Background
As the pace of life to speed up, coupled with unhealthy diet habits and lifestyles, people are suffering from more complex and intense stress in both physical and psychological aspects. Long-term exposure to these hazards has been considered as a major cause of many chronic diseases [
1]. It is well known that depression and diabetes, especially type 2 diabetes (T2D), were among the leading causes of global disability-adjusted life-years (DALYs) [
2]. According to reports from World Health Organization, more than 300 million human beings all over the world suffer from depression (refer to URL:
http://www.who.int/mediacentre/factsheets/fs369/en/), while approximately 422 million people worldwide have diabetes (refer to URL:
http://www.who.int/mediacentre/factsheets/fs312/en/). In brief, depression is a devastating mental disorder characterized by anhedonia, sadness, feelings of guilt or fatigue, disrupted sleep or loss of appetite, and poor concentration. At worst, it would lead to suicides. Apart from that, diabetes is usually diagnosed by elevated plasma glucose levels, whereas abnormalities that present before glucose changes may sometimes be undetectable. More individuals are in high-risk diabetic states—prediabetes, which is associated with the co-occurrence of insulin resistance (IR) and β-cell dysfunction [
3].
Surprisingly, it was discovered that depression, T2D, and even metabolic syndrome (MetS) are often comorbid. T2D was associated with a higher susceptibility to depression [
4], while depression increased T2D risk by 60% [
5‐
7]. Moreover, this bidirectional association cannot be attributed to the usage of antidepressants [
8]. Recently, abundant evidence hinted that depression and T2D might share biological origins. Researchers postulated that the underlying mechanism could be the chronic cytokine-mediated inflammatory response which was induced by dysregulation of hypothalamic-pituitary-adrenal (HPA) axis and overactivation of innate immunity [
9‐
11].
In fact, this hypothesis was based on various clinical and preclinical evidence. Among them, we found out some inspirational clues. A multitude of researches indicated that activation of NLRP3 (nucleotide-binding domain, leucine-rich-containing family, pyrindomain-containing-3) inflammasome and subsequent production of Interleukin-1β (IL-1β) contribute to various metabolic abnormalities, particularly IR-related T2D and MetS. It was demonstrated that ablation of genes of inflammasome components and IL-1β could protected mice from fatty acid-induced IR [
12‐
14]. Treating rodent diabetic models with anti-IL-1β therapy resulted in ameliorated insulin insensitivity as well as improved β-cell survival [
15]. Furthermore, Lee et al. [
16] discovered overactivation of the NLRP3 inflammasome in type 2 diabetic inpatients. Prior to that, Larsen [
17] and colleagues found that anakinra, a blockade of IL-1, improved glycemia and β-cell secretory function in patients with T2D. Hence, NLRP3 inflammasome was considered a bond between chronic inflammation and metabolism.
In addition, it seems that NLRP3 inflammasome and IL-1β also play pivotal roles in depression [
18]. According to Goshen et al. [
19], cerebral IL-1 mediated stress-induced depressive-like behavior in mice via suppressing hippocampal neurogenesis and activating HPA axis. Updated studies on this issue, including that from our laboratory, proved that depressive-like behaviors induced by CUMS required a functional NLRP3 inflammasome [
20‐
22]. Moreover, clinical evidence showed that NLRP3 inflammasomes in mononuclear blood cells of depressive patients were activated [
23]. Besides, depression was accompanied by an increase in pro-inflammatory cytokine levels including IL-1, while associated with the alleviation of immune overactivity when symptoms relived [
24]. Therefore, scholars across the world postulated that therapies targeting the NLRP3 inflammasome might be a novel strategy for depression [
25,
26].
Collectively, NLRP3 inflammasome activation and subsequent IL-1β generation were involved in the development of depression and T2D separately. We tentatively presume that this common pathogenesis might be the “shared biological origin.” That is to say, chronic hyperactivation of HPA axis and consequent overreaction of innate immunity were blamed for the comorbidity [
11,
27]. For instance, glucocorticoids (GC) could supervise immune system and prime the NLRP3 inflammasome [
28,
29], as well as induce apoptosis in β-cells via thioredoxin-interacting protein (TXNIP) [
30] and instigate IR through pathways including prompting immune reactions [
1,
31]. Additionally, it was reported that anti-diabetic drug glyburide was an effective inhibitor of NLRP3 inflammasome [
32,
33], while fluoxetine could also improve hyperglycemia [
34]. Herein, we designed this study to investigate the comorbidity of depression and T2D/IR, as well as unravel possible molecular mechanisms.
Methods
Animals and drugs
Male 6-week-old C57BL/6 mice were introduced from Experimental Animal Center of Second Military Medical University (Shanghai, China). All the animals were bred in a standardized animal room (temperature 22 ± 2 °C, lights on during 7 a.m.–7 p.m.), with free access to clean tap water and rodent chow. For acclimation, mice were treated with 1% sucrose solution (weight/volume) for 14 days according to our previous study [
21]. Thereafter, the rodents were randomly allocated to 5 independent groups, namely Control, CUMS, CUMS + Vehicle (Veh), CUMS + Glyburide (Glb) and CUMS + Fluoxetine (Flx) (
n = 8). Drugs used within the procedure were dissolved in 2% DMSO and 2% Tween-80 (Sigma-Aldrich, St. Louis, MO, USA) normal saline, then injected at the dose of 10 mg/kg/day intraperitoneally (i.p.) according to former studies [
32,
35‐
38]. Before each injection, drug solutions will be mixed thoroughly and heated to 37 °C. Glyburide was purchased from Sigma-Aldrich (#G0639, St. Louis, MO, USA) while fluoxetine hydrochloride was bought from MedChem Express (#HY-B0102A, Shanghai, China).
Chronic unpredictable mild stress protocol
The chronic unpredictable mild stress (CUMS) paradigm was adapted from our past research [
21,
39]. Briefly, it lasted for 12 weeks and contained diverse randomly assigned stressors: swimming at 4 °C for 5 min; 45 °C dry-heat stress for 10 min; cage vibration for 30 min; constraint for 2 h; 45° cage inclination for 12 h; damp bedding for 16 h; continuous illumination for 24 h; food or water deprivation for 24 h. During the stress procedure, only one stressor was performed per day, and no single stressor was conducted consecutively for 2 days. Animals that were subjected to CUMS were kept separately. Drug administrations were maintained throughout the whole CUMS procedure. Baseline data including specific behavioral and glucose metabolic parameters were acquired before and after the protocol.
Behavioral tests
The behavioral tests were all carried out in dark phase (18:00–22:00 p.m.). To eliminate olfactory interference, the trial chambers were wiped with 75% ethanol between two separated test sessions in tail suspension tests and open field tests, respectively.
Sucrose preference test
Sucrose preference test (SPT) is a widely used method of evaluating depressive-like behavior in animals. Generally, it represents anhedonia, one of the major symptoms of depression in human beings. According to literature, food and water were deprived 20 h before SPT [
21]. During the test, each animal would be provided with two bottles of the same size: one containing clean tap water and the other containing 1% sucrose solution. Fluid consumptions were checked after the 1-h test. Sucrose preference proportion = sucrose solution consumption/(sucrose solution consumption + tap water consumption) *100%.
Tail suspension test
Tail suspension test (TST) is another common behavioral test, which reflected the state of despair and helplessness. At the end of CUMS protocol, the mouse would be hung upside down on a hook by the tail in the PHM-300 tail suspension chamber (MED Associates Inc., St. Albans, VT, USA). After 1 min of adaptation to the apparatus, the immobility durations during the next 5 min’ test were recorded and analyzed with the TST software (lower threshold value = 0.25).
Open field test
Open field test (OFT) was used to measure spontaneous activity, aiming at identifying anxious and sickness behavior. It was generally believed that total traveling distance mainly reflected sickness behavior of animals, whereas time spent in central and periphery areas would indicate anxious status. During the tests, the mice were carefully placed in the middle of trial chambers, consisting of perforated acrylonitrile butadiene styrene copolymers (#RD1112-IOF, Mobile Datum Information Technology, Shanghai, China). The 5-min movement traces of the mice were recorded by infrared cameras and analyzed by OFT software.
Intraperitoneal glucose tolerance test
For the glucose tolerance tests (GTT), mice were fasted overnight (16 h) in clean homecages and then injected (i.p.) with 2 g/kg body weight of glucose solution. Blood samples for glucose measurements were collected gently from the tail veins and dropped immediately onto glucometer strips. MAJOR II Blood Glucose Monitoring System (Major Biosystem, Taiwan, China) were utilized for fast blood-glucose detection at 0, 30, 60, 90, and 120 min after glucose injection. The calculation of areas under curve (AUC) was based upon the polygonal lines joining glucose values for different time sections.
Intraperitoneal insulin tolerance test
For the insulin tolerance tests (ITT), mice were administrated (i.p.) with 1 IU/kg body weight of Novolin R (Novo Nordisk, Bagsvaerd, Denmark) after 6 h’ fasting. Same as GTT, blood samples were acquired from the tail veins and measured immediately. AUCs were calculated according to the connecting lines of blood glucose values at 0, 30, 60, 90, and 120 min after insulin injection.
Glucose-stimulated insulin secretion test in vivo
Mice were starved for 16 h and then injected (i.p.) with 1 mg/kg of glucose solution. Blood was collected softly from the tail veins using Microvette® CB 300 Lithium Heparin tubes (Sarstedt, Nümbrecht, Germany) at 0 and 30 min after glucose injection. The plasma samples were separated in a centrifuge at 2000
g for 5 min. The determination of plasma insulin would be conducted as described below. The homeostasis model of assessment for insulin resistance index (HOMA-IR) values were calculated according to a previous study [
40]: [fasting blood glucose (mg/dl)* fasting plasma insulin (μU/ml) /405]. Plasma insulin (μU/ml) was determined using titer of human insulin (26 U/mg), for that of mice was not defined uniformly.
Sacrifice and sample preparation
After completion of behavioral and metabolic tests, mice were sacrificed under general anesthesia with pentobarbital. Blood was collected, followed by coagulation at room temperature for 30 min and then centrifuged at 4000 rpm for 15 min. The supernatant serum was separated and preserved at −80 °C. Hippocampi were dissected and flash frozen in liquid nitrogen immediately after decapitation. Pancreases were cut into two parts along their long axes, one fixed with cold 4% paraformaldehyde and the other flash frozen and stored at −80 °C. Serum samples would be used for the detection of Corticosterone and Interleukin-1β, hippocampi and frozen pancreas were utilized in Western Blotting, whereas fixed pancreases were applied for immunostaining.
Enzyme-linked immunosorbent assays
Serum corticosterone and IL-1β levels were measured by Mouse Corticosterone ELISA kit (#F10246, Westang, Shanghai, China) and Mouse IL-1beta Platinum ELISA kit (#BMS6002, eBioscience, San Diego, CA, USA) according to the manufacturer’s instructions. Plasma insulin levels were determined by Ultra Sensitive Mouse Insulin ELISA Kit (#90080, Crystal Chem, Downers Grove, IL, USA) following the manufacturer’s protocol.
Western Blotting
The hippocampi and pancreases were homogenized in the ice-cold RIPA lysis buffer (#P0013B, Beyotime Biotechnology, Nantong, Jiangsu, China) with protease inhibitor cocktail (#B14001, Bimake, Shanghai, China) and PhosSTOP (#04906845001, Roche, Indianapolis, IN, USA) as previously described [
21]. The protein concentration was determined by Enhanced BCA Protein Assay Kit (#P0010, Beyotime Biotechnology, Nantong, Jiangsu, China). The centrifuged lysates were added with 5X loading buffer (#P0015, Beyotime Biotechnology, Nantong, Jiangsu, China) at 4:1 volume ratio, vortexed and then boiled for 10 min. Equal amounts of protein samples were loaded and separated on 10% or 12% SDS-PAGE gels (#P0012A, Beyotime Biotechnology, Nantong, Jiangsu, China). After electrophoresis, proteins were transferred onto Immobilon-P PVDF membranes (Millipore, Billerica, MA, USA) using a Bio-Rad (Hercules, CA, USA) wet transfer system. Afterwards, the blots were blocked with 0.1% Tween-20 solution (TBS-T) containing 3% bovine serum albumin (BSA) and incubated with appropriate primary antibodies and then specific IRDye conjugated secondary antibodies. Finally, the membranes were scanned, and the intensities of protein bands were quantified using Odyssey Infrared Imaging System (LI-COR, Inc., Lincoln, NE, USA) and Image J Software (NIH, Bethesda, MD, USA). Resultant values were normalized to grayscale intensities of the total (non-phosphorylated) protein levels and that of internal reference GAPDH to reduce inter- and intra-gel variability. Analyzed data were displayed as fold change vs. Control levels.
Primary antibodies for NLRP3 (#ab4207), IL-1β (#ab9722), and TXNIP (#ab188865) were bought from Abcam plc. (Cambridge, UK). Primary antibodies for Caspase-1 p10 (#sc-514), IRS1 (#sc-559) and phospho-IRS1Ser307 (#sc-33,956) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Rabbit anti-mouse primary antibodies for Akt (#1080–1) and phospho-AktSer473 (#2118–1) were bought from Epitomics (Burlingame, CA, USA). Rabbit anti-mouse GAPDH antibody (#D110016) was purchased from Sangon Biotech (Shanghai, China). IRDye secondary antibodies were introduced from LI-COR Biosciences (LI-COR, Inc., Lincoln, NE, USA).
Immunostaining and morphometric analysis
Paraffin-embedded pancreatic sections were incubated with specific primary and secondary antibodies, then observed under Axiovision Observer A1 fluorescence microscope (Carl Zeiss, Germany) and photographed using the accompanying software. In accordance with former researches [
38,
41], quantifications of insulin, glucagon and F4/80 positive area proportions per islet were performed using the manual histology tools in Image J Software (NIH, Bethesda, MD, USA). Four pairs of 4-μm-thick serial sections, 100 μm apart were analyzed. An average of 53 islets from 3 to 4 mice for each group were analyzed for insulin and glucagon areas, and 67 islets from 6 mice per group were assessed for F4/80 areas.
Primary antibodies for Insulin (#GB13121), Glucagon (#GB13097), F4/80 (#GB11027) and Cy3/Alexa 488 conjugated fluorescent secondary antibodies (#GB21301/#GB25303) were all purchased from Servicebio Biotechnology (Wuhan, Hubei, China). Fluorescent stain solution 4′, 6-diamidino-2-phenylindole dihydrochloride (DAPI) for nuclei labeling was bought from the same company.
Statistical analysis
The data were expressed as mean ± standard error (SEM) and differences were defined statistically significant only when p < 0.05. Data were mainly analyzed using one-way analysis of variance (ANOVA) followed by Bonferroni’s multiple comparison post hoc tests with GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA). Two-way ANOVAs for repeated measurements followed by Tukey’s or Bonferroni’s post hoc tests were used in the analyses of GSIS data and time section values of GTT and ITT.
Discussion
In the present study, we found that depressive-like behavior co-occurred with abnormalities of insulin secretion and signaling in a 12-week CUMS mouse model. This comorbidity may be due to chronic cytokine-mediated inflammatory responses, which characterized by activation of NLRP3 inflammasome and consequent IL-1β production, and induced by dysregulation of HPA axis. As an inhibitor of the NLRP3 inflammasome and antidiabetic drug of sulfonylureas, glyburide modulated not only depressive-like behavior, but also metabolic disturbance induced by chronic stress. We tend to attribute this effect mainly to the inhibition of NLRP3 inflammasome via suppressing its upstream molecule TXNIP, as coinciding with discoveries from Zhou et al. [
42]. Furthermore, it is also confirmed that fluoxetine exerts an inhibitory effect on NLRP3 inflammasome possibly through downregulating TXNIP, which may help explain why fluoxetine can ameliorate inflammatory biomarkers [
43]. Interestingly, fluoxetine can even regulate insulin signaling and secretion somehow.
Though many researchers believe cortisol dysregulation and chronic immune activation are the critical bidirectional links between stress, depression, and type 2 diabetes mellitus [
11,
27], there is still a lack of adequate evidence and specific molecules which function as the key node. As Busillo et al. [
29] elucidated, GC could sensitize the innate immune system by regulating NLRP3 inflammasome. The cleavage and maturation of pro-IL-1β caused by caspase-1 are largely mediated by inflammasome activation [
44]. A great number of animal and human studies have demonstrated the role of IL-1β in facilitating β-cell dysfunction and IR [
45]. Besides, Reich and colleagues [
30] suggested that TXNIP was a mediator of GC-induced β-cell apoptosis. ROS-induced TXNIP overexpression could drive hepatic inflammation as well as lipid accumulation, diabetic retinopathy and diabetic nephropathy through activation of NLRP3 Inflammasome [
46‐
48]. Apart from that, TXNIP was also explored in mediating CUMS-induced depression by activating inflammasome [
49]. It seems that GC/ROS-TXNIP-NLRP3 pathway may be an alternative option for the aforementioned key nodes in comorbidity.
Metabolic tests composed of IPGTT, ITT, GSIS, and HOMA-IR in our research indicated that stressed mice got insulin-resistant phenotypes, but were not obvious in hyperglycemia. This is not exactly consistent with previous studies. For example, Pan et al. found that after a 12-week CUMS regimen, rats exhibited depression-like behavior and glucose intolerance. He reckoned that the signal cross-talk between hypothalamic corticotrophin-releasing factor (CRF) system and insulin might be impaired, while fluoxetine treatment adjusting that system potentially prevented or healed depression and comorbid diabetes [
50]. Besides, similar CUMS-induced depressive co-morbid glucose intolerant phenotypes were also observed in some other studies and could be reversed by rosiglitazone, umbelliferone and curcumin [
51‐
53]. In contrast, another recent work reported down-regulated blood glucose in depression mouse model [
54]. Indeed, during the preclinical period of T2D, hyperinsulinemia and β-cell hyperplasia were often developed to compensate for insulin resistance [
55]. On this basis, we would rather define it as “depression comorbid with prediabetes” in our present research, because decreases in islet insulin area, increased infiltrations of F4/80 positive inflammatory cell and disturbances of insulin secretion were observed simultaneously. That coincided with papers published before, which suggested F4/80 positive macrophages invaded diabetic islets [
41,
56], CD68 positive macrophages could be found isolated or dispersed between adipocytes, throughout and around the pancreatic islets [
57], as well as another report insisted that macrophages mediate β-cell loss in T2D [
58].
Reversely, depressive-like behavior also showed up in mouse models of diabetes and other IR-related diseases. Ernst et al. [
59] found that db/db mice exhibited molecular alterations which were usually seen in neurological disorders. Sharma with his colleagues [
60] reported the occurrence of depression-, psychosis-like symptoms and anxiolytic behavior in db/db mouse strain. Moreover, it was demonstrated, in a brain-specific knock out of the insulin receptor mice model, that IR in the brain altered dopamine turnover and resulted in behavioral disorders [
61]. Within our study, hippocampal insulin signaling abnormalities also appeared, but whether it would cause dopamine turnover dysfunction needed further confirmation. Besides, high-fat diet could induce anhedonia by activating the purinergic P2X7 receptor-NLRP3 inflammasome pathway [
62,
63]. Even more, a clinical investigation of polycystic ovary syndrome (PCOS) found that HOMA-IR was significantly associated with the risk of depression [
64]. Meanwhile, depression-like behavior was also observed in a dehydroepiandrosterone-induced PCOS mouse model [
65]. In fact, the HOMA-IR index tended to increase in the stressed mice we studied.
Glyburide is the first identified compound that prevents NLRP3 inflammasome activation, but the exact mechanism of its pleiotropic effects remains obscure [
32]. Zhou el al [
42] reported that glyburide exerted its function by suppressing the induction of TXNIP. Controversially, Masters et al. [
66] doubted the role of TXNIP in altering the effect of glyburide on the inflammasome. Regardless of the mechanisms, the efficiency of glyburide was solid. For example, it could not only attenuate blood-brain barrier disruption in brain injuries induced by trauma [
67] and myocardial injury induced by lipopolysaccharides, but also prevent brain swelling [
68]. However, results from Lahmann et al. [
35] suggested that only little glyburide reached the central nervous system when given systemically. Therefore, we think that the anti-depressive and anti-IR effect of glyburide may be due to the inhibition of circulating inflammasomes, but its activity similar to peroxisome proliferator-activated receptor gamma (PPAR-γ) agonists cannot be excluded [
69]. Anyhow, there exists a certain peripheral-cerebral communication which leads to quantities of diseases [
70], which might help with the explanation. Nevertheless, chronic glyburide treatment in vivo may cause side effects including reversible loss of insulin secretory capacity due to β-cell hyperexcitability and other irreversible consequences [
71,
72]. Apart from that, PPAR-γ agonists were also indicated to assist the treatment of major depression [
73,
74], possibly owing to its anti-inflammatory effects, including inhibiting inflammasomes [
75].
As for the complicated actions of fluoxetine used in this study, it, of course, requires some clarifications. As a classical antidepressant, fluoxetine has displayed sound effects of counteracting depressive-like behaviors. According to conventional wisdom, this might attribute to regulating reuptakes of synaptic 5-HT. Actually, fluoxetine inhibited activation of the inflammasome to a lesser extent than glyburide. However, it appeared that there was no difference between the two drugs for depressive phenotypes. In our view, this can be ascribed to their dissimilar anti-inflammatory properties, which led to the same destination. According to pre-clinical and clinical evidence, fluoxetine and other SSRIs (selective serotonin-reuptake inhibitors) could induce reductions in several proinflammatory cytokines [
43], though not selectively. That is to say, the inhibitory effect on inflammasome may go through by-pass access, including the detrimental interactions between activation of inflammasomes and systemic inflammation [
21]. Nevertheless, glyburide prevented mice from inflammasome-related inflammation specifically, which blocked the “final common pathway” leading to the development of depressive disorder. Similarly, the abovementioned mechanism can also be applied to illustrate the complex impacts on insulin signaling.