Abstract

   Clinical and experimental studies have demonstrated that type 2
   diabetes mellitus (T2DM) affects the brain structure and function, in
   particular the hippocampus, leading to cognitive impairments. However,
   the molecular mechanisms underlying cognitive deficits induced by T2DM
   are not fully understood. In this study, we aimed to investigate the
   effects of T2DM on behavior, the proteome profile in the hippocampus,
   and the potential molecular pathways involved in the development of
   cognitive dysfunction in T2DM mice. We found that the diabetic mice
   exhibited cognitive impairment in the novel object location recognition
   test and the novel object recognition test. The proteomic analysis
   revealed that various molecular pathways were involved in this context.
   These included the upregulation of proteins in the protein synthesis
   and folding pathway (EIF5A, RSP24, and PPIB), endocytosis and cellular
   trafficking (VPS24, SNX12, and ARP2/3), cannabinoid receptor
   interacting (CRIP1), ubiquitination (SKP1), and oxidative stress
   response (NUDT3). Downregulated proteins were related to mitochondria
   function (ANT1), neuronal development (ELP1), protein glycosylation
   (RPN2), and endocytosis (VPS4). Our study shows that T2DM mice exhibit
   neurocognitive impairment, which is linked to the dysregulation of
   hippocampal proteins involved in various molecular pathways. These
   findings contribute to a better understanding of the pathophysiology of
   T2DM-related cognitive impairment and may identify molecular targets
   for drug development to treat T2DM-associated cognitive impairment
   conditions.

   Keywords: diabetes mellitus, hippocampus, mouse model, cognitive
   deficit, proteomics

Introduction

   Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder that
   affects millions of people worldwide, and its prevalence is increasing
   due to the epidemic of obesity.^ [33]1 T2DM is characterized by insulin
   resistance, hyperglycemia, and chronic inflammation, which are
   associated with various complications, including cardiovascular
   disease, renal disease, and retinopathy.^ [34]2 T2DM also increases the
   risk of developing mild cognitive impairment (MCI) and enhances the
   conversion from MCI to dementia.^ [35]3 Cognitive impairments affect
   the quality of life of patients who suffer from T2DM, and these
   conditions impose a heavy financial burden on families because of the
   need for home care.^ [36]4 Thus, better understanding of the
   neuropathological substrates of cognitive impairment induced by
   diabetes mellitus is urgent.

   Mounting evidence suggests that T2DM can also affect brain function and
   structure, particularly in the hippocampus, which plays a crucial role
   in learning and memory.^[37]5-[38]7 There is a growing body of evidence
   linking brain insulin resistance, inflammation, hypercortisolism and
   amyloid accumulation to cognitive impairments and the pathology of
   Alzheimer disease (AD) in experimental models of diabetes.^ [39]8
   Mitochondrial dysfunction and neurodegeneration, blood-brain barrier
   injury, Ca^2+ signaling disruption, and abnormal neuronal structure and
   function have also been proposed to contribute to DM-induced cognitive
   impairments.^[40]9,[41]10 However, the exact molecular mechanisms
   underlying cognitive impairments induced by T2DM remain unclear.

   Proteomics is a powerful tool for the comprehensive analysis of protein
   expression and modification of complex biological systems and has been
   used to investigate the molecular basis of neurodegenerative diseases.^
   [42]11 Despite the recognition of hippocampal involvement in
   T2DM-related cognitive dysfunction, the molecular mechanisms underlying
   this condition are not yet fully elucidated. To address this gap,
   researchers have employed proteomic analysis approaches to investigate
   protein expression profiles in the hippocampus of animal models.
   However, few studies are available in this context.

   Previous studies demonstrated that proteins involved in energy
   metabolism, oxidative stress response, protein transport, and
   inflammatory pathways were altered in the hippocampus of a
   streptozotocin/high fat diet-induced diabetic animal model, correlating
   with observed cognitive deficits.^[43]12,[44]13 Similarly, significant
   alterations in the levels of proteins implicated in the arrangement of
   myelin sheath and morphology of axons have been reported in the
   hippocampus of diabetic Zucker fatty rats.^ [45]14 Although these
   studies provide valuable insights, the proteomic data obtained from
   animal models are quite different in terms of protein alterations and
   molecular pathway involvement. In addition, these studies have
   predominantly used the 2D gel electrophoresis (2DE) approach. This
   method has limitations, such as the pH range and potential loss of some
   proteins during separation, which may result in an incomplete proteomic
   profile. Therefore, employing an additional proteomic approach, such as
   Gel-Liquid Chromatography (GeLC), is necessary to overcome these
   limitations and provide a more comprehensive understanding of the
   molecular mechanism underlying T2DM-related cognitive dysfunction.

   Therefore, the main objective of this study was to identify the changes
   in the hippocampus proteome of T2DM mice compared with controls. We
   aimed to use a GeLC shotgun proteomic approach to comprehensively
   analyze the hippocampus proteome and identify potential molecular
   pathways involved in the development of cognitive dysfunction in T2DM.
   We additionally applied the novel object location recognition test
   (NOLT) and novel object recognition test (NORT) to assess and confirm
   the effect of T2DM on cognitive function. The results obtained from
   this study contribute to a better understanding of the molecular basis
   of T2DM-related cognitive impairment in mouse models and may facilitate
   the development of new treatments of T2DM-associated cognitive
   dysfunction.

Materials and Methods

Animals and treatment

   Eight-week-old male C57BL/6N mice were obtained from Nomura Siam
   International, Thailand. The animals were housed in a
   climate-controlled environment with alternating 12 h light–dark cycles,
   4 to 5 animals per cage. After 1 week of acclimatization, animals were
   randomly assigned to receive either a control diet (n = 12) offering a
   total of 3.04 kcal/g (containing 4.5% crude fat, 24% crude protein)
   (the National Animal Center, Salaya Campus, Mahidol University,
   Bangkok, Thailand) or a high-fat diet (HFD; n = 12) supplemented with
   25% sucrose (containing 14.4% crude fat, 24.3% crude protein) (Quick
   fat; CLEA, Japan) offering a total of 4.11 kcal/g for 4 weeks.^ [46]15
   The diabetic animal model was generated by 2 intraperitoneal injections
   of streptozotocin (STZ), dissolved in cold citrate buffer (0.01 M, pH
   4.5) (Sigma-Aldrich, Saint Louis, USA) at the dose of 100 mg/kg with
   the injection volume of 10 mL/kg, while the normal control (ND) group
   received the same volume injection of citrate buffer (vehicle).^ [47]16
   The second injection was performed 48 h after the first injection. One
   week after the first injection of STZ, the fasting blood glucose (FBG)
   test was performed to confirm the diabetic model. The diabetic group
   (DM group) mice which met the criterion (FBG > 200 mg/dL) were included
   in the study.^[48]16,[49]17 Two mice from the DM group did not meet the
   criterion and were excluded for data analysis. In total, 12 normal
   control mice (ND) and 10 diabetic mice (DM) were included in the
   behavioral tests and metabolic profile analysis. The sample size was
   calculated by using the G*Power program, configuring t-test (2-tailed)
   with the effect size of 1.41^ [50]15 a significance level (α) of 0.05
   for type I error and a test of power (1-β) of 0.8. The resulting total
   sample size calculated was 9 animals per group. However, the attrition
   rate of 25% has been taken into account. Therefore, the total number of
   sample size was equal to 24 animals which were assigned into the normal
   control (ND) group (n = 12) and the diabetic group (DM) (n = 12). The
   behavioral tests were performed 4 weeks after the diabetes status
   confirmation. The experimenters were blinded to the treatment group.
   After the behavioral test, mice were euthanized, and hippocampi were
   dissected and frozen at -20°C for proteomic analysis. This study
   adheres to the ARRIVE guidelines for the reporting of animal
   experiments. All experiments were conducted in accordance with the
   Animal Care and Use Committee of Burapha University (IACUC Number
   006/2564).

Novel object location recognition test

   The protocol carried out was an adapted version used by previous
   studies.^[51]15,[52]18 Briefly, the test was performed in the white
   plastic open-field box (40 × 40 × 25 cm). The test consisted of 2
   phases: (1) the training phase; mice were allowed to freely explore the
   box containing 2 identical objects (glass bottle) for 10 minutes and
   (2) the retention phase was performed 3 hours after the training phase.
   Animals were allowed to re-explore the objects for 5 minutes where one
   object remained in the same position (nondisplaced object), but the
   location of the other object was changed (displaced object). The
   minimal time that animals had to explore the objects during both the
   training and test phases to be included in the analysis was 10 seconds.
   The percentage of recognition index was represented as the time spent
   exploring the displaced object divided by the total time spent
   exploring the displaced object and the nondisplaced object X 100. The
   counterbalancing was used to minimize the order of measurement effect.

Novel object recognition test

   The NORT was conducted as described previously.^[53]19,[54]20 Briefly,
   the test consisted of 3 phases: habituation phase, training phase, and
   test phase. During the habituation phase, mice were exposed to an
   open-field box in the absence of objects for 5 minutes. Mice were then
   returned to their home cage. On the following day (training phase),
   mice were placed in the same open field with 2 identical objects and
   were allowed to freely explore the environment and objects for
   10 minutes. Twenty-four hours later (test phase), mice were placed back
   in the open field and, at this time, allowed to explore a novel object
   and a familiar one for 10 minutes. During the experiment, the
   open-field box and objects were cleaned using 70% alcohol to eliminate
   the olfactory cue. Time spent exploring each object was collected in
   the test phase in which the mouse’s nose was within 2 cm of the object.
   The minimal time that animals had to explore the objects during both
   the training and test phases to be included in the analysis was 10
   seconds. The percentage of time spent exploring the novel object versus
   the total object exploration time was presented.

Measurement of body weight and blood glucose level

   Body weight was monitored weekly across the experiment. The blood
   glucose levels were evaluated at 1 week after STZ injection and 4 weeks
   after STZ injection. After a 6-hour fast, blood samples were collected
   from a small incision at the end of each mouse’s tail.^ [55]13 A drop
   of blood was placed onto a glucose strip and blood glucose was measured
   with a glucose meter according to the user instructions guideline
   (ACCU-CHEK Guide, Roche, Thailand). The counterbalancing was also used
   to minimize the order of measurement effect.

Plasma insulin measurement and insulin resistance determination

   Plasma samples from each group of mice were collected and stored at
   -20°C until they were thawed for the assay. Plasma insulin was detected
   using ELISA kits (MyBioSource, San Diego, CA, USA) according to the
   manufacturer’s instructions. The optical density (OD) value of the
   sample at 450 nm was collected by a microplate reader (Thermo Fisher
   Scientific, Waltham, USA). The sample concentration was calculated
   according to a corresponding standard curve.

   The homeostatic model assessment for insulin resistance (HOMA-IR) was
   calculated as previously described.^ [56]21

Proteomic analysis

Sample preparation

   The hippocampal proteome of T2DM mice was compared with the controls
   using the GeLC shotgun proteomic approach. Mice from the diabetic group
   and control group were selected based on 2 criteria including FBG and
   the NOLT values, which reflect hippocampus function. Four DM with the
   highest FBG levels and poorest cognitive function in NOLT were selected
   while 4 normal control mice (ND) with highest cognitive performance in
   NOLT were selected for hippocampal proteome analysis. Hippocampal
   proteins were extracted according to our previous work.^ [57]22 A total
   of 10 µg protein/individual was separated on a 12% sodium dodecyl
   sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel using 100 V
   for 15 minutes to obtain protein migration of approximately 0.5 cm in a
   separating gel and stained with Coomassie blue for gel band
   visualization.

In-gel digestion and MS/MS analysis

   Gel bands that appeared in the separating gel were cut into
   approximately 1 × 1 mm, washed and destained. Proteins fixed in the gel
   were reduced with dithiothreitol and alkylated with iodoacetamide.
   In-gel digestion was performed with trypsin (Trypsin Gold, Mass
   Spectrometry Grade, Promega, Madison, WI, USA) at a final trypsin
   concentration of 20 ng/µL in 50 mM aqueous ammonium bicarbonate and 5
   mM CaCl[2]. Digestion proceeded for 8 hours at 37°C. Afterward,
   peptides were extracted with 5 changes of 50 µL of 5% trifluoro acetic
   acid in 50% aqueous acetonitrile supported by ultrasonication for
   10 minutes per change. Extracted peptides were dried down in a vacuum
   concentrator (Eppendorf, Hamburg, Germany). Dried peptides were
   resuspended in 100 µL 0.1% trifluoroacetic acid (TFA) for Q Exactive
   Orbitrap LC-MS/MS analysis and 5 µL were injected in technical
   duplicates. All procedures for Q Exactive Orbitrap LC-MS/MS analysis
   were performed according to Gutiérrez et al.^ [58]23

   The GeLC shotgun proteomic approach offers significant advantages in
   enhancing protein separation and resolution by separating proteins in a
   gel prior to digestion and analysis. This prefractionation step reduces
   sample complexity and helps remove impurities and contaminants, leading
   to more accurate and detailed protein identification and
   quantification.^ [59]24 In addition, this method allows the
   identification of a broader range of proteins, including those with
   extreme isoelectric points (pI) or molecular weights that may not be
   well-resolved by 2DE proteomic approach, resulting in a more
   comprehensive proteome profile.

Database search and pathway analysis

   Database search was performed using the Proteome Discoverer Software
   2.4.0.305 (Thermo Fisher Scientific) with the following settings:
   Protein Database: Uniprot mouse (tx1009) ([60]www.uniprot.org) and the
   database of common contaminants (cRAP,
   [61]http://ftp.thegpm.org/fasta/cRAP/crap.fasta), Enzyme Name: Trypsin
   (Full), Max. Missed Cleavage Sites: 2, Precursor Mass Tolerance: 10
   ppm, Fragment Mass Tolerance: 0.02 Da, Dynamic Modification:
   Oxidation/+15.995 Da (M), Dynamic Modification: Deamidated/+0.984 Da
   (N, Q), N-Terminal Modification: Gln->pyro-Glu/-17.027 Da (Q),
   N-Terminal Modification: Acetyl/+42.011 Da (N-Terminus), N-Terminal
   Modification: Met-loss/-131.040 Da (M), N-Terminal Modification:
   Met-loss+Acetyl/-89.030 Da (M), Static Modification:
   Carbamidomethyl/+57.021 Da (C).

Bioinformatic and pathway analysis

   In the proteomic experiment, intensity-based, label-free quantification
   protein abundance raw values were generated in Proteome Discoverer
   software. Raw data were filtered for proteins containing no or at least
   4 values per group. After aggregation by the mean, further statistical
   analysis by t-test comparison was performed with the R programming
   language using the DEP package including normalization and imputation
   of missing values by zero. Proteins were considered differentially
   expressed when protein abundance differed by a factor of 2-fold change
   (upregulation when log[2]-fold change ⩾ 1 or downregulation when
   log[2]-fold change ⩽ 1), and an adjusted P value < .05.

   For pathway analysis, differential proteins were annotated using Gene
   Ontology (GO) information from Uniprot. Protein-protein interactions
   (PPI) among these differential proteins were predicted using the STRING
   database. Gene names were searched against Mus musculus with minimum
   required interaction score of high confidence (0.700). Metabolic
   pathway enrichment analysis was conducted using the Kyoto Encyclopedia
   of Genes and Genomes (KEGG).

Statistical analysis

   Data obtained from NOLT, NORT, the body weight and blood glucose,
   plasma insulin level, and HOMA-IR are presented as mean ± standard
   error of mean (SEM). The independent t-test was used to compare the
   means of parameters addressed between the 2 groups. Two-way analysis of
   variance (ANOVA) with repeated measures was used to analyze the body
   weight and the FBG at different time points by the GraphPad version 5.
   Shapiro-Wilk test was used to test the normality for data distribution
   and Levene test was used to test the homogeneity of variance. P
   value < .05 was considered as significant.

Results

The effect of diabetes mellitus on the body weight, FBG plasma insulin
levels, and insulin resistance index

   Two-way ANOVA with repeated measure for the body weight revealed a
   significance of time effect, F[(4,80)] = 3.395, P = .013, group effect,
   F[(1,80)] = 5.621, P = .028, and interaction between time and group
   effect, F[(4,80)] = 94.820, P < .001. A Bonferroni post hoc test
   revealed that after 4 weeks of HFD feeding when the injection of
   streptozotocin (STZ) was applied (W0), the diabetic group (DM) had a
   body weight significantly heavier than that of the normal control group
   (ND), t(20) = 4.317, P < .001. However, 4 weeks after STZ injection,
   the body weight of the DM group was significantly lower than that of
   the ND group, t(20) = 7.121, P < .001 ([62]Figure 1A). This result
   indicated that diabetes status induces body weight loss in this model.

Figure 1.

   Figure 1.
   [63]Open in a new tab

   The effects of diabetes mellitus on metabolic parameters and cognitive
   functions. Diabetes status induces body weight loss (A), increase in
   fasting blood glucose level (B), increase in plasma insulin level (C),
   augment HOMA-IR (D) and impaired cognitive function in novel object
   location recognition test (NOLT) (E) and affects cognitive function in
   novel object recognition test (NORT) (F). Normal control mice (ND)
   (n = 12), diabetic mice (DM) (n = 9-10). ***P < .001, **P < .01,
   *P < .05 when compared with the normal control group (ND), ###P < .001
   when compared with the chance level (50%) and ns = nonsignificant.

   We further investigated the effect of diabetic induction by a
   combination of HFD feeding and STZ injections on the FBG levels.
   Two-way ANOVA with repeated measure for the FBG revealed a significant
   time effect, F[(1,20)] = 8.291, P = .009, group effect,
   F[(1,20)] = 1003, P < .001, and interaction between time and group
   effect, F[(1,20)] = 12.290, P = .002. A Bonferroni post hoc test
   revealed that diabetic mice exhibited an increase in the FBG levels at
   both 1 week after STZ injection (W1), t(20) = 13.34, P < .001, and 4
   weeks after STZ injection (W4), t(20) = 19.35, P < .001 ([64]Figure
   1B). This result indicates that a combination of HFD feeding and STZ
   injections induces hyperglycemia in this model.

   We next investigated the effect of diabetic induction on plasma insulin
   level. The independent t-test revealed an elevation in plasma insulin
   levels of diabetic mice (DM) compared with the normal control group
   (ND), t(20) = 2.418, P = .025 ([65]Figure 1C). We further evaluated the
   insulin resistance in these mice and found that the diabetic group (DM)
   exhibited insulin resistance as indicated by the HOMA-IR value higher
   than 2.5. Moreover, the diabetic mice (DM) presented a significant
   increase in the HOMA-IR when compared with the normal control mice
   (ND), t(20) = 15.38, P < .001 ([66]Figure 1D). This result suggests
   that combination of HFD feeding and STZ injections induces
   hyperinsulinemia and insulin resistance in this model.

The effect of diabetes mellitus on cognitive functions

   The cognitive functions of these mice were evaluated in the NOLT and
   the NORT. In the NOLT, 1 sample t-test for the percentage of
   recognition index during test phase revealed that the normal control
   mice (ND) presented a good performance in object location memory
   retention, t(11) = 5.053, P < .001. In contrast, the diabetic mice (DM)
   did not show a preference for the object in the novel location,
   t(9) = 0.806, P = .441. Moreover, the percentage of recognition index
   of the diabetic mice (DM) was significantly lower than that of the
   normal control group (ND), t(20) = 3.290, P = .004 ([67]Figure 1E). The
   total exploration time of the objects during the training phase and the
   test phase was not different between group, t(20) = 0.513, P = .614 and
   t(20) = 0.268, P = .792 ([68]Supplementary material, Figure S1A and
   [69]S1B).

   We next evaluated the cognitive functions of these mice in the NORT.
   One sample t-test for the percentage of recognition index during the
   test phase revealed that the normal control mice (ND) presented a good
   performance in object recognition memory, t(11) = 6.095, P < .001,
   whereas the diabetic mice (DM) did not show preference for the novel
   object, t(8) = 1.583, P = .152. The percentage of recognition index in
   NORT significantly differed between 2 groups, t(19) = 3.602, P = .002
   ([70]Figure 1F). The total exploration time of the objects during the
   training phase and the test phase in NORT was not statistically
   different between groups, t(19) = 1.036, P = .313, and t(19) = 0.972,
   P = .343, respectively ([71]Supplementary material, Figure S1C and
   [72]S1D). This indicates that the diabetic mice exhibit deficits in
   both object location memory and object recognition memory.

   We further investigated the correlation between the FBG and the
   cognitive performance in NOLT and NORT tests. We found a negative
   correlation between FBG and the cognitive performance in NOLT
   (r = -0.53, P = .014) and a negative correlation between FBG and the
   cognitive performance in NORT (r = -0.53, P = .014) ([73]Supplementary
   material, Figure S2A and [74]S2B).

Proteomic results

   In this study, approximately 2430 hippocampal proteins were identified
   by Q Exactive Orbitrap LC-MS/MS. By differential analysis, 13 proteins
   were differentially expressed (adj. P value < .05; ±2-fold change). Of
   these, 9 proteins were upregulated while 4 proteins were downregulated
   in the hippocampus of diabetic mice as compared with the controls
   ([75]Figure 2). A list of differential proteins and detailed analysis
   of protein functions are shown in [76]Table 1. Upregulated proteins are
   composed of a variety of functions, including protein synthesis and
   folding (EIF5A, RSP24, and PPIB), endocytosis and retromer complex
   (VPS24, SNX12, and ARP2/3), cannabinoid receptor interacting (CRIP1),
   ubiquitination (SKP1) and oxidative stress response (NUDT3), while
   downregulated proteins are composed of proteins relating to
   mitochondria function (ANT1), neuronal development (ELP1), protein
   glycosylation (RPN2), and endocytosis (VPS4).

Figure 2.

   [77]Figure 2.
   [78]Open in a new tab

   Protein expression profile of T2DM mice as compared with the control.
   (A) Volcano plot of differential proteins (P > .05). The horizontal
   axis represents Log[2] fold-change, the vertical axis represents
   –Log[10] adjusted P value, and values closer to the center represents
   smaller differences. Upregulated and downregulated differential
   proteins are shown in red and green, respectively. (B) Expression
   levels of downregulated and upregulated proteins which were obtained
   from proteomic analysis.

Table 1.

   List of differential proteins (adjusted P value < .05; ±2-fold change)
   in hippocampus of type 2 diabetic mice with cognitive impairment as
   compared with the control.
   Protein IDs Protein name Protein function Gene name Peptides Unique
   peptides Log2 fold change Adj. P value
   Upregulation
   Protein synthesis and folding
   A0A0A0MQM0 Eukaryotic translation initiation factor 5A (Fragment)
   Translation elongation factor activity EIF5A 7 5 2.30 .0317
   [79]Q3TIF8 40S ribosomal protein S24 Small ribosomal subunit rRNA
   binding RPS24 2 2 1.79 .0413
   [80]P24369 Peptidyl-prolyl cis-trans isomerase B Assist protein folding
   PPIB 7 7 2.18 .0498
   Endocytosis and cellular trafficking
   [81]Q9QZ88 Vacuolar protein sorting-associated protein 29 Retromer
   complex binding VPS29 7 7 2.54 .0288
   [82]Q3TGS7 Sorting nexin-12 Intracellular trafficking SNX12 5 5 3.18
   .0282
   [83]Q9JM76 Actin-related protein 2/3 complex subunit 3 Formation of
   branched actin networks in the cytoplasm ARP2/3 4 4 2.63 .0485
   Cannabinoid receptor interacting
   [84]Q5M8N0 CB1 cannabinoid receptor-interacting protein 1 Suppresses
   cannabinoid receptor CNR1-mediated tonic inhibition of voltage-gated
   calcium channels CNRIP1 6 6 2.70 .0372
   Ubiquitination
   [85]Q9WTX5 S-phase kinase-associated protein 1 Protein ubiquitination
   SKP1 7 7 2.09 .0207
   Oxidative stress
   [86]Q9JI46 Diphosphoinositol polyphosphate phosphohydrolase 1 Oxidative
   stress may play a role in signal transduction NUDT3 4 4 1.11 .0443
   Downregulation
   Mitochondria dysfunction
   [87]P48962 ADP/ATP translocase 1 Plays a role in mitochondrial
   uncoupling by acting as a proton transporter SLC25A4 16 8 -1.03 .0003
   Neuron development
   [88]Q7TT37 Elongator complex protein 1 Migration and branching of
   projection neurons in the developing cerebral cortex ELP1, IKBKAP 3 3
   -1.43 .0195
   Protein glycosylation
   [89]Q3U505 Dolichyl-diphosphooligosaccharide–protein
   glycosyltransferase subunit 2 (Fragment) Protein glycosylation RPN2 6 6
   -1.53 .0022
   Endocytosis
   [90]Q3TN07 Vesicle-fusing ATPase Involved in late steps of the
   endosomal multivesicular bodies (MVB) pathway VPS4 2 2 -1.28 .0058
   [91]Open in a new tab

   Moreover, PPI enrichment results revealed that most differential
   proteins are involved in the ribosome pathway and endocytosis pathway
   ([92]Figure 3). KEGG pathway analysis also showed that many proteins
   involved in various steps of endocytosis are identified including
   clathrin-coated vesicle (ARP2/3), early endosome (VSP29, SNX12, and
   ARP2/3), and late endosome (VPS4) ([93]Figure 4).

Figure 3.

   [94]Figure 3.
   [95]Open in a new tab

   Enrichment PPI of differential proteins represents the majority of
   proteins related to the ribosome pathway (red circle) and endocytosis
   pathway (blue circle). (PPI enrichment P value = .00137).

Figure 4.

   [96]Figure 4.
   [97]Open in a new tab

   KEGG pathway maps. ARP2/3, VPS29, SNX3/12, and VPS4 were particularly
   mapped in the endocytosis pathway (Mus musculus).

Discussion

The effect of diabetes mellitus on the body weight, FBG plasma insulin
levels, and insulin sensitivity

   We found that the diabetic mice exhibited significant weight loss and
   hyperglycemia throughout the experiment. Our results were consistent
   with previous studies that demonstrated streptozotocin (STZ) treatment
   was sufficient to induce diabetes mellitus, characterized by
   significant weight loss and hyperglycemia.^[98]16,[99]25 Moreover, in
   our model, the diabetic mice induced by the combination of an HFD and
   STZ injection showed a significant increase in plasma insulin levels,
   which was similar to previous experiments.^ [100]26 Overall, these
   findings indicate that the combination of an HFD and STZ injection
   induces weight loss, hyperglycemia, hyperinsulinemia, and insulin
   resistance in mice.

The effect of diabetes mellitus on cognitive functions

   In this study, we demonstrated that diabetic mice exhibited cognitive
   deficits in object location recognition memory (NOLT) and in novel
   object recognition memory (NORT). These 2 types of memory are dependent
   on the hippocampus function.^ [101]27 Our finding is in accordance with
   previous studies showing that diabetes mellitus affected cognitive
   functions both in clinical studies^[102]28,[103]29 and in animal models
   of diabetes.^[104]15,[105]16,[106]30 This alteration in cognitive
   function was associated with changes in structure and function of the
   hippocampus, not only in humans,^[107]31,[108]32 but also in animal
   experiments of the diabetes model.^[109]33,[110]34 However, the
   molecular mechanisms underlying diabetes-induced cognitive impairments
   are not yet fully understood. Therefore, the GeLC shotgun proteomic
   approach was employed to investigate the proteome response in this
   context.

Proteomics

   In the current work, we demonstrated that T2DM affected cognitive
   functions evaluated by NOLT and NORT, which are hippocampal-dependent
   tasks. This opens the window for us to explore proteomic alteration of
   T2DM mice as compared with the controls to obtain a comprehensive view
   of the molecular mechanisms in the hippocampus underlying the cognitive
   deficits of T2DM.

   Based on the proteomic analysis, 13 differential proteins in the
   hippocampus are implicated in neurocognitive function in the context of
   T2DM. Seven of these proteins were previously reported to play an
   important role in neurocognitive function. These include Eukaryotic
   translation initiation factor 5A, Peptidyl-prolyl cis-trans isomerase
   B, Actin-related protein 2/3 complex subunit 3, CB1 cannabinoid
   receptor-interacting protein 1, S-phase kinase-associated protein 1,
   ADP/ATP translocase 1, and Elongator complex protein 1. However, the
   specific mechanisms of these proteins have never been explained in
   regard to cognitive impairment in T2DM context. The 5 novel proteins
   that are potentially associated with T2DM-induced cognitive dysfunction
   include 40S ribosomal protein S24, Vacuolar protein sorting-associated
   protein 29, Sorting nexin-12, Diphosphoinositol polyphosphate
   phosphohydrolase 1, Dolichyl-diphosphooligosaccharide-protein
   glycosyltransferase subunit 2, and Vesicle-fusing ATPase. All of these
   differential proteins are of particular interest as they will improve
   current knowledge of T2DM-associated cognitive dysfunction.

   The PPI enrichment results clearly demonstrated that the hippocampal
   proteins involved in the ribosome pathway (protein synthesis), and
   endocytosis pathway were among the most dysregulated in diabetic mice
   with cognitive impairment. Alteration of those proteins may
   theoretically affect the protein synthesis steps and molecular
   trafficking of cargo proteins within hippocampal neurons. In addition,
   some of the differential proteins were previously reported to be
   involved in neurodegenerative diseases.

Protein synthesis and folding

   In this study, proteins involved in protein synthesis pathway including
   40S ribosomal protein S24, (RSP24) and eukaryotic translation
   initiation factor 5A (EIF5A) were upregulated in T2DM mice with
   cognitive impairment. Moreover, a protein involved in the folding
   pathway, peptidyl-prolyl cis-trans isomerase B (cyclophilin B or PPIB),
   was also upregulated. These indicate that altered dynamics of protein
   production and modification affect hippocampal function.

   Upregulation of RSP24 and EIF5A also indicates altered protein
   production and quality control mechanisms of misfolded protein since
   RSP24 is involved in protein production and EIF5A is involved in
   protein synthesis and folding. A recent study of EIF5A in the mouse
   nervous system has shown that EIF5A is crucial for neurodevelopment and
   cognitive function. Knockout of EIF5A in the mouse nervous system led
   to impairments in growth, viability, neurodevelopment, and cognitive
   function in mice.^ [111]35 The evidence from the recent study showed
   that an increase of EIF5A is observed in the T2DM mice. Altogether, it
   revealed that EIF5A is important to neurocognitive function and an
   imbalance of EIF5A expression is involved in T2DM neurocognitive
   impairment.

   PPIB encodes a protein CypB, which is involved in the protein folding
   pathway. Our results showed that PPIB is upregulated in the hippocampus
   of diabetic mice. The CypB was shown to protect against Endoplasmic
   reticulum stress and play a role in synaptic transmission through
   synapsin signaling.^ [112]36 In addition, CypB was shown to be
   protective against amyloid beta (Aβ)-induced neurotoxicity through
   regulating ROS levels and activation of kinases phosphoinositide
   3-kinase and mitogen-activated protein kinase.^ [113]37 However, the
   function of PPIB regarding T2DM context and hippocampal function has
   not been reported. More research is needed to elucidate the precise the
   role of the hippocampal PPIB in cognitive function in this diabetic
   model.

Endocytosis and cellular trafficking

   Our findings highlight that proteins associated with endocytosis and
   the retromer complex pathway were among the most prominent groups in
   the context of diabetes-induced cognitive impairment in mice
   hippocampus ([114]Figure 4). The observed imbalance in these proteins
   such as Actin-related protein 2/3 complex subunit 3 (ARP2/3), Vacuolar
   protein sorting-associated protein 29 (VPS29), Sorting nexin 12
   (SNX12), and Vesicle-fusing ATPase (VPS4), indicates a disturbance of
   endocytosis, sorting and trafficking pathways. Specifically, the
   increase of ARP2/3, a key factor in dendritic spine plasticity in the
   hippocampus of diabetic mice suggests induction of the endocytosis
   pathway.^[115]38,[116]39 This increase signifies a potential
   augmentation in the dynamics of endocytosis within the neural context.
   Moreover, the increase of VPS29, a component of the retromer complex
   and SNX12, a component of sorting nexin, indicate a potential cascade
   of events. This includes the continuous fusion of endocytosis vesicles
   with lysosomes and the directed trafficking of cargo to specific
   cellular destinations.^ [117]40 These observed alterations suggest a
   complex interaction of molecular events and show that endocytosis and
   intracellular trafficking are involved in diabetic mice hippocampus
   dysfunction.

   VPS4 is a member of the ATPases, which play an important role in the
   lysosomal degradation pathway and intracellular protein trafficking. A
   recent study showed that VPS4B is mainly located in the cytoplasm of
   neurons and may play an important role in the adult rat hippocampus
   after middle cerebral artery occlusion (MCAO).^ [118]41 However, the
   role of VSP4 related to cognitive function induced by diabetes mellitus
   has never been reported. Our study introduces VSP4 as a novel player in
   the context of diabetes-induced cognitive dysfunction, warranting
   exploration of its specific involvement in this context.

Cannabinoid receptor interacting protein

   In this study, we found that the protein expression of cannabinoid
   receptor interacting protein 1 (CRIP1) was upregulated in the
   hippocampus of diabetic mice. This is the first study to reveal
   alteration of CRIP1 expression in the hippocampus of diabetic mice with
   memory deficits.

   CRIP1 is an important CB1 cannabinoid receptor-associated protein and
   it was shown that CRIP1 negatively modulates CB1 receptor function.^
   [119]42 In the hippocampus, CB1 receptor and CRIP1 co-expressed in both
   glutamatergic and GABAergic neurons.^ [120]43 However, the largest
   expression of CB1 receptor resides in GABAergic interneurons.^ [121]44
   Previous experiments demonstrated that episodic-like memory processes
   and synaptic plasticity are controlled by hippocampal CB1
   receptor.^[122]45,[123]46 In addition, it has been shown that CRIP1
   modulated CB1 receptor signaling in the hippocampus.^ [124]43
   Downregulation of CRIP1 protein expression in the hippocampus induced
   by subchronic pyridoxin treatment was associated with an enhancement of
   object recognition memory.^ [125]47 Furthermore, knockdown of CRIP1
   expression in the medial prefrontal cortex rescued cognitive
   flexibility in long-term estrogen-deprived female mice.^ [126]48 Taken
   together, these findings indicate that upregulation of hippocampal
   CRIP1 expression may participate in diabetes-induced cognitive
   deficits.

Ubiquitination

   S-phase kinase-associated protein 1 (SKP1) is a core component of an E3
   ubiquitin ligase (SCF complex) that is involved in protein
   ubiquitination. The ubiquitination of proteins is an important part of
   several biological processes including the regulation of protein
   homeostasis, autophagy, mitochondrial function, endoplasmic reticulum
   homeostasis, DNA repair, and cell cycle regulation.^ [127]49 Recent
   studies reported the involvement of SCF complex in neurodegenerative
   disease^ [128]50 via modulation of the expression of tau and Aβ, 2
   proteins which are involved in the pathogenesis of AD,^ [129]51
   alteration of mitochondria function,^ [130]52 and induction of
   neuroinflammation.^ [131]53 This finding suggests that diabetes
   mellitus may enhance the ubiquitination of hippocampal proteins, which
   in turn, may lead to the impairment of memory.

Mitochondrial dysfunction

   A detailed analysis of protein function and disease relevance revealed
   that ADP/ATP translocase 1 (ANT1) was downregulated in diabetic mice.
   ANT1, a protein in AD pathway (KEEG ID mmu05010) and Parkinson disease
   pathway (KEGG ID mmu05012), is involved in the exchange of cytosolic
   adenosine diphosphate (ADP) and mitochondrial adenosine triphosphate
   (ATP). ADP/ATP exchange by ANT1 is vital for the maintenance of ATP
   synthase activity, thus its deficiency may contribute to the
   mitochondrial pathogenesis.^ [132]54 It was proposed that diabetes
   mellitus is associated with mitochondrial dysfunction and mitochondrial
   dysfunctions have been identified as a key factor in cognitive
   impairment in diabetes mellitus.^[133]55,[134]56 A previous study
   reported the downregulation of ANT1 expression in the mouse brain,
   including the midbrain, the striatum, the brainstem, and the
   hippocampus, of a Parkinson disease model.^ [135]49 However, the direct
   implications of the hippocampal ANT1 in cognitive deficits induced by
   diabetes mellitus are unknown. Further study is needed to unravel the
   causal link between the cognitive impairments and the hippocampal ANT1
   in the context of diabetes mellitus.

Oxidative stress response

   In this study, we found that the protein expression of
   diphosphoinositol polyphosphate phosphohydrolase I (NUDT3) was
   upregulated in the hippocampus of diabetic mice. It was reported that
   Nud3, a subfamily of diphosphoinositol-pyrophosphate-polyphosphatases
   (DIPPs), is involved in the degradation of inositol pyrophosphates.
   Specifically, it has been demonstrated that NUDT3 plays a vital role in
   limiting DNA damage and maintaining cell survival upon the oxidative
   stress in the presence of Zinc (Zn^2 ^+).^ [136]57 It is well-known
   that hyperglycemia and diabetes mellitus promotes oxidative stress
   which is linked to cognitive deficits observed in both human and animal
   studies.^[137]58,[138]59 It seems that the upregulation of hippocampal
   NUDT3 in diabetic mice is the protective mechanism of neurons to
   prevent cellular damage caused by the DM-induced oxidative stress.

Neuronal development

   Our results also suggest that the development of hippocampal neurons
   might be disrupted via the downregulation of elongator complex protein
   1 (ELP1). Loss of ELP1 function has been linked to neurodevelopment
   disruption.^ [139]60 This may result in structural change of the
   hippocampus. ELP1 was identified as a noncatalytic scaffolding protein
   within the heterohexameric transcriptional elongator complex.^ [140]61
   Transcriptional factor function of ELP1 was proposed to be involved in
   alpha-tubulin acetylation which affects the regulation of cortical
   neuron migration and differentiation.^ [141]62 Recently, it has been
   further shown that ELP1 protein is localized in the cytoplasm of cells,
   where it is essential for retrograde nerve growth factor (NGF)
   signaling and neuron target tissue innervation and survival.^ [142]63

Protein glycosylation

   Dolichyl-diphosphooligosaccharide-protein glycosyltransferase subunit 2
   (RPN2) is an integral membrane protein found only in the rough ER.^
   [143]64 RPN2 is involved in protein glycosylation by catalyzing the
   transfer of an oligosaccharide from a lipid donor to a nascent protein
   during protein synthesis. This modification is important for protein
   folding, stability, and function. However, the importance of RPN2 in
   the hippocampus of T2DM mice regarding cognitive function is unknown.
   Since RPN2 plays an important role in protein glycosylation and quality
   control, regulation of RPN2 may be critical for the proper function of
   glycoprotein production in hippocampal neurons, with downregulation
   leading to misfolding and malfunctioning of glycoproteins in ER lumen.
   The alterations of the protein synthesis, folding and modification
   pathway suggest that they may play an important role in the
   pathophysiology of cognitive impairment in diabetic mice.

   Proteomic results in the context of T2DM-associated cognitive
   impairment have been previously reported as mentioned
   earlier.^[144]12-[145]14 In this study, we found 9 upregulated proteins
   which are involved in protein synthesis and folding (EIF5A, RSP24, and
   PPIB), endocytosis and trafficking (VPS24, SNX12, and ARP2/3),
   cannabinoid receptor interacting (CRIP1), ubiquitination (SKP1) and
   oxidative stress response (NUDT3), and 4 downregulated proteins which
   are involved in mitochondria function (ANT1), neuronal development
   (ELP1), protein glycosylation (RPN2), and endocytosis (VPS4).
   Downregulation of proteins with known function implicated in the
   arrangement of myelin sheath and morphology of axons (MBP), the
   structure of astrocytes (GFAP), and response to oxidative stress and
   inflammation (Apo-AI) have been reported in the hippocampus of
   spontaneous diabetic Zucker fatty rats when compared with the control
   littermates.^ [146]14 Also, significant alterations in hippocampal
   protein expressions involved in cytoskeletal regulation, oxidoreductase
   activity, protein deubiquitination, energy metabolism, GTPase
   activation, heme binding, hydrolase activity, iron storage,
   neurotransmitter release, protease inhibitor, and transcription factor
   have been observed in STZ/HFD induced-diabetic mouse model.^ [147]12
   Changes in protein expression with the functional role in the response
   to oxidative stress, energy metabolism and neuronal transport, which
   were associated with cognitive decline, have been found in the
   hippocampus of STZ/HFD induced-diabetic rat model.^ [148]13 To the best
   of our knowledge, our study is the first to show that T2DM-associated
   cognitive dysfunction is linked to alterations in protein synthesis,
   endocytosis and cellular trafficking pathways in the hippocampus.

   The differential proteins found in the hippocampus of T2DM mice in our
   study differ from those found in previous works. Several factors could
   explain this discrepancy in results obtained including the differences
   in technique used to analyze the proteomic profile, animal model of
   T2DM, and the onset age and duration of diabetes between studies.

   In this study, we employed the GeLC shotgun proteomic approach, whereas
   the previous study used 2DE-based proteomic approaches. Each technique
   may have different limitations, which can lead to different results.
   The GeLC shotgun proteomic approach is effective in obtaining a broad
   overview of proteome alteration, but offers semiquantitative data.
   Therefore, gel-free label-based quantitative proteomic approaches such
   as iTRAQ should be considered for future studies to confirm these
   findings.

   Regarding the diabetic animal model, the diabetic model in our study
   was induced by the combination of long-term HFD feeding and 2
   intraperitoneal injections of STZ at the dose of 100 mg/kg in mice,
   while the diabetic mouse model in the study of Matsuura et al^ [149]12
   was induced by intraperitoneal injections of STZ at the dose of
   50 mg/kg for 5 consecutive days. Shi et al^ [150]13 used the diabetic
   rat model by feeding rats with HFD for 4 weeks followed by a single
   injection of STZ (30 mg/kg, i.p.). Nam et al^ [151]14 investigated the
   hippocampal proteome profile in the diabetic Zucker fatty (ZDF) rats
   that develop advanced insulin resistance and hyperglycemia due to a
   genetic mutation of the leptin receptor gene. These diabetes model
   differences between studies may affect the type and timing of protein
   expression differences.

   In this study for the proteomic analysis, only 4 mice per group were
   used for the analysis and this could be considered a limitation of
   sample size. Further studies with a larger sample size are needed to
   confirm the hippocampal proteome results of T2DM mice.

   It is noted that all 5 novel proteins identified in this study
   including 40S ribosomal protein S24, Vacuolar protein
   sorting-associated protein 29, Sorting nexin-12, Diphosphoinositol
   polyphosphate phosphohydrolase 1,
   Dolichyl-diphosphooligosaccharide-protein glycosyltransferase subunit
   2, and Vesicle-fusing ATPase, have not been previously reported in the
   context of T2DM-associated cognitive dysfunction. These proteins are of
   particular interest as potential drug targets for the treatment of
   T2DM-associated cognitive dysfunction. However, further studies are
   required to validate their roles, as well as to investigate the causal
   relationships between these proteins and T2DM-associated cognitive
   dysfunction.

Conclusion

   Overall, this study provides a new insight into the molecular
   mechanisms underlying the impact of T2DM on hippocampal function and
   cognitive impairment. Utilizing the GeLC shotgun proteomic approach, we
   have discovered additional molecular pathways that have not been
   previously reported in this context. Our findings suggest that the
   alteration in hippocampus proteome of T2DM mice may contribute to
   cognitive dysfunction and provide potential therapeutic targets for
   T2DM-associated neurodegenerative disorders. These findings enhance the
   understanding of the pathophysiology of T2DM-related cognitive
   impairment and identify molecular targets for drug development to treat
   T2DM-associated cognitive impairment conditions.

Supplemental Material

   sj-docx-1-bbi-10.1177_11779322241306290 – Supplemental material for The
   Alteration of Proteomic Profiles in Hippocampus of Type 2 Diabetic Mice
   Associated With Cognitive Impairment
   [152]sj-docx-1-bbi-10.1177_11779322241306290.docx^ (141.4KB, docx)

   Supplemental material, sj-docx-1-bbi-10.1177_11779322241306290 for The
   Alteration of Proteomic Profiles in Hippocampus of Type 2 Diabetic Mice
   Associated With Cognitive Impairment by Yoottana Janthakhin, Sirikran
   Juntapremjit, Karin Hummel, Ebrahim Razzazi-Fazeli and Sutin Kingtong
   in Bioinformatics and Biology Insights

Acknowledgments