Preventive carnosine supplementation attenuates painful diabetic neuropathy via restoring mitochondrial function and suppressing inflammasome activation

Abstract

BACKGROUND

Painful diabetic neuropathy (PDN) is a common and debilitating complication of diabetes mellitus characterized by chronic pain resulting from peripheral nerve damage. Although PDN substantially impairs quality of life, its underlying mechanisms remain incompletely understood.

AIM

To investigate whether preventive carnosine supplementation attenuates PDN-related abnormalities and is associated with alterations in mitochondrial homeostasis and inflammasome-related signaling.

METHODS

This study used untargeted metabolomics to analyze dorsal root ganglia (DRG) from rat model of type 2 diabetic neuropathy. Transmission electron microscopy, JC-1 staining, whole-cell patch-clamp recordings, motor and sensory nerve conduction velocity measurements, and molecular analyses were used to evaluate mitochondrial function, neuronal excitability, peripheral nerve function, and inflammasome-related signaling in vivo. Complementary mechanistic studies were performed in high-glucose-stimulated PC12 cells with dynamin-related protein 1 (Drp1) knockdown or overexpression.

RESULTS

In the DRG of rats with PDN, mitochondrial ultrastructural abnormalities, altered expression of fission/fusion-related proteins, and reduced mitochondrial membrane potential were identified. Untargeted metabolomic analysis identified 147 upregulated and 211 downregulated metabolites, among which carnosine was significantly reduced. Preventive carnosine supplementation attenuated mechanical allodynia, thermal hyperalgesia, neuronal hyperexcitability, and peripheral nerve conduction deficits while improving mitochondrial abnormalities. In DRG tissues, carnosine treatment was associated with reduced Drp1 expression and NOD-like receptor pyrin domain-containing protein 3 (NLRP3) inflammasome activation. In high-glucose-stimulated PC12 cells, carnosine similarly ameliorated mitochondrial dysfunction and suppressed Drp1-related mitochondrial fission and NLRP3 inflammasome signaling.

CONCLUSION

Preventive carnosine supplementation attenuated PDN-related behavioral, electrophysiological, and mitochondrial abnormalities, and was associated with PDN and was associated with reduced Drp1- and NLRP3-related signaling in the DRG. These findings support further preclinical evaluation of carnosine as a potential early-intervention strategy for PDN.

Key Words: Painful diabetic neuropathy; Mitochondrial fission; Dynamin-related protein 1; NOD-like receptor pyrin domain-containing protein 3 inflammasome; Carnosine

Core Tip: Painful diabetic neuropathy (PDN) is characterized by impaired mitochondrial function and reduced carnosine levels in the dorsal root ganglia. Untargeted metabolomic analysis identified carnosine deficiency as a metabolic alteration associated with PDN. Preventive carnosine supplementation alleviated behavioral, electrophysiological, and mitochondrial abnormalities in diabetic rats and was associated with reduced dynamin-related protein 1-related mitochondrial fission and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation in dorsal root ganglia. These findings were further supported by complementary in vitro experiments, collectively supporting further preclinical and translational evaluation of carnosine as a potential early-intervention strategy for PDN.

INTRODUCTION

Painful diabetic neuropathy (PDN) is a prevalent and debilitating complication of diabetes mellitus (DM), which is characterized by chronic pain arising from peripheral nerve damage. It affects up to 51% of diabetics[1], significantly impairing their quality of life through sleep disruption, mood disturbances, and physical limitations[2], highlighting its clinical and societal burden. Despite its clinical burden, effective therapies remain limited, largely due to an incomplete understanding of its underlying mechanisms.

Emerging evidence demonstrates that mitochondrial dysfunction and neuroinflammation play pivotal roles in peripheral neuropathy[3,4]. Our previous study revealed that mitochondrial dysfunction contributes to the development of PDN[5]. The maintenance of normal mitochondrial dynamics is critical for preserving neuronal function[6,7]. Dysregulated mitochondrial fission, mediated by proteins, such as dynamin-related protein 1 (Drp1), has been implicated in a variety of neurodegenerative[8] and inflammatory conditions[9]. Aberrant mitochondrial dynamics, particularly excessive fission regulated by Drp1, has been linked to impaired neuronal bioenergetics and reactive oxygen species (ROS) accumulation[10,11].

Drp1-mediated mitochondrial fission has been shown to activate the NOD-like receptor pyrin domain-containing protein 3 (NLRP3) inflammasome, recognized as a key driver of neuroinflammation through the maturation of pro-inflammatory cytokines, such as interleukin-1β (IL-1β)[12,13]. This Drp1-NLRP3 axis has been implicated in various conditions, including myocardial injury[14], hepatic steatosis[15], acute ischemic stroke[16], and Alzheimer’s disease[17], while its role in PDN remains incompletely understood.

Carnosine, a naturally occurring dipeptide composed of β-alanine and L-histidine, exhibits potent antioxidant and anti-inflammatory activities[18,19]. This antioxidative action is complemented by its anti-inflammatory effects[19-21], which are crucial in mitigating the progression of chronic diseases[21,22]. Importantly, it has also been shown to regulate mitochondrial function and energy metabolism[23,24]. However, whether carnosine can modulate the Drp1-NLRP3 signaling axis in PDN has not been investigated.

In this study, untargeted metabolomic profiling of dorsal root ganglion (DRG) tissues from rats with PDN identified a significant reduction in carnosine levels. Based on this finding, it was hypothesized that preventive carnosine supplementation could attenuate PDN-associated abnormalities and be associated with improved mitochondrial homeostasis and reduced inflammasome-related signaling. Accordingly, the present study combined untargeted metabolomics with in vivo and in vitro functional analyses to investigate the potential role of carnosine in PDN and its association with Drp1-related mitochondrial dynamics and NLRP3 inflammasome activation.

MATERIALS AND METHODS

Animals

Healthy male Sprague Dawley rats (153-191 g, 6-7 weeks) were purchased from the Department of Experimental Animal Sciences, Peking University Health Science Center (Beijing, China). These rats were acclimated to their environment for 1 week before the experiment. They were housed at a temperature of 24 ± 2 °C and a humidity of 55% ± 5%, and under a 12-hour light/dark cycle, with free access to food and water. The study was approved by the Institutional Animal Care and Use Committee under the Medical Science Research Ethics Committee of Peking University Third Hospital, approval No. 2022-111-01 and performed according to the ARRIVE guidelines.

Induction of type 2 DM

Rats were randomly assigned to either the diabetes group or the control group. Consistent with previously published methods[25], rats were subjected to 8 weeks of high-fat diet-feeding using a diet containing 45% fat, 20% protein, and 35% carbohydrate (catalog No. D12451, Research Diets, New Brunswick, NJ, United States). Prior to the induction of diabetes, rats were fasted for 12 h and then injected with streptozotocin (STZ, Sigma-Aldrich, St Louis, MO, United States) intraperitoneally at a dose of 30 mg/kg. Blood glucose level was monitored weekly using a glucometer (Roche Holding AG, Basel, Switzerland), and rats with fasting glucose level > 11.1 mmol/L were considered diabetic[25]. In parallel, rats in the control group were fed standard chow for 8 weeks and administered an equivalent volume of vehicle, consisting of 0.1 mmol/L citrate/citric acid buffer (C1013, Solarbio, Beijing, China; pH 4.5).

Administration of carnosine

Given the significantly reduced levels of carnosine detected in the DRG of PDN rats (refer to the Results section), carnosine supplementation was performed as a treatment intervention. Carnosine (Sigma-Aldrich) was dissolved in saline and administered intraperitoneally at a dosage of 250 mg/kg/day. This dosing regimen was selected based on prior studies demonstrating its efficacy in rodent models of type 2 DM[26] and high-fat diet-induced obesity[27]. Carnosine treatment was initiated during the high-fat diet period, followed by STZ administration according to the same schedule as that used for the PDN group.

Intraperitoneal glucose tolerance test and insulin tolerance test

According to a previous study[5], on day 7 post-STZ injection, an intraperitoneal glucose tolerance test was performed by administering 50% glucose (2 g/kg) after 12-hour fasting. On day 14, an insulin tolerance test was conducted by intraperitoneal administration of regular insulin at 0.5 U/kg (Novo Nordisk, Denmark). For both tests, blood glucose level was measured at 0 minute, 30 minutes, 60 minutes, 90 minutes, and 120 minutes post-injection.

Assessment of pain behaviors

Pain behaviors, including mechanical and thermal pain, were assessed at the beginning of the experiment (baseline), 1 day before STZ injection, and on days 7, 14, 21, 28, and 35 following STZ injection. All the tests were performed in the same quiet room in a blinded fashion to ensure the objectivity of the measurements and the repeatability of the data.

Mechanical allodynia: As previously described, mechanical allodynia was assessed using the paw withdrawal threshold (PWT) test, during which a 50% PWT was employed to evaluate pain perception[5]. To determine the response to sets of von Frey filaments with respect to the 50% PWT, the “up and down” testing method was conducted[28]. Rats were placed individually in transparent plastic cages with metal mesh floors, and eight von Frey filaments with logarithmically graded bending forces were selected for testing. At the beginning of each trial, a 2-g von Frey filament was applied perpendicularly to the mid-plantar surface of the hind paw of each rat. A sudden paw withdrawal during stimulation or immediately after filament removal was recorded as a positive response.

Thermal hyperalgesia: In a previous study, the perception of thermal pain was quantified using paw withdrawal latency (PWL) following a thermal stimulus[29]. Rats were placed in transparent acrylic cages and allowed at least 30 minutes for adaptation to this environment. A light source was thereafter perpendicularly directed onto the plantar surface of the hind paw of each rat. When a rat lifted its leg to avoid the heat stimulus, the thermal stimulus was terminated, and the latency was recorded as the PWL. Each rat underwent three PWL measurements and the mean latency was calculated.

Measurement of the motor nerve conduction velocity and sensory nerve conduction velocity

At week 3 post-STZ injection, motor nerve conduction velocity (MNCV) of the sciatic nerve and sensory nerve conduction velocity (SNCV) of the digital nerve were measured following previously described methods in the literature[30], and the results were reported in meters per second.

Tissue sampling

Rats were anesthetized with sodium pentobarbital (40 mg/kg, intraperitoneal; Sigma-Aldrich) and subsequently euthanized. Bilateral L4/5 DRG samples were promptly dissected and collected for downstream experiments. Samples designated for western blotting (WB) were snap-frozen in liquid nitrogen and stored at -80 °C until further processing. For immunofluorescence imaging, rats were perfused through the ascending aorta with cold saline, followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4, 4 °C), as described in a previous study[31].

Transmission electron microscopy

DRG samples were fixed in 2.5% glutaraldehyde for 24 hours to preserve cellular structures. Following fixation, samples were embedded in Epon resin, and ultrathin sections of approximately 70 nm thickness were prepared. These sections were stained with uranyl acetate and lead citrate to enhance ultrastructural contrast. The ultrastructure was examined using a JEM-1400 transmission electron microscope (JEOL, Tokyo, Japan) at a magnification of 10000 ×. Images were captured for detailed analysis, and nerve fiber ultrastructure was assessed in three randomly selected fields per sample. Mitochondrial morphology was quantified by sampling multiple mitochondria from each animal, in which at least 50 mitochondria were evaluated per group overall, following established methods[32]. Per-animal mean values were used for statistical analysis. Ultrastructural evaluation specifically concentrated on axons and myelin sheaths in the DRG, representing peripheral sensory nerve fiber components.

Measurement of mitochondrial membrane potential in DRG

Mitochondrial membrane potential (MMP) was measured using the fluorescent dye 5,5’,6,6’tetrachloro1,1’,3,3’tetraethylbenzimidazolcarbocyanine iodide (JC-1; C2003S; Beyotime, Shanghai, China). Purified DRG mitochondria were incubated with JC-1 working solution, and fluorescence signals were continuously time-scanned on a fluorescence microplate reader at excitation and emission wavelengths of 485 nm and 590 nm, respectively. Quantitative red-to-green fluorescence intensity ratios were directly obtained using a microplate reader to evaluate changes in MMP.

WB

WB of proteins in DRG[31] and PC12 cell lysates[33] was performed as previously described. The following primary antibodies were used: Anti-Drp1 (sc-271583; Santa Cruz Biotechnology, Dallas, TX, United States, 1:500), anti-mitofusin1 (MFN1; sc-166644; Santa Cruz Biotechnology, 1:500), anti-mitofusin2 (MFN2; sc-100560; Santa Cruz Biotechnology, 1:500), anti-phosphorylated Drp1 (p-Drp1; S616, 3455S; Cell Signaling Technology, Danvers, MA, United States, 1:1000), anti-NLRP3 (ab263899; Abcam, Cambridge, United Kingdom, 1:1000), anti-cleaved caspase-1 (4199s; Cell Signaling Technology, 1:1000), anti-pro-caspase-1 (ab286125; Abcam, 1:1000), and anti-IL-1β (12703S; Cell Signaling Technology, 1:1000). β-actin antibody (HRP-60008-100UL; Proteintech, Chicago, IL, United States, 1:5000) was used as the loading control.

Immunofluorescence

Double immunofluorescence staining was performed on 5-μm-thick paraffin-embedded DRG sections. The sections were initially permeabilized with 0.2% Triton X-100 in phosphate-buffered saline for 15 minutes and subsequently blocked with 10% normal goat serum for 30 minutes. All experiments were conducted using immunofluorescence detection and mouse monoclonal antibodies against Drp1 (sc-271583; Santa Cruz Biotechnology, 1:100), mouse monoclonal antibodies against NLRP3 (AG-20B-0014-C100; AdipoGen Life Sciences, San Diego, CA, United States, 1:100), and rabbit polyclonal antibodies against neuronal nuclear antigen (NeuN) (26975-1-AP; Proteintech, 1:500). The entire sections were scanned using a × 20 objective lens, and images were collected under an optical microscope (Eclipse E100, Nikon, Tokyo, Japan) and analyzed using SlideViewer software (3DHISTECH Ltd., Budapest, Hungary).

Metabolomics analysis of DRG samples

DRG samples (n = 5 per group) were analyzed using mass spectrometry in both positive and negative ion modes to identify metabolic alterations. Thirty milligrams of DRG tissue were placed in 2-ml centrifuge tubes for metabolite extraction. A total of 900 μL of extraction solution (methanol: Water = 4:1) was added to each sample. The DRG tissues were pulverized using a cold tissue grinder for 6 minutes (-10 °C, 50 Hz) to ensure thorough homogenization. Following this, the samples underwent low-temperature ultrasonic treatment for 30 minutes (5 °C, 40 kHz) to enhance metabolite extraction. The mixtures were left to stand at -20 °C for 30 minutes and then centrifuged at 12000 rpm for 15 minutes at 4 °C. The resulting supernatants were transferred into sample vials fitted with insert tubes for further analysis.

Metabolomic profiling was conducted using a Q-Exactive HF mass spectrometer (Thermo Fisher Scientific, Waltham, MA, United States), as described previously[34]. Initial data-dependent acquisition was performed with a single full-scan at a resolution of 60000, targeting hydrophilic metabolites with mass-to-charge ratios ranging from 60 to 900. Lipid profiling was subsequently performed within an extended mass-to-charge ratio range of 300 to 1200. For further structural elucidation, 10 consecutive tandem mass spectrometry scans were conducted in a high-energy collision dissociation mode.

Stringent quality control measures were implemented throughout the metabolomic study to ensure the accuracy and reliability of the data generated. Metabolite annotation was carried out through tandem mass spectrometry (MS/MS)-based structural elucidation by matching spectral features against the Human Metabolome Database under stringent quality-control criteria.

Whole-cell patch-clamp recordings

Whole-cell patch-clamp recordings were performed on small- and medium-sized DRG neurons to assess neuronal excitability. Experiments were performed at room temperature using an EPC-10 amplifier and Patch-Master software (HEKA, Freiburg, Germany), following established protocols[5]. To assess firing characteristics, a depolarizing current pulse (1000 milliseconds, 2-fold rheobase) was delivered to evoke repetitive firing. Rheobase was used operationally to normalize the depolarizing stimulus across recorded neurons, and the number of evoked action potentials at 2 × rheobase was used as the primary measure of neuronal hyperexcitability in this study. The data were analyzed using OriginPro 9.0 software (OriginLab Corporation, Northampton, MA, United States).

In vitro experiments

Cell culture and treatments: PC12 cells, a neuron-like cell line commonly used to investigate intracellular signaling responses to metabolic stress, were employed as a simplified in vitro system to complement the in vivo mechanistic analyses of diabetic neuropathy[35,36]. The PC12 neuronal cell line was acquired from the Cell Resource Center, Institute of Basic Medical Sciences and cultured at 37 °C in a humidified incubator with 5% CO₂. The cells were cultured in a Roswell Park Memorial Institute-1640 medium supplemented with 5.5 mmol/L β-D glucose, 10% fetal bovine serum, 2 mmol/L glutamine, and 1% penicillin/streptomycin (Boster Biological Technology Co., Ltd., Wuhan, China). To simulate hyperglycemic conditions, PC12 cells were exposed to 30 mmol/L D-glucose (Sigma-Aldrich) for 24 hours, following established protocols for in vitro models of diabetic complications[37,38]. For carnosine intervention, cells were pretreated with varying concentrations of carnosine (1 mmol/L, 5 mmol/L, 10 mmol/L, 20 mmol/L, or 50 mmol/L), which was dissolved in saline for 2 hours prior to high-glucose exposure[39]. After pretreatment, cells were incubated in a high-glucose medium for an additional 24 hours before being subjected to biochemical and molecular analyses.

Assessment of cell viability: PC12 cells were seeded into 96-well plates at a density of 5000 cells per well. To control for potential osmotic effects, equivalent concentrations of mannitol were used as an osmotic pressure control. After treatment and incubation, 10 μL of cell counting kit-8 (CCK-8; CK04-500T, Dojindo, Mashiki, Japan) solution was added to each well, followed by a 2-hour incubation at 37 °C. The absorbance (A) of each well was then measured at 450 nm using a microplate reader (Tecan Group AG, Mannedorf, Switzerland), with wells containing no cells serving as blanks. Cell viability was calculated using the following formula: Cell viability (%) = [A (treated) - A (blank)]/[A (untreated) - A (blank)] × 100%.

Measurement of MMP in cells: Cells were cultured in a six-well plate. Upon reaching 70%-80% confluence, the culture medium was removed, and 1 mL of JC-1 staining working solution was added. The cells were then incubated at 37 °C for 20 minutes. Subsequently, they were washed twice with JC-1 buffer and examined using a laser confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany). The red-to-green fluorescence ratio was used as a marker of MMP integrity.

Enzyme-linked immunosorbent assay: The IL-1β concentration in the medium was measured by an IL-1β enzyme-linked immunosorbent assay kit (NeoBioscience, Shenzhen, China), following the manufacturer’s instructions. A 4-parameter logistic curve was applied for each assay. The limit of quantification was set at 31.25 pg/mL, and values below this threshold were excluded from statistical analysis[40]. Experiments involving IL-1β concentration measurements were performed in triplicate.

Infection of cells with lentivirus: To achieve stable knockdown or overexpression of Drp1, PC12 cells were infected with lentivirus-short hairpin RNA targeting Drp1 or lentivirus-Drp1 (Obio Technology, Shanghai, China). The lentiviruses were constructed using the cloning vectors pSLenti-U6-shRNA (Dnm1 L)-CMV-EGFP-F2A-Puro-WPRE for knockdown and pcSLenti-EF1-EGFP-P2A-Puro-CMV-MCS-3xFLAG-WPRE for overexpression. The viral titer was 1 × 10⁸ TU/mL. Approximately 1 × 10⁵ PC12 cells were seeded into each well of a 24-well plate. The volume of lentiviral suspension added to each well was calculated using the multiplicity of infection (MOI) and the following formula: Virus volume per well (μL) = MOI × number of cells at the time of infection × 10³/titer (TU/mL). When the cells reached 30%-40% confluence and were in an optimal condition, they were infected with the calculated volume of virus suspension. After 36 hours, green fluorescence was observed under a fluorescence microscope to evaluate the efficiency of lentiviral transduction. Cells demonstrating stable knockdown or overexpression of Drp1 were harvested for further validation. The efficiency of Drp1 knockdown or overexpression was confirmed through WB. The transduced cells were subsequently used in mitochondrial function assays to investigate the role of Drp1 in mitochondrial dysfunction associated with PDN.

Statistical analysis

For nested data, including multiple mitochondria, image fields, or cells obtained from the same biological sample, measurements were first averaged at the level of each animal or independent experiment before inferential statistical analysis. Accordingly, the biological replicate, rather than the technical replicate, was used as the statistical unit. Data normality was assessed using the Shapiro-Wilk test. Normally distributed data were presented as the mean ± standard error of the mean (SEM) and were analyzed using unpaired Student’s t-test for making comparisons between two groups. Abnormally distributed data were presented as the median with interquartile range and were analyzed using the Mann-Whitney U test for two-group comparisons or the Kruskal-Wallis test for comparisons among multiple groups. For analyses involving more than two groups or multiple time points, one-way or two-way analysis of variance, including repeated-measures analysis of variance where appropriate, was performed, followed by Tukey’s or Sidak’s post hoc multiple-comparison test. Linear mixed-effects models were applied to clustered measurements nested within individual animals, with animal ID as a random intercept. Animal ID was incorporated as a random intercept to adjust for correlated observations derived from multiple biological samples harvested from the same rat. Experimental group was defined as the fixed factor for intergroup comparisons, and pairwise post-hoc tests with Bonferroni correction were conducted to control type I error inflation from multiple comparisons across the three groups. Statistical significance was defined as P < 0.05. All statistical analyses were performed using Prism 10.0 software (GraphPad Software, La Jolla, CA, United States).

Metabolomic analysis of DRG samples was conducted using the ropls package (Version 1.6.2) in R (R Foundation for Statistical Computing, Vienna, Austria). Principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed using preprocessed metabolomic data matrices to evaluate differences in metabolic profiles between the PDN and control groups. Differential metabolites were identified according to the following criteria: Variable importance in projection (VIP) score > 1 derived from the PLS-DA model and P < 0.05 based on univariate t-test analysis for individual metabolites. Pathway enrichment analysis was conducted using the Kyoto Encyclopedia of Genes and Genomes database (https://www.kegg.jp/kegg/pathway.html) and a hypergeometric test. The resulting P values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure to control the false discovery rate, and an adjusted P value < 0.05 was considered statistically significant. Relationships between variables were assessed using Spearman’s rank correlation analysis.

RESULTS

PDN rats exhibited mitochondrial morphological and functional impairments in the DRG

After 8 weeks of feeding, PDN rats exhibited significant body weight gains compared to control rats. However, following STZ injection, the body weight of PDN rats gradually decreased and was significantly lower than that of control rats by 35 days post-injection [F _{(1, 20)} = 4.808, P = 0.04; week: F _{(1.389, 27.79)} = 647.4, P < 0.001; interaction: F _{(6, 120)} = 115.9, P < 0.001, Supplementary Figure 1A]. Persistent hyperglycemia (blood glucose level > 11.1 mmol/L) was found in PDN rats since day 7 after STZ administration (Supplementary Figure 1B). Intraperitoneal glucose tolerance test indicated significantly higher blood glucose levels in the PDN group than those in the control group (P < 0.001; Supplementary Figure 1C), and insulin tolerance test demonstrated a significant intergroup difference at 30 min post- insulin injection (P < 0.001; Supplementary Figure 1D) with a slow blood glucose decline in the PDN group, indicating insulin resistance. Supplementary Figure 1 collectively presents comprehensive metabolic and disease-model characterization data supporting the successful establishment of the PDN rat phenotype. Behavioral assessments revealed remarkable mechanical allodynia [group: F _{(1, 20)} = 88.26, P < 0.001; week: F _{(4.72, 94.39)} = 18.71, P < 0.001; interaction: F _{(6, 120)} = 15.43, P < 0.001, Figure 1A] and thermal hyperalgesia [F _{(1, 20)} = 255.0, P < 0.001; week: F _{(4.166, 83.32)} = 38.76, P < 0.001; interaction: F _{(6, 120)} = 40.67, P < 0.001, Figure 1B] in PDN rats, as evidenced by significantly reduced PWT from 21 days post-STZ injection and decreased PWL from 14 days post-STZ injection. Compared with control rats, PDN rats exhibited significantly reduced MNCV (t = 6.64, df = 20, P < 0.001; Figure 1C) and SNCV (t = 7.593, df = 20, P < 0.001; Figure 1C), reflecting functional impairment of the peripheral nerves. Structural analysis of peripheral nerve fibers and myelin sheath structures in the DRG revealed severe neuropathic damage, characterized by axonal degeneration and demyelination (Figure 1D). Electron microscopy of nerve fibers in the DRG revealed fragmented, small, spherical mitochondria with disrupted cristae and double membranes in PDN rats compared with the intact and elongated mitochondria in control rats (Figure 1E). In the PDN group, a statistically significant increase in the number of mitochondria per field of view was identified compared with the control group [F _{(1, 16)} = 6.37, P < 0.05, Figure 1F]. Furthermore, the proportion of abnormal mitochondria was higher [F _{(1, 16)} = 20.93, P < 0.001, Figure 1G], with significantly shorter mitochondrial length [F _{(1, 103)} = 26.02, P < 0.001; Figure 1H]. Functionally, MMP was significantly reduced in the DRG of PDN rats (t = 5.21, df = 14, P < 0.001; Figure 1I). The expression levels of the mitochondrial fission protein Drp1 and the fusion proteins MFN1 and MFN2 were also altered (Figure 1J). Drp1, a key regulator of mitochondrial fission, was significantly upregulated in PDN rats compared to control rats (t = 7.79, df = 10, P < 0.001, Figure 1K). In contrast, the fusion proteins MFN1 and MFN2 were significantly downregulated (t = 4.04, df = 10, P < 0.01, Figure 1L; t = 2.90, df = 10, P < 0.05, Figure 1M). Elevated Drp1 expression or activity may precede and contribute to imbalances in mitochondrial fission and fusion in various pathophysiological conditions by disrupting mitochondrial integrity and signaling pathways, potentially leading to downstream changes in MFN1 and MFN2[41,42]. These findings supported the subsequent investigation of Drp1-related signaling to further examine its potential involvement in PDN.

Figure 1 The rat model of painful diabetic neuropathy established through high-fat feeding combined with low-dose streptozotocin injection. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001 vs the control group; two-way analysis of variance with Sidak’s post-hoc test for Figure 1A and B; Unpaired t-tests, two-tailed for Figure 1C, I, K-M; Linear mixed-effects models were applied for two-group comparisons of data in panels F-H, with animal ID set as a random intercept to account for clustered observations within individual rats; Figure 1A-C: n = 11; Figure 1D-H: n = 3 rats per group, with at least 50 mitochondria counted in 3 random fields per sample. For Figure 1F and G, each field was the statistical unit; for panel H, each individual mitochondrion was the statistical unit for mitochondrial length quantification; I: n = 8; J-M: n = 6. A: Paw withdrawal threshold in painful diabetic neuropathy (PDN) rats; B: Paw withdrawal latency in PDN rats; C: Motor nerve conduction velocity and sensory nerve conduction velocity in PDN rats; D: Ultrastructural analysis of nerve fibers in the dorsal root ganglion (DRG) of PDN rats. Scale bar = 2 μm; E: Examination of mitochondrial structure in the DRG of PDN rats. Scale bar = 500 nm; F: Measurement of the number of mitochondria per field; G: Proportion of abnormal mitochondria to total mitochondria; H: Measurement of mitochondrial length; I: Assessment of mitochondrial membrane potential in the DRG of PDN rats, in which quantitative data were acquired by a fluorescence microplate reader; J-M: Western blotting showing the expression level of proteins associated with mitochondrial fission protein dynamin-related protein 1 and fusion proteins MFN1/MFN2, including representative electrophoresis bands and relative expression levels. PDN: Painful diabetic neuropathy; MNCV: Motor nerve conduction velocity; SNCV: Sensory nerve conduction velocity; STZ: Streptozotocin; Drp1: Dynamin-related protein 1.

DRG metabolomics identified reduced carnosine in PDN rats

Metabolomic analysis using mass spectrometry revealed distinct metabolic profiles between the control and PDN groups. PCA (Figure 2A) and PLS-DA (Figure 2B) indicated clear separation between the two groups, reflecting distinct metabolic patterns. A total of 147 metabolites were significantly upregulated, and 211 metabolites were downregulated in the PDN group compared with the control group (VIP > 1 and P < 0.05; Figure 2C).

Figure 2 Dorsal root ganglia metabolomic profiling and correlation analyses between carnosine levels and behavioral pain indices, including paw withdrawal threshold and paw withdrawal latency. A: Principal component analysis; green represents quality control, blue indicates the control group, and yellow represents painful diabetic neuropathy; B: Partial least squares discriminant analysis; green represents quality control, blue indicates the control group, and yellow represents painful diabetic neuropathy; C: Volcano plot of differential metabolites; orange indicates upregulated differential metabolites, blue indicates downregulated differential metabolites, and gray indicates metabolites detected but not significantly different; D: Kyoto Encyclopedia of Genes and Genomes topological analysis, the color gradient represents the significance of pathways arranged by P value (Y-axis, darker red indicates smaller P values), while the size of circles represents pathway impact values (X-axis, larger circles indicate higher pathway impact value scores); pathway enrichment was analyzed using a hypergeometric test, and P values were adjusted for multiple comparisons using the Benjamini-Hochberg method to control the false discovery rate (corrected P < 0.05); E and F: Correlation analysis between carnosine levels and behavioral pain indices, including paw withdrawal threshold and paw withdrawal latency. n = 5 animals per group. QC: Quality control; PDN: Painful diabetic neuropathy; KEGG: Kyoto Encyclopedia of Genes and Genomes.

The metabolites with the highest variations in VIP scores across ionization modes are listed in Supplementary Tables 1-4. Pathway enrichment analysis revealed significant dysregulation in several metabolic pathways in the DRG of PDN rats. The most affected pathways included histidine metabolism (featuring key metabolites, such as histamine, carnosine, L-histidine, ergothioneine, and urocanic acid), tryptophan metabolism, arginine and proline metabolism, lysine degradation, arginine biosynthesis, and vitamin B6 metabolism (Figure 2D).

Among the significantly altered metabolites, carnosine, a histidine metabolism-related dipeptide, was significantly reduced in the PDN group (anion mode; Supplementary Table 4). Correlation analysis further indicated a strongly positive association between carnosine level and behavioral measures of neuropathic pain in diabetic rats. Specifically, carnosine level was positively correlated with PWT (r = 0.94, P = 0.021; Figure 2E) and PWL (r = 0.81, P = 0.034; Figure 2F). Carnosine was selected for further investigation on the basis of its pathway relevance, previously documented antioxidant, anti-inflammatory, and mitochondrial-protective properties, and its positive correlations with PWT and PWL in the present dataset.

Preventive carnosine supplementation attenuated neuropathic pain and improved mitochondrial abnormalities

To investigate the effects of preventive carnosine supplementation, carnosine was administered during model induction, and behavioral and mitochondrial outcomes were subsequently assessed in PDN rats. The experimental timeline, encompassing the control, PDN, and carnosine treatment groups, is outlined in Figure 3A. Carnosine treatment significantly reduced the body weight of PDN rats [group: F _{(6, 70)} = 160.7, P < 0.001; week: F _{(1.899, 132.9)} = 69.80, P < 0.001; interaction: F _{(12, 140)} = 15.98, P < 0.001; Figure 3B]. Moreover, carnosine treatment significantly reduced fasting blood glucose levels in PDN rats at 1 week following STZ injection [group: F _{(2, 30)} = 279.7, P < 0.001; week: F _{(3.062, 91.85)} = 302.3, P < 0.001; interaction: F _{(12, 180)} = 82.25, P < 0.001; Figure 3C].

Figure 3 The impact of intraperitoneal injection of carnosine on behavioral, electrophysiological, ultrastructural, and neuronal functional abnormalities in painful diabetic neuropathy rats. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001 vs the control group; ^dP < 0.05, ^eP < 0.01, ^fP < 0.001 vs the painful diabetic neuropathy group. Two-way analysis of variance with Sidak’s multiple comparisons test for Figure 3B-E. One-way analysis of variance with Tukey’s multiple comparisons test for Figure 3F, H, and N; Linear mixed-effects models were used for statistical analyses of data shown in panels Figure 3K-M, with animal ID designated as the random intercept to account for clustered microscopic observations nested within individual rats. Bonferroni-adjusted pairwise comparisons were performed for intergroup comparisons across the three experimental groups. Figure 3B-F: n = 11; Figure 3G and H: n = 3; Figure 3I-M: n = 3 rats per group, with at least 50 mitochondria counted in 3 random fields per sample. For Figure 3K and L, each field was the statistical unit; for Figure 3M, each individual mitochondrion was the statistical unit for mitochondrial length quantification; Figure 3N: n = 8. A: Flowchart of the experiments; B: Changes in body weight after intraperitoneal carnosine injection; C: Changes in blood glucose level following intraperitoneal carnosine injection; D: Paw withdrawal threshold in rats post-intraperitoneal carnosine injection; E: Paw withdrawal latency in rats post-intraperitoneal carnosine injection; F: Motor nerve conduction velocity and sensory nerve conduction velocity in rats following intraperitoneal carnosine injection; G and H: Results of whole-cell patch-clamp recordings at week 3 after streptozotocin injection; G: Representative traces of neuronal action potential; H: Statistical analysis of the action potential number in dorsal root ganglia (DRG) neurons of rats (n = 3); I: Ultrastructural changes of the DRG post-intraperitoneal carnosine injection. Scale bar = 2 μm; J: Alterations in mitochondrial structure in the DRG post-intraperitoneal carnosine injection. Scale bar = 500 nm; K: The number of mitochondria per field; L: Proportion of abnormal mitochondria relative to the total mitochondrial count; M: Mitochondrial length; N: Changes in mitochondrial membrane potential in the DRG following intraperitoneal carnosine injection. PDN: Painful diabetic neuropathy; MNCV: Motor nerve conduction velocity; SNCV: Sensory nerve conduction velocity; STZ: Streptozotocin.

Additionally, compared with untreated PDN rats, carnosine treatment significantly improved PWT and PWL, indicating attenuation of mechanical allodynia [group: F _{(2, 30)} = 39.75, P < 0.001; week: F _{(4.946, 148.4)} = 41.15, P < 0.001; interaction: F _{(12, 180)} = 6.337, P < 0.001; Figure 3D] and thermal hyperalgesia [group: F _{(2, 30)} = 101.9, P < 0.001; week: F _{(5.075, 152.2)} = 70.50, P < 0.001; interaction: F _{(12, 180)} = 11.82, P < 0.001; Figure 3E]. Importantly, carnosine treatment significantly improved both MNCV [F _{(2, 30)} = 36.89, P < 0.001; Figure 3F] and SNCV [F _{(2, 30)} = 63.53, P < 0.001; Figure 3F] in PDN rats compared with untreated PDN rats. Subsequent assessments were conducted at 3 weeks following STZ injection, a timepoint at which therapeutic improvements were evident and persisted through 5 weeks post-injection. Furthermore, carnosine supplementation attenuated DRG neuronal hyperexcitability (Figure 3G), as evidenced by a significant reduction in the number of evoked action potentials [F _(2,27) = 26.47, P < 0.001; Figure 3H] compared with PDN rats.

Compared with PDN rats, those treated with carnosine demonstrated significantly reduced demyelination of myelinated nerve fibers (Figure 3I). The mitochondria in these rats were more elongated, exhibited fewer small, round forms, and had denser and more distinct cristae, with reduced membrane disruption (Figure 3J). Transmission electron microscopy revealed a decreased number of mitochondria per field [F _{(2, 24)} = 11.59, P < 0.001; Figure 3K] and fewer abnormal mitochondria in carnosine-treated rats [F _{(2, 24)} = 19.33, P < 0.001; Figure 3L], along with improved mitochondrial morphology, and increased mitochondrial length [F _{(2, 158)} = 13.53, P < 0.001; Figure 3M]. In addition, the MMP was significantly enhanced in the carnosine group [F _{(2, 21)} = 13.37, P < 0.001; Figure 3N]. Taken together, these results indicated that preventive carnosine administration attenuated neuropathic pain-related phenotypes and improved mitochondrial abnormalities in PDN rats.

Carnosine supplementation was associated with reduced mitochondrial fission and NLRP3 inflammasome activation in the DRG

Immunofluorescence revealed co-localization of Drp1 (Figure 4A and B) and NLRP3 (Figure 4C and D) with the neuronal marker NeuN in the DRG. Compared with control rats, the DRG neurons of PDN rats exhibited significantly higher expression levels of Drp1 (Figure 4A and B) and NLRP3 (Figure 4C and D) in DRG neurons compared with control rats, reflecting increased mitochondrial fission and inflammasome activation. Notably, carnosine supplementation reduced the expression levels of both proteins, as confirmed by quantification of immunofluorescence intensity [Drp1: F _{(2, 575)} = 1027.56, P < 0.001; NLRP3: F _{(2, 614)} = 767.41, P < 0.001].

Figure 4 Impact of intraperitoneal carnosine injection on mitochondrial fission and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation in rats with painful diabetic neuropathy. ^cP < 0.001 vs the control group; ^dP < 0.05, ^fP < 0.001 vs the painful diabetic neuropathy (PDN) group; Linear mixed-effects models were used for the analyses in Figure 4B and D, with animal ID included as a random intercept to account for neurons nested within individual rats. Bonferroni-adjusted pairwise comparisons were performed for post hoc analyses. One-way analysis of variance followed by Tukey’s multiple comparisons test was used for Figure 4F-K. For Figure 4A-D, n = 3 rats per group. Multiple image fields and neurons were analyzed for each animal. For Figure 4E-K, n = 6 rats per group. A: Representative immunofluorescence images of dynamin-related protein 1 and neuronal marker neuronal nuclear antigen in the dorsal root ganglia (DRG) of control rats, rats with PDN, and carnosine-treated PDN rats. Scale bar = 20 μm; B: Variations in immunofluorescence intensity of dynamin-related protein 1 in DRG neurons among control, PDN, and carnosine-treated PDN rats; C: Representative immunofluorescence images of NOD-like receptor pyrin domain-containing protein 3 (NLRP3) and neuronal marker neuronal nuclear antigen in the DRG of control, PDN, and carnosine-treated PDN rats. Scale bar = 20 μm; D: Changes in immunofluorescence intensity of NLRP3 in DRG neurons among control, PDN, and carnosine-treated PDN rats; E-K: Western blotting indicating the expression levels of proteins associated with mitochondrial fission and NLRP3 inflammasome pathway, including representative electrophoresis bands and relative expression levels. NLRP3: NOD-like receptor pyrin domain-containing protein 3; Drp1: Dynamin-related protein 1; PDN: Painful diabetic neuropathy; DAPI: 4’,6-diamidino-2-phenylindole; NeuN: Neuronal nuclear antigen; IL: Interleukin.

As shown in the representative western blot bands (Figure 4E), WB further demonstrated that the protein expression levels of Drp1 (Figure 4F) and p-Drp1 (Figure 4G) in the DRG of PDN rats were significantly elevated compared with control rats. Administration of carnosine significantly reduced the expression levels of both total and activated Drp1 [Drp1: F _{(2, 15)} = 15.30, P < 0.001; p-Drp1: F _{(2, 15)} = 76.10, P < 0.001], aligning with immunofluorescence results and an association between carnosine treatment and reduced mitochondrial fission-related abnormalities in PDN.

In addition, PDN rats exhibited increased protein expression levels of NLRP3 (Figure 4H), while pro-caspase-1 expression remained unchanged (Figure 4I). In contrast, cleaved caspase-1 (Figure 4J) and IL-1β (Figure 4K) were markedly increased, reflecting inflammasome activation. Carnosine administration significantly attenuated the expression levels of these proteins [NLRP3: F _{(2, 15)} = 14.67, P < 0.001; cleaved caspase-1: F _{(2, 15)} = 14.06, P < 0.001; IL-1β: F _{(2, 15)} = 43.96, P < 0.001], indicating that carnosine was associated with changes in inflammasome activation downstream of pro-caspase-1 expression level.

Carnosine could protect PC12 cells against high glucose-induced injury

PC12 cells exposed to high glucose (30 mmol/L) exhibited significantly reduced viability compared with the control group, indicating glucose-induced cytotoxicity. Mannitol treatment, used as an osmotic control, did not affect cell viability, confirming that the observed effects were glucose-specific [F _{(2, 6)} = 49.77, P < 0.001; Figure 5A]. Under normal culture medium conditions, carnosine treatment did not affect the viability of PC12 cells (P > 0.05; Figure 5B). Under high-glucose conditions, the effects of carnosine on cell viability varied by concentration. Only 5 mmol/L carnosine significantly improved the cell viability [F _{(6, 14)} = 16.16, P < 0.001; Figure 5C], while 1 mmol/L, 10 mmol/L, and 20 mmol/L had no significant impact. Therefore, 5 mmol/L was selected for subsequent experiments.

Figure 5 The screening of optimal carnosine concentration in PC12 cells. ^bP < 0.01, ^cP < 0.001 vs the control group; ^dP < 0.05, ^eP < 0.01, ^fP < 0.001 vs the high-glucose (HG) group. One-way analysis of variance with Tukey’s multiple comparisons test for Figure 5A-C, E and F. For Figure 5A-C, n = 3; for Figure 5D-F, n = 4. A: Cell counting kit-8 (CCK-8) assay of viability in PC12 cells treated with HG and mannitol; B: CCK-8 assay of viability in PC12 cells treated with various concentrations of carnosine in a standard culture medium; C: CCK-8 assay of viability in PC12 cells treated with different concentrations of carnosine in a HG culture medium; D-F: Western blotting indicating changes in the expression levels of mitochondrial fission protein dynamin-related protein 1 and NOD-like receptor pyrin domain-containing protein 3 inflammasome in PC12 cells treated with various concentrations of carnosine in a HG culture medium. NLRP3: NOD-like receptor pyrin domain-containing protein 3; Drp1: Dynamin-related protein 1; HG: High-glucose.

WB revealed that high-glucose exposure increased the expression levels of Drp1 and NLRP3 in PC12 cells, whereas treatment with 5 mmol/L carnosine significantly reduced the expression levels of both Drp1 [F _(6,21) = 5.625, P = 0.0013; Figure 5D and E] and NLRP3 [F _(6,21) = 10.53, P < 0.001; Figure 5D and F]. Other carnosine concentrations did not produce significant effects.

JC-1 staining revealed that mannitol did not affect MMP (Figure 6A), while high-glucose stimulation decreased the red/green fluorescence ratio, indicating mitochondrial dysfunction (Figure 6B). Carnosine treatment partially restored MMP [F _(3,16) = 133.1, P < 0.001].

Figure 6 Effects of carnosine intervention on mitochondrial membrane potential, mitochondrial fission, and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation in PC12 cells. ^cP < 0.001 vs the control group; ^dP < 0.05, ^eP < 0.01, ^fP < 0.001 vs the high-glucose group. One-way analysis of variance with Tukey’s multiple comparisons test for Figure 6B, D-I. For Figure 6A and B, n = 5; for Figure 6C-I, n = 4. A: Representative fluorescence images of JC-1 monomers/aggregates in PC12 cells. Scale bar = 20 μm; B: Mitochondrial membrane potential level in PC12 cells; C-H: Impact on the expression levels of proteins related to mitochondrial fission and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation in PC12 cells, including representative electrophoresis bands and relative expression levels; I: Enzyme-linked immunosorbent assay showing the effect on interleukin-1β expression level in PC12 cells. NLRP3: NOD-like receptor pyrin domain-containing protein 3; Drp1: Dynamin-related protein 1; HG: High-glucose; IL-1β: Interleukin-1β.

Furthermore, the expression levels of Drp1, phosphorylated Drp1, and NLRP3 inflammasome pathway proteins were measured. Additionally, high-glucose exposure significantly upregulated Drp1, p-Drp1, NLRP3, and cleaved caspase-1, accompanied by enhanced IL-1β release. These elevations were significantly attenuated by carnosine treatment [Drp1: F _{(3, 12)} = 81.09, P < 0.001; p-Drp1: F _{(3, 12)} = 50.89, P < 0.001; cleaved caspase-1: F _{(3, 12)} = 66.95, P < 0.001; IL-1β: F _{(3, 12)} = 27.93, P < 0.001; Figure 6C-I]. Neither high-glucose stimulation nor carnosine intervention affected pro-caspase-1 levels (Figure 6G).

Carnosine suppressed high-glucose-induced activation of the Drp1-NLRP3 inflammasome signaling pathway in PC12 cells

To explore the mechanistic role of Drp1 in NLRP3 inflammasome activation under hyperglycemia, PC12 cells were transduced with lentivirus-short hairpin RNA targeting Drp1. After 36 h, Drp1 protein level was reduced by approximately 70%-80% [F _{(2, 9)} = 33.03, P < 0.001; Figure 7A and B]. In contrast, control shRNA lentivirus had no effect (Supplementary Figure 2A and B). High glucose increased NLRP3 expression (Figure 7C), while pro-caspase-1 expression was not obviously altered by high-glucose stimulation but was markedly reduced after Drp1 knockdown (Figure 7D). Drp1 knockdown attenuated high glucose-induced elevations [NLRP3: F _{(2, 9)} = 24.15, P < 0.001; cleaved caspase-1: F _{(2, 9)} = 12.25, P = 0.0027; IL-1β: F _{(2, 9)} = 35.70, P < 0.001; Figure 7A-F], reflecting that Drp1 could promote inflammasome activation under hyperglycemic stress.

Figure 7 dynamin-related protein 1 knockdown inhibits high glucose-induced NOD-like receptor pyrin domain-containing protein 3 inflammasome activation, whereas dynamin-related protein 1 overexpression attenuates the inhibitory effect of carnosine in PC12 cells. For Figure 7A-F, ^bP < 0.01 vs the control group; ^cP < 0.001 vs the control group; ^dP < 0.05 vs the high-glucose (HG) group; ^fP < 0.001 vs the HG group. For panels H-O, ^gP < 0.05 vs the vector control group, ^hP < 0.01 compared with the vector control group, ⁱP < 0.001 vs the vector control group; ^jP < 0.05 vs the vector HG group, ^kP < 0.01 vs the vector HG group, ^lP < 0.001 vs the vector HG group; ^mP < 0.01 vs vector carnosine group, ⁿP < 0.001 vs vector carnosine group. One-way analysis of variance with Tukey’s multiple comparisons test for Figure 7B-F, H, J-O. For Figure 7A-F, I-O, n = 4, G-H, n = 5. A-E: Western blotting of mitochondrial fission and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation-related protein expression following lentivirus-short hairpin RNA targeting dynamin-related protein 1 knockdown, including representative electrophoresis bands and relative expression levels; F: Enzyme-linked immunosorbent assay showing interleukin-1β expression level; G: Representative fluorescence images of JC-1 monomers/aggregates in PC12 cells. Scale bar = 20 μm; H: Mitochondrial membrane potential level in PC12 cells; I-N: Western blotting of mitochondrial fission and NOD-like receptor pyrin domain-containing protein 3 inflammasome activation-related protein expression after lentivirus-dynamin-related protein 1 overexpression, including representative electrophoresis bands and relative expression levels; O: Enzyme-li O: Enzyme-linked immunosorbent assay indicating interleukin-1β expression level in PC12 cells. NLRP3: NOD-like receptor pyrin domain-containing protein 3; Drp1: Dynamin-related protein 1; HG: High-glucose; IL-1β: Interleukin-1β; LV-shDrp1: Lentivirus-short hairpin RNA targeting dynamin-related protein 1.

In contrast, overexpression of Drp1 was efficiently induced by lentiviral vectors at an MOI of 20, as validated by WB and quantitative analysis (Supplementary Figure 2C and D). Drp1 overexpression reduced the MMP of cells cultured in both standard and high-glucose media (Figure 7G), and abolished the MMP-restorative effect of carnosine [F _{(5, 24)} = 467.7, P < 0.001; Figure 7H].

Furthermore, Drp1 overexpression significantly increased the levels of these proteins and inflammatory cytokines [Drp1: F _{(5, 18)} = 35.43, P < 0.001; p-Drp1: F _{(5, 18)} = 194.9, P < 0.001; NLRP3: F _{(5, 18)} = 71.56, P < 0.001; pro-caspase-1: F _{(5, 18)} = 11.55, P < 0.001; cleaved caspase-1: F _{(5, 18)} = 192.4, P < 0.001; IL-1β: F _{(5, 18)} = 102.2, P < 0.001; Figure 7I-O]. Although carnosine suppressed these elevations under high-glucose conditions, pro-caspase-1 remained unaffected. This effect was partially diminished by Drp1 overexpression, supporting a role of Drp1 as an upstream regulator of NLRP3 inflammasome-related signaling and mitochondrial abnormalities in this model.

DISCUSSION

In the current study, metabolomic analysis revealed that carnosine levels were significantly reduced in PDN rats. Preventive carnosine supplementation could alleviate PDN, which was accompanied by improvements in mitochondrial function and reductions in inflammation. Mechanistically, these findings were associated with altered Drp1-mediated mitochondrial fission signaling and reduced NLRP3 inflammasome-related activation in DRG neurons, supporting an integrative framework linking reduced carnosine, mitochondrial alterations, and inflammasome-related signaling in PDN.

Given that DRG neurons represent the somata of primary sensory neurons innervating peripheral tissues, the present study specifically examined peripheral sensory mechanisms implicated in neuropathic pain processing, rather than mechanisms involving central brainstem structures. In this study, it was found that preventive carnosine supplementation was associated with improved mitochondrial and inflammasome-related readouts in vivo, while in vitro experiments supported a role of Drp1-NLRP3 signaling under high-glucose conditions. Notably, the absence of fasting hyperglycemia after 8 weeks of high-fat feeding reflects an early dysmetabolic phenotype, in which obesity and insulin resistance precede overt hyperglycemia before STZ-induced β-cell dysfunction[43]. In the rat model of type 2 DM induced by HFD combined with low-dose STZ, prominent mechanical allodynia, thermal hyperalgesia, and mitochondrial abnormalities were detected in the DRG, together with enhanced mitochondrial fission and inflammasome activation. Untargeted metabolomic profiling revealed a significant reduction in carnosine, a histidine-pathway metabolite, in DRG tissues from PDN rats. Notably, exogenous carnosine administration was associated with improved mitochondrial function, decreased Drp1 expression and activation, attenuated NLRP3 inflammasome activity, and consequent amelioration of neuropathic pain-related phenotypes. Mechanistic experiments using high-glucose-stimulated PC12 cells further supported the involvement of Drp1 inhibition and downstream inflammasome-related signaling in the protective effects of carnosine in this in vitro model.

Mitochondrial dynamics encompass continuous changes in mitochondrial shape, size, and function, including fission, fusion, autophagy, and translocation[6,44]. These dynamic processes are essential for maintaining mitochondrial integrity and bioenergetic homeostasis, particularly in neurons with high metabolic demands.

In the present study, electron microscopy revealed small, rounded, and fragmented mitochondria with disrupted cristae in the DRG neurons of PDN rats, reflecting excessive mitochondrial fission. This morphological damage was accompanied by a significant reduction in MMP and upregulation of Drp1, a key regulator of mitochondrial fission[45]. Persistent mitochondrial fission has shown to impair adenosine triphosphate (ATP) production, increase oxidative stress, and promote neuronal injury in various neurodegenerative conditions[46]. However, in the present study, mitochondrial dysfunction was directly assessed mainly by ultrastructural abnormalities, altered fission/fusion-related protein expression, and reduced MMP, rather than by direct measurements of ROS, ATP production, or respiration.

Drp1 plays a pivotal role in modulating mitochondrial division[45], and it may influence the balance between fission and fusion by modulating the expression and activity of mitochondrial fusion proteins, such as MFN1 and MFN2. In the presented model, Drp1 expression level was significantly elevated, whereas the expression levels of MFN1 and MFN2 were downregulated in the DRG of PDN rats, supporting a shift toward mitochondrial fragmentation.

Notably, similar regulatory patterns have been identified in myocardial ischemia-reperfusion injury, where inhibition of Drp1 phosphorylation restored MFN1/2 expression, promoted mitochondrial fusion, and improved mitochondrial function[41]. A comparable mechanism has also been implicated in diabetic retinopathy[42]. These findings highlight the broader relevance of Drp1 dysregulation in diabetic complications.

Based on this collective evidence, Drp1 was subsequently investigated as a candidate mediator of mitochondrial abnormalities in PDN, and alterations in Drp1-related signaling were found to be associated with improved mitochondrial homeostasis and attenuation of neuropathic phenotypes. Although the present study identified predominant neuronal localization of Drp1 and NLRP3 in the DRG using the pan-neuronal marker NeuN, future studies employing sensory neuronal subtype-specific markers, such as calcitonin gene-related peptide and isolectin B4, are required to further define the peptidergic and non-peptidergic neuronal populations involved in carnosine-mediated regulation.

Previous studies have indicated that modulation of mitochondrial dynamics and NLRP3 inflammasome activation alleviated diabetic neuropathy through diverse upstream mechanisms, including SIRT-dependent signaling and antioxidant activation[47,48]. Collectively, these mechanisms may converge on the Drp1/NLRP3 axis, contributing to improved mitochondrial function and reduced neuroinflammatory signaling. Distinct from prior mechanism-focused approaches, the present study adopted a metabolomics-driven strategy and identified decreased carnosine abundance as a PDN-associated metabolic alteration. These findings indicate that carnosine may be mechanistically associated with Drp1-mediated mitochondrial fission and NLRP3-related inflammasome signaling, thereby providing a metabolic link between disrupted histidine metabolism and mitochondrial-inflammatory remodeling in PDN.

Metabolomics has become a robust tool for dissecting disease pathogenesis, providing insights into systemic metabolic disturbances across various disease states. In diabetic complications, including PDN, both preclinical and clinical metabolomic studies have identified potential biomarkers and novel therapeutic targets[49]. In the present study, untargeted metabolomic analysis revealed broad metabolic changes in the DRG of PDN rats, including reduced carnosine level, a dipeptide implicated in neuroprotection.

Among significantly dysregulated pathways, histidine metabolism, involving carnosine, histamine, and ergothioneine, was one of the most prominently altered pathways, reflecting its relevance to PDN pathophysiology. Carnosine was not only significantly reduced, but also exhibited strongly positive correlations with behavioral indicators of neuropathic pain, namely PWT and PWL. These findings support the possibility that reduced carnosine may be linked to neuronal dysfunction and pain hypersensitivity in PDN, providing a rationale for further evaluation of carnosine supplementation.

Carnosine possesses diverse biological functions, including antioxidant[18,19], anti-inflammatory[19-21], and mitochondrial protective properties[23,24], making it a compelling candidate for addressing the multifactorial pathogenesis of PDN. These attributes align closely with the key mechanisms underlying diabetic neuropathy, such as oxidative stress, neuroinflammation, and mitochondrial dysfunction.

In the present study, intraperitoneal administration of carnosine significantly alleviated mechanical allodynia, thermal hyperalgesia, and DRG neuronal hyperexcitability in PDN rats. These effects are consistent with previous findings in various neuropathic pain models, including chemotherapy-induced peripheral neuropathy[50], trauma-induced peripheral neuropathy[51], and knee osteoarthritis[52]. In addition to the observed neurobehavioral effects, carnosine administration significantly reduced blood glucose levels in diabetic rats, reflecting that systemic metabolic improvement may have contributed to the overall phenotype. Previous studies have reported glucose-lowering effects of carnosine in experimental and clinical settings, including attenuation of hyperglycemia and hemoglobin A1c levels in type 2 DM mouse models[53], improvement of glucose metabolism in STZ-induced diabetic rats[54], and reduced fasting blood glucose in individuals with prediabetes and type 2 DM in randomized clinical trials[55]. A recent meta-analysis further confirmed its glycemic beneficial effects across human and animal studies[56]. Notably, similar dual metabolic and neuroprotective actions have been described for agents, such as imeglimin, improving mitochondrial function while conferring neuronal benefits[57,58]. However, the present study was not designed to determine whether the beneficial effects of carnosine on neuropathic phenotypes were independent of improved glycemic control or were mediated primarily through changes in mitochondrial dynamics and neuroinflammation. Future studies are needed to clarify this issue. Long-term management of diabetes and diabetic neuropathy relies on comprehensive metabolic control[59] and lifestyle-based interventions, including weight management and exercise[60]; however, these factors were beyond the scope of the present study. It should also be emphasized that the present dosing paradigm involved intraperitoneal administration initiated during the metabolic induction phase, remarkably differing from oral carnosine or related supplement use in patients with established neuropathy. Accordingly, the present findings should not be interpreted as evidence that clinically achievable oral carnosine exposure would reproduce the same effects in patients with established PDN. Pharmacokinetics, tissue exposure, and bioavailability under diabetic conditions require further investigation before clinical extrapolation. Mounting evidence indicates that non-pharmacological interventions, such as acupuncture, mitigate neuropathic pain in diabetic patients[61]. Given that the present study was not designed to compare carnosine’s efficacy against established pharmacological and non-pharmacological interventions, including acupuncture, our findings only support carnosine as a mechanistically relevant candidate rather than a definitive or clinically superior treatment.

In the present study, carnosine treatment was associated with improved mitochondrial morphology, enhanced MMP, and a reduced proportion of structurally abnormal mitochondria in DRG neurons. These effects reflect the central role of mitochondrial integrity in regulating neuronal excitability and pain processing. Previous studies have similarly highlighted carnosine’s capacity to preserve mitochondrial function in the central nervous system. For instance, carnosine has shown to protect against ischemia-induced brain injury[62] and to alleviate transcriptional deficiencies associated with neurodegenerative conditions in Parkinson's disease models[63].

Mitochondrial dysfunction is a potent trigger for NLRP3 inflammasome activation[64], a process that contributes to neuroinflammation and the progression of chronic pain. In the present study, carnosine supplementation was associated with improved mitochondrial readouts and reduced expression of Drp1- and NLRP3-related proteins in DRG neurons. WB confirmed reductions in phosphorylated Drp1, cleaved caspase-1, and IL-1β, indicating suppression of the Drp1-NLRP3-caspase-1 axis. As a key proinflammatory mediator in diabetic neuropathy, IL1β drives neuronal oxidative stress and injury and contributes to persistent pain. Previous studies have reported that IL-1β may enhance peripheral nociceptor sensitization by modulating pain-related channels, including transient receptor potential vanilloid 1 and sodium channel-dependent excitability[65].

To directly probe this mechanism, in vitro experiments were conducted using PC12 cells under high-glucose stimulation. High-glucose exposure induced remarkable upregulation of Drp1 and NLRP3 and promoted mitochondrial dysfunction, all of which were significantly attenuated by carnosine treatment. These in vitro findings indicate that Drp1 may function as an important upstream regulator of NLRP3-related signaling under high-glucose stress in PC12 cells. However, the Drp1-dependent nature of this mechanism requires direct validation in primary sensory neurons. Carnosine is known to exert potent antioxidant and carbonyl-scavenging activities, raising the possibility that reduction of oxidative stress may represent an upstream event in the suppression of Drp1-NLRP3 signaling found in the present study. ROS are recognized as upstream triggers of Drp1 activation and mitochondrial fission and can thereby promote mitochondrial dysfunction and downstream inflammasome activation[66,67]. In sensory neurons relevant to PDN, hyperglycemic stress has shown to engage Drp1-associated mitochondrial injury and fragmentation in DRG neurons[68].

In this context, it is plausible that attenuation of intracellular ROS by carnosine contributes to downstream suppression of Drp1 activation and NLRP3 inflammasome signaling. However, the precise temporal sequence between ROS modulation of the Drp1-NLRP3 regulation was not directly examined in the present study, and future experiments are required to clarify this causal relationship.

These findings are consistent with recent reports that excessive mitochondrial fission can lead to inflammasome activation through release of mitochondrial damage-associated molecular patterns and loss of MMP[69,70]. Taken together, the present study supports the relevance of carnosine to mitochondrial dynamics and inflammatory signaling in PDN.

Limitations of the study

Several limitations of this study should be acknowledged. Firstly, although preventive carnosine supplementation was associated with beneficial effects in the PDN model, its pharmacokinetics, tissue exposure, and bioavailability under diabetic conditions remain unclear. These parameters may vary under hyperglycemic and neuroinflammatory states, and additional pharmacodynamic profiling is needed before considering clinical application. Although mitochondrial ultrastructural abnormalities and reduced MMP were detected, direct measurements of mitochondrial ROS production, ATP synthesis, and respiratory function were not performed; therefore, broader conclusions regarding mitochondrial bioenergetic impairment should be interpreted cautiously. In addition, electrophysiological evaluation was primarily based on the number of action potentials evoked at 2 × rheobase, whereas other intrinsic excitability parameters, including rheobase, input resistance, and firing threshold, were not comprehensively characterized. Secondly, the PC12 cell line used in the in vitro assays does not fully recapitulate the molecular and electrophysiological features of primary sensory neurons. Validation using primary DRG neurons or human induced pluripotent stem cells-derived sensory neurons would enhance the robustness of the mechanistic conclusions. Future studies employing more physiologically relevant sensory neuronal models, such as primary DRG neurons or established sensory neuronal cell lines, will be valuable to further validate and extend the mechanistic insights of this study. Thirdly, although the present study concentrated on mitochondrial dysfunction-driven inflammatory signaling in DRG neurons, future studies examining whether carnosine modulates classical pain-related molecules, including calcitonin gene-related peptide, substance P, transient receptor potential vanilloid 1, and voltage-gated sodium channels, are important to further define its effects on peripheral sensitization in PDN. Fourthly, although the NLRP3 inflammasome was identified as a key downstream effector of Drp1 activation, the broader inflammatory and glial responses in the spinal cord or central nervous system were not investigated, even though these processes might contribute to the full spectrum of neuropathic pain in PDN. Finally, while carnosine treatment improved both glycemic control and diabetic neuropathy in the present study, the current experimental design did not allow us to determine whether its neuroprotective effects are independent of glucose modulation. Future studies incorporating glycemia-matched control groups or insulin-controlled experimental designs are required to clarify the role of metabolic control vs direct neuroprotection.

CONCLUSION

In conclusion, preventive carnosine supplementation was associated with improved pain-related behaviors, nerve conduction, mitochondrial morphology, and MMP in rats with PDN. This study further revealed reduced Drp1- and NLRP3-related signaling in the DRG, while mechanistic experiments in high-glucose-stimulated PC12 cells supported a role of the Drp1-NLRP3-caspase-1 axis. These findings support the relevance of mitochondrial dynamics and inflammasome-related signaling in PDN and demonstrate that carnosine warrants further mechanistic and preclinical evaluation. Accordingly, the present study should not be interpreted as demonstrating clinical efficacy for the treatment of established PDN in human patients.