Myeloid Cell-specific Disruption of Period1 and Period2 Exacerbates Diet-induced Inflammation and Insulin Resistance*

Background: Circadian clockworks gate macrophage inflammatory responses. Results: Myeloid cell-specific disruption of Period1 and Period2 exacerbates diet-induced adipose and liver inflammation and systemic insulin resistance. Conclusion: Macrophage circadian dysregulation contributes to diet-induced inflammation and metabolic phenotypes in adipose and liver tissues. Significance: Interactions between circadian clocks and pathways mediating adipose tissue inflammation are critical in the development and possibly treatment of obesity-associated metabolic disorders. The circadian clockworks gate macrophage inflammatory responses. Given the association between clock dysregulation and metabolic disorders, we conducted experiments to determine the extent to which over-nutrition modulates macrophage clock function and whether macrophage circadian dysregulation is a key factor linking over-nutrition to macrophage proinflammatory activation, adipose tissue inflammation, and systemic insulin resistance. Our results demonstrate that 1) macrophages from high fat diet-fed mice are marked by dysregulation of the molecular clockworks in conjunction with increased proinflammatory activation, 2) global disruption of the clock genes Period1 (Per1) and Per2 recapitulates this amplified macrophage proinflammatory activation, 3) adoptive transfer of Per1/2-disrupted bone marrow cells into wild-type mice potentiates high fat diet-induced adipose and liver tissue inflammation and systemic insulin resistance, and 4) Per1/2-disrupted macrophages similarly exacerbate inflammatory responses and decrease insulin sensitivity in co-cultured adipocytes in vitro. Furthermore, PPARγ levels are decreased in Per1/2-disrupted macrophages and PPARγ2 overexpression ameliorates Per1/2 disruption-associated macrophage proinflammatory activation, suggesting that this transcription factor may link the molecular clockworks to signaling pathways regulating macrophage polarization. Thus, macrophage circadian clock dysregulation is a key process in the physiological cascade by which diet-induced obesity triggers macrophage proinflammatory activation, adipose tissue inflammation, and insulin resistance.

tissues including the liver and skeletal muscle, leading to systemic insulin resistance (13)(14)(15)(16)(17)(18). In contrast, treatment with thiazolidinediones or supplementation with fish oils ameliorates adipose tissue inflammation, which in turn contributes to the reversal of diet-induced adipose tissue dysfunction and systemic insulin resistance (19 -21). As such, obesity-associated inflammation is key to the regulation of systemic insulin sensitivity.
In relation to the link between inflammation and metabolic dysregulation in obesity, there is increasing evidence that dysregulated macrophage functional plasticity and versatility (polarization) is a key component of the mechanism by which inflammation in adipose and liver tissues mediates the development of obesity-associated insulin resistance and metabolic diseases. For example, in diet-induced obesity, adipose tissue macrophage infiltration is increased, and polarization is shifted toward the proinflammatory M1 activation, resulting in increased production of proinflammatory cytokines and potentiation of adipose tissue inflammation that contribute to impaired systemic insulin sensitivity (22). Within macrophages, peroxisome proliferator-activated receptor ␥ and ␦ (PPAR␥/␦) are key transcription factors that stimulate macrophage alternative M2 (anti-inflammatory) activation (23)(24)(25). Importantly, myeloid cell-specific disruption of PPAR␥ and/or PPAR␦ increases proinflammatory activation of adipose tissue macrophages and exacerbates obesity-associated insulin resistance (6,23). In contrast, the effect of PPAR␥ activation on reversing HFD-induced insulin resistance is mediated at least in part by stimulation of alternative activation of macrophages in adipose tissue (20). Macrophage polarization is also regulated by Toll-like receptor 4 (TLR4) and/or c-Jun N-terminal kinase (JNK) such that their myeloid cell-specific disruption protects mice from diet-induced adipose tissue inflammation and systemic insulin resistance (26 -28). As such, these findings demonstrate how the inflammatory status of macrophages governs the outcome of adipose tissue inflammation and systemic insulin sensitivity.
Circadian clocks in peripheral tissues and cells drive daily rhythms and coordinate many physiological processes including inflammation and metabolism. Recent observations suggest that circadian clock dysregulation plays a key role in the development of metabolic diseases including obesity and diabetes. Studies using mice with genetic mutation or deletion of core clock genes correspondingly indicate that global and adipocytespecific disruption of circadian clock function produces obesity or significant alterations in metabolism (29 -31). However, the specific mechanism underlying the link between circadian clock-and metabolic-dysregulated phenotypes is unknown. As key components of inflammation in obesity, macrophages contain cell-autonomous circadian clocks that have been shown to gate macrophage inflammatory responses including rhythms in lipopolysaccharide (LPS)-induced cytokine secretion (32,33). Because HFD induces adipose tissue circadian clock dysregulation in conjunction with adipose tissue macrophage proinflammatory activation (34) and environment-mediated circadian disruption amplifies macrophage proinflammatory responses (35), our hypothesis is that over-nutrition causes circadian clock dysregulation, which in turn induces macrophage proin-flammatory activation in adipose tissue so as to exacerbate inflammation and fat deposition, thus leading to systemic insulin resistance. To test this hypothesis, we conducted a series of experiments to determine whether 1) HFD induces circadian clock dysregulation along with proinflammatory activation in indigenous and adipose tissue macrophages, 2) genetic disruption of the clock mechanism using mice with targeted mutations of the clock genes Period1 (Per1) and Per2 (Per1 ldc / Per2 ldc ) alters macrophage proinflammatory status, 3) PPAR␥ plays a role in the proinflammatory activation of Per1/2-disrupted macrophages, and 4) adoptive transfer or co-culture of Per1/2-disrupted macrophages potentiates inflammatory responses and pathophysiology of adipose tissue in diet-induced obesity or adipocytes in vitro.

EXPERIMENTAL PROCEDURES
Animal Experiments-Animals used in this study were derived from wild-type (WT) C57BL/6J mice and 129J mice obtained from The Jackson Laboratory (Bar Harbor, ME) and from breeding pairs of homozygous mPer2 Luc knockin mice (C57BL/6J background; generously provided by Dr. Joseph Takahashi, UT Southwestern Medical School, Dallas, TX) and homozygous mutant mice with targeted disruption of the clock genes, Per1 and Per2 (Per1 ldc /Per2 ldc ; generously provided by Dr. David Weaver, University of Massachusetts Medical School, Worcester, MA). In homozygous mPer2 Luc mice, a luciferase (Luc) gene was fused in-frame to the C terminus of the endogenous mPER2 coding sequence so as to enable continuous recording of Per2 oscillations via luciferase bioluminescence (36). Establishment, genotyping, and phenotype of Per1 ldc /Per2 ldc mutant mice (129J background) have been described previously (37). Per1 ldc /Per2 ldc mice are distinguished by a loss of circadian rhythmicity as well as altered expression and circadian regulation of other clock genes (37). In different peripheral tissues, Per1 ldc /Per2 ldc mutant mice consistently showed increased BMAL1 (ϳ600%) and increased CLOCK (ϳ65%) protein levels relative to WT controls. 5 All mice were maintained on a 12:12-h light-dark cycle (lights on at 06:00). At 5-6 weeks of age, male wild-type C57BL/6J mice were fed an HFD (60% fat calories, 20% protein calories, and 20% carbohydrate calories) or a low fat diet (LFD, 10% fat calories, 20% protein calories, and 70% carbohydrate calories) for 12 weeks as previously described (7,8). After the feeding regimen, mice were fasted for 4 h before sacrifice for collection of blood and tissue samples (38,39). Epididymal, mesenteric, and perinephric fat depots were dissected and weighed as abdominal fat mass (39). After weighing, adipose tissue and liver samples were either fixed and embedded for histological and immunohistochemical analyses or frozen in liquid nitrogen and stored at Ϫ80°C for further analyses. Some mice were fasted similarly and used for insulin and glucose tolerance tests and/or and insulin signaling analyses as described below. To examine the dietary effect on in vivo circadian clock rhythmic in relation to obesity-associated metabolic phenotypes, mPer2 Luc mice were subjected to HFD feeding. Briefly, female homozygous mPer2 Luc mice at 5-6 weeks of age were fed an HFD for 12 weeks. Age-and gender-matched mPer2 Luc mice were fed an LFD and used as controls. After the feeding regimen mPer2 Luc mice were subjected to the same assays used for C57BL/6J mice. Additionally, both HFD-fed and LFD-fed mPer2 Luc mice were subjected to examination of wheel-running locomotor activity rhythms as previously described (40). Some age-matched female mPer2 Luc mice were fed ad libitum and used for isolation of bone marrow cells.
To address a role for circadian clock rhythmic dysregulation in altering macrophage activation in relation to metabolic homeostasis, Per1 ldc /Per2 ldc mice were used for the present study. Accordingly, Per1 ldc /Per2 ldc mice (129J background) and wild-type 129J mice were maintained on a 12:12-h light-dark cycle (lights on at 06:00) and fed ad libitum. At 5-6 weeks of age, male Per1 ldc /Per2 ldc mice and wild-type 129J mice were anesthetized by ketamine (100 mg/kg)/xylazine (10 mg/kg of body weight). Immediately after euthanasia, Per1 ldc /Per2 ldc mice and wild-type 129J mice were subjected to isolation of bone marrow cells as previously described (24,27). The isolated bone marrow cells were used for in vitro analyses and served as donor cells for bone marrow transplantation (BMT). Briefly, male wild-type C57BL/6J recipient mice at 5-6 weeks of age were lethally irradiated and subjected to BMT to generate chimeric mice using the established method (24,26). Wild-type C57BL/6J mice that were transplanted with donor cells from Per1 ldc /Per2 ldc mice were referred to as BMT-Per1 ldc /Per2 ldc mice. Wild-type C57BL/6J mice that were transplanted with donor cells from wild-type 129J mice were referred to as BMT-WT mice. After recovery for 4 weeks, all chimeric mice were subjected for HFD feeding for 12 weeks followed by metabolic assays. All study protocols were reviewed and approved by the Institutional Animal Care and Use Committee of Texas A&M University.
Macrophage Differentiation and Characterization-Bone marrow cells were isolated from the tibias and femurs of LFDand/or HFD-fed mPer2 Luc mice as previously described (23). After differentiation with Iscove's modified Dulbecco's medium containing 10% fetal bovine serum and 15% L929 culture supernatant for 8 days, bone marrow-derived macrophages (BMDMs) were subjected to inflammatory assays and circadian rhythm analyses using a lumicycle. Similarly, bone marrow cells were isolated from chow diet-fed Per1 ldc /Per2 ldc mice and 129J WT control mice and differentiated into BMDM for FACS analyses using flow cytometry with fluorescence-conjugated antibodies against F4/80 and CD11b as well as CD11c and CD206. For PPAR␥ rescue experiment, Per1 ldc /Per2 ldc BMDM were treated with a PPAR␥2-expressing lentiviral vector or a control vector for 24 h according to procedures provided with commercial kits (GenTarget Inc, San Diego, CA). WT BMDM (129J background) were treated similarly with a control vector and served as the control (Ctrl). Untreated WT BMDM were also included as controls. After incubation for an additional 24 h, the treated BMDM were subjected to inflammatory analyses. To examine macrophage proinflammatory activation, BMDM were treated with LPS (100 ng/ml) or PBS for 30 min before cell harvest to examine the inflammatory signaling using Western blot analyses. Some cells were treated with or without LPS at the same dose for 6 h before harvest of RNA samples. Some cells were subjected to FACS analyses as described below.
Confirmation of Per1/2 Disruption-Genomic DNA was prepared from the tails of Per1 ldc /Per2 ldc mice and 129J wild-type mice as well as bone marrow cells, bone marrow-derived macrophages, and adipose tissue and liver samples of chimeric mice and subjected to PCR genotyping using previously described methods (41).
Insulin and Glucose Tolerance Tests-Mice were fasted for 4 h and intraperitoneally injected with insulin (1 unit/kg of body weight) or D-glucose (2 g/kg of body weight). For insulin tolerance tests, blood samples (5 l) were collected from the tail vein before and at 15, 30, 45, and 60 min after the bolus insulin injection. Similarly, for glucose tolerance tests blood samples were collected from the tail vein before and at 30, 60, 90, and 120 min after the glucose bolus injection (7,8).
Measurement of Metabolic Parameters-The levels of plasma glucose, triglycerides, and free fatty acids were measured using metabolic assay kits (Sigma and BioVision, Mountain View, CA). The levels of plasma insulin were measured using ELISA kits (Crystal Chem Inc., Downers Grove, IL).
Isolation of Stromal Vascular Cells from Adipose Tissue-Adipose tissue stromal vascular cells (SVCs) were isolated using the collagenase digestion method as previously described (7,20). After digestion and centrifugation, the pelleted cells were collected as SVC. Adipose tissue SVC from mPer2 Luc mice were subjected to bioluminescence analysis as previously described (41). SVC from HFD-fed BMT-Per1 ldc /Per2 ldc and BMT-WT mice were subjected to FACS analyses and other assays.
Circadian Properties of mPer2 Luc SVC and BMDM Cells-SVC and BMDM prepared from LFD-and/or HFD-fed mPer2 Luc mice were suspended in supplemented Dulbecco's modified Eagle's medium (DMEM) (for SVC) or Iscove's modified Dulbecco's medium (for BMDM) and then plated on 35-mm culture dishes (Corning). For both cell types, the medium was changed ϳ24 h after plating so as to reduce the FBS concentration to 5%, and about 12 h later cultures were exposed to medium containing 50% adult horse serum for 2 h. Thereafter, serum-shocked SVC and BMDM were maintained in serumfree recording medium containing 0.1 mM beetle luciferin (Promega), 25 units/ml penicillin, and 25 g/ml streptomycin for bioluminescence analysis as described previously (40). Individual cultures were sealed airtight with sterile glass coverslips (VWR) and sterile silicon grease (Dow Corning). The temporal patterns of mPER2::LUC bioluminescence were analyzed using an automated 32-channel luminometer (LumiCycle; Actimetrics, Wilmette, IL) that was maintained within a standard cell culture incubator at 35°C. Bioluminescence from individual cultures was continuously recorded with a photomultiplier tube (PMT) for ϳ70 s at intervals of 10 min. Because of the transient induction of bioluminescence after the medium change at the initiation of this analysis, the first cycle was excluded from data analysis. Bioluminescence data were analyzed using the Lumicycle Analysis program (Actimetrics). For each raw data set, base-line drift was removed by fitting a polynomial curve with an order equal to one less than the number of recorded cycles. Rhythm parameters were determined from base-line-subtracted data using Levenberg-Marquardt algorithm (damped Sine) fit.
Co-culture of Macrophages and Adipocytes-BMDM were prepared from Per1 ldc /Per2 ldc and 129J wild-type control mice. Adipocytes were differentiated from 3T3-L1 cells in DMEM supplemented with 10 g/ml insulin, 1 M dexamethasone, and 0.5 mM 3-isobutyl-1-methyl-xanthine for 48 h followed by incubation for an additional 6 -8 days in growth medium supplemented with 10 g/ml insulin (7,43). After differentiation, adipocytes were co-cultured with BMDM at a ratio of 10:1 based on the published method (24). To examine changes in inflammatory signaling, the cells were treated with or without LPS (10 ng/ml) for 30 min before cell harvest. Some cells were treated with or without insulin (100 nM) for 30 min before harvest for determination of changes in insulin signaling. Cell lysates were prepared and used to examine inflammatory and insulin signaling using Western blot analyses. Additionally, some co-culture cells were treated with or without LPS (10 ng/ml) for 6 h and subjected to preparation of RNA samples to quantify gene expression using real-time reverse transcription PCR.
Statistical Methods-Numeric data are presented as the means Ϯ S.E. Statistical significance was assessed by unpaired, two-tailed analysis of variance or Student's t tests. Differences were considered significant at the two-tailed p Ͻ 0.05.

HFD Feeding Induces Adiposity, Adipose Tissue Inflammation, Systemic Insulin Resistance, and Circadian
Clock Dysregulation-Consistent with established observations (4, 5, 7), HFD-fed C57BL/6J mice displayed increased adiposity, adipose tissue inflammation, and systemic insulin resistance (data not shown). To address a possible link between metabolic phenotype and alterations in circadian clock function, homozygous mPer2 Luc knockin mice (C57BL/6J background) were subjected to an identical feeding regimen. After HFD feeding for 12 weeks, mPer2 Luc mice exhibited a marked increase in body weight and abdominal fat mass compared with LFD-fed mice (Fig. 1, A and B). Additionally, HFD-fed mPer2 Luc mice exhibited a significant increase in the severity of insulin resistance and glucose intolerance (Fig. 1, C and D) as well as an increase in the period of the circadian rhythm of wheel-running behavior (Fig. 1E).

HFD Feeding Lengthens the Period of Clock Gene Oscillations in Adipose Tissue Stromal Vascular Cells and Bone Marrowderived Macrophages and Increases
Macrophage Proinflammatory Activation-Given the role of macrophages in inflammation (5, 7), we analyzed circadian oscillations of the clock gene Per2 in adipose tissue SVC, the immune cell-containing fraction of collagenase-digested visceral fat that has been widely used to examine the inflammatory status of adipose tissue macrophages (5,23), and BMDM derived from mPer2 Luc mice. SVC cultures from LFD-fed mice exhibited robust mPER2:: LUC rhythms with a circadian period of 25.06 Ϯ 0.4 h ( Fig. 2A). However, SVC cultures from HFD-fed mice were characterized by mPER2::LUC rhythms in which the period was significantly increased by ϳ9 h (Fig. 2A). Similarly, the period of BMDM rhythms in mPER2::LUC bioluminescence was significantly increased in cultures from HFD-fed mPer2 Luc mice relative to those from LFD-fed mice (Fig. 2B). BMDM cultures from HFDfed mPer2 Luc mice were also distinguished by robust decreases in the amplitude of the mPER2::LUC rhythm. When the inflammatory status was analyzed under basal conditions (PBStreated), the phosphorylation of JNK1 (p46, the key JNK isoform that mediates proinflammatory signaling) and NF-B p65 in BMDM from HFD-fed mPer2 Luc mice did not significantly differ from that in BMDM from LFD-fed mice. However, upon LPS treatment, the phosphorylation of JNK1 and NF-B p65 (Ser-536) in BMDM from HFD-fed mPer2 Luc mice was significantly increased by 2.1-and 1.6-fold, respectively, compared with that in BMDM from LFD-fed mice (Fig. 2C). Moreover, under both basal and LPS-treated conditions, IL-1␤, IL-6, and TNF␣ mRNA levels in BMDM from HFD-fed mPer2 Luc mice were increased significantly relative to their respective levels in BMDM from LFD-fed mice (Fig. 2D). Thus, over-nutrition increases macrophage proinflammatory activation in concert with its effects on altering the timekeeping function of the macrophage circadian clockworks.
Circadian Clock Disruption Increases Macrophage Proinflammatory Activation-To address a direct role for circadian clock disruption in altering macrophage inflammatory status, BMDM obtained from mutant mice with targeted disruption of the clock genes Per1 and Per2 (Per1 ldc /Per2 ldc , 129J background) characterized by a loss of circadian rhythmicity (37) and from WT 129J mice were subjected to FACS analyses. Among the analyzed cells, Ͼ95% of a total of 20,000 -30, 000 cells were mature macrophages (F4/80 ϩ CD11b ϩ cells) (Fig.  3A, quantitative data not included). Mature macrophages were then examined for activation status based on CD11c and CD206 expression (Fig. 3B). Compared with controls, BMDM from Per1 ldc /Per2 ldc mutant mice displayed a significantly higher percentage of proinflammatory (M1) macrophages (F4/ 80 ϩ CD11b ϩ CD11c ϩ CD206 Ϫ cells) and no difference in percentage of anti-inflammatory (M2) macrophages (F4/80 ϩ CD11b ϩ CD11c Ϫ CD206 ϩ cells) (Fig. 3C). When inflammatory FIGURE 1. Over-nutrition induces obesity-associated circadian clock dysregulation. At 5-6 weeks of age, mPer2 Luc mice were fed an HFD or a LFD for 12 weeks (n ϭ 5-10). A, body weight was recorded weekly during the feeding regimen. B, adipose mass. Abdominal fat mass was calculated from the sum of gonadal and perinephric fat content. C, insulin tolerance tests. D, glucose tolerance tests. E, representative circadian rhythms of wheel-running activity in LFDand/or HFD-fed mice during exposure to constant darkness. The bar graph depicts differences in the free-running period of the activity rhythm between treatment groups. For A-E, data are the means Ϯ S.E. in line and bar graphs. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, HFD versus LFD (E, right panel) for the same time point (A, C, and D) or the same fat pad (B). For B-D, mice were fasted for 4 h starting at the same time of the day before tissue collection or physiological assays. For C and D, mice were given a peritoneal injection of insulin (1 unit/kg) (C) or glucose (2 g/kg) (D).
signaling was examined under basal conditions (PBS-treated), BMDM levels of JNK1 phosphorylation remained low with no genotype-based differences (Fig. 3D). However, in response to LPS, the phosphorylation of JNK1 in BMDM from Per1 ldc / Per2 ldc mice was significantly increased compared with that in BMDM from WT mice. Additionally, under both basal and LPS-treated conditions, NF-B p65 phosphorylation (Ser-536) in BMDM from Per1 ldc /Per2 ldc mice was significantly higher than that in BMDM from WT mice. Among proinflammatory cytokines that were examined, IL-1␤ and TNF␣ mRNA expression in BMDM under both basal and LPS-treated conditions differed in accord with genotype (Fig. 3E). Compared with controls, BMDM from Per1 ldc /Per2 ldc mice exhibited significant increases in both basal and LPS-induced expression of IL-1␤ and TNF␣ mRNA. The levels of IL-6 mRNA did not differ between genotypes. Thus, circadian clock malfunction associated with targeted disruption of Per 1 and Per 2 recapitulates macrophage proinflammatory activation induced by over-nutrition (Fig. 2).
PPAR␥2 Overexpression Ameliorates Per1/2 Disruption-associated Macrophage Proinflammatory Activation-In subsequent mechanistic analyses, we next examined the expression of PPAR␥ because activation of this transcription factor stimulates macrophage alternative (anti-inflammatory) activation (20,23), and its promoter region contains multiple E-boxes that bind the basic helix-loop-helix protein complexes formed by the core elements of the molecular clock brain muscle ARNT-like protein 1 (BMAL1) and circadian locomotor output cycles kaput (CLOCK) (46). PPAR␥ protein levels in BMDM from Per1 ldc /Per2 ldc mice were significantly decreased by 40 -60% relative to that observed in untreated and control vectortreated WT BMDM (Fig. 4A). To address the potential involvement of PPAR␥ in Per1/2 regulation of macrophage activation, we next determined whether overexpression of PPAR␥2, one of the two PPAR␥ isoforms that effectively mediates PPAR␥ actions (47), ameliorates the inflammatory status of BMDM from Per1 ldc /Per2 ldc mice. In comparison with cultures infected with a control vector, PPAR␥2 overexpression in Per1 ldc / Per2 ldc -derived BMDM significantly increased PPAR␥ content by 2.5-fold (Fig. 4A) and significantly reduced the phosphorylation of JNK1 and NF-B p65 (Ser-536) in response to LPS treatment (Fig. 4B). IL-1␤ and TNF␣ mRNA levels were similar under basal conditions in Per1 ldc /Per2 ldc controls and PPAR␥2overexpressing Per1 ldc /Per2 ldc BMDM. However, the LPS-mediated induction of IL-1␤ and TNF␣, but not IL-6, mRNA levels in PPAR␥2-overexpressing Per1 ldc /Per2 ldc BMDM was significantly decreased relative to that found in control-treated Per1 ldc /Per2 ldc BMDM (Fig. 4C). Consistent with these observations, FACS analysis revealed that among analyzed mature macrophages, Per1 ldc /Per2 ldc controls showed a higher percentage of M1 macrophages compared with WT control cells. Importantly, PPAR␥2 overexpression in Per1 ldc /Per2 ldc BMDM induced a significant decrease in the percentage of M1 macrophages relative to that observed in control-treated cells ( Fig. 4D). Among three groups of BMDM (control-treated WT, control-treated Per1 ldc /Per2 ldc , PPAR␥2-overexpressing Per1 ldc /Per2 ldc ), the percentages of M2 macrophages did not differ significantly. Thus, PPAR␥ may play a role in the mechanism by which Per1/2 disruption up-regulates macrophage proinflammatory activation.
Myeloid Cell-specific Per1/2 Disruption Exacerbates Adipose Tissue Inflammation in Obesity-The inflammatory status of macrophages governs the outcome of adipose tissue inflammation and thereby systemic insulin sensitivity (22,23). To address the impact of macrophage-specific circadian clock on HFDinduced adipose tissue inflammation, bone marrow cells were isolated from Per1 ldc /Per2 ldc mutant and WT mice and then transplanted into lethally irradiated WT recipient (C57BL/6J) mice (BMT). After HFD feeding for 12 weeks and subsequent and metabolic characterization, bone marrow cells, BMDM, and adipose-derived SVC as well as adipose and liver tissues from chimeric mice were genotyped to identify WT and targeted alleles (Per1, Per2) and confirm bone cell repopulation. In BMT-WT mice (recipients, WT C57BL/6J mice; donors, WT 129J mice), genotyping of all genomic DNA samples revealed PCR products for only WT alleles (Fig. 5, A-E). In contrast, genomic DNA from all of these tissues/cells except BMDM in BMT-Per1 ldc /Per2 ldc mice (recipients, WT C57BL/6J mice; donors, Per1 ldc /Per2 ldc mice) showed PCR products for WT and mutant alleles. Genotyping of BMDM genomic DNA from BMT-Per1 ldc /Per2 ldc mice only yielded PCR products for mutant alleles (Fig. 5B). These results demonstrate the success of adoptive transfer of Per1/2-disrupted myeloid cells to WT recipient mice.
When adipose tissue inflammatory status was examined, the phosphorylation of JNK1 and NF-B p65 (Ser-536) in HFD-fed BMT-Per1 ldc /Per2 ldc mice was significantly increased (Fig. 6A) compared with that in HFD-fed BMT-WT. Additionally, adipose tissue IL-1␤, IL-6, and TNF␣ mRNA levels in HFD-fed BMT-Per1 ldc /Per2 ldc mice were significantly higher than their respective levels in HFD-fed BMT-WT mice (Fig. 6B), whereas BMT genotype had no significant effect on the mRNA levels of arginase 1 and IL-10, which are indicative of macrophage alternative activation. Next, adipose-derived SVC were subjected to FACS analyses. After HFD feeding, SVC of BMT-Per1 ldc / Per2 ldc mice displayed a 1.8-fold increase in mature macrophages (F4/80 ϩ CD11b ϩ cells) compared with those of BMT-WT mice (Fig. 6C). Upon analyzing the inflammatory status of the mature macrophages, adipose tissue SVC isolated from BMT-Per1 ldc /Per2 ldc mice showed a 2-fold increase in the overall percentage of proinflammatory macrophages (F4/80 ϩ CD11b ϩ CD11c ϩ CD206 Ϫ cells) relative to that observed in   BMT-WT mice (Fig. 6D). In contrast, the percentage of M2 macrophages in adipose tissue SVC did not differ between the WT and mutant groups. Thus, myeloid cell-specific Per1/2 disruption increases adipose tissue infiltration of proinflammatory macrophages, thereby exacerbating HFD-induced adipose tissue inflammatory responses.
Myeloid Cell-specific Per1/2 Disruption Exacerbates Adiposity and Adipose Tissue Dysfunction-Next, we examined the effects of myeloid cell-specific circadian clock disruption on adipose tissue metabolic responses in WT recipients. In response to HFD feeding, BMT-Per1 ldc /Per2 ldc mice gained much more abdominal fat mass and displayed a greater increase in adiposity than BMT-WT mice (Fig. 7, A and B). Consistent with these changes, the size of most adipocytes in adipose tissue was larger in BMT-Per1 ldc /Per2 ldc mice than in BMT-WT mice (Fig. 7C). Locally, adipose tissue of HFD-fed BMT-Per1 ldc / Per2 ldc mice showed a significant decrease in insulin-induced Akt phosphorylation (Ser-473) relative to that observed in HFD-fed BMT-WT mice (Fig. 7D), and this decrease was accompanied by decreased adipose tissue expression of mRNAs encoding adiponectin, an adipokine that improves systemic insulin sensitivity, and hormone-sensitive lipase (Fig. 7E), a rate-determining enzyme that controls adipose tissue fat deposition by hydrolyzing triglycerides stored in adipose tissue lipid droplets. Compared with HFD-fed BMT-WT mice, HFD-fed BMT-Per1 ldc /Per2 ldc mice also showed a significant increase in adipose tissue mRNA levels of resistin, an adipokine that promotes systemic insulin resistance (Fig. 7E); although adipose tissue mRNA levels of MCP1, whose increase in adipocytes attracts macrophage infiltration, and TLR4, an inflammatory receptor, did not differ between HFD-fed BMT-Per1 ldc /Per2 ldc mice and BMT-WT mice. Together, these results indicate that myeloid cell-specific Per1/2 disruption potentiates HFD-induced adipose tissue dysfunction.
Per1/2-disrupted Macrophages Increase Adipocyte Inflammatory Responses and Impair Adipocyte Functions-Given the critical role of macrophages in initiating or exacerbating adipose tissue inflammation, we co-cultured macrophages and adipocytes at a 1:10 ratio to address the direct effects of circadian clock disruption on macrophages and/or macrophage factors that regulate adipocyte function. Compared with co-cultures containing adipocytes and WT BMDM (Adi/WT BMDM), adipocytes cultured with Per1 ldc /Per2 ldc BMDM (Adi/Per1 ldc /Per2 ldc BMDM) showed a much greater increase in NF-B p65 (Ser-536) phosphorylation in response to LPS treatment (Fig. 8A). Adipocytes cultured with Per1 ldc /Per2 ldc BMDM also exhibited a significant increase in IL-1␤, IL-6, TNF-␣, and resistin mRNA levels and a decrease in adiponectin mRNA levels compared with their respective levels in Adi/WT BMDM co-cultures (Fig. 8B), indicating increased production of pro-insulin resistant and pro-hyperglycemic factors. When insulin signaling was analyzed, adipocytes cultured with Per1 ldc /Per2 ldc BMDM showed a significant decrease in insulin-induced Akt phosphorylation (Ser-473) compared with Adi/WT BMDM co-cultures (Fig. 8C). Taken together, these results suggest that Per1/2-disrupted macrophages directly increase adipocyte proinflammatory responses and decrease adipocyte insulin sensitivity.
Myeloid Cell-specific Per1/2 Disruption Exacerbates Diet-induced Insulin Resistance and Glucose Intolerance-Macrophage-driven adipose tissue inflammation contributes to the development of systemic insulin resistance (27,28). We examined systemic pathophysiology. Before HFD feeding, the body weight of BMT-Per1 ldc /Per2 ldc mice did not differ from FIGURE 7. Myeloid cell-specific Per1/2 disruption exacerbates HFD-induced adiposity and adipose tissue insulin resistance. Chimeric mice (BMT-WT and BMT-Per1 ldc /Per2 ldc ) were fed an HFD for 12 weeks (n ϭ 6 -10). Before tissue collection or insulin injection, mice were fasted for 4 h starting at the same time of the day. A, adipose mass. Abdominal fat mass was calculated as the sum of epididymal, mesenteric, and perinephric fat content. B, adiposity was calculated by normalizing abdominal fat content to body weight. C, H&E staining of adipose tissue. D, adipose tissue insulin signaling. After fasting, HFD-fed chimeric mice were given a bolus injection of insulin (1 unit/kg of body weight) or PBS into the portal vein. The levels of total and phosphorylated Akt1/2 were examined using Western blot analyses and quantified using densitometry. E, adipose tissue gene expression. The adipose mRNA levels of genes involved in adipose tissue inflammatory and metabolic responses were quantified using real-time PCR and plotted as relative expression. For bar graphs (A, B, D, and E), data are the means Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01 BMT-Per1 ldc / Per2 ldc versus BMT-WT (B) for the same fat pad (A) or under the same condition (insulin in D) or for the same gene (E). AU, arbitrary units.

FIGURE 8. Per1/2-disrupted macrophages increase adipocyte inflammatory responses and impair adipocyte insulin signaling in co-cultures.
Adipocytes were co-cultured with macrophages differentiated from bone marrow cells of wild-type (Adi/Wild-type) or Per1 ldc /Per2 ldc mice (Adi/Per1 ldc /Per2 ldc ) at a 10:1 ratio (adipocytes:macrophages) for 48 h (n ϭ 4 -6). A, NF-B signaling of co-cultured adipocytes. Before harvest, adipocyte-macrophage co-cultures were treated with LPS (100 ng/ml) or PBS for 30 min. The levels of total and phosphorylated NF-B p65 were examined using Western blot analyses and quantified using densitometry. B, cytokine and adipokine mRNA levels in co-cultured adipocytes were quantified using real-time PCR and plotted as relative expression. C, insulin signaling of co-cultured adipocytes. Before harvest, the co-cultures were treated with insulin (100 nM) or PBS for 30 min. The levels of total and phosphorylated Akt1/2 were examined using Western blot analyses and quantified using densitometry. For bar graphs, data are the means Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01 Adi/Per1 ldc /Per2 ldc versus Adi/Wild-type under the same conditions (LPS in A; insulin in C) for the same gene (B). AU, arbitrary units.
BMT-WT mice. After HFD feeding, BMT-Per1 ldc /Per2 ldc mice gained more body weight and consumed more food than BMT-WT mice (Fig. 9, A and B). However, feeding efficiency did not differ between two groups of chimeric mice (Fig. 9B). Compared with HFD-fed BMT-WT mice, HFD-fed BMT-Per1 ldc /Per2 ldc mice exhibited a significant increase in plasma insulin and glucose levels (Fig. 9, C and D); however, plasma triglyceride and free fatty acid levels remained unchanged (data not shown). In addition, HFD-fed BMT-Per1 ldc /Per2 ldc mice displayed a significant increase in the severity of insulin resistance and glucose intolerance indicated by insulin and glucose tolerance testes, respectively (Fig. 9, E and F). Thus, disruption of myeloid cell-specific circadian clock mechanism is sufficient to exacerbate HFD-induced insulin resistance and metabolic dysregulation.
Myeloid Cell-specific Per1/2 Disruption Exacerbates Liver Inflammatory Responses, Insulin Resistance, and Metabolic Dysregulation-During obesity, adipose tissue inflammation and dysfunction can produce distal effects on liver inflammatory and metabolic responses (43,48), which in turn also contribute to the control of systemic insulin sensitivity and glucose homeostasis. Thus, we used the same BMT-WT and BMT-Per1 ldc /Per2 ldc mice to concurrently analyze the effect of myeloid cell-specific circadian clock disruption on liver responses. Compared with HFD-fed BMT-WT mice, HFD-fed BMT-Per1 ldc /Per2 ldc mice exhibited a significant increase in liver inflammatory responses, indicated by the increases in the phosphorylation of JNK1 (1.7-fold) and NF-B p65 (2-fold) (Fig.  10A) and in the mRNA levels of IL-1␤, IL-6, and TNF␣ (Fig.  10B). Meanwhile, HFD-fed BMT-Per1 ldc /Per2 ldc mice displayed a significant increase in liver weight (Fig. 10C) and a corresponding increase in the severity of hepatic steatosis (Fig.  10D). When insulin signaling was analyzed, liver tissue from HFD-fed BMT-Per1 ldc /Per2 ldc mice showed significant decreases in insulin-induced Akt phosphorylation (Ser-473) (Fig.  10E). Consistent with increased severity of hepatic steatosis, HFD-fed BMT-Per1 ldc /Per2 ldc mice were distinguished from controls by a significant increase in liver mRNA levels of SREBP1c and FAS, which are, respectively, a critical transcription factor and a lipogenic enzyme that promotes lipogenesis and hepatic steatosis (Fig. 10F) relative to that found in control mice. These results indicate that similar to adipose tissue, HFDinduced liver metabolic dysregulation is amplified by myeloid cell-specific Per1/2 disruption. FIGURE 9. Myeloid cell-specific Per1/2 disruption exacerbates diet-induced insulin resistance and glucose intolerance. Chimeric mice (BMT-WT and BMT-Per1 ldc /Per2 ldc ) were fed an HFD for 12 weeks (n ϭ 6 -10). A, body weight was monitored before and after HFD feeding. B, food intake was recorded during the feeding period and calculated as food consumption (g) per day per mouse. Right panel, feeding efficiency was calculated as the ratio of food intake to body weight. C, plasma levels of insulin. D, plasma levels of glucose. E, insulin tolerance tests. F, glucose tolerance tests.

DISCUSSION
During obesity, macrophage proinflammatory activation initiates or aggravates inflammation in key metabolic organs including adipose and liver tissues, thereby leading to systemic insulin resistance. However, it remains to be elucidated exactly how over-nutrition induces proinflammatory activation of macrophages, particularly in those found in adipose tissue where increased inflammation plays a causal role in the pathogenesis of insulin resistance (4 -8). In parallel with this inductive effect on macrophage proinflammatory activation (5), HFD feeding has been shown to decrease the amplitude of adipose tissue clock gene rhythms in Bmal1, Clock, and Per 2 expression (34) and to lengthen the period of the Per2 rhythm in cultured adipose tissue (49). Taken together, these observations suggest that dysregulation of peripheral circadian clocks may link dietassociated obesity to adipose tissue inflammation. In support of this putative association, the present study demonstrates that in HFD-fed mice where minor increases in period of the activity of rhythm were similar to those observed previously (34), overnutrition caused dysregulation of peripheral circadian clocks (i.e. substantial increases in the period of Per 2 oscillations) specifically within key mediators of adipose tissue inflammation, macrophages in stromal vascular cells, and BMDM. Furthermore, the diet-related clock dysregulation in BMDM was accompanied by increased proinflammatory activation. It is noteworthy that our findings provide evidence for the direct link between circadian clock dysregulation and macrophage proinflammatory activation in diet-induced obesity. Specifically, clock-disrupted (Per1 ldc /Per2 ldc ) macrophages are distinguished by enhanced proinflammatory activation, and myeloid cell-specific clock disruption (Per1 ldc /Per2 ldc ) exacerbates dietinduced inflammation and insulin resistance in chimeric mice repopulated with only mutant bone marrow cells. As such, the current study indicates for the first time that dysregulation of the macrophage circadian clock (e.g. Per1/2 disruption) is a critical factor in the mechanism through which over-nutrition induces macrophage proinflammatory activation, thereby leading to adipose tissue inflammation in the context of systemic insulin resistance and hyperglycemia.
Per1 and Per2 are vital components of the circadian clock mechanism and function as negative regulators in feedback loops involving interactions between Bmal1, Clock, and the cryptochrome (Cry1 and Cry2) genes (50). Previously, genetic or environmental disruption of circadian clock function has been shown to alter macrophage proinflammatory responses. In macrophages, conditional targeting of Bmal1 abolishes the rhythmic gating of innate immune responses such that cytokine responsiveness to LPS stimulation remains constantly elevated (33). Environmental disruption of circadian rhythms after weekly 6-h shifts in the light-dark cycle similarly exacerbates inflammatory responses of peritoneal macrophages to LPS challenge in the absence of sleep loss or stress effects (35). In the present analysis of Per1 ldc /Per2 ldc mice that are characterized by a globally arrhythmic phenotype (37), circadian clock dysfunction did not influence macrophage differentiation, as the percentage of mature macrophages (F4/80 ϩ CD11b ϩ cells) differentiated from bone marrow cells of Per1 ldc /Per2 ldc mice was identical to that found in wild-type mice. However, among mature macrophages, Per1/2-disrupted cells showed a much higher percentage of CD11c ϩ CD206 Ϫ cells, indicating increased macrophage proinflammatory activation. The macrophage inflammatory status was corroborated by the finding that Per1/2 disruption increased JNK1 and NF-B p65 phosphorylation and IL-1␤ and TNF␣ expression in mature macrophages under LPS-stimulated conditions. Thus, Per1/2 disruption-associated circadian dysfunction increases macrophage proinflammatory activation.
How the Per1 and Per2 genes regulate macrophage activation remains to be determined. It is possible that Per1 and Per2 modify inflammatory gene expression through their feedback regulation of CLOCK/BMAL1 transcriptional activity. Indeed, CLOCK has been shown to function as a positive regulator of the transcription factor NF-B, which is a critical determinant of inflammatory responses (51). Consequently, the relative increases in CLOCK levels observed in peripheral tissues from mutant mice (data not shown) suggest that the loss of negative feedback regulation resulting from Per1/2 disruption may further enhance the activation of NF-B and ultimately the induction of NF-B-responsive genes such as IL-1␤, TNF␣, and other cytokines. Alternatively, Per1 and Per2 may alter macrophage inflammatory signaling through interactions with other components of the core or auxiliary feedback loops comprising the molecular clockworks because REV-ERB␣ has been shown to modulate macrophage TLR signaling (52) and function as an inhibitor of proinflammatory IL-6 release (33). Given that active PPAR␥ promotes macrophage anti-inflammatory activation (20,23,53) and this transcription factor is a molecular target of insulin-sensitizing and anti-diabetic drugs (54), PPAR␥ may play a role in Per1/2 regulation of macrophage activation. In fact, our data support this hypothesis because PPAR␥ levels were decreased in Per1/2-disrupted macrophages in conjunction with their enhanced proinflammatory activation and PPAR␥2 overexpression reversed, at least in part, this Per1/2 disruption-associated shift in inflammatory status. Collectively, these observations suggest a novel mechanism for macrophage polarization in which the circadian clock normally acts through PPAR␥ to inhibit macrophage proinflammatory activation. Because the PPAR␥ promoter contains multiple E-boxes (46), it will be important to determine whether negative feedback by Per1 and/or Per2 modulates CLOCK/BMAL1 regulation of E-box mediated transcription of PPAR␥.
The significance of circadian clock function in macrophage activation was further manifested by the finding that adoptive transfer of Per1/2-disrupted myeloid cells to wild-type mice exacerbated HFD-induced inflammation and insulin resistance. Notably, BMT-Per1 ldc /Per2 ldc mice displayed marked increases in adipose tissue inflammatory responses as indicated by enrichment of mature proinflammatory macrophages in adi-pose tissue and increased inflammatory signaling and proinflammatory cytokine expression. Consequently, inflammationassociated adipose tissue dysfunction (i.e. decreased insulin signaling and adipose tissue metabolic dysregulation) was markedly increased in BMT-Per1 ldc /Per2 ldc mice. Given the origin of the myeloid cells/macrophages, the increased severity of adipose tissue inflammatory responses and dysfunction in BMT-Per1 ldc /Per2 ldc mice solely reflects proinflammatory activation in clock-disrupted macrophages. This relation was substantiated by the finding that co-culture of adipocytes with Per1/2-disrupted macrophages increased adipocyte inflammatory response and decreased adipocyte insulin sensitivity. In similar fashion, Per1/2-disrupted macrophages appeared to mediate the increased severity of liver inflammatory responses and metabolic dysregulation. Altogether, these observations provide convincing evidence that macrophage circadian clock dysregulation is sufficient to potentiate over-nutrition-induced inflammation in adipose tissue and the liver, thereby increasing systemic insulin resistance.
In summary, the present data unveil a novel mechanism for obesity-associated insulin resistance in which circadian clock dysregulation is a key factor that links over-nutrition and macrophage proinflammatory activation to adipose tissue inflammation. At the cellular level, HFD-associated over-nutrition modulates circadian clock function in macrophages, which in turn induces the proinflammatory activation of macrophages, especially those in adipose tissue. When activated, clock-disrupted macrophages are sufficient to exacerbate dietinduced adipose tissue inflammation and systemic insulin resistance. Because adipose tissue inflammatory responses and corresponding insulin sensitivity in over-nutrition-related obesity are differentially engaged by specific types of fatty acids found in HFDs, future studies are warranted to compare the effects of saturated and polyunsaturated fatty acids on macrophage and adipocyte circadian clocks. Thus, it will be important to determine whether saturated fatty acids that activate proinflammatory signaling pathways and lead to impaired insulin sensitivity (e.g. palmitate) also modulate the circadian timekeeping function of macrophage and adipocyte clocks and whether this circadian clock dysregulation is blocked or abated by polyunsaturated fatty acids that repress adipose tissue inflammatory responses and have a beneficial impact on insulin sensitivity.