In the hypothalamus, some studies have been able to identify some inflammatory pathways, involving different cell types, to be relevant for weight loss.

At the molecular level, as previously mentioned, the literature related to the deregulation of the control of energy balance in the anorexia–cachexia syndrome has focused on the role of pro-inflammatory cytokines, such as interleukin (IL)-1β, IL-6, and TNF-α. It has been shown that pro-inflammatory cytokines, whose production can be induced by infectious agents affecting peripheral organs, notably by LPS—the most commonly used model for modeling disease behavior and anorexia associated with inflammation (27)—participate in the induction of sickness behavior.

Some studies deciphering the kinetic aspects of the effect of central or peripheral cytokine injection suggest that pro-inflammatory mediators exert their effect primarily at the central level versus periphery (28, 29). Numerous studies have focused on IL-1β and have shown that its intracerebroventricular (ICV) injection induces profound behavioral changes in rodents (30). Furthermore, the central role of cytokine signaling at the central level has been confirmed, for example, by the fact that the central injection of the IL-1β receptor antagonist prevents the effects induced by the peripheral injection of this cytokine (31, 32). In addition, some studies suggest that inflammatory inducers such as LPS are not capable of directly inducing sickness behavior and require downstream actors such as cytokines. Indeed, for example, unlike the central injection of LPS, central injection of IL-1 is able to induce sickness behavior in mice whose LPS toll-like receptor (TLR) 4 is non-functional (33).

As mentioned earlier, cytokines, involved in the induction and regulation of sickness behavior at the central level, can be released into the circulation by the immune cells (8) or directly produced by neurons and glial cells in the central nervous system (CNS) (34–39). In the context of systemic diseases, the central effects of these cytokines appear to be independent of the place of their secretion. As previously mentioned, when produced in the periphery, cytokines have the following two main pathways to act on the brain: a neural pathway and a humoral pathway. Most cytokines act in a paracrine manner at the site of infection, suggesting that neural afferents may be the target of pro-inflammatory cytokines. Interestingly, the perineural sheath of the vagus nerve contains immune cells that are capable of producing IL-1, in particular in response to LPS (40). Furthermore, the sensory neurons of the vagus nerve express IL-1 receptors and it has been shown that this cytokine stimulates the sensory activity of the vagus nerve (40). Vagotomy experiments have demonstrated the importance of the vagus nerve in the transmission of information from the periphery to the brain: for example, after injection of LPS or IL-1 at the periphery, vagal afferents are involved in the induction of the sickness behavior and in the neural activation of the brainstem, the hypothalamus, and the limbic structures (40).

Chronic administration of cytokines can reproduce the characteristics of anorexia–cachexia syndrome (38, 41–44), while blocking the signaling of one of them, such as TNF, by the use of neutralizing antibodies, inhibits its development (42, 45, 46). Similarly, administration of an IL-1 receptor antagonist prevents anorexia in cancer animal models (47). Some studies have suggested that endogenous brain IL-1 is a mediator of LPS-induced anorexia by acting on the expression of cytokines in the hypothalamus (9). In addition, it appears that IL-1β can act on ARC POMC neurons (48) and that TNF-α indirectly increases energy expenditure via β3 adrenergic signaling in the brown AT (thermogenesis) (49). By different approaches, it has been shown that interfering with mediators of inflammation coincides with a reduction in hypothalamic inflammation and prevents weight loss in animal models of anorexia: for example, the inhibition of the “adenosine monophosphate protein kinase” in the hypothalamus reduces hypothalamic inflammation, which is accompanied by an increase in food intake in the case of cancer-associated anorexia, allowing better overall survival (50). Interestingly, in the context of cancer-associated cachexia, the administration of ghrelin, an appetite-stimulating hormone, increases food intake and is accompanied by a decrease in IL-1β (51).

Focusing on the LPS-induced anorexia model, it is interesting to underline that some studies suggest that this loss of appetite is independent from the vagal afferents and that it only depends on central inflammatory mechanisms, where the central effects of peripheral LPS could be, among others, mediated by some cytokines and/or via receptors for LPS expressed by some brain cells (11, 52). Indeed, microglia expresses TLR4, through which LPS has been shown to exert some of its effects (53–56). Furthermore, Hines and colleagues have shown that disrupting TLR4 signaling prevents microglial activation, inhibits the production of cytokines, and prevents the development of the sickness behavior that are induced by peripheral LPS (57). In contrast, a recent study by Reis et al. has shown that if microglia and TLR4 are necessary for LPS to acutely induce inhibitory effects on orexigenic agouti-related peptide (AgRP)/neuropeptide Y (NPY) neurons, conversely, LPS can actually acutely increase the firing activity of anorexigenic pro-opiomelanocortin (POMC) neurons in a microglia/TLR4-independent manner (58). In addition, the same study presents results which further supports that mediators of inflammation such as microglia and TLR4 can, at least acutely, affect basal food intake and hormonal-dependent modulation of feeding behavior. Indeed, the inhibition of microglia alone, via ICV injection of minocycline, leads to an increase in food intake, comparable to the one induced by ICV ghrelin (58). Interestingly, the co-injection of ghrelin and minocycline does not synergistically increase food intake. This implies that the inhibition of microglia itself, as it increases food intake, interferes or prevents ghrelin’s orexigenic effects (58).

Changes in gene expression of hypothalamic anorexigenic and orexigenic neuropeptides were assessed in mice or rats that received LPS IP injections (59–61). Generally, peripheral LPS appears, initially, to reduce the expression of the orexigenic peptides and, on the contrary, to increase that of the anorexigenic peptides. In later stages, the expression of orexigenic peptides increases again, probably to promote food intake and compensate for weight loss.

Some authors have used other models to study to understand links between inflammation in the brain and the induction of sickness behavior.

For example, Jang and colleagues sought to determine the mechanisms behind anorexia and weight loss associated with illness via administration, in mice, of either bacterial or viral products: LPS and human immunodeficiency virus-1 transactivator protein (Tat), respectively (62). They found in both cases that pro-inflammatory cytokines such as IL-β, IL-6, and TNF-α were upregulated in the hypothalamus. Furthermore, AtT-20 and SH-SY5Y cells treated with either Tat or LPS exhibit increased POMC transcriptional activity. In addition, the injection of Tat or LPS in the hypothalamus of mice induces in both cases a decrease in food intake and in body weight. They identified POMC as a potential mediator of illness-induced anorexia and as a possible downstream target of NF-κB. Indeed, they showed through different approaches that the NF-κB pathway in the melanocortin system plays an important role in illness-induced anorexia and body weight loss: both administration of AgRP, an endogenous melanocortin antagonist and the inhibition of the NF-κB pathway specifically in the POMC neurons (Ikkβ^ΔPomc mice), significantly blunted the effects of Tat and LPS on food intake and body weight. Interestingly, the authors also present results suggesting that leptin-induced anorexia is also dependent on the NF-κB pathway (62).

Another study has shown that a pathway involving TLR2 participates in the induction of sickness behavior through a microglia-POMC neurons axis (63). In this publication, the authors show, by ICV injection of Pam3CSK4, a specific synthetic TLR2 agonist that TLR2 central signaling is able to trigger sickness behavior through activation of microglial cells, which express TLR2, and via the NF-κB and COX pathways. The authors also show that activation of central TLR2 is able to modify the synaptic architecture in the ARC, especially at the level of the POMC neurons: it reduces the rate of GABAergic contacts on POMC neurons, whereas it increases the vesicular glutamate transporter 2 (vGLUT2) contacts on POMC soma, translating an increased excitatory state, in correlation with an increased microglial occupancy.

In addition, Murray et al. used the peripheral administration of the viral mimetic poly I:C to induce type-I interferons (IFN-I) overexpression in the brain. Using IFN-I receptor 1 (IFNAR1)-deficient mice, the results of the authors present suggest that IFN-I are involved in the induction of sickness behavior, including anorexia, whereas IL-6 participates in sickness behavior but not in anorexia (64).

Furthermore, a local increase in serotonin levels in the hypothalamus has already been linked to anorexia and cachexia (65–67) and a recent study showed that inflammation induced by peripheral treatment by either IL-6 and/or TNF-α is associated with an impairment in local serotonin turnover in the hypothalamus, a decrease in NPY and AgRP gene expressions, and a decrease in food intake in comparison to control conditions (68). Authors were able to identify upstream inflammatory regulators including interferon gamma (IFN-γ), transforming growth factor beta (TGF-β), IL-6, and IKBKG, an enzyme crucial for the activation of the NF-κB pathway. This study thus suggests that peripheral inflammation reaches the hypothalamus where it impairs serotonin turnover, which is associated to a decrease in food intake.

Taken together, numerous studies support the hypothesis that inflammation at the hypothalamic level is able to disrupt the proper function of neuropeptidergic circuits of the hypothalamus and thus to induce an involuntary weight loss.

However, mediators linking inflammation and its consequences at the level of central systems regulating energy homeostasis, and ultimately on weight, have not been determined with certainty, and one may note that the nature of the mediators responsible for the central effects of IL-1β has been especially poorly described. Thus, chemokines can represent interesting candidates to study further understanding of the underlying mechanisms.

Studies have associated inflammation and obesity for a long time (79). Yet this inflammation is not “typical,” as it cannot be associated to the cardinal signs of redness, swelling, heat, and pain. Thus, the inflammation associated to obesity is of different nature. First, it is aseptic: it is caused by the overconsumption of specific nutrients (lipids and carbohydrates); thus, the trigger of this inflammation can be considered to be of metabolic nature. Interestingly, not only the trigger of this inflammation is metabolic, but this inflammation first targets the cells that specialized in metabolism: e.g., adipocytes, hepatocytes, pancreatic β-cells, and myocytes (but also the neurons of the peptidergic systems involved in the regulation of feeding behavior, as detailed further). Hence, this inflammation is often referred to as “metabolic inflammation” or “metainflammation” (79).

The characteristics of the inflammation associated with obesity, in periphery at least, are: (i) being of metabolic nature—it is induced by nutrients and is orchestrated by “metabolic” cells; (ii) being low grade, with moderate and localized overexpression of pro-inflammatory mediators; (iii) creating an impaired milieu where the tissue “composition” in terms of immune cells favors an inflammatory environment in the tissues; and (iv) being sustained in time, without any apparent resolution (79). Furthermore, peripheral metabolic inflammation associated with obesity is deleterious at many levels and, as a consequence of poor eating habits favoring the obese phenotype, inflammation itself promotes tissue dysfunctions, which also contribute to development of obesity, including resistance to insulin and leptin.

In the context of nutritional obesity, inflammatory pathways are not only activated in peripheral tissues but also activated in central areas involved in the control of energy metabolism, especially in the hypothalamus. Several questions then deserve to be asked and answered such as “What are the signals that induce this inflammation?” and “Is inflammation a cause or a consequence of obesity?”

The relationship between nutritional obesity and hypothalamic inflammation was first described by De Souza and his collaborators (83). The authors demonstrated in rats that a 4-month period of HFD feeding activates inflammatory pathways in the medio-basal hypothalamus (MBH), such as JNK and NF-κB, leading to the production of canonical pro-inflammatory cytokines such as IL-1β, TNF-α, and IL-6 and to deficiencies in insulin and leptin signaling. These observations on hypothalamic inflammation associated to nutritional obesity were confirmed in rats, obese mice, and non-human primates (82, 84–92). Interestingly, some studies have shown that the inflammation induced by the consumption of HFD, likely a response to overnutrition, develops much faster in the hypothalamus than in the peripheral tissues. Indeed, inflammation in the AT, for example, is only observable after several weeks or even months, whereas in the hypothalamus, the consumption of a HFD during a much more restricted period of 24–72 h is enough to induce overexpression of pro-inflammatory cytokines and gliosis—which corresponds to an activation and proliferation of glial cells such as microglia and astrocytes, an inflammation hallmark in the brain (82).

Thus, it appears that the inflammation induced by the excess of nutrients in the hypothalamus precedes the establishment of obesity, as well as peripheral inflammation and metabolic disturbances. It is worth mentioning that the gliosis induced by the consumption of a HFD for a short-period of time (2–3 weeks) is actually reversible, whereas it reappears after longer periods of HFD consumption (8 months) (82). This suggests that the initial short-term gliosis is rather a protective mechanism induced to protect against the “injury” induced by the overload of fat, which is in line with the known roles of these cells, and that with time, the consumption of HFD triggers a durable gliosis and inflammation that would then play a major role in the development of the pathophysiology induced by overnutrition. For example, a study by Dalvi et al. further supports that acute gliosis associated with a moderate period of HFD feeding would actually be protective: as long as the inflammation is restricted to glial cells, the organism seems to attempt to limit brain injury by overexpressing in parallel some protective factors such as the heat shock protein 70 and the ciliary neurotrophic factor. After a longer period of HFD feeding, overexpression of inflammatory mediators within the neurons suggests an exhaustion of neuroprotective mechanism in the hypothalamus, leading to a deregulation of the expression of certain neuropeptides that will favor a positive energy balance (93).

What is the signal that triggers hypothalamic inflammation? It has been suggested that it is the excess of nutrients, which are major physiological regulators of hypothalamic neural networks, that induces the establishment of inflammation at the level of the hypothalamus. Some studies attribute the triggering of hypothalamic inflammation to, more especially, saturated fatty acids (SFA), especially long-chain SFA, due to their accumulation in the hypothalamus when consuming HFD, which could, among others, induce inflammatory signaling by activating the TLR4 pathway. It is interesting to note that other fatty acids, such as long-chain polyunsaturated fatty acids, appear, conversely to long-chain SFA, to be beneficial, especially in the diet-induced obesity (DIO) context, as they have anti-inflammatory properties (94–97). In contrast, not only nutrients but also hormones seem to play a role in the regulation of glial cells function. Indeed, as illustrated by a study by Gao et al., showing that monogenic obese mice with deficient leptin signaling [ob/ob (leptin deficient) and db/db (leptin receptor mutation)] exhibit less microglial activation than wild-type controls both on chow and on HFD (98). The authors further demonstrated that the lack of leptin signaling affects also the microglial function in the hypothalamus as the expression of several inflammatory mediators is reduced (98). This indicates that the signaling of leptin, already described as a pro-inflammatory adipokine (99), could affect microglial activation. Other mechanisms have been proposed as mediators of the hypothalamic inflammation associated with overnutrition such as the endoplasmic reticulum stress (86, 100, 101). Similarly, oxidative stress could be an initiating factor and participate in the maintenance of the hypothalamic inflammation induced by nutrition. Indeed, at the periphery, oxidative stress, along with the production of reactive oxygen species (ROS), has been shown to precede severe metabolic disturbances and insulin resistance (102). Yet, the brain alone uses a large amount of the oxygen and calories consumed by the body, making it particularly vulnerable to excessive ROS production and oxidative stress. Indeed, mitochondria naturally produce ROS in physiological conditions, especially during the oxidation of nutrients such as glucose or fatty acids (103), and it has been shown that an excessive production of ROS in the hypothalamus occurs in the obese rat. This impairs not only the local detection of glucose but also the associated physiological response, such as peripheral insulin secretion (104). Autophagy is another possible pathway. It is a cellular process allowing the elimination of damaged cytoplasmic elements and organelles in order to maintain internal homeostasis and structural integrity and also plays a key role in cellular responses to metabolic stress. Under conditions of overnutrition, endoplasmic reticulum stress and oxidative stress can induce autophagy (105, 106). Impairing the pathways of autophagy can increase food intake and weight, in association with overactivation of the IKKβ/NF-κB pathway in the hypothalamus, illustrating the connection between autophagic and inflammatory pathways in the hypothalamus.

Over the recent years, the importance of the role played in hypothalamic inflammation in the onset of metabolic disorders in obesity was indeed underlined and the potential of counteracting inflammation in the hypothalamus as a strategy in the fight against obesity was highlighted.

For example, a recent study by Douglass and colleagues demonstrates that astrocytes in the MBH of HFD-fed mice mediate hypothalamic inflammation together with DIO via their own inflammatory signaling (107). Indeed, the authors showed that the inducible and specific deletion of IKKβ in astrocytes, which should blunt astrocytic inflammatory capacity, reduces the mice susceptibility to DIO (reduced weight gain and fat mass associated with a decrease in food intake and an increase in energy expenditure), improves glucose tolerance and insulin sensitivity (according to glucose and insulin tolerance tests), and finally, reduces HFD-induced astrocytosis in the MBH.

Furthermore, it has been shown that subcutaneous application of liraglutide or canagliflozin (an inhibitor of the sodium-glucose cotransporter 2) in obese and insulin-resistant rodents is able to disrupt the activation of microglial cells in the hypothalamus, which is associated with an improvement of insulin and glucose homeostasis. In addition, when the ICV injection of IL-4 in HFD-fed rats increases further HFD-induced inflammation in the hypothalamus and causes excessive weight gain, the ICV injection of an IKKβ/NF-κB blocker allows, on the contrary, to prevent hypothalamic inflammation, which is associated with both a decrease in body weight and fat mass and improvements in glucose metabolism and general energy homeostasis in DIO animals (86, 108–111).

Another study by André et al. shows that overnutrition will induce, in addition to an increase in body weight and in adiposity, an increase in the number of microglial cells in the ARC. The authors show the inhibition of microglia expansion in the ARC, achieved by the central delivery of an antimitotic drug, allows to limit food intake and the increase in body weight and in adiposity and also to restore leptin sensitivity. This is associated with a “predictable” inhibition of the upregulation of inflammatory pathways in the ARC that is normally associated with overnutrition, but also in periphery (112).

Taken together, studies studying the hypothalamic inflammation associated with obesity support that this inflammation plays a key role in the metabolic dysfunctions associated with obesity and represents a very interesting target.

Research aiming to understand how the central regulation of energy balance is altered in the context of overnutrition has mainly focused on neurons so far. Yet, as previously mentioned, since hypothalamic inflammation, especially its gliosis feature, has been associated to overnutrition, a strong interest in the role of glial cells (mainly microglia and astrocytes) emerged more recently. Interestingly, these glial cells, once activated, are capable of overproducing, locally, pro-inflammatory mediators such as cytokines and chemokines that can, in turn, affect the neuropeptidergic systems of the hypothalamus, potentially participating in the onset of obesity.

Interestingly, chemokines and their receptors have been demonstrated to be expressed by glial cells, and also by neurons. Especially, chemokine receptors appear to be more expressed by neurons than cytokine receptors in some nuclei of the hypothalamus, allowing to ask the following question: “Could not chemokines act as one of the latest inflammatory mediator that would link inflammation to the disruption of the proper functioning of the neuropeptidergic systems involved in the regulation of the energy balance to promote weight gain and thus the development of obesity?”