Introduction

The link between metabolic conditions such as diabetes and psychiatric disorders has been reported widely.1 In most cases, metabolic abnormalities have been identified in association with psychiatric disorders as a side effect induced by commonly prescribed anti-psychotic medications.2 However, recent studies have shown that such changes can occur at disease onset even before medications have been applied. For example, hyperinsulinemia and insulin resistance have been reported in first onset, antipsychotic naive schizophrenia patients in several studies over the past decade.3, 4 This indicates that there may be an etiological link between metabolic and psychiatric conditions, which could be an important consideration in downstream treatment approaches.

The db/db mouse has been used as a model of type 2 diabetes mellitus and other metabolic conditions, such as obesity and dyslipidemia.5 This model was first described in 1965 by Hummel et al.,6 who identified random mutations in mice associated with obesity and excessive hunger. The models were later segregated as ob/ob mice with gene mutations for leptin, and db/db mice that have mutations for the gene encoding the long isoform of the leptin receptor, localized mainly in the hypothalamic region of the brain.7 The db/db mouse model mirrors the clinical picture of type 2 diabetes mellitus.

A recent study found that hippocampal inflammation in db/db mice is associated with increased anxiety-like behavior.8 It has been reported that patients with diabetes have an increased probability of developing major depressive disorder9 and Alzheimer’s disease or cognitive impairments10 compared with non-diabetic people. It has also been suggested that type 2 diabetes and Alzheimer’s disease share similar biochemical and molecular features.11 Consistent with the link between psychiatric and metabolic disorders, recent studies have shown that db/db mice exhibit behavioral abnormalities, such as impairments in memory function and long-term potentiation, as well as depression and psychosis-like behaviors.12, 13 Furthermore, treatment of db/db mice with anti-diabetic medications such as rosiglitazone has led to improved insulin signalling along with a decrease in measures of depression but not in psychosis-like behaviors.14

In this study, we have carried out molecular profiling of plasma samples from db/db mice in order to identify peripheral molecular pathways, which are altered in this model. In addition, we have carried out the first proteomic profiling analysis of brain tissues from these mice in order to gain insight into the altered molecular pathways associated with the observed schizophrenia- and depression-like behaviors.

Materials and methods

Animals

Male db/db mice (10-week old; background strain C57BL/KsJ (BKS-Cg-Dock7m +/+ Leprdb/J)) and age-matched non-diabetic lean control C57BL/KsJ mice (Jackson Laboratories; Bar Harbor, ME, USA) were housed singly in plastic cages with wooden shavings in a temperature controlled room (22–23 °C) on a 12-h/12-h light/dark cycle (lights on at 0500 hours) (Table 1). Single housing of db/db mice is routinely practiced in our lab and by most of the labs worldwide. Housing at this temperature range results in no change in diurnal rhythm of db/db mice compared with control mice.15 To assess the metabolic status of the mice, body weight, food consumption and water intake were monitored weekly (Table 2). Blood samples were taken from a cut made on the tip of the tail from each mouse and glucose concentrations were determined using an Accu-Check Advantage Blood Glucose Monitor (Roche Diagnostic Corporation, Indianapolis, IN, USA) to confirm development of diabetes in db/db mice compared with that in lean controls. Body fat composition was measured as described previously.13 The EchoMRI whole-body composition analyzer (Houston, TX, USA) was used to determine fat and lean body mass.16 Fat and lean mass were calculated as percentage of total mass. A total of 20 mice (10 db/db and 10 lean control mice) were killed according to schedule 1, decapitated and trunk blood was collected in ice-chilled tubes containing EDTA and centrifuged at1100 g, 4 °C for 15 min. Plasma was immediately separated and stored frozen at −80 °C for later use. Frontal cortex and hippocampal regions of the brain were dissected on ice and immediately frozen at −80 °C. All experiments were carried out in full compliance with the principles of laboratory animal care (NIH publication no. 85–23, revised 1985), and the protocols were approved by the Wright State University Animal Care and Use Committee.

Table 1 Physical and metabolic characteristics of db/db and control mice
Full size table
Table 2 Food and water consumption of db/db and control mice
Full size table

Plasma insulin measurements

Plasma insulin levels were determined at Mouse Metabolic Phenotyping Centre (Cincinnati, OH, USA) by using the commercial immunoassay kits (Millipore, St Charles, MO, USA) according to the manufacturer’s specifications. Briefly, standards, quality controls and samples were incubated for 2 h in microplate wells pre-coated with mouse anti-rat insulin antibodies (80 μl per well). After washing, biotin-labeled polyclonal antibodies were added (100 μl per well) followed by incubation for 30 min. After washing, horse radish peroxidase conjugate and the substrate 3,3’,5,5’-tetramethylbenzidine solution (100 μl per well) were added for a 15-min incubation. The reactions were halted by addition of 100 μl per well 0.3 M HCl, and the absorbance of the resulting color product was measured at λ=450 nm. The concentrations of the experimental samples were determined by comparison of the absorbance values with those of a standard curve, constructed using 0.2, 0.5, 1, 2, 5 and 10 ng ml−1 insulin.

Multiplex immunoassay profiling

Plasma samples were analyzed using the RodentMAP, Rat MetabolicMAP and Rat KidneyMAP platforms comprising multiplexed immunoassays for a total of 89 analytes (86 proteins and 3 steroid hormones) (Supplementary Table S1) in a Clinical Laboratory Improved Amendments (CLIA)-certified laboratory at Myriad-RBM (Austin, TX, USA) as described recently.17 Immunoassays were calibrated using duplicate standard curves for each analayte and raw intensity measurements converted to protein concentrations using proprietary software. Multiplexed calibrators (eight levels per analyte) and controls (three levels per analyte) are developed to monitor key performance parameters, such as lower limit of quantification, precision, cross-reactivity, linearity, spike-recovery, dynamic range, matrix interference, freeze–thaw stability, and short-term sample stability are established for every assay as described by the manufacturer (http://www.myriadrbm.com/technology/data-quality/). Data analyses were performed using the statistical software package R (http://www.r-project.org) and the levels of analytes were determined. Analyses were conducted under blinded conditions with respect to sample identities, and samples were analyzed in random order to avoid any sequential biases.

Liquid chromatography mass spectrometry (LC-MSE) profiling

Frontal cortex and hippocampus tissues from db/db (n=10) and control (n=10) mice were sonicated in a buffer comprised of 7 M urea, 2 M thiourea, 4% chaps, 2% ASB14, 70 mM dithiothreitol and protease inhibitors (5:1 (v/w) ratio) at a 5:1 (v/w) ratio and mixed for 30 min. Samples were centrifuged for 3 min at 17 000 g at 4 °C and the supernatants recovered. Protein concentrations of the lysates were determined using a Bradford assay (Bio-Rad; Hemel Hempstead, UK). Proteins (approximately 100 μg) were precipitated using acetone, centrifuged for 3 min at 17 000 g at 4 °C and the pellets suspended in 50 mM ammonium bicarbonate. Disulfide bonds were reduced in 5 mM dithiothreitol at 60 °C for 30 min, and free sulfhydryl groups were alkylated by incubation in 10 mM iodacetamide in the dark at 37 °C for 30 min. Protein digestion was conducted using porcine trypsin at a 1:50 (w/w) ratio for 17 h at 37 °C, and reactions were stopped by addition of 8.8 M HCl at a 1:60 (w/w) ratio. Quality control samples consisting of frontal cortex and hippocampus protein extracts from a wild-type mouse were prepared as above and underwent all experimental steps in parallel with the test samples to monitor machine performance.

LC-MSE analyses were carried out in triplicate of individual samples as described previously.18 Samples were diluted in 0.1% formic acid (Sigma-Aldrich; Poole, UK) to a final peptide concentration of 0.12 μg μl−1 and reverse phase LC separation was performed using a nanoACQUITY system (Waters Corporation; Milford, MA, USA) comprised of a C18 trapping column (180 μm × 20 mm; 5 μm particle size) and a C18 BEH nanocolumn (75 μm × 200 mm; 1.7 mm particle size). The buffers were as follows: (A) H2O+0.1% formic acid and (B) acetonitrile (Sigma-Aldrich)+0.1% formic acid. Initial buffer concentrations were 3% B (97% A) followed by 3–30% B over 90 min, 30−90% B over 25 min, 90−97% B over 5 min, constant 97% B for 10 min and 97−3% B over 1 min. The column was coupled online to a nanoESI emitter on a quadrupole time-of-flight (Q-TOF Premier) mass spectrometer (Waters Corporation). Approximately 500 fmol μl−1 of Glu Fibrinopeptide B (Waters Corporation) was infused every 30 s using a lock spray for mass calibration. The mass spectrometer was operated in V mode and analyses were performed using positive nanoESI ion mode. The collision energy was 5 eV for low-energy scans and ramped from 17 to 40 eV for high-energy scans with a cycle time of 1.3 s. The low collision energy generates intact precursor ion data, and the high collision energy provides information about the peptide fragments.

Data analysis

Analysis of multiplex immunoassay profiling data was conducted using nonparametric Wilcoxon rank-sum tests to account for any uneven distribution of the data. P<0.05 was considered as statistically significant. LC-MSE data were processed using ProteinLynx Global Server (PLGS) v.2.4 (Waters Corporation) and Rosetta Elucidator v.3.3 (Rosetta Biosoftware; Seattle, WA, USA) for time and mass/charge alignment of ion fragments as described previously.18 The Mus musculus complete proteome FASTA (sourced from www.uniprot.org) sequence Integr8 database was used for the appropriate protein searches. Acceptable modifications were set to include oxidation of methionine and phosphorylation of serine, threonine or tyrosine residues. The criteria for valid protein identifications were set at a minimum of three fragment ions/peptide ion and seven fragment ions/protein and detection of peptides in at least two out of three replicates. In addition, proteins had to be identified in 60% of samples to ensure biological reproducibility. The results were imported into Elucidator for alignment and annotation of the features. The end result was a matrix that included peptide intensities for each sample. Proteins identified by ⩾2 peptides were considered for further analyses (Supplementary Table S2).

Statistical analyses

Significant differences in protein expression generated by the LC-MSE analyses, between db/db and control mice, were determined using Wilcoxon signed-rank tests (P<0.05). The false discovery rate was calculated, although no adjustments were made for multiple comparisons in order to not exclude possible true positives. Nevertheless, only those molecules with a false discovery rate <0.4 and a fold change cut off of 0.10 for the LC-MSE analyses were applied in order that only those molecules with the most robust changes were considered.

Classification of differentially expressed proteins

The UniProt accession codes of significantly altered proteins found in brain tissues from db/db and control mice were uploaded to the Ingenuity Pathway Analysis Knowledge Base (IPKB; Ingenuity Systems; Mountain View, CA, USA). Networks and pathways that were most significant to the data set were determined by overlaying the uploaded proteins onto predefined pathway maps in the IPKB. A right-tailed Fisher’s exact test was used to calculate the P-values associated with the identified canonical pathways.

Results

Plasma analysis

Multiplex immunoassay profiling of plasma samples showed that 31 molecules were present at different levels between db/db and control mice (Table 3). As expected, leptin levels were significantly higher (approximately 18-fold) in db/db compared with the control mice. Several other molecules were also present at higher levels in the db/db mice, including insulin, C-reactive protein, serum amyloid P component, immunoglobulin A, interleukin 18, monocyte chemotactic proteins 1 and 3 and fibrinogen. In addition, other proteins showed decreased levels in the db/db mice peptide YY, glucagon, prolactin, apolipoprotein A, monocyte chemotactic protein 5, osteopontin, haptoglobin, interleukin 10, beta 2 microglobulin and lymphotactin.

Table 3 Multiplexed immunoassay of plasma from db/db and control mice
Full size table

Frontal cortex analysis

Brain proteome analyses of the db/db and control mice were performed using LC-MSE to determine whether changes occurred as described in some post-mortem studies of neuropsychiatric and neurodegenerative diseases. LC-MSE analysis of whole frontal cortex extracts led to the identification of 478 proteins. Of these proteins, 13 were found to be significantly altered in db/db compared with control mice (Table 4). Four of these proteins were increased and nine decreased. The highest increase was observed for the insulin-like growth factor 2 mRNA binding protein 1 and the calcium calmodulin-dependent protein kinase type II alpha chain (CAMK2A). The most robust decreases were observed for stathmin 2 and neuromodulin. In silico functional analysis using the IPKB software analysis showed that 12 of the 13 altered proteins could be incorporated into a single interaction network (Figure 1). The top disease identified with the uploaded proteins was neurological disorder (P=7.1E-5–4.4E-2) and Parkinson’s disease-related signalling was the top canonical pathway (P=7.4E-5; 2/16 proteins).

Table 4 LC-MSE analysis of frontal cortex from db/db and control mice
Full size table
Figure 1
figure 1

In silico functional analysis of proteomic changes in the frontal cortex of db/db mice. UniProt accession codes were uploaded into the Ingenuity Pathway Analysis Knowledge Base to determine the most over-represented networks and canonoical pathways associated with the data. Straight lines=direct connections; dotted lines=indirect connections; green colour=increased; red colour=decreased.

Full size image

Hippocampus analysis

Proteome analysis of the hippocampus from db/db and control mice led to identification of 441 proteins. Differential analysis showed that more proteins were altered in this tissue compared with the frontal cortex, with 56 proteins found to be present at significantly different levels between db/db and control mice (Table 5). The highest increases were observed for tubulin beta 5 chain and Thy 1 membrane glycoprotein. The largest decreases were found for groups of proteins involved in metabolic processes (malate dehydrogenase mitochondrial, cytochrome b c1 complex subunit 2, malate dehydrogenase cytoplasmic) or intracellular transport (alpha internexin, heat shock 70 kDa protein 1L, dihydropyrimidinase related protein 1, ras-related protein rab 39B). IPKB analysis of the altered proteins identified a network comprising 14 of the target proteins (Figure 2). Most of the altered proteins were involved in cellular signalling, cell structure and maintenance, neural functioning and energy metabolism. Three members of the 14-3-3 protein family, conserved regulatory molecules expressed in all eukaryotes, were among the group of altered proteins. Furthermore, several molecules linked to regulation of calcium signalling were also altered. The top disease associated with the altered hippocampal proteins was hematological disease (P=8.4E-5–1.7E-2) and the top canonical pathway was clathrin-mediated endocytosis signalling (P=3.0E-6; 7/181 proteins).

Table 5 LC-MSE analysis of hippocampus from db/db and control mice
Full size table
Figure 2
figure 2

In silico functional analysis of protein changes in the hippocampus of db/db mice. UniProt accession codes were uploaded into the Ingenuity Pathway Analysis Knowledge Base to determine the most over-represented networks and canonoical pathways associated with the data. Straight lines=direct connections; dotted lines=indirect connections; green colour=increased; red colour=decreased.

Full size image

Discussion

This is the first proteomic profiling study investigating plasma and brain tissues from the db/db mouse model. The advantage of using the multiplex profiling approach is the potential to identify differences in hundreds of molecules simultaneously, which can lead to the identification of changes in functional pathways. Profiling of plasma from db/db mice showed the expected robust increase in leptin levels, along with significantly higher levels of insulin compared with the levels in control mice. In line with these findings, we also found decreased levels of other hormones such as peptide YY, glucagon and prolactin, which suggested that lack of leptin receptors in the db/db mice can lead to either altered biosynthesis or secretion of other major hormones.

The use of the multiplex immunoassay profiling analysis resulted in identification of several novel serum molecules, which have not been described previously in association with this model. The finding of lower peptide YY levels is consistent with the increased appetite and body weight of the db/db mice and with previous studies which have shown that obese humans secrete lower levels of this hormone.19 Lower levels of peptide YY have also been found in drug-free schizophrenia patients in a biomarker study of cerebrospinal fluid, suggesting another link between metabolic regulation and psychiatric disorders.20 The finding of decreased circulating prolactin levels in db/db mice has also not been reported previously. Prolactin is secreted by the anterior pituitary and this process can be inhibited by release of dopamine from the arcuate nuclease of the hypothalamus.21 This is interesting as one of the major theories of how schizophrenia develops involves increased dopaminergic activity in the brain.22 Therefore, further studies are warranted to determine whether hypothalamic–pituitary dopamine signalling is altered in this model.

In addition to the effects on hormonal pathways, we also identified alterations in immune and inflammation-related molecules in plasma from db/db mice. Several studies have shown that immune system perturbations occur in type 2 diabetes mellitus,23 as well as in psychiatric and neurodegenerative disorders, including schizophrenia,17 depression24 and Alzheimer’s disease.25 However, the overlap of immune-related molecules across these conditions with the db/db mice model is not surprising given that these pathways are affected in a variety of diseases. In the case of the db/db mouse, the observed perturbations in immune system molecules could be due to the high circulating levels of leptin, as this hormone can also act as pro-inflammatory adipokine.26 Leptin has been tested as a possible treatment for autoimmune and/or inflammatory diseases in a recent study27 and preclinical studies have shown that the depression-like symptoms in the chronic stress and social defeat rat models was associated with lower serum leptin concentrations, and the behavioral effects could be reversed by systemic leptin administration.28

Comparison of circulating molecules changing in the db/db mouse model to those identified in our previous study of first onset schizophrenia patients shows that 12 proteins were altered in common and 6 of these showed the same directional changes (insulin, serum amyloid P, interleukin 18, serum glutamic-oxaloacetic transaminase, apolipoprotein A1 and epidermal growth factor).17 In addition, seven circulating proteins (insulin, serum glutamic-oxaloacetic transaminase, prolactin, cystatin C, stem cell factor, macrophage colony-stimulating factor and beta-2 microglobulin) were changed in common between the db/db mouse and the acute PCP rat model of schizophrenia, using the same molecular profiling platform.29 Taken together, these findings suggest that the db/db model shows an overlap of molecular profiles with schizophrenia and other psychiatric or neurodegenerative disorders.

The most novel findings of the current study involved the identification of changes in brain proteins as these have not been described previously in association with this model. Consistent with the effects on behavior seen in the db/db mice,13, 14 we identified multiple proteins that were altered in brain tissues extracted from db/db mice compared with control mice. The finding that more proteins were altered in the hippocampus of the db/db mouse is consistent with the reported effects on cognition and learning in this model. In line with this, we found changes in three members of the 14-3-3 protein family in db/db mouse hippocampi. These proteins are highly conserved and have an important role in neurodevelopment due to their involvement in cellular processes, such as cell signalling, growth and apoptosis.30 Also, perturbations in these proteins have been found in schizophrenia, autism, bipolar disorder and in subjects with impaired sensory motor gating and memory.30, 31 The latter finding is consistent with the impairments seen in paired-pulse inhibition in the db/db mice. Furthermore, our findings of altered expression of Ca2+-regulatory proteins, such as CAMK2B, CAMK2D and calcineurin subunit B type 1, are in line with studies that identified impaired cardiac function in db/db mice through suppression of Ca2+ permeability.32 This also supports the possibility that the defect found in the db/db mice is systemic, with effects seen in both the brain and periphery.

Fewer proteomic changes were found in the frontal cortex region of the db/db mice using the LC-MSE proteomic profiling approach. However, in silico analysis of the 13 significantly altered proteins resulted in identification of ‘neurological disorder’ as the top associated disease and ‘Parkinson’s signalling’ as the top canonical pathway. Again, this indicated potential effects on dopaminergic pathways as seen most notably in Parkinson’s disease and schizophrenia. In addition, one of the brain proteins found to be altered in both the db/db and the acute PCP rodent model described above,29 was protein DJ-1 (also known as PARK7). DJ-1 is an oxidative stress response protein that defends cells against reactive oxygen species. Furthermore, CAMK2A was found to be changing in the db/db mice as found in our previous study of the acute PCP model of schizophrenia.29 Previous studies have also found perturbed expression of these proteins in post-mortem frontal cortex tissues from schizophrenia patients33, 34 as well as in disorders associated with effects on learning and memory.35, 36, 37

One limitation of this study was the small numbers of animals used to generate the data. However, we have attempted to compensate for this through comparison of the findings with those in the literature on the same mouse model, the acute PCP model, drug-naive schizophrenia subjects and on protein pathways altered in schizophrenia post-mortem brain tissues. Nevertheless, these studies require repetition in independent cohorts of db/db mice, and future studies should attempt to further explore the specificity of the effects through comparison with other animal models and other human psychiatric and neurodegenerative diseases. Also, the db/db mouse has been mainly used in type 2 diabetes research, and it should be noted that the alterations in central nervous system could result from the change of leptin receptor signalling in the brain or though complications of diabetes itself.

In conclusion, this study resulted in the identification of novel central nervous system proteomic alterations that link the metabolic abnormalities found in the db/db mouse model to neurodegenerative and neuropsychiatric diseases. This has laid the ground work for further studies of the db/db mouse model in testing whether treatment with drugs such as antidepressants, antipsychotics or anti-diabetic compounds has a positive effect on the behaviors and brain proteome changes observed. Sharma et al.14 have already carried out a study which showed that treatment of db/db mice with rosiglitazone could affect depression measures, but this had no effect on readouts for psychosis-like behaviours. Taken together with previous investigations on the behavioural effects in this model, the current study supports the use of the db/db mouse as a potential preclinical model for schizophrenia and potentially other psychiatric or neurodegenerative disorders. Also, the combined use of this model with incorporation of high throughput assays for the plasma and brain molecular candidates identified here would potentially offer a useful novel tool for drug discovery, with a focus on the metabolic aspects of mental disorders and/or the psychopathology associated with diabetes.