Abstract

Background

   Cognitive dysfunction in diabetes significantly impairs quality of
   life, yet effective therapies remain limited. Our previous work showed
   that α-methyltryptophan (α-MT), an inhibitor of the amino acid
   transporter SLC6A14 and indoleamine 2,3-dioxygenase 1 (IDO1),
   ameliorates diabetic nephropathy by modulating renal metabolism. Given
   its role in metabolic regulation, we hypothesized that α-MT may also
   protect against cognitive decline in diabetes by influencing gut-brain
   metabolic crosstalk.

Methods

   Db/db mice were used as an in vivo model of type 2 diabetes,
   complemented by high-glucose–treated Caco-2 cells in vitro. Cognitive
   function was assessed using the Morris water maze. Gene expression
   related to synaptic plasticity (c-FOS, ARC, BDNF, EGR1) was measured by
   RT-qPCR. Colon morphology and SLC6A14 expression were evaluated by H&E,
   AB-PAS, immunohistochemistry, and Western blotting. ¹H NMR–based
   metabolomics was applied to profile metabolic changes in the colon and
   hippocampus.

Results

   α-MT treatment significantly reduced hyperglycemia, restored intestinal
   barrier integrity, and improved cognitive performance in db/db mice. It
   also downregulated SLC6A14 overexpression under hyperglycemic
   conditions in both models. Metabolomic analysis revealed that α-MT
   induced significant metabolic reprogramming in the colon, with amino
   acid metabolism as the most affected pathway. Notably, metabolite
   alterations in the colon were positively correlated with those in the
   hippocampus, and both were negatively associated with increased
   expression of c-FOS, ARC, BDNF, and EGR1, suggesting coordinated
   gut-brain metabolic responses.

Conclusions

   These findings indicate that α-MT alleviates diabetes-associated
   cognitive impairment by promoting gut-brain metabolic remodeling.
   Targeting intestinal amino acid transport may represent a promising
   nutritional and therapeutic strategy for neuroprotection in diabetic
   encephalopathy (DE).

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12986-025-01024-w.

   Keywords: Cognitive dysfunction, Diabetes, SLC6A14, α-MT, Metabolite
   profiles

Background

   Diabetic encephalopathy (DE) is a prevalent central nervous system
   complication of diabetes, characterized by progressive cognitive
   decline that significantly impairs quality of life [[44]1, [45]2].
   Epidemiological evidence suggests that individuals with both type 1
   diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM) exhibit
   mild-to-moderate declines in cognitive function compared to
   non-diabetic controls [[46]3]. While glycemic control remains a
   cornerstone in the management of diabetes and its associated
   complications, it may be insufficient to fully prevent or reverse
   cognitive decline in DE [[47]4]. Although some therapeutic strategies
   used for Alzheimer’s disease (AD) - a neurodegenerative disorder that
   shares overlapping pathogenic mechanisms with DE - have been explored
   in DE, their efficacy in modifying disease progression remains limited
   [[48]5, [49]6]. Therefore, novel therapeutic approaches specifically
   targeting the underlying mechanisms of DE are urgently needed.

   The gut-brain axis has emerged as a key regulator of cognitive
   function, with gut-derived metabolites increasingly recognized as
   critical mediators of brain health [[50]7, [51]8]. Metabolites such as
   short-chain fatty acids (SCFAs) [[52]9–[53]11], produced or modulated
   by the intestinal microbiota, can influence neuroinflammation, synaptic
   plasticity, and neuronal metabolism via neural, immune, and endocrine
   pathways. In diabetes, dysbiosis and intestinal barrier dysfunction
   alter the colonic metabolic environment, and accumulating evidence
   links these changes to hippocampal impairment and cognitive decline.
   For instance, reduced acetate levels have been shown to decrease
   hippocampal synaptophysin and accelerate cognitive deficits in type 1
   diabetes [[54]12], while elevated gut-derived glutamate disrupts
   hippocampal signaling and correlates with cognitive dysfunction in type
   2 diabetes [[55]13].These findings highlight the potential of targeting
   intestinal metabolism to modulate central nervous system outcomes.

   α-Methyltryptophan (α-MT) is a promising candidate for targeting
   gut-brain metabolic axis dysfunction in diabetes-associated cognitive
   impairment. α-MT is a well-established inhibitor of SLC6A14 (ATB⁰^,⁺),
   a broad-specificity amino acid transporter that mediates the uptake of
   nearly all amino acids except glutamate and aspartate [[56]14–[57]16].
   This transporter is expressed in colon, where L-type enteroendocrine
   cells that secrete GLP-1, peptide YY, and pancreatic polypeptides are
   present [[58]17]. By modulating intestinal amino acid uptake, α-MT may
   reshape colonic metabolic output and systemic amino acid availability -
   key factors in gut-brain communication. Dysregulated amino acid
   homeostasis in diabetes may contribute to central nervous system
   dysfunction, and targeting intestinal transport offers a strategy to
   indirectly influence brain metabolism. Additionally, α-MT also inhibits
   indoleamine-2,3-dioxygenase-1 (IDO1) [[59]18], an enzyme catalyzing the
   initial step in tryptophan degradation via the kynurenine pathway.
   Increased IDO1 activity elevates kynurenine metabolites, which are
   implicated in neuroinflammation, oxidative stress, and neurotoxicity
   [[60]19–[61]21]. While both mechanisms may contribute to
   neuroprotection, the gut-restricted expression of SLC6A14 makes it a
   compelling target for probing gut-initiated metabolic influences on
   cognition. Notably, α-MT has demonstrated beneficial effects in
   diabetic nephropathy through modulation of renal and urinary
   metabolism, with modest improvements in glycemic control [[62]22],
   underscoring its broader metabolic regulatory properties. Preclinical
   studies further support its safety, with no significant adverse effects
   reported on behavior, pharmacokinetics, or organ function
   [[63]23–[64]25]. Together, these properties justify its investigation
   as a modulator of gut-brain metabolic crosstalk in diabetic cognitive
   dysfunction.

   Despite these pharmacological properties, a critical knowledge gap
   remains: whether α-MT–induced changes in colonic metabolism translate
   into coordinated metabolic remodeling in the brain and ultimately
   improve cognitive function in diabetes. The hippocampus, a brain region
   essential for spatial learning, memory formation, and integration
   [[65]26], is particularly vulnerable in DE. To address this gap, we
   employed ¹H NMR–based metabolomics to comprehensively profile
   metabolite signatures in both the colon and hippocampus of db/db mice.
   We hypothesized that α-MT treatment would induce metabolic
   reprogramming in the gut that correlates with improvements in
   hippocampal biochemistry and cognitive behavior, thereby providing
   mechanistic insights into the gut-brain metabolic crosstalk underlying
   DE.

methods

Cell cultures and treatments

   The human colonic epithelial cell line Caco-2 was cultured in DMEM
   supplemented with 10% FBS and 1% penicillin/streptomycin. For
   experiments, cells were treated for 24 h under the following
   conditions: (I) Control (5.5 mM glucose); (II) High glucose (HG, 25 mM
   glucose); (III) Control + 1.25 mM α-MT; (IV) Control + 2.5 mM α-MT; (V)
   HG + 1.25 mM α-MT; (VI) HG + 2.5 mM α-MT. The high-glucose treatment
   conditions were established based on previously reported studies
   [[66]27–[67]29].

Animals and experimental design

   Eight-week-old male db/db mice (C57BLKS/J-Lepr^db/db) and their
   wild-type (WT) littermates (C57BLKS/J-Lepr+/+) were obtained from Model
   Animal Research Center of Nanjing University. Mice were housed under
   specific pathogen-free (SPF) conditions with a 12-hour light/dark cycle
   and ad libitum access to food and water. All experimental procedures
   were approved by the Institutional Animal Care and Use Committee of
   Wenzhou Medical University (Approval No.: xmsq2022-1008).

   After one week of acclimatization, mice were randomly assigned to four
   groups (n = 7/group): (1) WT + vehicle (0.9% saline, 0.2 mL, i.p.); (2)
   WT + α-MT (DL-α-methyltryptophan, 2.5 mg per mouse in 0.2 mL saline,
   i.p.); (3) db/db + vehicle; (4) db/db + α-MT. The DL-form of α-MT (CAS
   No.: 153-91−3; C₁₂H₁₄N₂O₂: 218.25 g/mol; Shenzhen Qimeike Biotechnology
   Co., Ltd.) was administered every 48 h for 8 weeks. This treatment
   duration was designed to cover the critical period of
   diabetes-associated cognitive decline in db/db mice, which typically
   occurs between 16 and 20 weeks of age [[68]30, [69]31]. The dose and
   regimen were selected based on previous studies demonstrating its
   efficacy in modulating glucose metabolism without overt toxicity
   [[70]22]. Injections were performed between 9:00 AM and 11:00 AM to
   minimize circadian variation. We designed and illustrated a schematic
   diagram of the experimental workflow, as shown in Fig. [71]1A.

Fig. 1.

   [72]Fig. 1
   [73]Open in a new tab

   α-MT improves glucose homeostasis in db/db mice (n = 7). (A)
   Experimental Design Arrow Diagram. (B) Random blood glucose level; (C)
   IPGTT; (D): ITT. The treatment α-MT (i.p., 2.5 mg every other day)
   began at 9 weeks of age and was continued until 17 weeks of age.
   **p < 0.01, and ***p < 0.001 versus WT mice; ^###p < 0.001versus db/db
   mice. Statistical analysis: two-way repeated-measures ANOVA with
   Bonferroni post hoc tests

Biochemical analysis

   Body weight and random blood glucose were monitored weekly.
   Intraperitoneal glucose tolerance tests (IPGTT) and insulin tolerance
   tests (ITT) were performed at week 17 after 4–6 h of fasting. For
   IPGTT, mice were injected with glucose (2 g/kg, i.p.), and blood
   glucose was measured at 0, 30, 60, and 120 min. For ITT, mice received
   insulin (0.75–1.0 U/kg, i.p.), and blood glucose was monitored at the
   same time point.

Morris water maze test

   Spatial learning and memory were assessed using the Morris water maze
   as previously described [[74]32]. The assay was conducted in a circular
   pool (120 cm diameter) filled with opaque water (22 ± 1 °C), with a
   hidden platform (10 cm diameter) submerged 1 cm below the surface in a
   fixed quadrant. Mice underwent four training days with four trials per
   day. Each trial had a maximum duration of 60 s, during which mice were
   allowed to locate the platform. If unsuccessful, they were guided to
   the platform and allowed to remain for 15 s. Inter-trial intervals were
   approximately 15 min. Starting positions varied pseudorandomly across
   trials to avoid bias. On the fifth day, a probe trial was conducted
   with the platform removed; mice were released from a novel starting
   point and allowed to swim freely for 60 s. Performance was
   video-tracked and analyzed using EthoVision XT (Noldus Information
   Technology). Key metrics included escape latency, path efficiency, time
   spent in the target quadrant, and number of platform crossings. The
   experimenter was blinded to genotype and treatment groups throughout
   all behavioral procedures.

Tissue collection and histopathology

   Mice were anesthetized with isoflurane. For histology, a subset of mice
   (n = 3/group) underwent transcardial perfusion with saline followed by
   4% paraformaldehyde (PFA). Colon and brain tissues were post-fixed,
   paraffin-embedded, and sectioned for staining with Hematoxylin and
   Eosin (H&E), Alcian Blue/Periodic Acid-Schiff (AB-PAS), and
   immunohistochemistry (IHC) for ZO-1 and Occludin. Histopathological
   scoring of colon sections was performed by a blinded investigator using
   a established system evaluating inflammatory infiltration, tissue
   damage, crypt architecture, and goblet cell depletion (scores 0–3 per
   parameter, total score 0–12) [[75]33]. For molecular and metabolomic
   analyses, tissues from non-perfused mice (n = 4/group) were rapidly
   dissected, flash-frozen in liquid nitrogen, and stored at −80 °C.

Metabolites extraction and metabolomic analysis

   Metabolites were extracted from colon and hippocampus tissues (~ 20 mg)
   using a methanol-chloroform-water method [[76]34]. After homogenization
   and centrifugation (10,000 g, 4 °C), supernatants were lyophilized and
   reconstituted in 0.6 mL D₂O containing 0.05% TSP-d4. ^1H NMR spectra
   were acquired at 298 K on a Bruker Avance III 600 MHz spectrometer with
   a 5 mm TXI probe. Hippocampus and colon samples were analyzed using
   zgpr and cpmgpr1d sequences, respectively, with 256 scans, 64,000 data
   points, a spectral width of 12,000 Hz, and a relaxation delay of 6 s.
   Spectra were processed in TopSpin 3.0: phased manually,
   baseline-corrected, and referenced to lactate (δ 1.31). Residual water
   (δ 4.70–5.00) and noisy regions (δ < 0.5, δ >9.5) were excluded. Data
   were binned (0.01 ppm for multivariate analysis; 0.0015 ppm for
   quantification), normalized to total spectral sum, and analyzed using
   PLS-DA in SIMCA-P 12.0. Metabolite concentrations were normalized to
   tissue weight and expressed relative to TSP in relative units (r.u.).
   MetaboAnalyst 6.0 ([77]https://www.metaboanalyst.ca/) was used to
   perform metabolic pathway analysis, integrating both pathway enrichment
   analysis and pathway topology analysis.

RNA extraction and RT-qPCR analysis

   Total RNA was extracted from Caco-2 cells, colon and hippocampal
   tissues using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according
   to the manufacturer’s instructions [[78]35]. RNA quality was assessed
   by measuring A260/A280 ratios (>1.8) on a NanoDrop spectrophotometer.
   cDNA was synthesized from 2 µg of total RNA using the GoScript Reverse
   Transcription System (Promega) with oligo(dT) primers. Real-time
   quantitative PCR (RT-qPCR) was performed using SYBR Green Master Mix
   (Promega) on a CFX Connect Real-Time PCR Detection System (Bio-Rad).
   Each 20 µL reaction contained 0.2 µM of each primer and 2 µL of cDNA
   template. Primer efficiencies were validated using standard curves from
   serial cDNA dilutions. Relative gene expression was calculated using
   the 2⁻^ΔΔCt method with normalization to β-actin. All primer sequences
   (designed via NCBI Primer-BLAST and synthesized by Sangon Biotech) are
   provided in Table [79]1.

Table 1.

   Primer sequences for RT-qPCR
   Gene    Forward primer 5′−3′   Reverse primer 5′−3′
   SLC6A14 CAAATCGTCTGGCAAGGTGG   GCAACTAAGCCACCCCAA
   c-FOS   AGGGGCAAAGTAGAGCAGCTA  CAATCTCAGTCTGCAACGCA
   EGR1    AACCGGCCCAGCAAGACACC   GGGCACAGGGGATGGGAATG
   ARC     ACCATATGACCACCGGCGGC   TCCAGCATCTCAGCTCGGCAC
   BDNF    CCATAAGGACGCGGACTTGTAC AGACATGTTTGCGGCATCCAGG
   β-actin CCTCACTGTCCACCTTCC     GGGTGTAAAACGCAGCTC
   [80]Open in a new tab

Western blot analysis

   Protein lysates were prepared from colon tissues using RIPA lysis
   buffer supplemented with protease and phosphatase inhibitors. Total
   protein concentration was determined using a BCA assay, and 30 µg of
   protein per sample was separated by 10% SDS-PAGE. Proteins were
   transferred to a PVDF membrane and incubated overnight at 4 °C with a
   primary antibody against SLC6A14 (ab316193, Abcam). After washing with
   TBST, the membrane was incubated with an HRP-conjugated secondary
   antibody (ab6721, Abcam) for 2 h at room temperature. Immunoreactive
   bands were visualized using Pierce ECL substrate and analyzed with
   ImageQuant 5.2 software. β-actin (4970 S, Cell Signaling Technology)
   was used as a loading control.

Statistical analysis

   Data are presented as mean ± standard deviation (SD) or median with
   interquartile range, as appropriate. Normality and homogeneity of
   variances were assessed using Shapiro-Wilk and Levene’s tests,
   respectively. Data meeting assumptions were analyzed by two-way ANOVA
   (factors: genotype and treatment) or two-way repeated-measures ANOVA
   (for longitudinal data), followed by Bonferroni’s post hoc test.
   Non-parametric data were analyzed by Kruskal-Wallis test with
   Bonferroni correction. Correlation analyses were performed using
   Pearson’s correlation with FDR correction. Statistical significance was
   set at p < 0.05. All analyses were performed using SPSS 19.0.

Results

α-MT improves glucose homeostasis in db/db mice

   α-MT treatment significantly influenced various metabolic parameters,
   with both main effects and interaction effects observed over time
   (Table [81]S1, Additional file 1). Specifically, α-MT significantly
   reduced random blood glucose levels in db/db mice compared to untreated
   diabetic controls at both 13 and 17 weeks (Fig. [82]1B and 13 weeks:
   p < 0.001, Cohen’s d = 6.955; 17 weeks: p < 0.001, Cohen’s d = 6.881).
   No significant changes were observed in WT mice, indicating that α-MT
   selectively targets hyperglycemia in diabetic animals. α-MT also
   improved glucose tolerance in db/db mice, as evidenced by IPGTT. At all
   tested time points (30, 60, and 120 min), α-MT-treated diabetic mice
   exhibited significantly lower blood glucose levels compared to
   untreated controls (Fig. [83]1C; all p < 0.001, Cohen’s d values
   ranging from 11.136 to 15.86). Similarly, ITT showed enhanced insulin
   sensitivity in α-MT-treated db/db mice at 30, 60, and 120 min
   (Fig. [84]1D; all p < 0.001, Cohen’s d values ranging from 5.230 to
   8.077). These findings indicate that α-MT effectively ameliorates
   hyperglycemia and improves glucose metabolism in db/db mice. The
   selective improvement in diabetic animals suggests that α-MT may target
   specific pathways involved in diabetes-associated metabolic
   dysregulation.

α-MT ameliorated cognitive function deficits in db/db mice

   To assess the impact of α-MT on cognitive function in db/db diabetic
   mice, we employed Morris water maze tests and RT-qPCR analysis. The
   swimming trajectories during the water maze test (Fig. [85]2A) revealed
   that db/db mice exhibited disorganized and less efficient paths
   compared to WT controls, indicative of cognitive deficits. Notably,
   α-MT treatment significantly improved the spatial navigation abilities
   of db/db mice, as evidenced by more organized and efficient motor
   trajectories. Significant effects of diabetes, time, and a three-way
   interaction (diabetes × treatment × time) on escape latency were
   observed (Table [86]S1, Additional file 1). Post hoc comparisons
   (Fig. [87]2B) further revealed that on day 4 of training, α-MT-treated
   db/db mice had significantly shorter escape latencies compared to
   untreated db/db mice (p < 0.001, Cohen’s d = 5.544), approaching levels
   similar to WT mice (p = 0.688, Cohen’s d=−0.875). This indicates that
   α-MT effectively ameliorates spatial learning deficits in db/db mice.
   In the platform crossing test assessing spatial memory retention, a
   significant interaction between diabetes and treatment was observed
   (p = 0.034, partial η² = 0.175, Table S2, Additional file 1). Simple
   effects analysis confirmed that α-MT specifically enhanced performance
   in db/db mice, with treated db/db mice exhibiting more platform
   crossings compared to untreated controls (p = 0.007, Cohen’s d=−1.562,
   Fig. [88]2C). No such effect was observed in WT mice.

Fig. 2.

   [89]Fig. 2
   [90]Open in a new tab

   α-MT improves cognitive function deficits in db/db mice. (A): Swimming
   trajectory in the water maze (n = 7); (B): Escape latency period
   (n = 7) (C): Number of platforms crossing (n = 7); mRNA levels for
   c-FOS (D), EGR1 (E), BDNF (F), and ARC (G) in hippocampus (n = 4). (H):
   The HE staining image of the hippocampus and a magnified image of the
   CA1 region. NG: neuronal degeneration. Statistical analysis:
   repeated-measures ANOVA (B), two-way ANOVA (C–G), with Bonferroni’s
   post hoc test

   Further analysis of hippocampal gene expression related to cognitive
   function provided insights into the molecular mechanisms underlying
   these improvements. Diabetes and its interaction with α-MT treatment
   had significant effects on key genes associated with neuroplasticity,
   including c-FOS, EGR1, BDNF, and ARC (Table S2, Additional file 1).
   Specifically, α-MT significantly upregulated c-FOS (p = 0.001, Cohen’s
   d=−3.109, Fig. [91]2D), EGR1 (p < 0.001, Cohen’s d=−5.615,
   Fig. [92]2E), BDNF (p < 0.001, Cohen’s d=−3.737, Fig. [93]2F), and ARC
   (p < 0.001, Cohen’s d=−3.739, Fig. [94]2G) expression in db/db mice,
   without affecting WT mice. Histological examination via HE staining of
   the hippocampus (Fig. [95]2H) revealed gliosis and neuronal deformities
   in the CA1 region of db/db mice (indicated by arrows in the figure).
   Following α-MT treatment, both gliosis and neuronal morphology showed
   significant improvement. In summary, our results demonstrate that α-MT
   effectively improves spatial learning and memory retention in db/db
   diabetic mice, as evidenced by reduced escape latency and increased
   platform crossings. The improvement in cognitive function may involve
   the upregulation of neurotrophic factors and early-response genes in
   the hippocampus.

α-MT protects against colonic barrier damage and inhibits the expression of
SLC6A14 colon of db/db mice

   Histological analysis revealed severe colonic damage in db/db mice,
   including submucosal edema, crypt loss, reduced goblet cells, and
   inflammatory infiltration (Fig. [96]3A–B). α-MT treatment markedly
   attenuated these abnormalities, as reflected in significantly improved
   histological scores (Fig. [97]3H). Goblet cell density and mucus
   production, assessed by AB-PAS staining, were also reduced in db/db
   mice but restored by α-MT (Fig. [98]3C–D, I). Statistical analysis
   confirmed significant main effects of diabetes and treatment, as well
   as a significant interaction (all p < 0.01, partial η² >0.50, Table S2,
   Additional file 1), indicating that α-MT specifically ameliorated
   structural damage in diabetic mice.

Fig. 3.

   [99]Fig. 3
   [100]Open in a new tab

   α-MT ameliorates colon damage and reduces the expression of SLC6A14 in
   db/db mice. (A) Representative H&E staining images of the colon and (B)
   a magnified view of the local area. a: submucosal edema; b: crypt
   injury; c: loss of goblet cells; d: inflammatory cell infiltration. (C)
   Alcian Blue–Periodic Acid-Schiff (AB-PAS) staining images of the colon
   and (D) a magnified view of the local area. (E) Representative
   immunohistochemical (IHC) staining images of ZO-1 and (F) Occludin in
   the colon (n = 4). (G) Expression levels of SLC6A14 protein in the
   colon. (H) Histological scores of the colon (n = 4). Quantitative
   analysis of (I) AB-PAS-positive area (n = 4), (J) ZO-1-positive area
   (n = 4), and (K) Occludin-positive area (n = 4). (L) Quantification of
   western blot analysis of SLC6A14 (n = 3). (N) Expression of SLC6A14
   mRNA in the colon (n = 3). (M) Expression of SLC6A14 mRNA in Caco-2
   cells (n = 4). Note: Con: Normal medium; Con-1.25 α-MT: Normal medium
   with 1.25 mmol/L α-MT; Con-2.5 α-MT: Normal medium with 2.5 mmol/L
   α-MT; HG (high glucose medium): Normal medium with 25 mmol/L glucose;
   HG-1.25 α-MT: High glucose medium with 1.25 mmol/L α-MT; HG-2.5 α-MT:
   High glucose medium with 2.5 mmol/L α-MT. Statistical analysis:
   Kruskal-Wallis H tests with Bonferroni correction (L-M), two-way ANOVA
   with Bonferroni’s post hoc test (H-K)

   IHC showed decreased expression of tight junction proteins ZO-1 and
   Occludin in db/db mice, which was reversed by α-MT treatment
   (Fig. [101]3E–F, J–K). Significant interactions between diabetes and
   treatment were observed for both proteins (p < 0.001, partial η² >0.70,
   Table S2, Additional file 1), supporting a protective effect of α-MT on
   intestinal barrier integrity. Given that α-MT inhibits the amino acid
   transporter SLC6A14, we examined its expression in the colon. Both
   protein and mRNA levels of SLC6A14 were elevated in db/db mice and
   significantly reduced by α-MT (Fig. [102]3G, L, N; p < 0.05, H > 9.0).
   In vitro, high-glucose conditions upregulated SLC6A14 mRNA in Caco-2
   cells, an effect significantly suppressed by α-MT (Fig. [103]3M;
   p < 0.05, H = 16.322). Collectively, these results demonstrate that
   α-MT alleviates diabetes-induced colonic barrier dysfunction,
   associated with downregulation of SLC6A14 and restoration of tight
   junction and mucosal integrity.

α-MT treatment reverses diabetes-induced alterations in colon and hippocampal
metabolite profiles

   To investigate the impact of diabetes and α-MT treatment on metabolic
   profiles, we performed ^1H-NMR analysis of colon and hippocampus
   tissues from WT and db/db mice, with or without α-MT treatment.
   Representative spectra are shown in Fig. [104]4A and B. A total of 37
   metabolites were identified, including short-chain fatty acids
   (acetate, propionate, butyrate, formate), energy-related compounds
   (glucose, UDP-glucose, lactate, fumarate, succinate, creatine, AMP,
   IMP, nicotinamide), amino acids (alanine, glycine, taurine, glutamate,
   glutamine, aspartate, leucine, isoleucine, valine, methionine,
   phenylalanine, tyrosine, histidine, GABA, 4-aminobutyrate,
   N-acetylglutamine, 3-hydroxyisovalerate), choline and its derivatives
   (choline, phosphocholine, dimethylamine, myo-inositol), and purines and
   pyrimidines (xanthine, adenine, hypoxanthine, inosine, uracil, uridine,
   pyridoxine). Notably, similar metabolic pathways were altered in both
   colon and hippocampus, suggesting a potential gut–brain axis
   involvement in diabetes-related metabolic dysregulation.

Fig. 4.

   [105]Fig. 4
   [106]Open in a new tab

   Metabolic analysis of the colon and hippocampus of WT and db/db mice
   with and without α-MT treatment. (A-B): Typical 600 MHz ^1H NMR spectra
   for metabolites from colon (A) and hippocampus (B) of db/db mice.
   (C-H): The changes in metabolic phenotypes observed in the colon and
   hippocampus of WT and db/db mice after treatment with α-MT. The PLS-DA
   score plot was used to identify the variations in the colon metabolomic
   profile among the four groups (C), between WT and db/db mice (D), and
   between db/db and α-MT-treated db/db mice (E). The PLS-DA score plot in
   the hippocampal metabolomic profile among four groups (F), between WT
   and db/db mice (G), and between db/db and α-MT-treated db/db mice (H).
   1, Leucine; 2, Valine; 3, Isoleucine; 4, Propionate; 5, Ethanol; 6,
   Alanine; 7, Butyrate; 8, Acetate; 9, Glutamate; 10, Glutamine; 11,
   Methionine; 12, Aspartate; 13, Trimethylamine (TMA); 14, Creatine; 15,
   Ethanolamine; 16, Choline; 17, Phosphocholine (PC); 18, Taurine; 19,
   Isocitrate; 20, Glycine; 21, Myo-inositol; 22, Lactate; 23,
   3-Hydroxyisovalerate (3-HV); 24, Glucose; 25, Uracil; 26, Uridine; 27,
   UDP-Glucose; 28, Fumarate; 29, Tyrosine; 30, Phenylalanine; 31,
   Inosine; 32, Pyridoxine; 33, Xanthine; 34, Adenine; 35, Histidine; 36,
   Inosine monophosphate (IMP); 37, Formate; 38, Adenosine monophosphate
   (AMP); 39, Niacinamide; 40, N-Acetylglutamine; 41, 4-Aminobutyrate; 42,
   Dimethylamine 43. Hypoxanthine. Note: hippo: hippocampus

   The PLS-DA model was used to determine changes in the metabolite
   patterns of the WT, db/db, and α-MT-treated groups. Score plots based
   on the colon and hippocampus metabolomes (Fig. [107]4C and F) clearly
   showed differences between the WT, db/db, and α-MT-treated db/db
   groups. Notably, there was a certain degree of overlap in the
   distribution of the WT and α-MT-treated WT groups in the PLS-DA score
   map, suggesting that α-MT does not cause significant metabolite profile
   alterations in mice under normal physiological conditions. The
   metabolite profile of the α-MT-treated db/db group was closer to that
   of the WT group than that of the db/db group, suggesting that α-MT
   treatment reverses the diabetes-induced changes in colon and
   hippocampal metabolite profiles. Subsequently, we performed PLS-DA
   analysis by comparing the db/db group with the WT group, and the
   db/db + α-MT group with the WT group. Significant differences were
   observed between the db/db and WT groups in both colon (Fig. [108]4D)
   and hippocampal (Fig. [109]4G) metabolomes. Additionally, distinct
   metabolic profiles were detected between the db/db and α-MT-treated
   db/db groups in both colon (Fig. [110]4E) and hippocampus
   (Fig. [111]4H). These results indicate that α-MT treatment ameliorates
   the metabolic profiles in the colon and hippocampus induced by
   diabetes, affecting multiple metabolic pathways.

Amino acid metabolism is a key target of α-MT in diabetic colon and
hippocampus

   To further investigate the key metabolic pathways modulated by α-MT in
   diabetic mice, we analyzed metabolite profiles in the colon and
   hippocampus. Heatmap visualization revealed widespread elevations of
   multiple metabolites in both colon (Fig. [112]5A) and hippocampus
   (Fig. [113]5B) of db/db mice compared to WT. These aberrant metabolite
   levels were broadly attenuated following α-MT treatment, suggesting a
   significant metabolic regulatory effect of α-MT.

Fig. 5.

   [114]Fig. 5
   [115]Open in a new tab

   Heatmap for variations in metabolites after α-MT treatment. The heatmap
   for the colon (A) and hippocampus (B) of the four groups of mice (WT,
   WT + α-MT, db/db, db/db + α-MT). The shades of red (color scale > 0)
   represent a relative higher concentration of metabolites, while the
   shades of blue (color scale < 0) indicate a relative lower
   concentration of metabolites. Abbreviation: 3-HV: 3-Hydroxyisovalerate;
   AMP, Adenosine monophosphate; IMP: Inosine monophosphate; DMA,
   Dimethylamine

   Pathway enrichment analysis was subsequently performed. In the colon
   (Fig. [116]6A), key differential metabolites included taurine,
   phenylalanine, tyrosine, alanine, aspartate, glutamate, and
   glutamine—all of which are associated with amino acid metabolism. In
   the hippocampus (Fig. [117]6B), key metabolites comprised nicotinamide,
   fumarate, alanine, glutamate, glutamine, 4-aminobutyrate, glycine,
   choline, creatine, and taurine. Notably, amino acid metabolism emerged
   as a significantly enriched pathway in both tissues, indicating it as a
   central node targeted by α-MT (Fig. [118]6A–B).

Fig. 6.

   [119]Fig. 6
   [120]Open in a new tab

   Identification of Key Metabolic Pathways. Metabolic pathways in the
   colon (A) and hippocampus (B) were identified using MetaboAnalyst,
   based on p-values from pathway enrichment analysis and pathway impact
   values from pathway topology analysis. (C) Metabolic network diagram of
   selected metabolites from the colon and hippocampus as analyzed by
   Metabo Analyst. Bars with black borders represent metabolites
   identified in the colon; bars with yellow borders represent metabolites
   identified in the hippocampus. Black font indicates metabolites
   specific to the colon; Yellow font indicates metabolites specific to
   the hippocampus. Red font indicates metabolites present in both the
   hippocampus and the colon. Statistical notations: *p < 0.05,
   **p < 0.01, ***p < 0.001 versus WT mice; ^#p < 0.05, ^##p < 0.01,
   ^###p < 0.001 versus db/db mice. Abbreviation: GABA, γ-aminobutyric
   acid. Statistical analysis: Kruskal-Wallis H tests with Bonferroni
   correction (C)

   To integrate and visualize the metabolic crosstalk between gut and
   brain, we constructed a network mapping of the key metabolites
   identified in both colon and hippocampus (Fig. [121]6C). All
   metabolites in this network—except aspartate—showed an overall
   p-value < 0.05, with H-values ranging from 7.691 to 11.868. Compared to
   WT mice, all selected key metabolites (except aspartate) were
   significantly elevated in db/db mice (all p < 0.05) and markedly
   reduced following α-MT intervention (all p < 0.05). Furthermore,
   network analysis revealed that taurine, alanine, glutamate, and
   glutamine were overlapping key metabolites shared between colon and
   hippocampus, whose levels were effectively normalized by α-MT
   treatment. These findings suggest that amino acid metabolism in the
   colon and hippocampus may contribute to the beneficial effects of α-MT
   in DE.

Gut–brain metabolic coordination is evident in diabetic mice

   To investigate functional coupling between the gut and brain in
   diabetes, we performed Pearson correlation analysis between key
   metabolites in the colon and hippocampus (Fig. [122]7A). A widespread
   positive correlation was observed, indicating coordinated metabolic
   dysregulation across the gut–brain axis. Notably, colonic taurine—a
   metabolite elevated in db/db mice and normalized by α-MT - showed
   strongest associations with hippocampal metabolism, including glycine
   (r = 0.81, p < 0.001), niacinamide (r = 0.80, p < 0.001), and fumarate
   (r = 0.81, p = 0.001). It also correlated significantly with multiple
   other hippocampal metabolites (alanine, glutamate, GABA, creatine;
   r = 0.70–0.77, p < 0.01). Other colonic amino acids, including alanine,
   tyrosine, and phenylalanine, similarly showed broad positive
   correlations with hippocampal metabolites involved in neurotransmission
   and energy metabolism (e.g., glutamine, fumarate, niacinamide;
   r = 0.70–0.79, p < 0.01), suggesting a systemic gut–brain metabolic
   linkage in diabetes.

Fig. 7.

   [123]Fig. 7
   [124]Open in a new tab

   Pearson correlation analysis involving colonic and hippocampal
   metabolites and cognitive-related genes. (A) Heatmap of correlations
   between colonic metabolites (columns) and hippocampal metabolites
   (rows). (B) Heatmap of correlations between colonic metabolites
   (columns) and gene expression (c-FOS, EGR1, BDNF, ARC; rows). (C)
   Heatmap of correlations between hippocampal metabolites (columns) and
   gene expression (c-FOS, EGR1, BDNF, ARC; rows). Cell values represent
   Pearson correlation coefficients (range: −1 [blue] to 1 [red]);
   asterisks denote statistical significance (*p < 0.05, **p < 0.01,
   ***p < 0.001)

Metabolic dysregulation is inversely associated with cognitive gene
expression

   We next examined the relationship between metabolite levels and
   hippocampal expression of neuroplasticity-related genes (BDNF, c-FOS,
   EGR1, ARC). Both colonic and hippocampal metabolites showed
   predominantly negative correlations with these cognition-associated
   transcripts (Fig. [125]7B–C). In the colon (Fig. [126]7B), glutamine,
   taurine, and phenylalanine were strongly and negatively correlated with
   cognitive gene expression. Glutamine exhibited the most consistent
   associations (r = − 0.71 to − 0.75, p < 0.01 across all four genes),
   while taurine and phenylalanine showed moderate-to-strong negative
   correlations (r = − 0.52 to − 0.74, p < 0.05). In the hippocampus
   (Fig. [127]7C), the negative associations were even more pronounced.
   BDNF expression was widely and strongly inversely correlated with
   metabolites such as alanine (r = − 0.85), glycine (r = − 0.82),
   glutamate (r = − 0.84), GABA (r = − 0.81), and choline (r = − 0.83; all
   p < 0.001). Similarly, ARC and c-FOS showed significant negative
   correlations with multiple key metabolites, reinforcing the link
   between metabolic imbalance and impaired synaptic plasticity. EGR1
   exhibited moderate negative correlations with key hippocampal
   metabolites, excluding glutamine, with Pearson’s r values between − 0.5
   and − 0.6, all statistically significant (p < 0.05). Taken together,
   the widespread negative correlations between gut and hippocampal
   metabolite levels and cognition-associated gene expression suggest that
   modulating gut–brain metabolism may enhance the expression of
   neuroplasticity-related genes and improve cognitive function.

Proposed mechanism of α-MT in ameliorating cognitive dysfunction

   Collectively, these findings support a gut–brain metabolic axis in
   diabetic cognitive impairment (summarized in Fig. [128]8). α-MT
   treatment downregulates colonic SLC6A14, reducing excessive amino acid
   uptake and restoring colonic metabolic homeostasis. This gut-level
   correction is positively correlated with improved metabolic profiles in
   the hippocampus, particularly in pathways linked to amino acids,
   energy, choline and nicotinamide metabolism. In turn, the normalization
   of these brain metabolites is inversely associated with the expression
   of neuroplasticity markers (BDNF, c-FOS, EGR1, ARC), which are
   upregulated following α-MT treatment. These data suggest that α-MT may
   indirectly support cognitive function through modulation of gut-brain
   metabolic crosstalk, although direct causal mechanisms require further
   investigation.

Fig. 8.

   [129]Fig. 8
   [130]Open in a new tab

   Potential mechanisms by which α-MT ameliorates diabetes-induced
   cognitive decline in mice

Discussion

   Diabetes mellitus is a complex metabolic disorder characterized by
   chronic hyperglycemia due to impaired insulin secretion or action.
   Chronic glucose dysregulation extends beyond peripheral complications,
   contributing to DE and impairments in learning, memory, and executive
   function. These cognitive deficits significantly impair quality of
   life, highlighting the need for neuroprotective strategies.
   Diabetes-associated cognitive decline involves multiple mechanisms,
   including metabolic dysregulation, inflammation, oxidative stress, and
   mitochondrial dysfunction [[131]36, [132]37]. Hyperglycemia promotes
   the formation of advanced glycation end products (AGEs), activation of
   protein kinase C (PKC), and oxidative stress, leading to synaptic
   dysfunction and neuronal damage [[133]38]. Emerging evidence suggests
   gut barrier disruption in diabetes may allow microbial products like
   LPS to enter circulation, contributing to systemic inflammation and CNS
   effects through gut-brain interactions, independent of glycemic control
   [[134]39, [135]40]. Given that α-MT ameliorates diabetic nephropathy
   through metabolic modulation [[136]21], we investigated its potential
   neuroprotective effects in DE.

   Our findings suggest that α-MT improves cognitive function through
   mechanisms largely independent of direct glycemic control. Cognitive
   improvements, assessed by Morris water maze performance and
   memory-related gene expression (BDNF, c-FOS, ARC, EGR1), were more
   pronounced than changes in blood glucose levels. Metabolomic analyses
   identified amino acid metabolism as the most significantly altered
   pathway in both the colon and hippocampus, with additional changes in
   energy, choline, and nicotinamide metabolism in the brain. The absence
   of glycolytic or glucose-related pathways among the significantly
   enriched pathways supports that α-MT’s effects are not primarily
   mediated by glucose modulation. α-MT also improved intestinal
   morphology and barrier integrity in diabetic mice. Positive
   correlations between colonic and hippocampal amino acid levels, and
   their association with memory-related gene expression, suggest that
   α-MT may influence brain function through systemic metabolic changes,
   potentially involving gut-brain interactions.

   α-MT attenuated the high-glucose-induced upregulation of SLC6A14 in the
   colon, and this attenuation was associated with improved colonic and
   hippocampal amino acid metabolism in db/db mice, suggesting a systemic
   metabolic effect. Amino acids are critical for neurotransmission,
   energy metabolism, and neuroplasticity, all of which support cognitive
   function [[137]41, [138]42]. We observed significant dysregulation of
   hippocampal amino acids—including glutamate, glutamine, alanine,
   glycine, GABA, and taurine—in db/db mice, which were partially
   normalized by α-MT treatment. Glycine, a neurotransmitter and
   neuromodulator involved in neuroprotection and metabolism, plays an
   important role in cognitive function, and its dysregulation has been
   linked to cognitive impairment [[139]43, [140]44]. Glutamate and GABA
   are the primary excitatory and inhibitory neurotransmitters,
   respectively, and their metabolic coupling with glutamine is essential
   for cognitive function; disruption of this cycle is associated with
   cognitive impairment in diabetic models [[141]45, [142]46], a
   phenomenon already substantiated in mouse models of diabetes-associated
   cognitive dysfunction. Taurine modulates the glutamatergic system and
   synaptic plasticity, and its dysregulation may contribute to learning
   and memory deficits [[143]47].

   Notably, the concentrations of most amino acid metabolites in the
   hippocampus showed a significant negative correlation with the
   expression levels of genes such as BDNF, c-FOS, ARC, and EGR1. BDNF, a
   key regulator of synaptic plasticity [[144]48], was downregulated in
   db/db mice and negatively correlated with elevated glutamate and
   glutamine levels, suggesting that metabolic disturbances may impair
   neurotrophic signaling. c-FOS, a marker of neuronal activation
   [[145]49], was downregulated in diabetic mice and was upregulated by
   α-MT treatment. ARC and EGR1, which regulate synaptic remodeling and
   memory [[146]50, [147]51], were also dysregulated and negatively
   correlated with amino acid levels, reinforcing the link between
   metabolic imbalance and cognitive dysfunction. In summary, hippocampal
   amino acid dysregulation may contribute to diabetes-associated
   cognitive dysfunction, and α-MT appears to alleviate these deficits by
   restoring metabolic balance and supporting the expression of
   neuroplasticity-related genes.

   Beyond amino acid dysregulation, impaired energy metabolism also
   contributes to cognitive deficits in diabetic mice. The brain primarily
   depends on glucose, which is metabolized via glycolysis and oxidative
   phosphorylation to produce ATP. Efficient energy metabolism supports
   cognitive functions such as attention, memory, and learning [[148]52].
   Cognitive disorders including Alzheimer’s disease [[149]53] and
   vascular dementia [[150]54] are associated with impaired brain energy
   metabolism. These include reduced glucose uptake, insulin resistance,
   and mitochondrial dysfunction, which limit energy production and
   increase oxidative stress. Such disturbances are linked to neuronal
   dysfunction, synaptic loss, and neurodegenerative pathology,
   contributing to cognitive decline [[151]55]. We found elevated
   hippocampal levels of fumarate and creatine in db/db mice, which were
   significantly reduced by α-MT treatment. Fumarate, a TCA cycle
   intermediate, plays a role in energy production and redox homeostasis.
   Elevated fumarate is associated with mitochondrial dysfunction and
   oxidative stress, impairing neuronal energy metabolism [[152]56]. The
   reduction of fumarate by α-MT may reflect improved TCA cycle function
   and reduced metabolic stress in the diabetic hippocampus.

   Creatine supports rapid energy buffering through phosphocreatine (PCr),
   especially during high-energy processes like synaptic transmission
   [[153]57, [154]58]. The PCr/creatine system acts as a transient ATP
   buffer, helping maintain energy homeostasis in the brain, particularly
   in the hippocampus. In db/db mice, elevated hippocampal creatine was
   not linked to improved cognition and was modestly associated with
   reduced neuronal activity. Elevated creatine levels may not be
   functionally beneficial and could instead indicate metabolic
   dysregulation in the diabetic brain. α-MT treatment significantly
   reduced creatine levels, normalizing them toward control values; this
   restoration of metabolic balance may support improved synaptic function
   in the hippocampus. The functional significance of elevated creatine in
   db/db mice remains unclear and warrants further investigation to
   determine whether it is compensatory, neutral, or detrimental. We also
   observed negative correlations between hippocampal energy metabolites
   and the expression of the synaptic plasticity genes, suggesting a link
   between metabolic dysregulation and impaired neuronal plasticity.

   Nicotinamide, a precursor of NAD+, participates in redox reactions, DNA
   repair, and cellular signaling. In this study, hippocampal nicotinamide
   levels were significantly higher in db/db mice than in controls, and
   α-MT treatment significantly reduced these levels, suggesting improved
   NAD⁺ metabolism and redox homeostasis. Correlation analysis revealed
   positive associations between hippocampal nicotinamide levels and
   colonic concentrations of taurine, phenylalanine, tyrosine, and
   alanine—amino acids involved in neuroprotection, neurotransmission, and
   energy metabolism [[155]59–[156]61]. Additionally, hippocampal
   nicotinamide levels were inversely correlated with BDNF and c-FOS
   expression, indicating a potential link between elevated nicotinamide
   and impaired synaptic plasticity. Thus, the reduction in nicotinamide
   following α-MT treatment might contribute to improved gene expression
   and cognition, but this causal relationship requires experimental
   validation.

   Choline is a precursor of acetylcholine, a key neurotransmitter in
   cognitive processes, and is particularly important in the hippocampus
   [[157]62]. Choline metabolism interacts with glutamatergic and
   GABAergic systems, contributing to excitatory-inhibitory balance in the
   brain. Dysregulation of choline metabolism disrupt this balance and
   impair cognitive function [[158]63]. Hippocampal levels of choline and
   related metabolites were significantly altered in db/db mice, with
   choline, phosphorylcholine, and dimethylamine all significantly
   elevated. α-MT treatment significantly reduced these elevated
   metabolite levels, normalizing them toward control values. Notably, we
   observed a significant negative correlation between hippocampal choline
   levels and the expression of neuroplasticity-related genes. The
   positive correlation between hippocampal choline and colonic taurine
   suggests a potential gut-brain metabolic link and reflects systemic
   metabolic changes influencing brain function. Furthermore, hippocampal
   choline levels were negatively correlated with BDNF and c-FOS
   expression, indicating that elevated choline may be associated with
   impaired synaptic plasticity. Collectively, these results suggest that
   dysregulated hippocampal choline metabolism may contribute to cognitive
   impairment in diabetic mice.

Conclusion

   Our metabolomic analysis indicates that α-MT is associated with
   improved cognitive performance in diabetic mice. α-MT reduced the
   overexpression of the colonic amino acid transporter SLC6A14, which was
   linked to improved gut barrier integrity and normalized colonic amino
   acid metabolism. These peripheral changes correlated with improved
   hippocampal metabolism across multiple pathways, consistent with a
   potential gut-brain interaction. Metabolic improvements were associated
   with partial restoration of synaptic plasticity gene expression (BDNF,
   ARC, EGR1, C-FOS) and cognitive function. Although these findings
   suggest systemic effects of α-MT, the underlying causal mechanisms
   require further investigation. This study supports the exploration of
   intestinal nutrient metabolism as a potential target for addressing DE.

Limitations and future perspectives

   First, the study used only male mice, had a limited sample size, and an
   8-week intervention period. Future studies should include both sexes,
   larger cohorts, and longer durations to enhance generalizability.
   Second, correlations do not establish causation; thus, mechanistic
   studies—such as gut microbiota profiling or SLC6A14 knockout models—are
   required to validate the gut-brain metabolic link. Finally, the
   pharmacokinetics, safety, and human translatability of α-MT remain
   poorly characterized. Further research is needed to assess its clinical
   potential and the translatability of α-MT and related analogs.

Supplementary Information

   [159]Supplementary Material 1.^ (12.9KB, xlsx)

Acknowledgements