Fractalkine (CX3CL1) Is Involved in the Early Activation of Hypothalamic Inflammation in Experimental Obesity

A complex network of neurons that are responsive to hormonal, neural, and nutritional cues tightly regulates body adiposity (1,2). Cells of the medium-basal hypothalamus act as first-order neurons in this system, and many genetic, pharmacological, and environmental approaches that lead to the damage or loss of some of these neurons can affect body energy homeostasis (1,2). In many experimental models, hypothalamic dysfunction that results from local inflammation plays a central role in the pathogenesis of obesity (3,4). In addition, recent studies using neuroimaging have provided strong evidence for the existence of inflammation and dysfunction in the hypothalamus of obese humans (5,6).

Dietary long-chain saturated fatty acids are the main triggers of hypothalamic inflammation in obesity (7). These fatty acids act through toll-like receptor 4 (TLR4) (7) and induce endoplasmic reticulum stress (4,7), leading to the activation of intracellular inflammatory signaling pathways through Jun NH₂-terminal kinase (JNK), nuclear factor-κB, and protein kinase C-θ (PKC-θ) (3,4,8). The increased hypothalamic expression of inflammatory cytokines is detectable as early as 1 day after the introduction of a fat-rich diet to rodents (6). Upon prolonged high-fat feeding, signals of cellular damage become evident, such as gliosis (6), apoptosis (9), and defects in the potential for neurogenic recovery (10). Thus, diet-induced hypothalamic inflammation follows a classical path of the inflammatory response, which is detectable a few hours after the exposure to the threatening stimulus, and progresses with a wide array of damaging/recovering outcomes.

An important missing link between the exposure to dietary fats and the induction and perpetuation of inflammation in the hypothalamus is cells first sensing the presence of the fatty acids and cells effectively generating and maintaining the inflammation. In at least one study, hypothalamic neurons were unable to respond to palmitate-producing cytokines (11). Thus, glial cells, particularly microglia, may be an important requirement of the system. In other inflammatory conditions of the central nervous system, the activation of microglia plays an important role in the progression of disease (12). Both resident and bone marrow–derived monocytic cells can, to a variable degree, exert the inflammatory and immunosuppressive actions that are frequently observed in chronic neuroinflammatory conditions (12). The recruitment and activation of microglia to the site of damage depends on the expression of chemokines and their receptors (13). Fractalkine (CX3CL1) is one of the most important chemokines that are involved in the regulation of neuroinflammatory conditions (13,14). In this context, fractalkine is produced and released by neurons, acting as a mediator in their talk with glial cells (13,14).

In the current study, we first show that bone marrow–derived cells are involved in the activation of diet-induced inflammation of the hypothalamus in obesity. Next, we show that saturated fatty acids induce an early expression of fractalkine in hypothalamic neurons and that the inhibition of this chemokine reduces diet-induced hypothalamic inflammation and body mass gain in an obesity-prone strain of mice.

Six-week-old (4-week-old for mice that underwent bone marrow transplantation) male Swiss, Balb-c, C57BL/6J, TNFRp55^−/−, TNFRp55^+/?, C3H/HeJ, C3H/Hepas, IL10^−/−, or IL10^+/? mice were used in this study, which was approved by the ethics committee of the University of Campinas. Swiss mice are an outbreed related to the obesity- and diabetes-prone AKR; Balb-c mice are inbred and partially protected from metabolic diseases; C57BL/6J is wild-type control for TNFRp55^−/− and IL10^−/−; TNFRp55^−/− is a knockout for the p55 tumor necrosis factor (TNF)α receptor, and C3H/HeJ is a loss-of-function mutant for the TLR4 (HeJ/Hepas is its wild-type control) (both are partially protected from diabetes and obesity due to a reduced inflammatory response when fed a high-fat diet [HFD]); finally, the IL10^−/− mouse is a knockout for interleukin (IL)10 and is prone to diabetes and obesity due to increased inflammation in response to dietary fats. In some experiments, mice were randomly selected for feeding on chow or an HFD (35% fat, measured in grams; 5.2 kcal/g) for 2–8 weeks. In some experiments, mice were stereotaxically instrumented using a Stoelting stereotaxic apparatus set on the coordinates anteroposterior, 0.0 mm; lateral, 0.0 mm; and depth, 4.8 mm—or bilaterally, to the arcuate nucleus, according to the following coordinates: anteroposterior, −1.7 mm; lateral, 0.3 mm; and depth, 5.6 mm. Single doses of a small interfering RNA (SI) against fractalkine (sense: GCC GCG UUC UUC CAU U; antisense: ACA AAU GGA AGA ACG C) or a scrambled control (SC) (sense: CAG GCU ACU UGG AGU G; antisense: AUA CAC UCC AAG UAG C) were delivered through the cannula, which was immediately withdrawn. The adequacy of the cannulation procedure was ascertained by randomly selecting some animals for methylene blue injection and posterior anatomical evaluation and by the capacity of the SI to inhibit the expression of fractalkine only near the injection site.

The neuron cell line, mHypoA 2/29 CLU189, was cultivated in DMEM containing 25 mmol/L glucose and 10% FBS to reach 70% confluence. The cells were exposed to either palmitate or TNFα according to doses and times as described in results. After incubation, the cells were harvested for RNA preparation for real-time PCR. The macrophage cell line, Raw 264.7, was cultivated in DMEM containing 11 mmol/L glucose and 10% FBS. The cells were exposed to either palmitate or TNFα according to doses and times as described in results.

During the preparation for radiation, mice were treated with a daily dose of sulfametoxazole (4.0 mg/kg) plus trimethoprim (0.8 mg/day) for 4 days. A sublethal dose of radiation (8 Gy) was used, and after 2 h, the mice received an intravenous (tail vein) injection of a bone marrow cell preparation that contained 5.0 × 10⁶ cells. The details of the protocol have previously been published (15).

Glucose consumption was assessed after a 12-h fast using a method previously described (16).

Mice were anesthetized, and the body fat mass and lean mass contents were measured using a DEXA system (Discovery Wi QDR Series; Hologic Apex Software, Hologic Inc.).

Liver specimens were homogenized and samples containing 50–125 µg protein were used in immunoblot experiments as previously described (9). Ser⁴⁷³ phospho-Akt was detected in the membranes using specific antibodies (pAKT sc7985-R; Santa Cruz Biotechnology, Santa Cruz, CA). Loading was evaluated by reblotting the membranes with an anti-Akt antibody (sc8312; Santa Cruz Biotechnology).

Hypothalamus was dissected on ice-cold phosphate buffer and chopped in very small fragments that were transferred to a tube and centrifuged gently for 30 s. The anatomical limits for the dissection of hypothalamus were as follows: rostral, optic chiasm; caudal, mammillary bodies; lateral, optic tracts; and superior, apex of the hypothalamic third ventricle. Trypsin (250 mg/mL) was added, and incubation was carried out at 37°C for 10 min under gently shaking conditions. DMEM and FCS were added to the suspension, which was thereafter filtered through a 100 μm nylon mesh. Cells were recovered after centrifugation and washed three times in ice-cold phosphate buffer. After counting, cells were incubated with specific antibodies and analyzed using a Becton-Dickinson FACSCalibur Cytometer (San Jose, CA) with the CellQuest Pro software (Becton-Dickinson) (17). For each measurement, cells were obtained from the hypothalamus of one mouse. Cell suspensions were used to determine DNA fragmentation and membrane integrity. The method was based on propidium iodide staining and fluorescence, which was measured using a flow cytometer. First, 50 μL isotonic solution containing propidium iodide (8.0 μg/mL in PBS) was used to homogenize 200 μL cell suspension. Fluorescence was measured using the FL2 channel (orange-red fluorescence; 585/42 nm).

TNFα, IL1β, MCP1, CCR2, CX3CL1, CX3CR1, CD11b, CD36, CD163, and TLR4 mRNAs were measured in hypothalamus or cell culture by real-time PCR (ABI Prism 7500 detection system; Applied Biosystems, Grand Island, NY). The anatomical limits for the dissection of hypothalamus were as follows: rostral, optic chiasm; caudal, mammillary bodies; lateral, optic tracts; and superior, apex of the hypothalamic third ventricle. The intron-skipping primers were obtained from Applied Biosystems: TNFα, Mm00443258_m1; IL1β, Mm00434228_m1; MCP-1, Mm99999056_m1; CCR2, Mm00438270_m1; CX3CL1, Mm00436454_m1; CX3CR1, Mm00438354_m1; CD11b, Mm00434455_m1; CD36, Mm01135198_m1; CD163, Mm00474091_m1; and TLR4, Mm00445273_m1. GAPDH (cat. no. 4352339E, Applied Biosystems) was used as endogenous control. Real-time data were analyzed using the Sequence Detector System 1.7 (Applied Biosystems).

For histological evaluation, hypothalamic tissue samples were fixed in paraformaldehyde (4% final concentration in PBS) and processed routinely for embedding in a paraffin block. Five-micrometer paraffin sections were processed for immunofluorescence staining using the following primary antibodies: F4/80 (sc25830), CD11b (sc28664), CD11c (sc26693), Iba1 (sc28530), GFP (sc5385), AgRP (sc18634), and POMC (sc20148) (all from Santa Cruz Biotechnology, Santa Cruz, CA); fractalkine (CX3CL1) (cat. no. ab25088; Abcam, Cambridge, MA); and the secondary antibodies conjugated to FITC or rhodamine (cat. nos. sc2777 and sc2092, respectively; Santa Cruz Biotechnology). In some experiments, the neuron cell line mHypoA 2/29 CLU189 was cultivated in glass slides for further fixation in 4% paraformaldehyde and preparation for indirect immunofluorescence staining with anti-AgRP and anti-fractalkine antibodies.

Results are presented as means ± SE. Levene test for the homogeneity of variances was initially used to check the fit of data to the assumptions for parametric ANOVA. All results were analyzed by t test or one-way ANOVA and complemented by the Tukey test to determine the significance of individual differences. The level of significance was set at P < 0.05.

A deficiency of TLR4 protects against diet-induced hypothalamic inflammation and obesity (7). In the central nervous system, TLR4 is predominantly expressed in microglia cells (18), which potentially place these cells in a pivotal position as triggers of hypothalamic inflammation in obesity. Here, we performed bone marrow transplants to generate chimeras that expressed functional TLR4 in all cells of the body except those derived from the bone marrow and chimeras that expressed functional TLR4 only in bone marrow–derived cells. Supplementary Fig. 1A depicts the protocol that was used to generate the chimeras and controls, and Supplementary Fig. 1B depicts PCR-amplified, BstOI-digested fragments that were generated from transcripts of TLR4-deficient and functional TLR4–expressing cells, which were used to monitor the efficiency of the transplants at 6 weeks of age. Four weeks on an HFD lead to the activation of microglia only in mice that express a functional TLR4 (Fig. 1A). However, chimeras that only express a functional TLR4 (and GFP) in bone marrow–derived cells present a hypothalamic infiltration of active microglia cells upon high-fat feeding (Fig. 1B). Chimeras that only express nonfunctional TLR4 in bone marrow–derived cells are protected from diet-induced obesity and hypothalamic inflammation (Fig. 1C–G and Supplementary Fig. 1C), whereas chimeras that only express functional TLR4 in bone marrow–derived cells adopt the obese phenotype with a deposition of fat in the visceral territory and increased expression of inflammatory cytokines in the hypothalamus (Fig. 1C–G and Supplementary Fig. 1C).

For investigation of chemokines and other inflammatory factors that are potentially involved in the hypothalamic recruitment of bone marrow–derived monocytic cells during the induction of obesity, we compared an obesity-prone strain, the Swiss mouse, and an obesity-protected strain, the Balb-c mouse. Supplementary Fig. 2A–L depicts the main metabolic and endocrine differences between the two strains. In contrast to experimental models of obesity (3,6,7), Balb-c mice present only a mild and transient increase in the expressions of hypothalamic TNFα (Fig. 2A) and IL1β (Fig. 2B). While obesity-prone Swiss mice present an early (1 day) activation of several markers of inflammation in the hypothalamus, which include two chemokines, MCP1 and fractalkine (Fig. 2C–J), Balb-c mice present a delayed activation of only some of these markers, which are MCP1 (Fig. 2C) and its receptor, CCR2 (Fig. 2D), CD11b (Fig. 2G), CD36 (Fig. 2H), and CD163 (Fig. 2J). Importantly, Balb-c mice present no activation of the expression of fractalkine (Fig. 2E) or its receptor, CX3CR1 (Fig. 2F). Because fractalkine is an important chemokine that is involved in neural inflammation (13,14) we decided to focus on its potential role in the induction of diet-induced hypothalamic inflammation in obesity and as a factor involved in the recruitment of bone marrow–derived monocytic cells to the hypothalamus.

Obesity-prone Swiss mice that were fed for 4 weeks on an HFD exhibit the preferential expression of fractalkine in neurons of the hypothalamus (Fig. 3A–D); however, some few cells expressing CD11b (Fig. 3A) or F4/80 (Fig. 3B) seem to express small amounts of fractalkine as well. In addition, both palmitate and TNFα induce a dose- and time-dependent increase in fractalkine in a hypothalamic neuronal cell line (Fig. 3E, I, K, and L). This increase in fractalkine is accompanied by the increased expression of MCP1 (Fig. 3F and J). Remarkably, whereas TNFα is capable of inducing the expression of IL1β but not of TLR4, palmitate produces an increase in TLR4 but not in IL1β (Fig. 3G and H). Finally, immunofluorescence staining reveals that palmitate produces changes not only in the expression of fractalkine but also in the morphology of the cells and in the distribution of AgRP granules, which become more concentrated in the perinuclear area (Fig. 3K and L). In contrast to neurons, a monocyte-macrophage cell line does not express fractalkine even after a stimulus with palmitate (Fig. 3M). CX3CR1, the fractalkine receptor, is expressed by the monocyte-macrophage cell line but undergoes no change after palmitate stimulus (Fig. 3M). Conversely, both TLR4 and TNFα are induced in response to palmitate in the same monocyte-macrophage cell line (Fig. 3N and O).

Because fractalkine is expressed only in neurons, which produce no inflammatory response when exposed to fatty acids (11), we hypothesized that cytokines, particularly TNFα, which is produced by microglia in response to fatty acids, would mediate the spreading of inflammation from cells of the innate immune system toward the neurons. For testing of this hypothesis, TNFR1 knockout mice were fed for 8 weeks on an HFD, and inflammatory markers were evaluated in the hypothalamus. As previously reported (19), TNFR1 knockout mice are protected from diet-induced obesity (Fig. 4A). Although hypothalamic TNFα is increased in mice that are fed the HFD (Fig. 4B), this increase was accompanied by no changes in the expression of other markers of inflammation, such as IL6 (Fig. 4C), IL1β (Fig. 4D), and IL10 (Fig. 4E); additionally, no changes in the expression of either fractalkine (Fig. 4F) or its receptor were observed (Fig. 4G). Conversely, when the inflammation-prone IL10 knockout mice were fed an HFD for 8 weeks, body mass increased significantly (Fig. 4H), and the expression of inflammatory cytokines (Fig. 4I–K) and fractalkine (Fig. 4L) and its receptor was stimulated (Fig. 4M). Additional controls for these experiments are presented in Supplementary Fig. 3.

Using an SI approach, we obtained up to a 70% reduction in the expression of fractalkine in the AgRP-expressing mHypoA 2/29 CLU189 neuron cell line (Fig. 5A). A protocol was conducted according to Fig. 5B; thus, a single dose of the SI was injected on the same day that the HFD was introduced; some mice were used in experiments after 2 days, while the remaining mice were monitored for 2 weeks before additional experiments were performed. The inhibition of hypothalamic fractalkine resulted in the reduction of the diet-induced activation of several inflammatory markers in the hypothalamus of Swiss mice that were fed the HFD for 2 days (Fig. 5C–N); this was not accompanied by significant changes in caloric intake and body mass (not shown). After 2 weeks, the inhibition of hypothalamic fractalkine (Fig. 6A) was still insufficient to produce significant changes in body mass (Fig. 6B and C); however, there was a reduction of body fat mass (Fig. 6D) but not lean mass (Fig. 6E). In addition, a glucose tolerance test (Fig. 6F) and an insulin tolerance test (Fig. 6G) revealed a beneficial effect of inhibiting hypothalamic fractalkine, which was confirmed by a reduction in the glucose area under the curve (Fig. 6F) and in the constant for glucose decay (Fig. 6G). The levels of fractalkine receptor (Fig. 6H) and TNFα (Fig. 6I) transcripts were no longer lower; however, IL1β was still significantly reduced (Fig. 6J).

Some mice were randomly selected for a bilateral medium-basal hypothalamic injection of SI or SC against fractalkine according to the protocol illustrated in Fig. 6K. This protocol resulted in the inhibition of fractalkine expression in the medium-basal hypothalamus (Fig. 6L) but not in the frontal cortex (Fig. 6M). There was a trend for body mass reduction (not shown), and glucose tolerance was improved as determined by the evaluation of glucose area under the curve during a glucose tolerance test (Fig. 6N).

Using a protocol similar to the one depicted in Fig. 5B, we inhibited the expression of fractalkine in the hypothalamus of wild-type mice that were previously submitted to a bone marrow transplant from GFP (Fig. 7A–E). Two weeks after the intracerebroventricular injection of the SI and the introduction of the HFD, the number of cells that expressed fractalkine was greatly reduced, and there were virtually no GFP-expressing cells in the arcuate nucleus (Fig. 7A–C) compared with the controls that were treated with the SC. Moreover, there were only few Iba1-expressing cells in the hypothalamus of SI-treated mice (Fig. 7D and E). In additional experiments, Swiss mice were treated with SI against fractalkine, and after 2 weeks on the HFD, the phenotype of the microglia/macrophage population in the hypothalamus was evaluated by flow cytometry. As depicted in Fig. 7F, the inhibition of fractalkine resulted in an increased number of cells that expressed the anti-inflammatory marker CD206.

In the 2-week protocol, as depicted in Fig. 5B, no significant change in body mass was detected; however, a trend, as shown in Fig. 6B, prompted us to extend the protocol. Therefore, we repeated the intracerebroventricular treatments with either SI against fractalkine or its scrambled control and monitored the mice for 6 weeks (Fig. 8A). The extended time resulted in significant changes in body mass (Fig. 8B and C and Supplementary Fig. 4) and adiposity (Fig. 8D) without any significant modification in caloric intake (Fig. 8E).

Body mass stability is defended by a complex and integrated neuronal circuitry that controls caloric intake and energy expenditure (1). Neurons of the hypothalamus are an important part of this circuitry because they are capable of sensing the signals that reflect the amount of energy that is stored in the body and of integrating this information with effector neurons of other brain regions (1). At least parts of these signals are delivered by leptin and insulin, and resistance to these hormones plays an important role in obesity (1,2,20–22). For most people, and for experimental animals, changes in the defended set point of adiposity are rapidly followed by adaptations that gradually lead to a return to the original body mass (20,23). However, environmental and genetic factors can act in concert to disrupt this stability (1,24). Dietary fats, particularly long-chain saturated fatty acids, can trigger hypothalamic inflammation through the activation of TLR4 signaling and through the induction of endoplasmic reticulum stress (4,7). Prolonged exposure to this inflamed milieu results in the loss of neurons and in the impaired capacity of neuronal regeneration (9,10). Therefore, it is currently accepted that hypothalamic dysfunction that results from prolonged inflammation is an important mechanism that leads to obesity (24).

A recent study has shown that hypothalamic inflammation is rapidly induced after the introduction of a fat-rich diet (6). However, upon continuous and prolonged exposure to dietary fats, the expression of inflammatory markers undergoes oscillation, fading away during the second and third weeks and returning and remaining elevated thereafter (6). This oscillatory pattern of inflammatory activity has been described in other contexts and is the result of a balance between pro- and anti-inflammatory factors that are allied to the recruitment of new cells to the affected anatomical site (25,26). In the current study, we initially asked whether bone marrow–derived cells would be recruited to the hypothalamus during the induction of diet-induced obesity. To test this hypothesis, we took advantage of the paradigm that a functional TLR4 is necessary for the complete induction of diet-induced hypothalamic inflammation (7). Using bone marrow transplantation, we produced chimeras that expressed functional TLR4 only in bone marrow–derived cells or only in non–bone marrow–derived cells. Taken together, the results of these experiments showed that bone marrow–derived cells with a functional TLR4 are important for the full induction of diet-induced hypothalamic inflammation and the obese phenotype.

In other inflammatory conditions of the central nervous system, the recruitment of bone marrow–derived cells is known to play important roles in the progression of these diseases. For instance, in experimental Parkinson disease, bone marrow–derived microglia can express the inducible form of nitric oxide synthase, which is an important effector of neurodegenerative damage (27). However, cells that migrate from the bone marrow do not only deliver damaging effects; a recent study has shown that, in another model of experimental Parkinson disease, bone marrow–derived cells can act protectively by delivering neurotrophic factors to the site of injury (28).

Alzheimer disease is yet another neurodegenerative condition that is associated with inflammation (29). Interestingly, epidemiological studies indicate an increased risk for the development of Alzheimer disease in patients with obesity and type 2 diabetes (30), and a recent study has provided a molecular basis for this association showing that amyloid β peptide oligomers can activate TNFα/JNK signaling, which leads to the inhibitory Ser phosphorylation of brain insulin receptor substrate 1 (31). This is very similar to the mechanisms linking inflammation to insulin resistance in peripheral tissues that are classic targets for insulin action (32,33). As in Parkinson disease, Alzheimer disease is characterized by the recruitment of bone marrow–derived microglia that can exert both pro- and anti-inflammatory roles (29).

In the second part of this study, we evaluated the potential role of fractalkine as a chemokine that is involved in the recruitment of bone marrow–derived cells to the hypothalamus during the induction of diet-induced obesity. Fractalkine signals through the CX3CR1 selective receptor to recruit bone marrow–derived cells to the site of inflammation (34). In this study, fractalkine was chosen because it is expressed in neurons (35) and is involved in the modulation of inflammatory activity in distinct neurological conditions (36).

Initially, in working with mouse strains with different predispositions to obesity, we demonstrated that upon high-fat feeding, fractalkine was the inflammatory marker with the most remarkable difference between the strains. In fact, even after prolonged exposure to dietary fats, Balb-c mice, which are resistant to diet-induced obesity and diabetes, did not present an increased expression of this chemokine in the hypothalamus. In the obesity-prone Swiss mouse, the introduction of dietary fat was rapidly followed by an increase in the expression of fractalkine in hypothalamic neurons. No stimulus for fractalkine expression was detected in other brain regions (not shown), which is in accordance with previous reports that showed the anatomical selectivity of the brain inflammatory process in obesity (3,6,7,9). In cell culture experiments, the expression of fractalkine was limited to neurons, whereas the monocyte cell line expresses only the fractalkine receptor. We used a cell line expressing AgRP, mHypoA 2/29 CLU189; the main purpose of the experiment was to determine whether a neuron cell line would express fractalkine in response to an inflammatory cytokine and/or fatty acids. The combination of results suggests that monocytes are more responsive to palmitate expressing TNFα, whereas the neuron cell line is more responsive to TNFα producing fractalkine and other inflammatory markers. This observation is in accordance with a previous study (11) and suggests that neurons are not the primary targets for dietary fats, which are affected only after glial cells are engaged, as previously proposed (6).

Next, we inhibited fractalkine. In the obesity-prone Swiss mouse, the inhibition of hypothalamic fractalkine resulted in the reduction of the inflammatory activity and in the recruitment of bone marrow–derived cells to the hypothalamus. These effects were accompanied by a reduction of adiposity and a reduction of glucose intolerance. In several studies that explored the inflammatory mechanisms that lead to an anomalous function of the hypothalamus in obesity, the pharmacological and/or genetic inhibition of distinct components of the inflammatory machinery resulted in the reduction of the obese phenotype. This result was obtained by targeting proteins that are involved in intracellular signaling, such as JNK (37), inhibitor of nuclear factor κB kinase (IKK) (4), and PKC-θ (8); a protein that is involved in the regulation of the inflammatory signal, SOCS3 (38); membrane receptors, such as TLR4 (7) and TNFR1 (19); endoplasmic reticulum stress (4,7); and at least one cytokine, TNFα (39). One important outcome resulting from the inhibition of inflammatory signaling in the hypothalamus of obese animals is the improvement of glucose intolerance. As explored recently by our group, this regulation can occur, at least in part, through parasympathetic connection with the liver (39). Thus, we suspect that the improved glucose homeostasis occurring under hypothalamic fractalkine inhibition, may, as well, be mediated at least in part by neural connection with the liver. Notably, 2 weeks after fractalkine inhibition, when glucose tolerance was improved, there were no changes in body mass; however, there was a reduction in adiposity.

All previous work exploring hypothalamic inflammation in experimental obesity has focused on cytokines or proinflammatory intracellular mechanisms. Now, we provide the first evidence for the role of a chemokine in this process. Within the multistep progression of most types of inflammatory responses, the recruitment of bone marrow–derived cells to the site of damage is an important event that warrants the equilibrium between effector and regulatory activities (40). The initial descriptions of fractalkine actions in inflammatory conditions of the brain suggested that it predominantly exerted an anti-inflammatory activity (41). However, subsequent studies provided evidence for its role in the recruitment of proinflammatory cells as well (42,43). Here, in the diet-induced inflammatory process of the hypothalamus, fractalkine is also involved in the induction of the inflammatory activity.

Two lines of evidence support the proinflammatory role of fractalkine in the hypothalamus of obese mice. First, in the context of cytokine expression, the inhibition of fractalkine resulted in the reduction of TNFα and IL1β. This observation is in line with previous studies that showed an important role for both of these proinflammatory cytokines in diet-induced hypothalamic inflammation (3,4,19). Second, in the context of cell infiltration, a reduction in the expression of fractalkine led to an inhibition of the migration of bone marrow–derived cells toward the hypothalamus and to an increased relative presence of monocytic cells of an anti-inflammatory phenotype, which expressed CD206. In other inflammatory conditions of the brain, an increased presence of CD206 cells was related to the reduction of local inflammatory activity (44).

It is noteworthy that recent studies have implicated fractalkine, directly or indirectly, in whole-body metabolic control. The TACE-TIMP3 dyad, as well as the ADAM17/TACE metalloprotease, are anomalously regulated in humans with type 2 diabetes and obesity, which can, potentially, modulate the levels of bioactive fractalkine (45–47). Moreover, fractalkine is involved in the regulation of insulin secretion by pancreatic β-cells (48). Thus, fractalkine may be an important modulator of inflammation and metabolic activity not only in the brain but also in other tissues.

A recent study has shown that even in mice that are not fed an HFD, bone marrow–derived cells can modulate food intake and energy expenditure by delivering brain-derived neurotrophic factor (BDNF) to the hypothalamus (49). The defective expression of BDNF in hematopoietic cells resulted in obesity, which was corrected by wild-type bone marrow transplantation. This observation raises the intriguing question of whether the hypothalamus, as a sensor for peripheral signals that indicate the variations in whole-body energy stores, is continuously monitored by bone marrow–derived immune cells. If that scenario is indeed the case, it may explain why there is such an anatomical specificity in brain inflammation in obesity. We propose that fractalkine is one of the earliest regulators of this process by controlling the homing of pro- and anti-inflammatory macrophages to the hypothalamus. In this model, resident microglia are rapidly activated in response to dietary fats by producing signals that induce the expression of fractalkine by neurons. If the exposure to fats is rapidly interrupted, then the first wave of inflammation can be self-contained (6); however, upon the persistence of the dietary stimulus, a permanent wave of inflammation is warranted by the recruitment of bone marrow–derived cells.

In summary, this study shows that bone marrow–derived cells are important for the progression of the inflammatory process in the hypothalamus in diet-induced obesity and that fractalkine plays an important role in this process.

Acknowledgments. The authors thank Josh Thaler and Michael Schwartz from the University of Washington in Seattle for thoughtful discussions in the context of the work presented in the article. The authors thank Erika Roman, Gerson Ferraz, and Marcio Cruz from the University of Campinas for technical assistance.

Funding. The support for the study was provided by Fundação de Amparo a Pesquisa do Estado de São Paulo and Conselho Nacional de Desenvolvimento Cientifico e Tecnologico. The laboratories of Cell Signaling and Experimental Endocrinology belong to the Obesity and Comorbidities Research Center and National Institute of Science and Technology–Diabetes and Obesity.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.M. and J.C.M. performed most of the experiments and wrote the manuscript. G.F.A., M.J.S., and I.L.-C. provided facilities and expertise in some steps of the work and revised the text. L.F.N., R.F.d.M., D.R., C.S., D.G., G.S., A.H.M., and V.D.P. performed some experiments guiding J.M. in unfamiliar steps. N.T. and C.D.R. performed the evaluation of body composition using DEXA. L.A.V. designed the study, proposed experiments, obtained grants, and wrote the manuscript. L.A.V. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.