Diabetes and Cognitive Impairment: A Role for Glucotoxicity and Dopaminergic Dysfunction

Diabetes mellitus (DM) is a chronic metabolic disorder characterized by hyperglycemia, responsible for the onset of several long-term complications. Recent evidence suggests that cognitive dysfunction represents an emerging complication of DM, but the underlying molecular mechanisms are still obscure. Dopamine (DA), a neurotransmitter essentially known for its relevance in the regulation of behavior and movement, modulates cognitive function, too. Interestingly, alterations of the dopaminergic system have been observed in DM. This review aims to offer a comprehensive overview of the most relevant experimental results assessing DA’s role in cognitive function, highlighting the presence of dopaminergic dysfunction in DM and supporting a role for glucotoxicity in DM-associated dopaminergic dysfunction and cognitive impairment. Several studies confirm a role for DA in cognition both in animal models and in humans. Similarly, significant alterations of the dopaminergic system have been observed in animal models of experimental diabetes and in diabetic patients, too. Evidence is accumulating that advanced glycation end products (AGEs) and their precursor methylglyoxal (MGO) are associated with cognitive impairment and alterations of the dopaminergic system. Further research is needed to clarify the molecular mechanisms linking DM-associated dopaminergic dysfunction and cognitive impairment and to assess the deleterious impact of glucotoxicity.


Introduction
Diabetes mellitus (DM) is a heterogeneous chronic metabolic disorder, characterized by hyperglycemia, representing a global epidemic public health problem [1]. Premature morbidity and mortality of DM are due to long-term diabetic complications [2], including retinopathy, nephropathy, peripheral vascular disease, and heart disease (micro-and macro-vascular disease) [3]. Hyperglycemia brings together type 1 and type 2 diabetes, the two most common forms of DM, which differ in both epidemiology and etiology. Type 1 diabetes (T1D) is mostly a juvenile-onset disease due to autoimmune destruction of pancreatic beta cells, leading to an absolute deficiency in insulin production. Type 2 diabetes (T2D) includes 90-95% of diabetes cases and is a pathology typical of the elderly, resulting from insulin resistance accompanied by progressive beta cells deficit [4]. Interestingly, since 1922 [5], the idea that both T1D [6] and T2D [7] are accompanied by a worsening of

Dopamine and Cognition
DA is historically known for its key role in the regulation of behavior and movement. Indeed, in 1973, for the first time, the strong undeniable association between striatal DA depletion and motor deficits characteristic of Parkinson's disease (PD) emerged [55]. Later, signs of abnormal dopaminergic function were found in several diseases, such as schizophrenia [56] and attention deficit hyperactivity disorder [57]. The idea that DA Figure 1. The main signaling pathways modulated by DA upon binding to specific G proteincoupled receptors.

Dopamine and Cognition
DA is historically known for its key role in the regulation of behavior and movement. Indeed, in 1973, for the first time, the strong undeniable association between striatal DA depletion and motor deficits characteristic of Parkinson's disease (PD) emerged [55]. Later, signs of abnormal dopaminergic function were found in several diseases, such as schizophrenia [56] and attention deficit hyperactivity disorder [57]. The idea that DA could play a key role in cognitive function arises from the observation that these pathologies, due to anomalies of the dopaminergic system, are also characterized by cognitive impairment [58][59][60]. In particular, PD is characterized by alterations of several cognitive domains, including executive functions [61], attention [62], verbal fluency, visuospatial skill [63], episodic memory [64], and reasoning [65]. Interestingly, some of these cognitive domains are altered in DM, too [66][67][68][69]. Imaging studies performed in PD patients revealed a positive correlation between decreased DA levels and cognitive impairment [70][71][72][73]. Interestingly, aging-associated cognitive decline is also accompanied by various modifications of the dopaminergic system [74]. The observed changes concern a reduction of DA receptor and transporter density [75][76][77][78]. In more detail, in healthy subjects, molecular imaging studies highlighted an age-related decrease in striatal dopamine transporter (DAT) density, paralleled by a worse performance on several tasks, including episodic memory, executive functioning, and verbal learning tasks [79,80].
Similarly, healthy aging is also characterized by a loss of D2 receptors in both striatal and extrastriatal areas and by simultaneous cognitive deficits [81][82][83]. An increase of DA catabolism [84] and alterations of DA synthesis [85] were observed too. Thus, dopaminergic dysfunction could mediate the association between aging and cognitive decline [86]. Moreover, impaired DA transmission was observed in diseases featuring cognitive deficit, such as Alzheimer's disease [87], autism [88], and Huntington's chorea [89]. Several in vivo data obtained in different experimental models undoubtedly link DA release to cognitive function. Experiments of in vivo microdialysis allowed evidence of increased DA release in the prefrontal cortex (PFC) of rats [90] and monkeys [91] during working memory tasks. Accordingly, the application of DA modulates neuronal "memory field" activity of PFC neurons [92]. DA relevance for cognition was then confirmed by both experimental manipulation of the dopaminergic system in animal models and by pharmacological studies in humans. Several studies that aimed to clarify DA involvement in cognitive function were performed in rodents. Simon and coworkers showed that in rats, the bilateral injection of 6-hydroxydopamine (6-OHDA) into the lateral septum selectively abolishes dopaminergic innervation and leads to deficits in spatial-memory tasks, without significantly damaging endogenous noradrenergic and cholinergic systems [93,94]. Similarly, in rhesus monkeys, depletion of DA in PFC severely impairs working memory [95]. The entity of the deficit is comparable to that observed when the PFC itself is ablated [96,97]. Interestingly, no alterations of working memory were observed in monkeys subjected to the depletion of other neurotransmitters [98]. Moreover, the deleterious effect of depletion of DA in PFC was reverted by treatment with DA receptor agonists, supporting the selective relevance of DA for working memory [95,99]. This finding was further confirmed by results obtained in rhesus monkeys when selective antagonists of the D1 dopamine receptor were locally injected into the PFC and altered mnemonic processes [100]. Experiments with agonists and antagonists of the D1 dopamine receptor further strengthen the idea that cognitive processing strongly depends on an optimal level of DA. Indeed, both excessive and inadequate activation of D1 receptor impairs working memory ability in both monkeys and rodents [101][102][103]. Analogous studies, performed in humans, revealed that the treatment of healthy human subjects with the selective D2R agonist bromocriptine facilitates spatial working memory [104]. Similarly, the administration of pergolide, an agonist for both D1 and D2 family receptors, improved performances in working memory tasks [105]. More detailed dose-response experiments with dopaminergic drugs supported the hypothesis that the complex functional relationship between DA and working memory is regulated by a nonlinear inverted U-shaped dose-response curve [106], where both low and excessive doses of DA impair working memory performance [98]. This trend is likely influenced by baseline levels of DA [107] and may depend on the differential effects of DA receptor activation in the striatum and PFC [108]. More recently, effects of D1 and D2 receptors agonists have been better investigated across multiple tasks exploring several cognitive domains, such as memory, flexibility, and learning in non-human primates, unveiling dose-and taskspecific actions and strongly suggesting distinct cognitive functions of DA receptors in the PFC and striatum [106,108]. Indeed, cognitive control deficit is often due not only to malfunctioning of PFC, but also to impaired striatal DA transmission. Studies performed with 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) in both animal and human models disclosed the relevance of dopaminergic signaling in the striatum for cognitive functions. Stern Y and coworkers in 1990 investigated general intellectual function, construction, language, memory, executive function, attention, and reaction time in MPTP-exposed individuals, characterized by reduced uptake of labeled 6-fluorodopa into the striatum. They featured significantly worse performances in a specific set of cognitive functions mediated by the dopaminergic system [109]. Similarly, in rats, intranigral administration of MPTP causes a partial lesion in the substantia nigra, compact part (SNc), and a specific loss of DA in the striatum, inducing habit learning and working memory deficits [110,111]. Memory acquisition and retention processes were impaired, too [112]. The idea that an impaired nigrostriatal system participates in cognitive dysfunction was further supported by bilateral lesion obtained by 6-OHDA injection into the ventrolateral neostriatum, leading to altered working memory, accompanied by striatal DA depletion [113]. Interestingly, intrastriatal administration of D2 receptor agonist to rats leads to improved cognitive performance [114,115]. Similar results were obtained in MPTP-lesioned monkeys upon systemic injection of D1 receptor agonist [116], suggesting that mnemonic processes need normal stimulation of striatal DA receptors. Studies performed in PD patients highlighted that DAT availability in the caudate, anterior putamen, and ventral striatum was also directly associated with attention/working memory, frontal/executive, and visuospatial functions [117]. Some experimental evidence supports a role for DA in the regulation of attentional function [118], too. In rats, bilateral 6-OHDA lesions of the terminal area of the nigrostriatal DA system induce a partial DA denervation of the striatum, leading to attentional deficits [119,120]. DA's role in the modulation of attentional processes was further confirmed by the observation that the stimulant drug methylphenidate improves attention in ADHD patients by enhancing DA signaling in the ventral striatum [121]. Interestingly, in intrastriatal bilateral 6-OHDA rats, the altered attentional performance on a reaction time task was significantly improved by co-administration of L-DOPA with piribedil, an agonist of D2/D3 receptors [122]. Several pharmacological studies then allowed better clarification of the role of DA receptors in attentional control, showing that the D2 receptor seems to be more involved in attentional control than the D1 receptor [123,124]. More recent findings in mice revealed that DA released in the dorsal hippocampus from the locus coeruleus has a role in increasing selective attention to relevant features of the environment, promoting spatial learning via D1/D5 receptor [125]. Moreover, attentional performances in healthy individuals are affected by genetic factors, such as SNPs in dopaminergic genes. Several studies suggest that dopaminergic polymorphisms influence selective aspects of cognition [126,127]. For instance, COMT val158met polymorphism modulates dopaminergic signaling affecting the function of the cingulate cortex during attentional control [128]. In addition, the allelic mRNA expression analysis of human brain autopsy tissues, followed by the SNP scanning of the DRD2 locus, allowed the discovery of regulatory polymorphisms modulating DRD2 splicing, working memory network, and cognitive performance in healthy humans [129]. Similarly, polymorphisms in several other dopaminergic genes, such as DRD1 and DRD4, modulate the executive function and working memory processes in healthy subjects [130]. Finally, dopaminergic system integrity is also necessary for several other cognitive functions, such as intertemporal choice [131], task-switching [132], response inhibition [133], and arousal [134].

Alterations of Monoamines System in Diabetes
An undeniable cross talk between the brain monoamine neurotransmitter system and glucose homeostasis has been widely described in the literature. Both T1DM and T2DM are indeed associated with deleterious changes in the brain monoaminergic system that play a role in the pathogenesis of DM and of DM-associated neurodegeneration [135][136][137]. In particular, several research groups described alterations in the metabolism and content of adrenaline, norepinephrine, serotonin, and DA in different specific brain areas of diabetic rodents and humans [138][139][140][141][142]. However, DA and serotonin appear to be the main regulators of cognitive function [143].

Serotonergic System
Studies performed in different experimental models evidenced alterations of the serotonergic system in DM. Indeed, in the CNS of T2DM patients, several perturbations of the serotonergic system were observed, such as a decreased free tryptophan amount, impaired serotonin biosynthesis, and alterations of serotonin receptors [144][145][146]. Changes of the serotonin content in the medial and lateral hypothalamus were evidenced in diabetic patients [139] as well as in specific brain regions of STZ-and alloxan-treated rats [138,147]. In more detail, in STZ-induced T1DM rodent models, in vivo intracerebral microdialysis studies under free-moving conditions evidenced a significant decrease of serotonin levels in the hypothalamus, hippocampus, brainstem, cortex, and ventromedial hypothalamus [148][149][150][151][152]. Different molecular mechanisms have been proposed to explain a reduction of the serotonin amount in STZ models, including increased activity of MAO-A/MAO-B, higher serotonin reuptake by serotonin transporter SERT, and elevated plasma levels of branched-chain amino acids, which compete with tryptophan to cross the blood brain barrier [153]. Similarly, a murine model of T2DM induced by a prolonged high-fat diet features a reduction of serotonin extracellular levels in hippocampal and a hypersensitization of inhibitory 5-hydroxytriptamine A1 serotonin autoreceptors in dorsal raphe nuclei, leading to an inhibition of the serotonergic circuit. These deleterious changes contribute to T2DM-associated mood and eating disorder [154] and lead to lower insulin sensitivity and metabolic dysfunctions, playing a role in T2DM exordium [135]. Indeed, serotonin and serotonergic drugs ameliorate peripheral glucose uptake, glucose tolerance, and insulin sensitivity in diabetic rat models and in diabetic patients [155][156][157][158]. Moreover, a critical role of serotonin in cognitive processes was pointed out by experimental manipulation of tryptophan levels in murine models, primates, and humans [159][160][161][162][163] and by the finding that anomalies of serotonin content and signaling are involved in cognitive decline associated with Alzheimer's disease and ageing [164,165]. Reduced serotonin transmission leads to impaired learning and memory function, while augmented serotonin transmission ameliorates cognitive performance both in rodents and humans [166]. Pharmacological strategies that aim to restore serotonin levels have beneficial effects both in diabetic patients and in T2DM animal models, improving cognitive function and metabolic parameters [144][145][146].
A key role for serotonergic alterations in cognitive dysfunction also emerged in rats in the later stage of T1DM, characterized by reduced spatial memory and learning ability [167]. An interesting topic is the intricate cross talk between serotonergic and dopaminergic systems, clearly highlighted by neurological and pharmacological studies and by evidence obtained in knockout murine models [168][169][170][171]. The serotonergic system modulates dopaminergic transmission by 5-HT2A receptors and there is also evidence of the existence of heteromers 5-HT2A/D2 receptors expressed in dopaminergic cells of different brain areas [172,173]. A further level of serotonergic and dopaminergic interaction occurs at the postsynaptic level within the PSD (post synaptic density), involving several scaffolding protein and signaling molecules [174]. The interplay between dopamine and serotonin systems obviously has both physiological and pathological implications and research is actually focused on its relevance for antipsychotic action and for the discovery of innovative therapeutic target for psychosis [175]. At variance, its contribution to DM-associated cognitive dysfunction has not been investigated yet.

Dopaminergic System
Some experimental data highlighted an interplay between glucose metabolism and the dopaminergic system. Indeed, it has been shown that, both in rodents and humans, modulation of striatal and systemic DA levels impinges on whole body glucose metabolism [176] and energy homeostasis [177,178]. In murine models, optogenetic activation of nucleus accumbens (NAc) cells expressing DRD1 improved glucose tolerance and insulin sensitivity [176]. In addition, DRD2 is implicated in the modulation of insulin secretion [179] and different authors showed that systemic treatment with bromocriptine, a DRD2 agonist, improves insulin sensitivity [180] and glucose tolerance in humans [181]. Similar results were observed in obese hamsters, too [182]. In contrast, systemic DA depletion leads to a decrease of striatal DA levels and in turn to a reduction of insulin sensitivity in healthy subjects [176], as well as antipsychotics, inhibiting DA receptors, and inducing hyperinsulinemia and glucose intolerance [183,184]. Moreover, striatal DA receptors regulate the expression of insulin receptor and of the neuron-specific glucose transporter GLUT-3 in streptozotocin diabetic rats [185].
On the other hand, the presence of DM promotes neurodegeneration and impairs dopaminergic neurotransmission [186]. This is consistent with the finding that the two major players of DM, such as hyperglycemia and relative insulin deficiency, can alter the dopaminergic system. Indeed, insulin is a key regulator of both neurons' survival and DA metabolism. First of all, insulin protects rat hippocampal cells in culture by oxygenglucose deprivation [187] and has neuroprotective action against H2O2 in retinoic acid (RA)-differentiated SH-SY5Y cells [188]. Similarly, in rats, insulin protects dopaminergic neurons of substantia nigra against 6-OHDA toxicity [189]. Importantly, impaired insulin signaling alters DA homeostasis [190][191][192][193] and the ablation of insulin receptors in dopaminergic neurons interferes with DA action on control of food intake [194]. Accordingly, it was recently shown that in ex vivo differentiated human dopaminergic neurons and in SH-SY5Y cells in culture, insulin resistance is accompanied by mitochondrial dysfunction, increased ROS levels, and increased expression of alpha-synuclein [195]. Insulin is a known modulator of DA synthesis and turnover, too. As proof of this, NIRKO mice, carrying a brain-specific knockout of the insulin receptor, feature increased DA turnover in the striatum and NAc, resulting in decreased DA signaling [196]. Moreover, in non-diabetic rats, insulin injection increases DA levels in NAc [197]. Interestingly, insulin is able also to regulate the expression of TH [198] and increase DA uptake by DAT [190,199,200]. Similarly, opportune glucose intake to the brain is crucial for both dopaminergic neurons homeostasis and DA metabolism. Studies focusing attention on the hyperglycemia effect in dopaminergic neurons revealed that they are prompted to apoptosis by chronic glucose exposure through oxidative damage [201][202][203]. In PC12 cells, chronic incubation with high glucose augmented depolarization-induced DA release [204], and in healthy human subjects, blood glucose levels are related to cerebrospinal fluid concentrations of the DA metabolite homovanillic acid [205]. In rats, variations of ambient glucose levels in substantia nigra, obtained by use of microdialysis probes, produce different effects on DA release, depending on both the concentration and duration of infusion. Glucose action seems to also involve ATP-sensitive K+ channels and regulate the efflux of other neurotransmitters, too. However, in the nigrostriatal pathway, glucose infusion seems to increase DA release when glucose availability is low while decreasing DA release when glucose is abundant [206]. Interestingly, the huge impact of glucose and insulin on the dopaminergic system has recently been observed in Caernorhabditis elegans, too [207]. Thus, given the key role of insulin and glucose in DA homeostasis, it is not surprising that dopaminergic function is altered in DM. Studies evidencing DM-associated dopaminergic dysfunction were performed in DM animal models for the vast majority. At variance, few studies about dopaminergic dysfunction have been conducted in diabetic patients, thus it is not clear yet if there are substantial differences in dopaminergic alterations between T1DM and T2DM patients. Some authors described an increase of DA levels during DM in specific brain regions of alloxan-or streptozotocin (STZ) rats [138,208], as well as diabetic patients [139].
The selectivity of DA content alterations was further confirmed by Ezzeldin et al. They found a reduced DA amount in the cerebral cortex, midbrain, and brainstem regions but augmented in the cerebellum and thalamus/hypothalamus [140]. However, in later years, there are more detailed studies supporting a reduction in DA levels in different brain areas during DM. In particular, in the hippocampus of STZ rats and spontaneously diabetic WBN/Kob rats (WBN rat), a reduction of DA levels and release was observed [151]. Interestingly, the reduced DA content in the hippocampus of STZ diabetic rats is paralleled by compensatory upregulation of DRD1 and DRD2 expression and contributes to a cognitive deficit [209]. Gallego et al. observed a selective reduction of DA content in the dopaminergic nigrostriatal system in STZ rats, also highlighting that the alterations of catecholamine metabolism depend on the severity and duration of DM [210]. Very recently, dopaminergic alterations induced by long-term hyperglycemia were investigated in detail in STZ rats. The glucose amount was increased in the midbrain and striatum, but preferential neurodegeneration of the nigrostriatal pathway, accompanied by astrogliosis and loss of microglial cells, was observed with aging. The higher vulnerability of the nigrostriatal pathway to long-term hyperglycemia probably results from an elevated basal oxidative burden paralleled by low levels of antioxidant defense [211]. Similar results were obtained by Pérez-Taboada et al., who found decreased levels of DA and related metabolites in the striatum of both STZ-treated mice and diabetic db/db mice. A specific reduction of the expression of protein regulating DA neurotransmission and stimulus-dependent striatal DA release, such as DAT, VMAT2, and Girk2, was observed too [186]. It is worthy of notice that the expression of several proteins, involved in DA synthesis and degradation, including TH, MAO, COMT, and SNCA, is deregulated in DM. DM deleterious effects on TH function have been known since the 1980s, when modifications of the amount of aminoacids precursors and TH activity, leading to reduced striatal DA metabolism, were observed in STZ diabetic rats [212]. Several authors obtained similar results. Indeed, a progressive decrease in TH activity was observed in STZ-treated Sprague-Dawley rats by Bitar and coworkers [213]. Moreover, in STZ-treated rats, TH mRNA was increased in the locus coeruleus but decreased in the ventral tegmental area/substantia nigra pars compacta [214]. Similarly, reduced TH activity in terminal fields for noradrenergic and dopaminergic neurons was observed in experimental diabetes [150,151,215], while genetically diabetic Wistar rats feature decreased levels of immunoreactive TH [216,217], too. Interestingly, MAO shows significantly increased activity in diabetics' platelets [218] and an augmented expression in NIRKO mice [196]. Finally, it has been shown that some functional polymorphisms in the COMT gene, responsible for the modulation of its enzymatic activity in PFC, are significantly associated with T2DM [219][220][221]. DM's deleterious effect on the dopaminergic system was recently confirmed in human studies, too. Indeed, diabetic patients feature striatal dopaminergic deficits and elevated levels of proteins involved in neurodegeneration, such as tau and SNCA, in cerebrospinal fluid [222].

Glucotoxicity Role in Dopaminergic Dysfunction and Cognitive Impairment
Chronic hyperglycemia, typical of DM, seriously damages organs and tissues, leading to the onset of diabetic complications and giving rise to glucotoxicity. Among the involved mechanisms, overactivation of the hexosamine and polyol pathways, activation of protein kinase C (PKC), and increased intracellular formation of advanced glycation end products (AGEs) have been described [223]. The glucotoxicity condition is characterized by abnormal intracellular accumulation of reactive dicarbonyls, such as methylglyoxal (MGO), glyoxal, and 3-deoxyglucosone [224]. The α-ketoaldehydes MGO represents the most potent glycating agent, promoting the endogenous non-enzymatic glycoxidation of proteins, lipids, and nucleic acids and the consequent formation of advanced glycation end products (AGEs). In more detail, MGO targets arginine and lysine residues of proteins and deoxyguanosine in DNA, leading to the formation of AGEs and MGO-derived DNA adducts, respectively. MGO mainly comes from spontaneous degradation of triosephosphates deriving from glycolysis and, to a lesser extent, from the catabolism of threonine, catabolism of ketone bodies, degradation of glycated proteins, and lipid peroxidation. MGO and MGO-derived AGE plasma levels are higher both in T1DM and T2DM. Indeed, hyperglycemia increases glycolytic flux and/or decreases the activity of MGO detoxifying systems, promoting MGO accumulation [225]. The undeniable role of MGO and AGEs in DM and its vascular complications has been extensively reviewed elsewhere [226,227]. Briefly, it has been shown that MGO affects insulin secretion [228] and promotes insulin resistance in different tissues, such as skeletal muscle [229] and endothelium, both in vitro and in vivo [230]. Different mechanisms underlie MGO deleterious action on endothelial function, including the downregulation of specific miRNAs [231,232] and the increased accumulation of the antiangiogenic factor HoxA5 [233]. Several recent studies highlighted the relevance of MGO and AGEs not only in micro-and macrovascular DM-associated complications, but also in neurodegenerative diseases and in cognitive dysfunction [234][235][236][237]. A great deal of evidence in the literature demonstrates the deleterious effects of MGO in neuronal cells. Most of the studies have been performed in neuronal cells from the hippocampus, a brain region essential for cognitive processes. Upon MGO exposure, hippocampal neurons obtained from fetal hippocampi of Sprague-Dawley rats undergo apoptosis through both mitochondrial and Fas receptor-mediated pathways. This phenomenon is accompanied by an unbalance of the cytokines network and by a significant alteration of antioxidant capacity and detoxification mechanisms. In addition, other authors describe MGO-induced inhibition of catalase enzymatic activity and protein expression and an increase of NGF and proinflammatory cytokine IL-1beta levels in this cellular model. Similar results were obtained ex vivo in slices of the cerebral cortex and hippocampus from the neonatal rat brain, where MGO elicited its toxicity through both a ROS-dependent ERK1/2 pathway and ROS-independent p38 and JNK pathways [238]. Incontrovertible proof of the impact of glucotoxicity on DM-associated cognitive dysfunction in vivo comes from both animal and human studies. Huang and coworkers showed that in STZ diabetic rats, the increase of blood glucose levels correlates with increased serum MGO. High MGO levels increase the percentage of apoptosis in hippocampal neurons, altering the amount of cleaved caspase-3, Bcl-2, and Bax [239]. Subsequent animal studies further confirmed that neurotoxicity due to an increased amount of MGO may play a key role in DM-associated cognitive decline. Indeed, in Wistar rats, intracerebroventricular infusion of MGO impairs GLO1 (glyoxalase 1) activity, increases AGE content, and leads to cognitive deficit, altering the hippocampus but not the frontal cortex. In more detail, MGO injection impairs discriminatory memory without affecting learning-memory processes and locomotion behavior [240]. In addition, the novel object recognition task and Y-maze test showed that short-and long-term memory and short-term spatial memory are impaired by intracerebroventricular injection of MGO in rats [241]. Similarly, dietary AGEs can worsen learning and memory and induce mitochondrial dysfunction in mice [242]. Glucotoxicity relevance for neurodegeneration has been explored in human studies, too. First of all, a role for MGO and MGO-derived AGEs in neurodegenerative diseases, such as Alzheimer's disease and Parkinson's pathogenesis, has been evidenced [243,244]. In particular, protein glycation adduct levels are increased in CSF of Alzheimer's disease patients and MGO levels are increased in the serum of individuals with mild cognitive impairment [245]. Importantly, in non-demented elderly subjects, higher serum MGO amount [246] and dietary AGEs [247] are associated with a faster cognitive decline and faster rate of decline in memory, respectively. Moreover, increased serum MGO levels are associated with poorer memory, worst executive function, and lower gray matter volume [248], supporting the idea that glucotoxicity is involved in cerebral atrophy and cognitive dysfunction. Interestingly, experimental data suggest that glucotoxicity and MGO in particular can impinge on the dopaminergic system, too. First, in frozen human brain tissue, a neurotoxin called ADTIQ (1-acetyl-6,7-dihydroxy-1,2,3,4-tetrahydro-isoquinoline) was identified, deriving from the reaction of methylglyoxal with dopamine and particularly abundant in the putamen and caudate nucleus regions of Parkinson's patients [249]. Interestingly, ADTIQ has neurotoxic properties, and its levels are significantly increased in a cell model of hyperglycemia, diabetic rat brain [250], and transgenic mice expressing mutant forms of alpha-synuclein [251]. Interestingly, MGO treatment of N2A cells overexpressing α-syn induces in the cytoplasm the formation of α-syn aggregates positive for anti-CML antibody staining. Similarly, in mice, unilateral stereotaxic administration of MGO into the substantia nigra leads to the formation of α-syn aggregates accompanied by a significant reduction in protein levels of TH and of DJ-1, a protein with deglycase activity and that works as a sensor of oxidative stress [252]. Very recently, de Almeida and coworkers, performing the tail suspension test and Y maze spontaneous alternation test, discovered that MGO treatment induces depression-like behavior and impairs working memory in mice, inducing in parallel a significant reduction of dopamine and serotonin levels in the cerebral cortex [253]. Similar results were obtained by Szczepanik JC et al., who showed impairment, anxiolytic, and depressive-like behavior in Swiss mice memory. A daily administration of MGO for 11 days decreased dopamine levels and the Glo1 amount in the prefrontal cortex [254]. Finally, at the molecular level, it is known that MGO can regulate dopamine levels and the expression of dopaminergic genes, such as TH and DAT, in SH-SY5Y cells [255].

Conclusions and Perspectives
To date, cognitive decline undoubtedly represents a new emerging long-term complication of DM, leading to a lack of diabetes self-management and poor glycemic control. DM-associated cognitive deficit has a wide deleterious effect on life quality and significant consequences on the public healthcare system. However, its pathogenesis still remains obscure. In parallel, the relevance of the dopaminergic system for cognitive function has recently emerged, too. Hyperglycemia is known to be responsible for tissue damage leading to classical diabetic complications and impinges on dopaminergic neurons' homeostasis, too. Indeed, hyperglycemia induces greater glucose permeation in the brain [256] and an increase of intracellular and extracellular glucose concentrations in the midbrain and striatum [212], leading to neuronal glucotoxicity through different mechanisms, such as mitochondrial dysfunction, oxidative stress, polyol pathway, hexosamine pathway, and accumulation of the glycating agent MGO, a precursor of AGEs [257]. In particular, recent evidence strongly suggests a key role for MGO in both DM-associated cognitive decline and dopaminergic dysfunction. Indeed, MGO impairs dopaminergic neurons' survival and regulates dopamine levels in animal models. However, MGO's deleterious effects on cognition and dopaminergic function probably involve several molecular mechanisms that still remain unexplored. For instance, the hypothesis that MGO could modify dopaminergic genes' expression has not been deeply explored yet. It is noteworthy that epigenetic changes could potentially be relevant for MGO's deleterious effect on cognitive and dopaminergic functions. However, this point has never been investigated. The elucidation of molecular mechanisms underlying MGO's action is crucial for both the early identification of DM patients at risk of cognitive decline and the development of innovative therapeutic strategies for its treatment. Indeed, epigenetic modifications induced by MGO could represent novel potential biomarkers of the risk to develop dopaminergic dysfunction and cognitive impairment. Interestingly, epigenetic alterations can be induced or reverted by environmental factors, such as dietary factors. This feature provides an intriguing possibility to manipulate epigenetic mechanisms via dietary nutrients, thus improving both cognitive and dopaminergic dysfunction associated with DM.
Author Contributions: Conceptualization, F.C.P. and A.D.; study selection P.M., C.N., G.P. and L.U.; writing-original draft preparation, F.C.P. and A.D.; writing-review and editing, C.M., R.N. and F.F.; F.B. and P.F. critically revised the manuscript, supervision, R.N. and F.F.; F.C.P. and A.D. equally contributed to this study as first authors. R.N. and F.F. equally contributed to this study as last authors. All authors have read and agreed to the published version of the manuscript.