of dietary glucose, we reasoned that there is a clear need to re-examine this phenomenon, especially during early time points following HFD feeding, and a need for more robust phenotyping of muscle macrophages during overnutrition conditions. Further, the inflammatory status of muscle macrophages has not been described even in the studies that find positive evidence of their presence in this tissue in conditions of obesity and insulin resistance, nor have these cells been isolated or characterized. Hence, it is imperative to analyze the inflammatory status of those cells in skeletal muscle, if found. Finally, it is unknown if there is chemoattraction and net recruitment of leukocytes to muscle in the course of HFD feeding, but precedence for this exists in injured muscle, where macrophages are recruited in response to the chemokine CCL2/MCP-1 .
Table 1. Reports studying macrophages in muscle in HFD and obesity
Lean IS vs. obese and IS vs. obese IR (vastus lateralis)
qPCR: CD11b, CD11c
Here we test the hypothesis that a distinct spectrum of leukocytes is present within muscle beds when mice are subjected to HFD, and that proinflammatory leukocytes are recruited in response to CCL2/MCP-1. To this end, we investigated the immune cell populations in mouse muscle during the first weeks of HFD-induced insulin resistance through analysis of gene expression, immunohistochemistry (IHC), cell extraction and quantification, along with characterization of their phenotypic polarization in both WT and CCL2-knockout mice. The results reveal an early increase in gene expression of inflammatory markers and a selective increase in macrophages, monocytes and neutrophils, which is mitigated in CCL2-KO mice. We buttress these results with evidence of inflammatory macrophages in muscle of obese, insulin-resistant individuals.
Mice and experimental protocols
All animal protocols were approved by the animal care committees at the University of California (experiment 1) and the Hospital for Sick Children (experiments 2 and 3). C57BL/6J wild-type (WT) and CCL2-KO mice (B6.129S4-Ccl2tm1Rol/J) were from, The Jackson Laboratory (Bar Harbor, ME, USA). Mice were fed control diet (CD, 10% energy from fat) or high fat diet (HFD, 60% energy from fat) from Research Diets, Inc. (New Brunswick, NJ, USA). Experiment 1 was previously described  and muscle was analyzed from mice fed HFD for 3 days, 1, 2, 5, and 10 weeks. Experiment 2 consisted of 3 groups, fed CD for 10 weeks; CD for 9 weeks followed by 1 week HFD; and HFD for 10 weeks, respectively. Experiment 3 compared WT and CCL2-KO mice fed CD or HFD for 1 week. Where indicated, glucose tolerance tests (GTTs) were performed 3 days prior to euthanasia. Mice were fasted for 6 hours before i.p. injection of 10% glucose in 0.9% saline (1 g/kg body weight), and tail vein blood glucose was measured. For insulin tolerance tests (ITTs), mice were injected i.p. with 1 U/kg body weight Humulin R (Eli Lilly, Indianapolis, IN, USA) in 0.9% saline. Within each experiment, mice were euthanized at same age (17 weeks old for HFD feeding up to 10 weeks; 11-weeks old for 1 week of HFD feeding in CCL2-KO and WT animals), blood was collected, and quadriceps muscle and epididymal white adipose tissue (eWAT) were excised.
Human muscle biopsies
The study was approved by the Ethics committee of the University of Leipzig, Germany. Individuals were divided into groups of normal glucose tolerance (NGT, n = 10) and type 2 diabetes mellitus (T2DM, n = 4) on the basis of 75 g oral GTTs (OGTTs) according to the American Diabetes Association criteria. Muscle biopsies and blood samples were collected after overnight fasting. Insulin sensitivity was assessed by hyperinsulinemic-euglycemic clamps  and expressed as glucose disposal rate (Rd). Skeletal muscle biopsies (<100 mg) were obtained under local anesthesia from the right vastus lateralis muscle and immediately snap-frozen.
Quadriceps muscle and eWAT obtained at 0 (n = 10 mice), 1 (n = 10 mice), and 10 weeks (n = 5 mice) of high-fat feeding were enzymatically digested at 37° with 0.2 WU/ml Liberase TM (Roche, Laval, Quebec, Canada) for 30 and 60 min, respectively, and strained through 70-µm pore filters. In eWAT and peripheral blood suspensions, erythrocytes were lysed prior to analysis. Samples were analyzed on a FACS Canto II (BD Biosciences) using antibodies to: CD16/CD32, CD3-PE, CD4-PerCP-Cy5.5, CD8-APC-eFluor780, CD11b-PE-Cy7, CD11c-PerCP-Cy5.5 (eBioscience); CD45-FITC, Ly6C-HorizonV450, Ly6G-HorizonV450, SiglecF-PE (BD Biosciences, San Jose, CA, USA); F4/80-APC and Ly6B.2-PE (AbD Serotec, Oxford, UK).
RNA was isolated from muscle obtained at 0, 3 days, 1, 2, 5, and 10 weeks (n = 4 mice) of high fat feeding using Trizol/chloroform, diluted 1:1 in RNeasy RLT buffer and run on RNeasy columns (Qiagen, Valencia, CA, USA). Single cell suspensions obtained as for flow cytometry were subjected to Lympholyte M (Cedarlane, Burlington, ON, Canada) gradient centrifugation to obtain leukocytes. These were lysed in RLT buffer and run on RNeasy columns. cDNA was synthesized using the VILO kit (Life Technologies). qRT-PCR was performed using gene expression assays (Supporting Information Table S1) and Gene Expression Mastermix (Life Technologies). Mouse Abt1 was employed as reference gene, selected within the Genevestigator database  as the least regulated gene in mouse muscle in microarray studies. For human samples, RIT1 was the reference based on similar criteria.
Muscle tissue was fixed in formalin, stained with antibodies against mouse: CD3 (polyclonal, Sigma, St. Louis, MO, USA), F4/80 (clone CI:A3-1, Abcam, Cambridge, UK), Ly6B.2 (“anti-mouse neutrophils” clone 7/4, Cedarlane), CD11c (clone N418, Abcam) or human: CD68 (clone Y1/82A, BD Biosciences), and CD11c (clone 3.9, eBioscience) followed by appropriate secondary antibodies, Vectastain ABC System and DAB substrate (Vector Laboratories, Burlingame, CA, USA).
Insulin-stimulated phosphorylation of Akt in vivo
CCL2-KO and WT mice (n = 4) were fed HFD or CD for 9 days, then fasted for 4 h and i.p. injected 1 U insulin/kg (or saline). After 20 min, mice were euthanized, quadriceps muscles collected, lysed and analyzed by SDS-PAGE. Akt and pAkt S473 were probed by immunoblotting (antibodies #9271 and #2920, Cell Signaling, Danvers, MA, USA). Fluorescent goat anti-mouse IRDye 800CW and antirabbit IRDye 680LT antibodies (#926-32210 and #926-68021, LI-COR, Lincoln, NE, USA) were used to detect pAkt and Akt in the same blot. Bands were visualized by Odyssey 9120 Infrared Imaging System and quantified with Odyssey Application Software 2.1 (LI-COR).
Mouse gene expression and GTT data were analyzed by one-way ANOVA. Flow cytometry data on quadriceps and eWAT tissue were analyzed by two-way ANOVA. Human IHC data were analyzed by Mann–Whitney U test, and human expression data by Spearman Rank correlation analysis, all using GraphPad Prism 6. P < 0.05 was considered statistically significant.
Expression of inflammatory cell genes increases in muscle during high fat diet feeding
We recently analyzed inflammation in AT of C57BL/6J mice fed HFD from 3 days to 10 weeks . By 3 days of HFD, these mice showed marked reductions in insulin sensitivity indices (glucose infusion rate (GIR) and insulin-stimulated glucose disposal rate (IS-GDR) by hyperinsulinemic-euglycemic clamps, and these values further dropped at 1-10 weeks, suggesting that muscle already has lower insulin response at early times of HFD. Here we used quadriceps muscles from the same mice to analyze their inflammatory markers. We observed a 10-fold increase in muscle gene expression of Ly6b (surface marker of neutrophils, monocytes and macrophages)  after only 3 days and throughout the duration of HFD. Gene expression of the proinflammatory cytokine Tnf, the chemokine Ccl2 and its receptor Ccr2, was elevated throughout the diet and peaked at 1 week (Figure 1A). Expression of Emr1 (encoding the macrophage marker F4/80) rose >2-fold, peaking at 1 week and 10 weeks of HFD. The same was true for the hematopoietic marker B2m. Tgfb1 (produced by alternatively activated macrophages) and Tgfb2 (highly expressed in skeletal, heart and smooth muscle ) peaked at 1 week and remained elevated throughout the study (Figure 1B). Itgax (CD11c) and Mgl2 (CD301b), denoting respectively “M1” (classically activated) and “M2” (alternatively activated) macrophages, showed a significant 2-3-fold increase at 10 weeks of HFD (Figure 1C). CD11c also tended to increase early paralleling F4/80, but the change was not statistically significant (Figure 1C). On the other hand, Mpo (present in neutrophils), Cd3e (present in T-cells), Il10, and hypoxia markers Vegfa and Hif1a did not change (not shown). Overall, our results suggest that early on during HFD feeding, skeletal muscle gains leukocytes along with a proinflammatory (Tnf, Ccl2) and to a lesser degree an alternatively activated (Tgfb1) gene expression profile. By 10 weeks, there was a clear increase in markers of both M1 and M2 macrophage polarization within skeletal muscle.
Inflammatory cells are present in skeletal muscle in mice fed a HFD
To identify the inflammatory cells within muscle and compare them with those in eWAT, in a separate experiment, mice were fed CD for 10 weeks, CD for 9 weeks, followed by 1 week HFD, or HFD for 10 weeks. Mice fed HFD had significant body weight gain at 1 and 10 weeks. Glucose tolerance was significantly impaired at 1 week, and severely at 10 weeks of HFD (Figure 2A). Quadriceps muscle weight was not changed by HFD feeding, but eWAT weight increased as expected at 1 and 10 weeks (Figure 2B, left panel). The number of leukocytes (CD45+ cells) detected by flow cytometry after collagenase tissue-digestion was elevated in both quadriceps muscle and eWAT after 10 weeks of HFD (Figure 2B, right panel). IHC of muscle from all CD and 1-week or 10-week HFD mice revealed cells expressing Ly6B.2 and F4/80 antigen (macrophages) located between muscle fascicles (fiber bundles), fibers and in fat cell depots within muscle (Figure 2C). The T-cell specific molecule CD3ε was mainly apparent in the muscle vasculature (not shown).
To quantify muscle leukocytes, single cell suspensions of quadriceps tissue and eWAT were analyzed for expression of macrophage, monocyte, and T cell markers by flow cytometry. Within the leukocyte population (CD45+ cells), the main immune cells in muscle from all animals were CD11b+F4/80+ macrophages, which on average represented more than 50% of the leukocytes (Table 2, representative experiment in Figure 3A). This population constituted a smaller proportion of the leukocytes in eWAT of CD-fed animals (30%), but rose to a percentage comparable to that in muscle (47%) by 10 weeks of HFD (Table 2, representative experiment in Figure 3A). Initial numbers of CD11b+F4/80+ macrophages/g tissue were comparable between quadriceps and eWAT; but by 10 weeks HFD, eWAT was populated by twice as many macrophages as muscle (Figure 3B, left panel). A significant increase in number of CD11c-expressing pro-inflammatory macrophages within the CD11b+F4/80+ cells in muscle occurred by 1 week HFD, and remained stable for 10 weeks (Figure 3B, right panel). In contrast, this CD11c subset of proinflammatory macrophages within the CD11b+F4/80+ population of macrophages rose steadily in eWAT, to 4-fold at 10 weeks HFD (Figure 3B, right panel). Due to low numbers of isolated cells, we did not assess M2 markers by flow cytometry; instead, to further characterize macrophage polarization, leukocytes were first isolated by gradient centrifugation from skeletal muscle- and AT-derived cell suspensions and then processed for qPCR. Consistent with the results from total tissue RNA, Mgl2 expression in muscle-derived leukocytes peaked at 10 weeks HFD, but declined in eWAT between 1 and 10 week(s) of HFD (Figure 3C). Itgax (CD11c) expression was elevated in both muscle and eWAT leukocytes at 10 weeks of HFD. Finally, CD11c+ cells were detected by IHC in all experimental groups, within and between muscle fiber bundles, and in adipocyte deposits adjacent to muscle fibers (Figure 3D).
Table 2. Cell types in quadriceps muscle, epididymal white adipose tissue and peripheral blood, expressed as percentage of CD45+ leukocytes.
Percentage of CD45+ cells (mean and range)
Significant changes with diets are indicated, *P < 0.05 compared to Control diet (CD), **P < 0.05 compared to high-fat diet 1 week (HFD1). Significant differences (P < 0.05) between cell percentages in quadriceps muscle and epididymal white adipose tissue (WAT) indicated with ***.
CD: 56 (48-65)%
CD: 30 (21-41)% ***
CD: 2.4 (1.5-4.0)%
HFD1: 60 (54-67)%
HFD1: 33 (27-42)% ***
HFD1: 2.3 (0.72-4.3)%
HFD10: 53 (51-58)%
HFD10: 47 (40-57)% *,**
HFD10: 1.4 (0.80-2.1)%
CD11b+F4/80− Myeloid” cells
CD: 16 (12-20)%
CD: 29 (24-22) %***
CD: 25 (20-29)%
HFD1: 17 (13-21)%
HFD1: 35 (28-44)% *,***
HFD1: 18 (6-30)%
HFD10: 17 (15-19)%
HFD10: 28 (23-33)% **,***
HFD10: 28 (25-32)%
CD: 3.5 (2.6-4.8)%
CD: 5.7 (4.7-7.3)% ***
CD: 1.8 (1.3-2.2)%
HFD1: 3.5 (2.5-5.6)%
HFD1: 6.2 (3.9-8.1)% ***
HFD1: 1.7 (0.65-2.5)%
HFD10: 4.6 (3.3-5.9)%
HFD10: 5.2 (4.7-6.5)%
HFD10: 5.2 (4.9-5.5)%*,**
CD: 1.7 (1.2-3.9)%
CD: 0.31 (0.064-0.61)% ***
CD: 1.8 (1.1-1.6)%
HFD1: 2.1 (0.71-4.6)%
HFD1: 0.73 (0.17-1.7)% *,***
HFD1: 4.6 (1.6-10)%
HFD10: 2.2 (1.9-2.6)%
HFD10: 0.25 (0.16-0.36)% ***
HFD10: 13 (11-15)%*,**
CD11b−F4/80− Non-myeloid” CD45+ cells (including T cells)
CD: 25 (18-32)%
CD: 39 (32-45)% ***
CD: 71 (66-77)%
HFD1: 20 (16-23)%
HFD1: 30 (24-39)%*,***
HFD1: 79 (67-93)%
HFD10: 27 (23-29)%**
HFD10: 24 (19-30)%*
HFD10: 70 (66-74)%
CD3+ T cells
CD: 11 (6.7-13)%
CD: 25 (19-31)% ***
CD: 30 (28-33)%
HFD1: 32 (31-33)%
HFD10: 12 (9-14)%
HFD10: 23 (17-32)%
CD3+CD4+ T cells
CD: 5.6 (3.3-6.7)%
CD: 14 (11-17)% ***
CD: 19 (17-21)%
HFD1: 6.0 (3.1-8.6)%
HFD1: 10 (8-12)%*,***
HFD1: 19 (19-21)%
HFD10: 5.4 (3.8-8.6)%
HFD10: 6.4 (5.0-7.1)% *,**
HFD10: 23 (17-32)%
CD: 0.94 (0.38-1.4)%
CD: 6.4 (4.5-8.4)% ***
CD: 1.2 (0.89-1.4)%
HFD1: 2.7 (1.2-5.8)%*
HFD1: 5.0 (2.9-8.4)% ***
HFD1: 1.2 (0.97-1.5)%
HFD10: 2.4 (1.6-4.0)%
HFD10: 4.2 (3.3-6.0)%
HFD10: 0.57 (0-0.89)%
Monocyte and neutrophil count rises in mouse muscle, adipose tissue, and blood in response to HFD
Monocytes and neutrophils are recruited to AT from blood during HFD feeding . By FACS analysis, the frequency of Ly6B.2hi monocytes (also expressing Ly6C, data not shown) and neutrophils within the CD11b+F4/80− myeloid cells was greatly enhanced in blood during the course of HFD feeding (Table 2 and Figure 4A). The CD11b+F4/80- myeloid cells were less frequent in muscle than in eWAT irrespective of diet (Table 2 and Figure 3A), and although absolute numbers of total CD11b+F4/80− myeloid cells increased in both muscle and eWAT after 10 weeks of HFD, they remained ∼3-fold higher in eWAT than muscle in all three conditions (Figure 4B). The gain in Ly6B.2hi monocytes/g showed similar time- and tissue-dependent patterns for muscle and eWAT, yet monocyte numbers were greater in eWAT (Figure 4C). In contrast, Ly6B.2+ Ly6G+ neutrophils/g were more numerous in quadriceps muscle than eWAT, peaking at 10 weeks of HFD, whereas eWAT neutrophil numbers peaked at 1 week (Figure 4D). Neutrophils constituted 1.7% of all leukocytes in muscle of CD-fed mice and this fraction remained higher in muscle than in eWAT in all diet groups (Table 2). Not only denoting M1 macrophages, in the absence of F4/80, CD11c is also a dendritic cell marker; however, CD11c+F4/80− (dendritic) cells did not increase in muscle, while they were elevated in eWAT (Figure 4E). Interestingly, HFD lowered the percentage of T cells (CD3+) within the CD45+ leukocytes in eWAT but not in muscle (Table 2). In contrast, CD3+CD4−CD8− (potentially NKT cells) increased from 1.0% to 2.4% selectively in muscle after 1 week of HFD (Table 2). Overall, leukocytes within muscle also included monocytes and neutrophils, and like macrophages, the numbers of monocytes and neutrophils increased in muscle and eWAT with HFD.
CCL2 knockout prevents the HFD-induced increase in muscle macrophages
We hypothesized that CCL2 may be involved in the increased macrophage and myeloid cell presence in HFD-fed mice, since gene expression of both Ccl2 and its receptor Ccr2 was higher after 3 days and 1 week of HFD compared to CD (Figure 1A). To test this prediction, we measured macrophages and other myeloid cells by flow cytometry in muscle of CCL2-KO mice after 1 week on HFD compared to WT and CCL2-KO mice fed CD.
Despite no increase in muscle weight (Figure 5A), the number of total macrophages (CD11b+F4/80+) and inflammatory macrophages (CD11c+CD11b+F4/80+) was lower in HFD-fed CCL2-KO than in WT mice in both muscle and eWAT (Figures 5B and 5C). The number of total CD11b+F4/80− myeloid cells and Ly6B.2hi monocytes was not significantly higher in quadriceps muscle after 1 week of HFD irrespective of mouse genotype, whereas these cell types increased in eWAT in WT on HFD, as did Ly6B.2hi monocytes in CCL2-KO mice on HFD (Figures 5D and 5E). Although a drop in eWAT eosinophil count has been implicated in the metabolic disturbances caused by HFD , these cells (SiglecF+) did not change significantly in either muscle or eWAT; however, a small reduction occurred in quadriceps and eWAT of HFD-fed WT mice, which was reversed in CCL2-KO mice fed HFD (Figure 5F). In summary, HFD induced a rise in macrophages in WT muscle and eWAT, but this as well as the increase in their CD11c+ subset was eliminated in CCL2-KO mice.
As expected, WT mice on HFD showed intolerance to both glucose and insulin compared to CD-fed mice (not shown). Despite having unrestored glucose and insulin tolerance (Figures 5G and 5H), CCL2-KO mice on HFD showed improved quadriceps insulin signaling compared to HFD-fed WT mice, assessed as pAkt/Akt 20 min following insulin injection (Figure 5I). The latter result is in line with our previous analysis of several different immuno-compromised mouse models (macrophage-depleted mice, hematopoietic cell-specific JNK KO mice, and Rag1 KO mice), which although not yet protected from short term (1 week) HFD-induced glucose intolerance and insulin resistance, were clearly protected from long term HFD-induced insulin resistance .
Macrophage presence in muscle of obese individuals
The mouse feeding study was complemented by IHC analysis of myeloid cells in vastus lateralis of obese patients with T2DM and control obese patients with normal glucose tolerance (NGT). Individuals' characteristics are provided in Supporting Information Table S2. None of the subjects were taking any medication as they were newly diagnosed, and hyperglycemic-euglycemic clamps were performed in all individuals. CD68+ cells (macrophages) were 80% higher in muscle of T2DM patients compared to NGT controls, although this difference did not reach statistical significance (P = 0.09, Figures 6A and 6C) (possibly due to the small cohort size). CD11c+ cells were also detected by IHC (Figure 6B) but due to the limited amount of biopsy material available it was not possible to accurately quantify them; however, gene expression levels of CD68 and ITGAX (CD11c) across all 14 subjects were informative when correlated to metabolic characteristics (Figure 6D). CD68 expression correlated negatively and significantly with glucose disposal during clamp (Rd) and positively with body mass index (BMI), fasting plasma insulin (FPI), fasting plasma glucose (FPG) and age. CD11c expression likewise correlated negatively with Rd and positively with FPI as well as positively with the percent body fat. These results further extend our previous findings of correlation of CD11c gene expression with FPG and hemoglobin A1c in a separate cohort of insulin-resistant obese individuals .
Skeletal muscle is the main tissue responsible for dietary glucose disposal, and as such it is a major determinant of glycemia in normal physiological and disease states. Despite its central importance in shaping whole-body insulin resistance, in particular during obesity and HFD feeding, the underlying causes for the changes in muscle metabolism remain poorly defined. The realization that a low-grade, chronic and systemic inflammatory state coincides with the emergence of whole-body insulin resistance has called for an in-depth analysis of the interplay between the innate immune system and metabolically relevant tissues.
In contrast to numerous studies analyzing AT, immune cell infiltration and inflammation in skeletal muscle has remained poorly defined and overlooked. Importantly, skeletal muscles show clear indices of local inflammation (i.e., increased Tnfa and Ccl2) in HFD-fed mice (Figure 1) [15, 22] and in human obesity . That such inflammatory signals might arise in part from immune cells is supported by the emerging evidence of increased macrophage presence in skeletal muscle in those models in the form of F4/80 or CD68 IHC detection and Itgax (CD11c) gene expression (see Table 1). Complementarily, we recently showed that anti-inflammatory macrophage gene markers in skeletal muscle from human subjects correlate inversely with HbA1c and fasting blood glucose ; however, reports failing to detect macrophage presence or association of CD68+ cells with BMI [23, 24] beg a careful re-examination and in depth characterization of immune cell presence in skeletal muscle in the context of obesity. Moreover, the inflammatory polarization of immune cells in muscle of obese animals and humans remains unknown; therefore, we tested the hypothesis that local infiltration of skeletal muscle beds by innate immune cells occurs early on during HFD feeding, and explored the possible contribution of CCL2 and parallels with human obesity. We found that 1) Skeletal muscle displays an absolute gain in macrophages (gene expression markers and cells) within 10 weeks of HFD feeding, and macrophages represent the major type of immune cells at this time; 2) both inflammatory and anti-inflammatory macrophage gene markers are elevated already after 1 week of HFD feeding; 3) the innate immune cell population isolated from skeletal muscle differs in composition from that of eWAT in the same mice, muscle having proportionally fewer T cells and monocytes but more neutrophils than eWAT; 4) the gain in muscle and eWAT inflammatory macrophages is significantly lower in HFD-fed CCL2-KO compared to WT mice on the same diet; and 5) muscles of obese, insulin-resistant humans also show increased inflammatory macrophage presence.
Immune cell populations in mouse skeletal muscle during HFD feeding
Using a combination of IHC and cell sorting analysis, we found that macrophages are the predominant leukocytes in quadriceps muscle of HFD-fed mice, and of these, a substantial proportion are M1 proinflammatory (CD11c+CD11b+F4/80+) macrophages. These findings parallel our observations in eWAT taken from the same mice and the results of others studying AT of HFD-fed mice [9, 22, 25]. Strikingly, however, the immune cell populations and percent representation in quadriceps is markedly different from those seen in eWAT or blood (Table 2), thus immune cell presence in muscle is not simply a reflection of the presence of adipocytes or blood in muscle beds. For example, Ly6Bhi monocyte frequency did not change in quadriceps but increased in blood after 10-week HFD; nonmyeloid cells increased in muscle and eWAT by 10-week HFD but did not change in blood; and in particular, CD3+CD4−CD8− cells increased in muscle with 1 week HFD but did not change in blood.
Using flow cytometry, we found macrophage accumulation amounting to 20,000/g in skeletal muscle and 40,000/g in eWAT after 10 weeks of HFD feeding. Nonetheless, the composition and dynamics of innate immune cells in muscle and eWAT differed. The CD11c+ M1 macrophages outnumber the “resident”-type macrophages expressing M2 markers [3, 26] in eWAT of HFD-fed mice. Interestingly, whereas expression of Mgl2 dropped from 1 to 10 weeks of HFD in eWAT leukocytes, Mgl2 expression in muscle leukocytes increased along with Itgax (CD11c) expression. This suggests that quadriceps muscle from HFD-fed mice contains both M1 and M2 macrophages. Notably, the gain in CD11c+ cells in muscle with HFD was mainly of macrophages whereas in eWAT both CD11c+ macrophages and dendritic cells increased 4-fold after 10 weeks of HFD.
Ly6B.2 is highly expressed by neutrophils and monocytes . By IHC, muscle showed brightly stained Ly6B.2+ cells in all conditions, and expression of the gene encoding this antigen rose rapidly after initiation of HFD. A pioneering report detected neutrophils in AT early during HF feeding , and these cells were recently linked causatively to insulin resistance . In line with those findings, we observed that AT neutrophil numbers peak at 1 week of HFD and then decline slightly by 10 weeks. In contrast, skeletal muscle neutrophil numbers continued to increase from 1 to 10 weeks of HFD feeding to reach 3-fold higher numbers in muscle than in fat. The significance of this prominent difference remains to be investigated, but it is tempting to speculate that neutrophils continue to sustain an immune cell presence in muscle due to unique cues in skeletal muscle that are absent from AT.
CCL2 is required for macrophage gain in skeletal muscle
The initial events triggering attraction of monocytes/macrophages into AT are hotly debated. CCL2 can chemoattract monocytes to muscle in the context of infection and tissue injury and repair [14, 29]. While controversial, CCL2 appears to be a contributing chemokine to the gain in AT macrophages during HFD feeding [3, 30-33]. Notably, we found that expression of Ccl2 and its receptor Ccr2 in muscle increased sharply within 3 days of HFD feeding. These findings prompted us to examine the possible relevance of CCL2 to the subsequent gain in macrophages described earlier. CCL2-KO mice displayed markedly lower macrophage infiltration in eWAT compared to HFD-fed WT mice, which was accompanied by an equally lessened macrophage count in skeletal muscle. In addition, we observed improved insulin signaling in muscle of CCL2-KO mice on HFD compared to WT mice although glucose and insulin tolerance was not yet restored. Although it remains to be formally proven, the lower macrophage number (CD11c+ macrophages in particular) in both skeletal muscle and eWAT may contribute to the improved insulin sensitivity of CCL2-KO mice after 12 weeks of HFD . It is possible that early infiltration of inflammatory macrophages is linked to early muscle insulin signaling defects, that only later evolve into whole body insulin resistance.
Using a tissue-specific knockout of CCL2 or bone marrow transplantation from WT or CCL2-KO mice, future studies will help to discern whether the CCL2 responsible for this chemoattraction is produced by parenchymal (muscle fibers) or by ancillary cells within the muscle bed (e.g., resident immune cells , microvascular endothelial cells, adipocytes). Of note, the recruitment of Ly6B.2hiLy6C+ “recent immigrant” monocytes to eWAT is not reversed in CCL2-KO mice. This raises the possibility that these cells may be recruited by other chemokines, not essential for the differentiation into AT macrophages.
In summary, we report a quantitative gain in macrophages of inflammatory polarization within skeletal muscle early on in the course of HFD feeding in mice. Other innate immune cells such as neutrophils also rise within this tissue, and the composition of the immune cell compartment in muscle differs from that in AT of the same mice. Macrophage recruitment to muscle appears to obey chemoattraction by CCL2. Macrophages were visualized in obese human muscles and ITGAX (CD11c) expression in this tissue correlated with lower glucose disposal index. Future studies should examine the metabolic consequence of local infiltration of muscle by macrophages during HFD feeding, given the prominent role of skeletal muscle in whole body insulin action.
We thank Dr. Assaf Rudich for his initial stimulus of this study, Zhi Liu for excellent technical assistance, and Dr. Daniel Winer for valuable suggestions on the manuscript.