Smith et al. (2011). analyzed data from different age groups, i.e., children and adolescents, adults to the elderly. They found that obesity and cognitive deficits were bidirectionally connected in the different age groups. Several reviews and meta-analyses have shown that obese individuals behave worse in tasks that examine executive function, verbal fluency, learning, and memory ability (Siervo et al., 2011; Fitzpatrick et al., 2013; Table 1). Higher BMI and older individuals have a lower executive function, memory performance, and processing speed due to lower GMV in corresponding brain regions. Specifically, memory decay has been associated with lower GMV in the frontal and thalamic BMI-associated regions (Kharabian Masouleh et al., 2016). However, studies on obese elderly individuals have shown their cognitive conditions to be complex. Some results claim that higher BMI in older individuals is not predictive of cognitive decline (Sturman et al., 2008). As for children and adolescents, recent studies have indicated that obesity is inversely associated with poorer cognitive competence such as executive function, attention, visuospatial skills, and motor skills, and may affect their academic achievements (Liang et al., 2014). Although a causal relationship between obesity and neurocognition is gradually being recognized, only a few studies have been performed. The study limitations are the narrow focus on executive function with a lack of verbal or memory tests for cognition measurement, as well as the overreliance on BMI as an indicator for obesity (Hartanto et al., 2019).

Actually, besides the use of BMI as a marker of overall obesity, various parameters, including waist-to-hip ratio (WHR), waist circumference (WC), lipid accumulation products, and visceral adiposity index, are applied to identify abdominal obesity, an indispensable subgroup of obesity that also significantly influences the cerebral structure and cognitive performance (Wan et al., 2020). Studies found that higher WHR is independently linked with lower GMV volume in the network, which has a correlation with worse memory performance (Kharabian Masouleh et al., 2018). WHR and WC are also associated with slower processing speed, worse executive function, and declined linguistic fluency (Gunstad et al., 2010; Gardener et al., 2020). Comparatively, the limitation in using BMI as a measurement is that the abdominal and overall fat is hardly distinguished, as one hypothesis pointed out that central obesity was more clinically relevant to brain health (Gardener et al., 2020). Studies suggested that abdominal fat conferred an increased risk of dementia and cognitive impairment, while overall obesity might be protective (West and Haan, 2009). To be more specific, the influence of visceral adiposity (or intra-abdominal adiposity) rather than abdominal subcutaneous fat on CNS has been underscored in recent years, but experts’ opinions and experimental results have not reached a consensus yet (Smith, 2015). Some studies agreed on the deleterious impact of excessive visceral adiposity on the brain, suggesting that a high visceral adiposity index was associated with lower gray matter density in the caudal anterior cingulate cortex (Iceta et al., 2021). Moreover, elevated adiponectin mRNA expression in visceral adipose tissue (VAT) exhibited a positive correlation with the motor deficit, declined memory, and depression- and anxiety-like behavior (Metwally et al., 2019). Nevertheless, some studies examined VAT and gray and white matter, performed several cognition tests, and reported no significant relationship among them (Hsu et al., 2016). Further, some researchers found no association between visceral fat with dementia, but a close link between increased abdominal subcutaneous fat and decreased likelihood of dementia in women in late-life (Hsu et al., 2016). In summary, more studies are anticipated to compare the influence of visceral adiposity and abdominal subcutaneous fat on brain structure and function and identify potential reasons behind the difference.

Genetic and environmental factors give rise to the occurrence of obesity, and the latter involves insufficient exercise and an unbalanced diet. Western diet, which usually contains high fat, high sugar, and other high caloric nutrients, is negatively associated with cognitive function. From the disease-developing aspect, in a population-based case-control study, it is reported that high caloric intake was associated with a nearly two-fold increased odds of having an amnesic mild cognitive impairment (MCI) as compared to the reference group and amnesic MCI is likely to be a prodromal stage of AD (Geda et al., 2013). Additionally, autonomic nervous system (ANS) dysfunction including the accumulation of α-synuclein in PD is exacerbated by obesogenic diets and ameliorated by intermittent energy restriction (Griffioen et al., 2013). On the other side, higher adherence to a Mediterranean-type diet, which calls for fruit and vegetable intakes as well as replaces saturated fatty acid (SFA) with unsaturated fatty acid (UFA), was conducive to slower cognitive decline and the progression of MCI and AD (Bousquet et al., 2012).

From a behavioral aspect, data proved that abundant SFA intake from milk products in midlife brought about poorer global cognitive function and prospective memory (Eskelinen et al., 2008). From structural aspects, an overnutrition pattern diet also reduced the volume of hippopotamus (Jacka et al., 2015). From biological aspects, high nutrition diet is nowadays proposed as an inflammatory factor. Dietary inflammatory index (DII) is applied to assess, which associated with individuals’ inflammatory markers like C-reactive protein and IR (Shivappa et al., 2015; Saghafi-Asl et al., 2021).

Depression is also closely linked with obesity and diet patterns. With regard to obesity, it was found that obesity measured using various anthropometric indices was related to somatic-affective instead of cognitive-affective symptoms (Wiltink et al., 2013). The resulting metabolic and inflammatory dysfunction of obesity played a role in developing depression with an impact on the neuroimmune status and neural circuits controlling mood and emotional states (Fulton et al., 2022). Recently, some researchers combined this adverse metabolomics and inflammatory alterations with atypical, energy-related depressive symptoms and proposed an “immune-metabolic depression” (IMD) dimension (Alshehri et al., 2021). Moreover, the severity, age of onset, duration, and comorbidities of depression were associated with abdominal obesity, with a potential mechanism referring to neutrophil gelatinase−associated lipocalin (NGAL), a recently discovered adipokine (Marijnissen et al., 2014). Not only does obesity lead to a high risk of major depressive disorder (MDD), but also diet quality is independently related to depression or depressive symptomatology. Several randomized controlled trials proved that the Mediterranean diet could improve depressive symptoms as an adjunctive intervention and also reduce the incidence of the disease as beneficial prevention (Sánchez-Villegas et al., 2013; Parletta et al., 2019; Marx et al., 2021). However, these protective effects might fade away in older age if high adherence to a Mediterranean diet is not implemented (Psaltopoulou et al., 2013). Similarly, a small or traditional diet comprising fruit, vegetables, and fish positively influenced brain function and behavior (Francis et al., 2019). On the contrary, a western diet was significantly associated with psychiatric disorders (Jacka et al., 2010). The different effects were possibly due to different energy consumption in these dietary styles (Quirk et al., 2013; Rao et al., 2020; Yin et al., 2021). Furthermore, the underlying mechanisms underlying the effects of obesity and diet on depression and cognition are similar. The former involves inflammation, oxidative stress, epigenetics, mitochondrial dysfunction, gut microbiota, tryptophan–kynurenine metabolism, hypothalamic–pituitary–adrenal (HPA) axis, neurogenesis, and so on (Marx et al., 2021).

Studies have traditionally reckoned that excessive nutrition causes obesity and subsequent cognition impairments. Rats fed on a high saturated fat diet and sugar resulting in weight gain and body fat had difficulties in hippocampal-dependent serial feature-negative and hippocampal-independent simple discrimination problems (Davidson et al., 2012). Behavior tests have also been utilized to examine cognition, such as novel object recognition, attentional set-shifting task, Morris water maze, etc. in rats. Each of these entails the use of different brain areas for optimal performance. Rats in the obesity groups showed poor ability to perform complicated discrimination, interdimensional and extradimensional shift tasks (Bocarsly et al., 2015; Lewis et al., 2019; see Table 2). Multiple factors that influence obesity-related cognition deficits have been gradually revealed. It was found that depriving estrogen aggregated cognition and synaptic dysfunction. Furthermore, aging was found to play a role in diet-induced weight gain and the negative outcomes resulting from obesity (Pratchayasakul et al., 2015; Moser et al., 2019).

However, conflicting results regarding the relation between high nutrition and cognitive impairment have also been reported. Some studies have demonstrated that excessive dietary fat for long periods does not induce learning deficits or spatial memory deficits, although HFD-induced obesity may have a detrimental influence on cognitive flexibility (Leyh et al., 2021).

Could high nutrition affect cognitive behavior? Recent studies have investigated the influence of obesity and high nutrition and suggested that overnutrition alone could affect cognition. Sobesky et al. (2014) found that mice fed on an HFD performed worse in contextual pre-exposure fear conditioning, which reflects memory function. Interestingly, when the HFD diet was reversed, the negative impact on memory was alleviated. This demonstrated that a high nutrient diet, not weight gain mediated the negative effects on memory (Sobesky et al., 2014). Additionally, studies have found that HFD mice demonstrated hyper-locomotion in the open-field test and Y-maze test, and impaired sucrose preference in the sucrose consumption test. These behavioral traits above were not associated with body weight or body weight gain (Yoshizaki et al., 2020). In conclusion, the above studies demonstrated that obesity was not a necessary factor that contributed to cognitive impairment induced by an excessive nutritional diet.

High nutrition-induced cognition decline and relevant brain structural alterations mainly affect the hippopotamus. Short-term high fat and sugar exposure or high sugar diet alone induced hippocampal-dependent place recognition memory before weight changes and were the result of hippocampal inflammation and oxidative stress (Beilharz et al., 2014). High fat and high refined sugar diet for 2 months were shown to be detrimental to spatial learning performance with reduced hippocampal levels of brain-derived neurotrophic factor (BDNF) (Molteni et al., 2002). In addition to a high fat, high-fructose-high-coconut oil consumption significantly affected the hippocampal-dependent spatial memory behavior determined using the Morris Water Maze tasks. This was closely related to reduced expression levels of leptin protein, as well as mRNA expression levels of leptin receptor and SCD1 mRNA in the hippocampus (Lin et al., 2017).

The negative effects of a high nutritional diet were not limited to one generation. Maternal HFD exposure brought long-term and detrimental effects on cognition across three subsequent generations. Impaired cognition was most evident in prepulse inhibition and the two-way active avoidance paradigm. Furthermore, the observed neurochemical changes like reduced dopamine, glutamate, and gamma-aminobutyric acid (GABA) levels suggest altered dopaminergic, glutamatergic, and GABAergic signaling. This may provide some explanation for the behavior of children from mothers consuming HFD (Sarker and Peleg-Raibstein, 2018).

With regards to the brain, the reduction of gray matter, compromised white matter integrity, and BBB hyperpermeability are well-known effects induced by obesity. Higher BMI is associated with lower GMV or cortical thickness within similar areas of the frontal lobes, cingulate gyrus, amygdala, dorsal striatum, thalamus, cerebellum, and affects children, adolescents, and adults (Kharabian Masouleh et al., 2016; Westwater et al., 2019; Laurent et al., 2020). The data above was based on magnetic resonance imaging (MRI), which was able to discriminate gray and white hypothalamic structures and highlight finer structures like brain nuclei (Baroncini et al., 2012). Interestingly, the volume of subcortical gray matter structures in men appears to be more affected by the total body fat (TBF) ratio compared to women. This is probably due to the positive effect of estrogen on the MetS (Mauvais-Jarvis, 2015). Apart from the volume and thickness of gray matter, studies using diffusion tensor imaging (DTI) to create apparent diffusion coefficient (ADC) maps of gray matter, have found that gray matter ADC in the left and right amygdala and the right parietal region was significantly and positively associated with higher fibrinogen levels among overweight/obese individuals (Cazettes et al., 2011). Moreover, lower gray matter density in the caudal anterior cingulate cortex was strongly associated with a high visceral adiposity index and metabolic alterations (Iceta et al., 2021).

As for white matter integrity, global and tract-specific white matter integrity were measured using fractional anisotropy and mean diffusivity in DTI. Hyperintensity of white matter has been shown to have better local network efficiency, information processing speed, and cognitive function (Vergoossen et al., 2021). It has been observed that obesity is related to higher coherence but the lower magnitude of white matter microstructures and higher TBF were related to reduced commissural fibers, association fibers, and projection fibers (Dekkers et al., 2019). Hyperpermeability of BBB in obese individuals has been observed and accounts for increased albumin content in the hippocampal, reduced claudin-5 protein levels, and collagen IV immunostaining, which eventually results in the transmigration of peripheral immune cells (de Aquino et al., 2018). Since BBB leakage measured by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a potential biomarker for bipolar disorder and dementia (Kamintsky et al., 2020; Chagnot et al., 2021), we believe that BBB integrity will be a useful indicator of obesity-related cognitive impairments in the future. Additionally, concerning brain functional connectivity measured using functional magnetic resonance imaging (fMRI), it was observed that global brain connectivity (GBC) was altered in obesity, which included reduced GBC in prefrontal and feeding circuits and increased GBC in the dorsal attention network (Geha et al., 2017).

In conclusion, MRI and other methods clearly showed negative changes in the gray matter, white matter, BBB, and functional connectivity in the brain under DIO and hence demonstrated a structural foundation for cognitive dysfunction and behavioral abnormality.

Under short-term HFD exposure before the obvious weight gain, reactive gliosis was observed in the hypothalamic arcuate nucleus (ARC) of rats and mouse models. Although T2-weighted imaging (T2WI) to study the effects of obesity and HFDs on brain tissue is rare, the results demonstrated that T2 hyperintensity was associated with the presence of gliosis in different pathologies (Lizarbe et al., 2020). It has been observed that microglia numbers were increased, enlarged, and transformed to a more activated morphology, as well as astrocytes forming a dense fibrous network. Furthermore, by examining gliotic changes below the visual detection threshold, clinical studies have demonstrated similar results that obese individuals have associated gliosis in the mediobasal hypothalamus (MBH) (Thaler et al., 2012). Negative effects on the synapse have been shown to be accompanied by microglial activation. Studies have demonstrated that in dietary obese mice, activated microglia induced a series of synaptic alterations including the impairment of hippocampal synaptic plasticity, reduction of dendritic spine density and sites of excitatory synapses, and promoted synaptic stripping (Hao et al., 2016; Cope et al., 2018). A negative correlation between BMI and synaptic density was also observed by [11C] UCB-J positron emission tomography (PET) imaging. In this approach, synaptic vesicle density was measured through synaptic vesicle glycoprotein 2A (SV2A), a transmembrane glycoprotein ubiquitously and homogeneously expressed in presynaptic vesicles, while PET was used to measure [11C] UCB-J binding to SV2A (Asch et al., 2022).

Concerning neuronal responses to diet or obesity, DIO and HFD consumption induced injury and loss of Pro-opiomelanocortin (POMC) neurons in the ARC and Sim1 neurons of the paraventricular nucleus (PVN). While the former was more specific to male mice, the latter shared features of exposure to HFD for both genders. POMC acts as an agonist for neurons expressing the type IV melanocortin receptor (MC4R) to increase energy expenditure and reduce food intake. Thus, POMC injury accelerated the progression of DIO (Iqbal et al., 2021). Studies have demonstrated that inhibition of POMC neurons by melanin-concentrating hormone (MCH) through the SIRT1/FoxO1 pathway resulted in hyperphagia and obesity (Al-Massadi et al., 2019). In addition to imaging techniques to study neuronal impairment like POMC, new methods are being or have been developed. For example, Lactam bridge-cyclized α-melanocyte-stimulating hormone analogs have been investigated for their suitability for MC1R-targeted imaging, especially for single-photon emission computed tomography (SPECT) (Zhang et al., 2017). Dopamine plays an important role in eating behavior and is one of the main reasons for obesity. Measured using dopaminergic PET and SPECT, reduced dopamine signaling in individuals with elevated BMI has been observed to be consistent with reduced dopamine synthesis capacity in the dorsal striatum (Val-Laillet et al., 2015). PET is also able to reflect brain substrate metabolism, albeit relatively limited and technically different from PET imaging of human diabetes and obesity (Guzzardi and Iozzo, 2019).

Neuroimaging technologies such as PET, SPECT, fMRI, and pharmacogenetic fMRI have been useful for identifying obesity-related abnormalities in a wide range of brain areas in obese and high nutritional subjects. This provides advantages for scientific studies, clinical diagnosis, and targeted treatments.

It is well-known that insulin, a peptide hormone secreted by pancreatic β-cells, regulates glucose metabolism in peripheral tissues. Regarding the central source of insulin, peripheral insulin enters the brain via an insulin-mediated pathway and triggers two distinct signaling cascades, namely the phosphoinositide 3-kinase (PI3) pathway and the mitogen-activated protein kinase (MAPK) pathway, which is crucially involved in insulin sensitivity (Begg, 2015; Kullmann et al., 2020a). However, whether the brain locally synthesizes insulin is unclear. Previous studies suggested the presence of immunoreactive insulin in the brains of humans and rodents, which was opposed by subsequent studies believing that “little or no insulin is produced in the CNS” (Banks, 2004; Blázquez et al., 2014). Recently, emerging evidence showed that insulin was produced by a subpopulation of inhibitory neurons of the cerebral cortex to rapidly regulate the energy homeostasis of the neural network (Csajbók and Tamás, 2016). Overall, the source of insulin in the brain entails further exploration to draw a definite conclusion. Moreover, a systematic review of patients with T2DM suggested that improving glycemic control by peripheral insulin affects brain function (Geijselaers et al., 2015). In addition, other studies have proved that the brain inherently contains insulin, which is involved in metabolic modulation and neural circuit function.

Insulin regulates peripheral metabolism, including regulation of appetite, white fat mass, hepatic glucose output, etc. The influence of CNS insulin on decreasing dietary consumption and body weight is one of the earliest discoveries on the topic. For instance, researchers reported that intracerebroventricular (ICV) injection of insulin could inhibit food intake, which is partly regulated by unexcited orexigenic agouti-related peptide (AgRP) neurons and activated ATP-dependent potassium channels in hypothalamic neurons (Könner et al., 2007; Mc Allister et al., 2015). Similarly, Koch et al. (2008) found that chronic ICV insulin treatment promotes fat mass, fat cell size, and adipose tissue lipoprotein lipase expression.

Besides, neurons express insulin receptors to maintain synapse density, dendritic plasticity, and normal functions (Chiu et al., 2008). Insulin receptor signaling promotes dendritic spine establishment and excitatory synapse development through activating PI3K/Akt/mTOR and Rac1 signaling pathways (Lee et al., 2011). Insulin signaling also benefits cognitive competence. Insulin/IGF-1 signaling charges spatial learning and memory in the hippocampus (Soto et al., 2019). In contrast, serine/threonine-protein kinase phosphorylation, a marker of IR is correlated with a lower level of global cognitive function (Arvanitakis et al., 2020). Furthermore, central insulin is involved in AD and other psychotic disorders. For AD patients, insulin signaling ameliorates amyloid-β (Aβ) fibrils and tau protein hyperphosphorylation. Claxton et al. (2015) found that intranasally administered insulin augments their verbal and visuospatial working memory (see Figure 1).

Insulin resistance refers to the impaired ability of insulin to exert its influence on target tissues (Kullmann et al., 2016). In the periphery, systematic IR is a hallmark feature of obesity. However, it has to be pointed out that the brain is an insulin-sensitive organ, and IR has been observed in the CNS of individuals with obesity and metabolic diseases like T2DM.

Recent advances in neuroimaging techniques have further promoted studies on brain IR in obese people. For example, a study on magnetoencephalography first reported brain insensitivity to exogenous insulin in obese adults in 2006. Tschritter et al. (2006) found that insulin infusion stimulates spontaneous cortical activity in healthy subjects, which was not observed in obese subjects. Moreover, Pegueroles et al. (2020) found a higher BMI augmented brain glucose metabolism and reduced cortical thickness after performing imaging brain glucose metabolism using fluorine-labeled fluorodeoxyglucose ([18F]-FDG) PET. Apart from obesity, visceral fat or body fat distribution is also linked with central insulin sensitivity (Kullmann et al., 2020b). This is achieved by an increase in functional connectivity between the dorsal medial prefrontal cortex and hippocampus after insulin administration (Kullmann et al., 2017). Therefore, central insulin sensitivity is vital to prevent visceral fat accumulation, which not only causes white matter lesions but also increases the risk of developing dementia (Ozato et al., 2021).

Obesity has long been known as a product of nutritional excess. However, independent of obesity, overnutrition can directly blunt central insulin sensitivity, even before peripheral insulin signaling is impaired. After short-term HFD exposure, rats were found to be unresponsive to 3rd-cerebral ventricle insulin administration, which is embodied by decreased ability to reduce food intake and activate hypothalamic signal transduction (Posey et al., 2009; Clegg et al., 2011). Gray and colleagues found that HFD feeding diminishes brain endothelial cell insulin uptake and transcytosis (Gray et al., 2017). Yet, whether brain insulin delivery is affected by IR needs to be further explored. Some studies have reported reduced insulin transportation into the cerebrospinal fluid (CSF) in several IR states (Heni et al., 2014). The underlying mechanisms of overnutrition leading to central IR have also been unveiled. Zhang et al. (2008) delivered IKKβ^{rm CA} bilaterally into the MBH of C57BL/6 mice and found that lIKKβ/NF-κB interrupts central insulin signaling in the hypothalamus because IKKβ^{rm CA} decreases insulin-stimulated Akt activation and PIP production and hinders tyrosine phosphorylation in insulin receptor and insulin receptor substrate 2 (IRS2).

Previous studies explored the relationship between cognition diseases and IR. An 11-year follow-up study observed that IR predicted poorer verbal fluency and a steeper decline in verbal fluency. It indicated that IR might serve as a precursor to cognitive disorders and intranasal insulin might become a potential therapy to hinder cognitive function decline (Kullmann et al., 2016; Ekblad et al., 2017). At present, the influence of central insulin on neuropathological characteristics of AD has been widely studied. High levels of pSer312-IRS-1 (ineffective insulin signaling) and lower levels of p-panTyr-IRS-1 (effective insulin signaling) are the exosomal biomarkers of brain IR. Studies proved that these biomarkers were associated with brain atrophy in AD (Mullins et al., 2017). Besides, in the cerebellar cortex of patients with AD and without T2DM, the response to insulin signaling in the IR→IRS-1→PI3K signaling pathway is markedly reduced. Moreover, increased biomarkers of brain IR are positively associated with aggregation of oligomeric Aβ plaques and dysfunction of working memory, which demonstrates that brain IR is highly related to cognition decline (Talbot et al., 2012).

Knowing that brain IR acts as a shared pathological feature of metabolic dysfunction and cognitive disturbances and that overnutrition and obesity both affect cognition, brain IR may be a bridge in the influence of nutritional excess and obesity on cognitive function. Nowadays, multiple studies combined factors of HFD, brain IR, and cognition impairment and found that HFD and a high-fat/high-sucrose (HFS) diet increases telencephalic IR [increased expression of IR marker IRS1-pS (616)], as well as amyloid-beta deposition and neurofibrillary tangle formation, having a deleterious effect on synaptic integrity and cognitive behavior (Arnold et al., 2014; Kothari et al., 2017). However, few of these studies truly illustrated underlying mechanisms and casual relationships, which should be addressed by future research.

The brain can modulate functions and environments of the gut, which then influence enteric microbiota via a series of parallel outflow systems, including sympathetic and parasympathetic branches of the ANS, the HPA axis, and endogenous pain-modulation systems. For example, neuronal and neuroendocrine signaling like norepinephrine released in the intestine under stress or trauma increases the virulence of Campylobacter (Cogan et al., 2007).

At the same time, the gut microbiome produces various signaling molecules which connect CNS through endocrine, immune, and neural signaling mechanisms (Rhee et al., 2009). For the endocrine aspect, enterochromaffin cells work as signal transducers. These cells promote CaV-dependent serotonin release and form synaptic-like structures with 5-HT3R-expressing nerve fibers for transmitting signals (Bellono et al., 2017). Additionally, tryptophan catabolism produced by Edwardsiella tarda and other gut microbiome stimulates enteroendocrine cells, which activates vagal sensory ganglia and secretes 5-HT (Ye et al., 2021). As for immune pathways, the colony of diverse bacterial communities in the gut induces IL-17-producing γδ T cells that are transferred to the brain under a specific circumstance like an ischaemic injury, where they exacerbate inflammation and tissue impairment (Powell et al., 2017). At last, although the gut microbiota communicates with the brain via endocrine and immune pathways, vagus nerve signaling is probably the fastest and most direct approach (Fülling et al., 2019). Goehler et al. (2005) found that live Campylobacter jejuni activates c-Fos in neurons in the vagal sensory ganglia and the solitary tract nucleus in mice. Moreover, Bravo et al. (2011) suggested that Lactobacillus rhamnosus reduces GABA (Aα2) mRNA expression in the prefrontal cortex and amygdala, which were not found in vagotomized mice. Similarly, Limosilactobacillus reuteri improves social interaction-induced synaptic plasticity in a vagus nerve-dependent manner (Sgritta et al., 2019).

In both the human and mouse models of obesity, the most acknowledged alteration in the gut microbiome is an increase in the relative abundance of Firmicutes and a decrease in Bacteroidetes compared with lean controls (Leigh and Morris, 2020). Clinical studies suggested that obese participants with more marked adiposity and increased fat mass are more likely to have low bacterial genes in the gut (<480,000 genes) (Le Chatelier et al., 2013). Also, higher BMI has been associated with lower alpha diversity, species richness, and beta diversity (Geisler et al., 2021).

Based on that, pertinent therapeutic methods targeting the gut-brain axis have gradually emerged. Erchen Decoction helps with obesity and lipid metabolism disorder by adjusting gut microbiota’s structure, abundance, and function following reduced metabolite -propionic acid (Zhao et al., 2021). Ussar et al. (2015) suggested that transplanting microbiota from obesity-resistant mice to germ-free mice with obesity susceptibility can ameliorate their specific metabolic consequences of obesity. Also, Nicolucci et al. (2017) found that prebiotics decreases body weight z-score, percent body fat, percent trunk fat, and serum level of IL-6 in children with overweight or obesity via selectively adjusting intestinal microbiota.

Clinical studies pointed out that gut microbiota could rapidly respond to an altered diet. Compared to a plant-based diet, a greater change in gut microbial abundance was found in an animal-based diet with high fat and high protein (David et al., 2014). Similar results were also observed in animal experiments. HFD increased the abundance of Firmicutes in different strains of mice. Among these operational taxonomic units (OTUs), Clostridiales, Roseburia intestinalis, and Ruminococcaceae were positively correlated with specific metabolic phenotypes (Ussar et al., 2015).

Interestingly, weight gain is not a necessary factor contributing to cognitive impairment induced by a high nutrition diet. For example, Ravussin et al. (2012) found no difference in weight between mice fed with a high-fat diet and mice fed with a relatively low-fat diet; yet, the abundance of OTUs belonging to the Lachnospiraceae and the Deferribacteraceae was different, which also explained different circulating leptin and inflammation markers between them. Hildebrandt et al. (2009) found a decrease in Bacteroidetes and an increase in both Firmicutes and Proteobacteria in obese mice or RELMbeta knockout mice who remained comparatively lean when fed with HFD. Maternal HFD, but not maternal obesity, reconstructed the offspring’s intestinal microbiome (Liu et al., 2021). However, a recent study pointed out that refined HFD promotes the alpha diversity in the cecum and colon and decreases the Firmicutes: Bacteroidetes ratio. This is inconsistent with previous studies, which is probably because of the higher content of fiber in refined HFD (Wang et al., 2020).

Gradually, surging evidence in animal studies revealed that interventions affecting the enteric microbiome and targeting specific bacterial species could influence cognition. The alteration of the gut microbiome prevented, ameliorated, and even restored brain function after obesity or diet-induced cognitive deficits. It also benefited cognitive performance in healthy models. In the future, similar microbiome interventions are expected to apply in clinical treatment as well as benefit patients suffering from cognition degeneration.

Diet-induced obesity propels neuroinflammation, as proved that DIO-induced morphological changes in the hypothalamic microglia and macrophages were similar to those of reactive microglia induced by peripheral nerve injury (Kälin et al., 2015). Interleukin plays a key role in obesity while associated with neuroinflammation. In the ARC of the hypothalamus of adult mice, IL6 and its canonical receptor are detected, and elevated expression of IL6 protects mice from DIO (Bobbo et al., 2021). Most recently, scientists figured out IL-27-IL-27Rα signaling in protecting against DIO and ameliorating IR (Wang Q. et al., 2021) but whether IL-27 is associated with inflammation alleviation in CNS as the base of common characteristics of obesity and neuroinflammation still remains further explore.

If we intend to link adiposity and cognitive disorders from an immune perspective, HIV-infected patients with immune defects would be a special group to explore. HIV directly or indirectly targets white adipose tissue, thus contributing to systematic inflammation or inflammation of distant organs such as the brain through adipokine secretion and lipotoxicity (Bourgeois et al., 2021). Neuroinflammation (choline and myoinositol) was detected in patients diagnosed with HIV-associated neurocognitive disorders (HAND), which was accompanied by less cortical gray matter and more abnormal white matter. It explained the comorbidity of acquired immune deficiency syndrome (AIDS) and cognitive decline to some extent (Saloner et al., 2019). Moreover, central obesity-related to AIDS predicts a higher prevalence of neurocognitive impairment, appearing as a decline in motor function, learning, and memory (McCutchan et al., 2012; Rubin et al., 2019). Therefore, it is possible that adiposity, especially central obesity, induces systematic and neuronal inflammation and causes cognitive deficits.

However, recent studies proved that overnutrition might cause inflammatory responses in the CNS, which appear earlier than the onset of obesity. Surprisingly, even after 1-day HFD exposure, increased microglial reactivity and pro-inflammatory cytokines like TNF have been observed in the MBH in mice and rats (Thaler et al., 2012). Diet-induced neuroinflammation is a key factor associated with obesity. It acts by regulating food intake behavior as well as microglial activation, thus consequently altering hypothalamic neurons at the earliest stage of obesity (see Figure 3).

A great amount of experimental data show strong bonds between overnutrition-induced cognitive impairment and hypothalamic or extra-hypothalamic inflammation. Hsu et al. (2015) found that a high fructose corn syrup-55 intake damage hippocampal-dependent spatial learning and memory in adolescent rats, which is probably associated with elevated expression of pro-inflammatory cytokines (IL6, IL1β) in the dorsal hippocampus. Moreover, Spencer et al. (2017) found that short-term exposure to HFD impaired hippocampal-dependent and amygdala-dependent memory in aged rats, accompanied by elevated IL-1β protein. When IL-1RA was centrally administrated, previous memorial deficits were alleviated, which indicated that IL-1β related neuroinflammation has a role in cognitive damage (Spencer et al., 2017). Interestingly HFD does not affect the basal expression of pro-inflammatory cytokines at the periphery or in the brain but enhances the expression of IL-1β and TNF-α, specifically in the hippocampus (Boitard et al., 2014).

Overfeeding mice with an SFAs diet caused an accumulation of microglia in the MBH and activated to M1 phenotype, followed by increased hypothalamic SFA levels (Beilharz et al., 2014). Moreover, a cascade of inflammatory signaling triggered by SFAs has been discovered. After combining with TLR4 and then activating myeloid differentiation primary response gene 88 (MyD88), its downstream signaling, including IKK complex and NF-κB, was triggered, which eventually led to the expression of proinflammatory genes in the hypothalamus (Milanski et al., 2009). Except for TLR/NF-κB signaling, IKKβ/NF-κB signaling, a master switch and main monitor of innate immunity in overnutrition-induced neuroinflammation was observed. The activation process firstly targeted IKKβ activated by phosphorylation at S177 and S181, which triggered the phosphorylation, ubiquitination, and proteasomal degradation of its substrate IκBα. NF-κB was then released in combination with IκBα and transported to the nucleus to transcript its relative genes (Hayden and Ghosh, 2008). Interestingly, Zhang et al. (2008) proved that up-activated NF-κB in the hypothalamus was elevated by 2-folds after exposure to HFD, and mRNA levels of IKKβ and IκBα in the hypothalamus were also enhanced. Additionally, studies proposed TNF-α as an activator and a gene product of IKKβ/NF-κB (Cai, 2013). Furthermore, inhibited weight gain, glucose intolerance, and IR were found in HFD-fed mice with a tamoxifen-inducible astrocyte-specific knockout of IKKβ (Douglass et al., 2017).

High nutritional diet and obesity can lead to cognitive dysfunction through the process of neuroinflammation, as formerly discussed; therefore, some consider it as one of the mechanisms of diet and obesity-induced cognitive impairment.

Microglia and astrocytes are the brain’s resident immune cells. As discussed above, overnutrition induces the expression of inflammatory cytokines as well as triggers microglia to a proliferated and activated state. Microglia surveil and phagocytose synapses regulate neural circuit formation and function during this process. During interaction with synapses, specific presynaptic (SNAP25) and postsynaptic (PSD95) proteins have been identified (Miyamoto et al., 2013; Miller and Spencer, 2014). Apart from structural reconstruction, microglia influence synaptic transmission, and functional maturation. LPS, a classic ligand of TLR4, activates microglia to release ATP, which works as a primary messenger to recruit astrocytes by purinergic signaling. Astrocytes then release glutamate to promote synaptic transmission. Furthermore, when microglia are inhibited by an anti-inflammatory called agentminocyclin, neuron remains silent to LPS stimulation, which further suggests that active microglia are indispensable to ALPS-mediated neuronal spontaneous excitatory postsynaptic currents (EPSCs) (Tikka and Koistinaho, 2001; Pascual et al., 2012).

In addition, systemically-derived pro-inflammatory cytokines also affect neuronal function, especially in neurodegenerative disorders. A meta-analysis showed elevated CSF TGF-β, MCP-1, and YKL-40 levels in AD patients and increased CSF TGF-β1, IL-6, and IL-1β levels in PD patients (Chen et al., 2018). TNFα is a vital neurodegenerative factor that is associated with M1 microglial polarization. Treating 12-month-old APP/PS1 mice with an ICV injection of anti-TNFα mAb triggered reductions in both Aβ1–42 load and plaque deposition. Similarly, by removing NLRP3 in APPSWE/PS1ΔE9 mice, age-dependent elevation in IL-1β was ablated, which alleviated the Aβ burden and rescued spatial learning and memory impairments (Minter et al., 2016).

Various medications have been proved to improve neuroinflammation. Atypical antidepressant agomelatine (Ago) could reverse gliosis in specific brain regions, which also augmented its curative effects on depression (Atanasova et al., 2021). Since circulating adiponectin (APN) deficiency could trigger neuroinflammatory reactions, the APN receptor agonist adipoRon was applied to produce anti-inflammatory effects by decreasing microglial and astrocyte activation as well as restricting cerebral cytokine levels, which is also reckoned as a potential treatment for AD (Ng et al., 2021). Besides, traditional medications such as honokiol are also involved in neuroprotection and neuroinflammatory alleviation (Talarek et al., 2017). Therapies targeting neuroinflammation probably improve psychiatric symptoms to some extent due to the close relationship between these symptoms with neuroinflammation and psychiatric diseases (Najjar et al., 2013; Meyer et al., 2020). More importantly, they may avoid the side effects of obesity or overweight caused by many typical psychiatric drugs. A large number of studies confirmed that antipsychotic drugs, mood stabilizers, and antidepressants had negative effects on weight gain, even for short-term usage (Wirshing, 2004; Chao et al., 2019). Therefore, for psychiatric patients with weight problems, periodic monitoring of BMI and substitute medications are necessary.

In conclusion, obesity and overnutrition stimulate central inflammation via various mechanisms, which can be presented in gliosis and many inflammatory mediators. In addition, increased central immune cells and cytokines target synaptic modification and pathology of neurodegenerative diseases, which negatively impact cognitive function.