Insulin resistance in brain alters dopamine turnover and causes behavioral disorders

Significance

Both types 1 and 2 diabetes are associated with increased risks of age-related decay in cognitive function and mood disorders, especially depression. Insulin action has been shown to regulate neuronal signaling and plasticity. Here we investigate whether brain-specific knockout of insulin receptor (NIRKO) in mice causes behavioral changes and how these are mechanistically linked. We find that NIRKO mice exhibit age-related anxiety and depressive-like behavior. This is due to altered mitochondrial function, aberrant monoamine oxidase (MAO) expression, and increased dopamine turnover in the mesolimbic system, and can be reversed by treatment with Mao inhibitors. Thus, brain insulin resistance alters dopamine turnover and induces anxiety and depressive-like behaviors. These findings demonstrate a potential molecular link between central insulin resistance and behavioral disorders.

Abstract

Diabetes and insulin resistance are associated with altered brain imaging, depression, and increased rates of age-related cognitive impairment. Here we demonstrate that mice with a brain-specific knockout of the insulin receptor (NIRKO mice) exhibit brain mitochondrial dysfunction with reduced mitochondrial oxidative activity, increased levels of reactive oxygen species, and increased levels of lipid and protein oxidation in the striatum and nucleus accumbens. NIRKO mice also exhibit increased levels of monoamine oxidase A and B (MAO A and B) leading to increased dopamine turnover in these areas. Studies in cultured neurons and glia cells indicate that these changes in MAO A and B are a direct consequence of loss of insulin signaling. As a result, NIRKO mice develop age-related anxiety and depressive-like behaviors that can be reversed by treatment with MAO inhibitors, as well as the tricyclic antidepressant imipramine, which inhibits MAO activity and reduces oxidative stress. Thus, insulin resistance in brain induces mitochondrial and dopaminergic dysfunction leading to anxiety and depressive-like behaviors, demonstrating a potential molecular link between central insulin resistance and behavioral disorders.

As life expectancy in humans has increased, we are faced with a worldwide epidemic of age-related diseases such as type 2 diabetes (T2D) and Alzheimer’s disease (1). These parallel epidemics may not be coincidental. Indeed, studies have demonstrated an association between diabetes and a variety of brain alterations including depression, age-related cognitive decline, Alzheimer’s disease, and Parkinson’s disease (2, 3). In addition, individuals with both type 1 and type 2 diabetes have been shown to have a variety of abnormalities in brain imaging, including altered brain activity and connectivity by functional MRI (4, 5), altered microstructure by diffusion tensor imaging (6, 7), and altered neuronal circuitry in the striatum (8). Conversely, patients with Alzheimer’s disease show signs of central insulin resistance with increased insulin receptor substrate (IRS) 1 serine phosphorylation in the brain and decreased insulin concentrations in the cerebrospinal fluid (9, 10). Furthermore, pilot clinical trials of intranasal insulin administered to individuals with Alzheimer’s disease suggest decreased rates of cognitive decline (11).

These observations in humans have been mechanistically supported by studies in rodents and cultured cells, which have shown that insulin receptor signaling in brain has an important role in central regulation of metabolism and may also be crucial for proper brain function (12–14). We have previously demonstrated that mice with insulin resistance in brain due to targeted deletion of the insulin receptor (NIRKO mice) develop hyperphagia, mild obesity, reduced fertility, and decreased counterregulatory response to hypoglycemia (15, 16). NIRKO mice also display glycogen synthase kinase 3 beta (GSK3-beta) activation, resulting in hyperphosphorylation of tau protein, a hallmark of early Alzheimer’s disease (17). Other animal studies have demonstrated that insulin has direct effects on the hypothalamic food reward system, synaptic plasticity, signal transmission, and neuroprotective functions (18–22). On the other hand, knockout of IRS-2 has been shown to have protective effects on brain pathology in a mouse model of Huntington’s disease (23).

In addition to the association between diabetes and accelerated cognitive decline, there is growing support for a link between diabetes and mood disorders, especially depression (24–26). The mechanisms by which T2D influences depression are not known, but insulin resistance states are associated with increased inflammation and cytokine production in some brain regions (27, 28). Furthermore, ablation of insulin receptor in catecholaminergic neurons attenuates insulin-induced excitability in dopaminergic neurons (29), whereas insulin administration into the central nervous system (CNS) of rats has been shown to increase dopamine transporter protein expression (30). The latter may be important, because alterations in the activity of dopamine and/or serotonin systems have been linked to depression (31–33).

Mechanistically, one potential link between insulin action and changes in brain function might be alterations in mitochondrial function. Insulin resistance and type 2 diabetes are well documented to be associated with mitochondrial dysfunction in classical metabolic tissues, such as muscle (34) and liver (35), and we have recently demonstrated that obesity-induced insulin resistance is also associated with altered mitochondrial function in the hypothalamus (36). Patients with major depression have also been shown to exhibit mitochondrial dysfunction in the brain (37). Although it is still unknown how mitochondrial dysfunction might be linked to depression, it is worth noting that the two enzymes degrading monoamine neurotransmitters, monoamine oxidase (MAO) A and B, whose dysregulation has been linked to depressive behaviors, reside in the outer mitochondrial membrane (38–40). These data suggest that the relationship between mood disorders and diabetes could be the result of altered insulin regulation of mitochondrial function and monoamine homeostasis.

In the present study, we demonstrate that insulin receptor deficiency in the brain results in brain mitochondrial dysfunction. Whereas initially this is not accompanied by behavioral changes, with aging, NIRKO mice exhibit signs of anxiety and depressive-like behaviors. These changes are secondary to decreased dopamine signaling in the striatum and nucleus accumbens, which in turn is a consequence of increased levels of MAO A and B, leading to increased dopamine turnover. In vitro data indicate that this is due to a loss of insulin effect on expression of MAO A and MAO B in neuronal and glial cells, and is further supported by the finding that these depressive behaviors are reversed by treatment with the MAO inhibitors. Thus, central insulin resistance causes altered dopamine turnover and age-related behavioral changes, creating a direct link between mood disorders and insulin-resistant states like type 2 diabetes.

Results

Age-Dependent Anxiety and Depressive-Like Behavior in NIRKO Mice.

NIRKO mice were created by breeding IR^lox/lox mice and Nestin-Cre transgenic mice as previously described (15). Quantitative PCR (qPCR) analysis of 4-mo-old mice revealed a 95% reduction of Ir mRNA throughout the brain, including in isolated hypothalamus (HTM), hippocampus (HCA), prefrontal cortex (PFC), striatum, nucleus accumbens (NAC), and ventral tegmental area (VTA), which was paralleled by decreases in insulin receptor protein (Fig. S1A). Although NIRKO mice show mild obesity and metabolic syndrome early in life (15), by 10 mo of age, NIRKO mice exhibited no significant differences in body weight, blood glucose levels, or food intake compared with controls (Fig. S1 B–F), and this continued until at least 24 mo of age.

To determine whether insulin receptor deficiency in the brain might lead to alterations in behavior, we performed a panel of behavioral tests on 10- and 17-mo-old NIRKO mice, representing middle age and older animals. Consistent with previous studies examining brain function in 4–6 mo old mice (17), the tail suspension test, which assesses depressive-like behavior as indicated by immobility time, revealed no difference in 10-mo NIRKO male and female mice compared with control. However, by 17 mo of age, female NIRKO mice exhibited a 33 ± 9% increase in immobility time in the tail suspension test and a 30 ± 5% increase in immobility time in the forced swimming test, both of which are considered signs of depressive behavior (both P < 0.05, Fig. 1 A and B). As with their metabolic abnormalities, male NIRKO mice showed a milder phenotype with a 35 ± 13% increase in immobility in the tail suspension test and a 16 ± 8% increase in immobility time during the forced swimming test, neither of which quite reached statistical significance (P = 0.06 and P = 0.15, Fig. S1 G and H).

Fig. 1.

Age-dependent anxiety and depressive-like behavior in NIRKO mice. (A and B) Assessment of depressive-like behavior. (A) Immobility time was assessed using the tail suspension test for 10- and 17-mo-old female control (n = 7) and NIRKO mice (n = 9). (B) Immobility time during a forced swimming test in 17-mo-old female control (n = 6) and NIRKO mice (n = 6). Values in this panel and all subsequent figures are mean ± SEM; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001, Student’s t test, unless otherwise stated. (C and D) Assessment of anxiety via light/dark box. (C) Time spent in dark compartment and (D) number of transitions during light/dark box test for 17-mo-old female control (n = 11) and NIRKO mice (n = 12). (E) Assessment of conflict-based anxiety as time to enter the center during novelty suppressed feeding test for 17-mo-old female control (n = 6) and NIRKO mice (n = 6). (F and G) Assessment of exploratory drive. (F) Number of entries and (G) latency to enter the center during open field test for 17-mo-old female control (n = 10) and NIRKO mice (n = 14).

Older NIRKO mice also exhibited signs of increased anxiety and stress response. Using a dark/light box, we found that 17-mo-old female NIRKO mice had increased anxiety as evidenced by staying in the dark compartment 19 ± 6% longer and exhibiting 40 ± 6% fewer transitions between compartments than control mice (Fig. 1 C and D). Both male and female NIRKO mice (17 mo old) also exhibited increased serum corticosterone levels after restraint stress compared with control, and in females corticosterone was significantly increased even in the basal state (Fig. S1 I and J).

To further analyze behavioral changes in NIRKO mice, we assessed anxiety using the novelty-suppressed feeding test. In this test, mice are fasted for 16 h and then placed in a large animal box containing a white disk with food in its center. They are therefore challenged to choose between going for the food and avoiding the unusual white disk on which it rests. Delay in entering the center of the field is considered to have increased anxiety. Again, in this test, 17-mo-old, but not 10-mo-old, female and male NIRKO mice exhibited increased anxiety levels as demonstrated by a twofold increase in the time to enter the disk containing food (Fig. 1E and Fig. S2 A–C). Likewise, an open field test, which assesses changes in exploratory drive, revealed that 17-mo-old NIRKO mice had 45 ± 5% fewer entries to the center and a 3.6-fold increase in latency in entering the center compared with age-matched control littermates (all P < 0.05) (Fig. 1 F and G and Fig. S2 D and E). Age- and sex-matched mice carrying only the Nestin-Cre transgene had no alterations in any of these parameters (Fig. S3 A–D). Thus, using multiple tests, 17-mo-old NIRKO mice of both genders showed clear signs of increased anxiety and depressive-like behavior.

NIRKO Mice Display Central Mitochondrial Dysfunction.

It has been shown that insulin resistance in the brain (36), like insulin resistance in skeletal muscle (34), is associated with mitochondrial dysfunction, and mitochondrial dysfunction has been associated with neurological disorders (41). Examination of isolated mitochondria from brains of control and NIRKO mice using a Seahorse XF24 Extracellular Flux Analyzer revealed a 28 ± 9% reduction in basal oxygen consumption rate (OCR) compared with control (P < 0.05), even by 4 mo of age. This defect was further exacerbated in 24-mo-old NIRKO mice, with a 38 ± 6% reduction compared with control (P < 0.05) (Fig. 2 A and B and Fig. S4A). To investigate whether reduced mitochondrial function was a result of decreased mitochondrial biogenesis, we analyzed key transcriptional regulators of mitochondrial biogenesis in dorsal striatum and NAC, regions implicated in anxiety and depressive disorders. This analysis revealed overall unaltered expression of Pgc1a (peroxisome proliferator-activated receptor gamma coactivator 1), Pgc1b, and Tfam in young and old NIRKO mice, with the exception of temporarily increased mRNA levels of Pgc1b in striatum of NIRKO mice at 4 mo of age (Fig. S4 B–G). Western blot analysis of striatal samples from 4-mo-old control and NIRKO mice also revealed reduced levels of complex I subunit NDUFB8, complex II subunit SDHB, complex III subunit UQCRC2, complex IV subunit MTCQ1, and complex V ATP synthase subunit alpha (Fig. 2C and Fig. S4H). Analysis of mitochondrial morphology in the striatum using electron microscopy revealed a 53 ± 8% reduction in average mitochondrial cross-sectional area and a 32 ± 4% increase in mitochondrial number (Fig. 2 D and E), with no change in the level of expression of enzymes involved in mitochondrial fusion and fission (Opa1, Mfn1, Mfn2, Drp1, and Mtfr1) or phosphorylation of Drp1 at serine 616 and 637 (Fig. S5 A–H). Thus, NIRKO mice exhibit altered brain mitochondrial morphology and mitochondrial dysfunction with reduced levels of proteins of the electron transport chain and reduced basal respiration early in life before any behavioral changes can be detected.

Fig. 2.

NIRKO mice display mitochondrial dysfunction. Mitochondrial activity was measured using isolated brain mitochondria in a Seahorse Flux Analyzer as described in Materials and Methods. Basal respiration measurements, displayed as area under the curve, of isolated brain mitochondria from (A) 4-mo-old and (B) 24-mo-old control and NIRKO mice using the XF24 Extracellular Flux Analyzer from Seahorse Bioscience. A total of six control and six NIRKO mice were used. (C) Determination of striatal protein levels of electron transport chain complexes. Representative Western blots are shown of nuclear- and mitochondrial-encoded genes (IR beta, CV-ATP5A, CIII-UQCRC2, CIV-MTCQ1, CII-SDHB, and CI-NDUFB8) using striatal samples from 4-mo-old female control (n = 6) and NIRKO mice (n = 6). β-Actin served as a loading control. (D) Transmission electron microscopy of brain striatum showing mitochondria of 4-mo-old female control and NIRKO mice. This is a representative example of 12 fields in four animals of each genotype. Note the decreased mitochondrial diameter. Original magnification 1:10,000. (E) Morphometric analysis of electron microscopy using ImageJ. Average mitochondrial area and total mitochondrial number of control and NIRKO mice (n = 4 mice of each genotype with 100–200 mitochondria analyzed).

NIRKO Mice Suffer from Oxidative Stress.

Mitochondrial dysfunction is often associated with oxidative stress. Consistent with this association, there was a 1.4-fold increase in lipid peroxidation in brains of NIRKO mice as determined by thiobarbituric acid reactive substances (TBARS) as early as 4 mo of age, and this increase persisted at 24 mo of age (Fig. 3 A and B). Protein oxidation, as assessed by derivatization of protein carbonyl groups with 2,4-dinitrophenylhydrazine (DNPH), was also increased in NIRKO brains (Fig. 3C). Interestingly, although mitochondrial dysfunction is often associated with an inflammatory response and increased cellular apoptosis, the mild oxidative stress in NIRKO mice did not induce inflammation or apoptosis by multiple measures (Fig. S6 A–D).

Fig. 3.

NIRKO mice suffer from oxidative stress. (A and B) Lipid peroxidation, as a measure of oxidative stress, was measured using thiobarbituric acid reactive substances (TBARS). Quantification of lipid peroxidation using TBARS assay of (A) 4-mo-old and (B) 24-mo-old control (n = 6) and NIRKO mitochondria (n = 6). (C) Protein carbonylation, as a measure of oxidative stress, was assessed by antibody staining against DNP. Representative immunohistochemical staining of protein carbonylation in the striatum and nucleus accumbens of NIRKO mice (n = 4) compared with control (n = 4).

NIRKO Mice Exhibit Decreased Dopamine Signaling.

The monoamine hypothesis of depression postulates that the deficiency of neurotransmitters, such as dopamine, serotonin, and catecholamines, in the brain is responsible for the manifestations of depression (42). Assessment of the levels of serotonin and catecholamines, as well as their degradation products, in the brains of 17-mo-old NIRKO mice revealed no differences from control (Fig. S7 A–D). Likewise, blood pressure and heart rate were unchanged in control and NIRKO mice, suggesting similar activation of the sympathetic nervous system (Fig. S7 E–G). There was a small increase in tryptophan hydroxylase 2, the rate-limiting enzyme in the synthesis of serotonin, in the raphe nucleus of 4-mo-old NIRKO mice, but this level returned to normal in older NIRKO animals and was not observed in the other brain regions analyzed (Fig. S7 H and I). Furthermore, there was no consistent alteration in the expression of the serotonin reuptake transporter Slc6a4 nor in serotonin receptor 4, both genes important for serotonin signaling and regulation of behavior (Fig. S7 J and K).

Dopamine signaling in the striatum and mesolimbic system has been shown to be particularly important in depression (32, 33). Because there was no change in the levels of mRNA for tyrosine hydroxylase, the rate-limiting enzyme for dopamine synthesis (see Fig. S9 A and B), we further explored this pathway by assessment of electrically evoked dopamine release using carbon fiber amperometry. This assessment revealed no change in the peak amplitude of electrically evoked dopamine release in brains of 10- and 17-mo-old NIRKO mice in the dorsal striatum (Fig. S8 A–G). However, there was a 40 ± 9% decrease in the average width of the evoked dopamine signal and a 44 ± 10% decrease in t1/2 (duration of the signal at 50% of its peak amplitude) in the dorsal striatum as measured by the peak width at half height (both P < 0.05) (Fig. 4A and Fig. S8H). This resulted in a 39 ± 14% decrease in area under the peak in 17-mo-old NIRKO mice. Similarly, recordings obtained from NAC revealed a 53 ± 8% decrease in the average width of the evoked dopamine signal and a 53 ± 9% decrease in area under the peak (Fig. 4B), whereas recording in the prefrontal cortex, which is not a part of the mesolimbic system, did not reveal alterations in dopamine signaling (Fig. S8 I and J). Likewise, catecholamine release from the adrenal glands of NIRKO mice was unaltered compared with control (Fig. S9C). These data indicate that the increase in dopamine turnover in NIRKO mice is restricted to the dorsal striatum and NAC and not part of a general defect in neurosecretory granule release or turnover.

Fig. 4.

NIRKO mice exhibit decreased dopamine signaling. (A and B) In vivo determination of dopamine release in the (A) striatum and (B) nucleus accumbens. (Upper) Average peak width and (Lower) average area of electrically evoked dopamine release of striatal slices using carbon fiber amperometry of 17-mo-old female control (slices = 11, recordings = 45, mice = 7) and NIRKO mice (slices = 11, recordings = 43, mice = 6). (C–E) Representative Western blots and densitometric analysis of MAO A and B protein levels in the striatum of 4-mo-old control (n = 5) and NIRKO mice (n = 7). (F) Gene expression analysis of Maoa and Maob in hypothalamic GT1–7 cells under basal conditions (n = 4) and following stimulation with 100 nM insulin (n = 4) for 24 h. (G and H) Gene expression analysis of (G) Maoa and (H) Maob in primary glial cells under basal conditions (n = 4) and following stimulation with 100 nM insulin (n = 4) for 24 h.

The reduced dopamine signaling in the dorsal striatum and NAC could be the result of increased reuptake by the dopamine reuptake transporter (DAT) or increased degradation by mitochondrial Mao A and B. Whereas the levels of Dat mRNA in striatal and NAC samples from NIRKO mice were unaltered (Fig. S9 D and E), mRNA levels of Maoa (P < 0.05) and Maob (did not quite reach statistical significance, P = 0.09) were increased in striatal and MAO A levels in NAC samples (Fig. S9 F and G) by 40–50%. This finding was confirmed by Western blot analysis, which showed a twofold increase of MAO A protein (P < 0.01) and a 1.3-fold increase in MAO B protein (P = 0.05) in the striatum of NIRKO mice by 4 mo of age (Fig. 4 C–E).

To test whether increased MAO A and B levels were a direct effect of reduced insulin signaling in NIRKO mice, we assessed the effect of insulin on neurons in vitro. Stimulation of the hypothalamic neuronal GT1–7 cell line with 100 nM insulin for 24 h caused 23 ± 2% and 62 ± 2% reductions in Maoa and Maob mRNA levels and 46 ± 8% and 73 ± 2% reductions of MAO A and B protein, respectively (all P < 0.05, Fig. 4F and Fig. S9H). A similar effect of insulin on Maoa and Maob expression was observed in primary neurons (Fig. S9I). Glial cells can also participate in dopamine degradation, and insulin stimulation of primary glial cells in vitro resulted in 52 ± 4% reduction in Maoa mRNA levels, but did not affect expression levels of Maob (Fig. 4 G and H). Thus, the increased levels of MAO A in brain of NIRKO mice appear to be the direct consequence of the loss of insulin receptor signaling in both neurons and glia, whereas the increase in MAO B is due primarily to the loss of insulin signaling in neurons.

Depressive-Like Behavior in NIRKO Mice Is Reversible with Antidepressant Treatment.

To further explore the relationship between the depressive-like behavior, mitochondrial dysfunction, and altered dopamine signaling in NIRKO mice, we subjected these mice to treatment with the antidepressant imipramine, which has been shown to target MAO A and B, as well as reduce oxidative stress secondary to mitochondrial dysfunction (43, 44). Similar to the experiment shown in Fig. 1A, saline-treated NIRKO mice exhibited a 31 ± 13% increase in immobility time during a tail suspension test compared with saline-treated controls, consistent with depressive-like behavior (Fig. 5A). Following imipramine treatment, there was a modest, nonsignificant reduction in immobility in controls but a significant 52 ± 9% reduction in immobility time in NIRKO mice (P < 0.003), such that the imipramine-treated NIRKO mice were indistinguishable from control (Fig. 5A). A similar effect was observed with the irreversible MAO A and B inhibitor phenelzine (Fig. 5B). Thus, the age-dependent depressive-like behavior in NIRKO mice was reversible with antidepressant treatments, which target dopamine turnover and mitochondrial dysfunction.

Fig. 5.

Depressive-like behavior in NIRKO mice is reversible with drug treatment. (A) Immobility time during tail suspension test of 24-mo-old female control and NIRKO mice treated with saline or imipramine (16 mg/kg) 1 h prior to the assessment (each n = 6). (B) Immobility time during a tail suspension test of 24-mo-old female control and NIRKO mice treated with saline or phenelzine (20 mg/kg) for 1 h prior to the assessment (n = 9–11). Significance was determined by a one-way ANOVA followed by Fisher post hoc analysis. (C) Model of insulin signaling effects on mood and behavior in healthy mice (Left) and mice with brain insulin resistance (Right).

Discussion

Both type 1 and type 2 diabetes are associated with a variety of CNS complications, including increased rates of cognitive decline, altered brain imaging, increased risk of neurodegenerative disease, and increased rates of depression; however, how alterations in insulin signaling might contribute to these complications is still poorly understood (5, 45–47). Using mice with a brain-specific knockout of insulin receptor, we show that depressive-like behavior and anxiety can be a direct consequence of insulin resistance in the brain. These alterations occur in a progressive, staged process. Thus, young NIRKO mice exhibit brain mitochondrial dysfunction and oxidative stress, especially in the dorsal striatum and NAC. NIRKO mice also exhibit increased levels of the dopamine-degrading enzymes Mao A and B, and these increased levels result in increased dopamine clearance and decreased dopamine signaling. Furthermore, the in vivo and in vitro experiments presented here show that altered MAO A and B expression is a direct result of a loss of insulin signaling in neurons and glia. As NIRKO mice age, these alterations in CNS metabolism and dopaminergic signaling result in multiple signs of anxiety and depressive-like behavior, which can be reversed by inhibiting MAO activity (schematized in Fig. 5C). Thus, a decrease in insulin action in the brain can directly alter dopamine turnover and lead to depressive behaviors via effects at the mitochondrial level.

The loss of insulin receptor signaling in the brain affects mitochondrial function in at least three ways: (i) decreased mitochondrial activity due to decreased expression of electron transport chain proteins; (ii) increased monoamine oxidase levels due to a loss of insulin action to suppress MAO gene expression; and (iii) changes in the morphology of mitochondria in brain, with smaller and more numerous mitochondria. Whereas the first two ways appear to be direct effects of a loss of insulin action, the exact mechanism for the third is unclear. Diet-induced obesity can reduce mRNA expression of mitofusin 2, a key enzyme in mitochondrial fusion, in proopiomelanocortin neurons in the hypothalamus, suggesting that central insulin resistance can affect mitochondrial dynamics (48). However, we did not detect any differences in expression levels of regulators of mitochondrial fission/fusion in brains of NIRKO mice. Serine phosphorylation of Drp1, a regulator of mitochondrial fission, alters its activity and thereby regulates mitochondrial fission. However, NIRKO mice exhibit no changes in phosphorylation of Drp1 at serine residue 616, which stimulates mitochondrial fission, or serine residue 637, which inhibits fission. Recent work has shown that Drp1 can also be phosphorylated at Ser-693 by GSK3-beta (49), and GSK3-beta is a known downstream target of insulin receptor signaling, such that insulin induced phosphorylation of the enzyme reduces its activity. We have previously shown that NIRKO mice display increased GSK3-beta activity (17), but whether this affects Ser-693 phosphorylation of Drp1 and mitochondrial dynamics is unclear, because no antibodies specific to this site exist.

Mechanistically, with regard to depression and anxiety, it is important to note that NIRKO mice show increased rates of dopamine clearance. This is secondary to increases in MAO A and B in the striatum and NAC of NIRKO brains. The MAO enzymes reside in the outer mitochondrial membrane and have also been shown to be increased in human depression (39). Dopamine is predominantly degraded by MAO A in rodents and by MAO B in humans (50, 51). Based on in vitro studies, it appears that insulin can directly down-regulate both MAO A and B in neurons and MAO A in glial cells. Thus, reduced insulin action in brain results in up-regulation of MAO A and B, increased dopamine clearance, and decreased dopamine signaling. Patients with depression also exhibit decreased half-life of dopamine in the striatum (52), similar to our findings in NIRKO mice. It is unclear why MAO dysregulation was detected in the striatum and NAC of NIRKO mice, and not in the prefrontal cortex. This may suggest that discrete compartments in the brain differ in their response to insulin (53), or that different cellular compositions in discrete brain areas result in different effects of insulin resistance (54). MAO A and B also degrade serotonin, thus affecting serotonin signaling. As altered serotonin signaling is also associated with depression, this pathway may also contribute to the pathogenesis of mood disorders (55). We have not observed any general differences of serotonin metabolite concentrations in the brain or changes in the rate-limiting enzymes for its synthesis in various brain regions. However, we cannot rule out that specifically altered serotonin signaling in distinct brain compartments may also play a role in this scenario. Further research will be needed to address these questions.

It is interesting to note that despite the alterations in brain mitochondrial function in the NIRKO mouse, the resulting mitochondrial stress does not induce a proinflammatory response or measurable signs of apoptosis, even in a prooxidative dopaminergic system (56). This finding is consistent with previous studies indicating that ablation of the insulin receptor in dopaminergic neurons does not itself result in reduced number of neurons (29) and suggests that insulin signaling in the brain acts primarily to fine tune brain activity, rather than as a survival factor for neurons. This is different from the situation in obese/diabetic mice in which obesity, hyperglycemia, and systemic insulin resistance are associated with hypothalamic inflammatory changes (57, 58). Because old NIRKO mice are neither obese nor hyperglycemic, and have no brain inflammation, but still develop mood disorders, it seems clear that brain insulin resistance, without these other abnormalities can lead to behavioral phenotypes. Importantly, this depressive phenotype can be reversed by treatment with antidepressive drugs. Previous studies have shown that treating obese, diabetic mice with the insulin sensitizer rosiglitazone, which has been shown to reduce reactive oxygen species formation in the brain (59), can also decrease depressive-like behavior (60). Whether this decrease is through increasing insulin sensitivity in the brain or systemically remains to be determined.

Our data demonstrate that insulin resistance in the brain can lead to alterations in mitochondrial function, increased levels of monoamine oxidases, and increased dopamine clearance (Fig. 5C). This is consistent with other studies suggesting that mitochondrial dysfunction can contribute to depression (61). Of course, these alterations may also interact with other factors, which can contribute to depression and anxiety, including genetic susceptibility, alterations in circadian rhythms, or changes in content of monoamines such as serotonin, norepinephrine, and dopamine (62). Thus, the increased incidence of depression in patients with diabetes can be a consequence of central insulin resistance. Improving central insulin signaling could lead to new therapeutic approaches for treatment of mood disorders, especially in patients with diabetes or metabolic syndrome.

Materials and Methods

Animal Care.

NIRKO mice were generated as previously described (15) and were backcrossed into a C57BL/6J background for at least nine generations before the study. All mice were housed in a mouse facility on a 12-h light/dark cycle in a temperature-controlled room and maintained on a standard chow diet containing 22% calories from fat, 23% from protein, and 55% from carbohydrates (Mouse Diet 9F 5020; PharmaServ) beginning at ∼4 wk of age. Mice were allowed ad libitum access to water and food. Animal care and study protocols were approved by the Institutional Animal Care and Use Committee of Brandeis/Joslin Diabetes Center and were in accordance with the National Institutes of Health guidelines.

Behavioral Tests.

To assess depressive-like behavior, we performed the mouse tail suspension test and forced swimming test on naïve 10-, 17-, and 24-mo-old mice. Importantly, depressive-like behavior testing was performed only once in a lifetime per mouse to avoid any learning effect in mice. To assess anxiety, novelty suppressed feeding test, dark/light box test, and stress by restraint test were performed. Open field test was performed to evaluate exploratory drive. Results of the open field test were quantitated using ANY Maze video tracking software (Stoelting).

Slice Electrophysiology.

Coronal slices from fresh mouse brains were used for slice electrophysiology and carbon fiber amperometry as previously described (63). Amperometric peaks were identified as events greater than 3.5 times the rms noise of the baseline. The event width was the duration between (i) the baseline intercept of the maximal incline from the baseline to the first point that exceeded the cut off and (ii) the first data point following the maximal amplitude that registered a value of ≤0 pA. The maximum amplitude (imax) of the event was the highest value within the event.

High-Performance Liquid Chromatography with Electrochemical Detection.

From each striatal sample, 20 μL of a 1:15 dilution with nanopure H₂0 were injected into an amperometric Antec Intro High-Performance Liquid Chromatography with Electrochemical Detection (HPLC-EC) system (GBC) with a 10-cm Rainin column and a phosphate mobile phase buffer, which allows the separation and detection of serotonin, norepinephrine, and the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA).

Statistical Methods.

Data sets were analyzed for statistical significance using a two-tailed unpaired Student’s t test or one- or two-way ANOVA, as appropriate. P values <0.05 were considered statistically significant.

Western blot and qPCR analysis, EM, lipid peroxidation, and mitochondrial respiration assays, analytical procedures, immunohistochemistry, and cell culture procedures are detailed in SI Materials and Methods.