Diabetic peripheral neuropathy (DPN), a major complication of diabetes mellitus (DM), affects approximately half of patients with DM (Sloan et al., 2021). DPN primarily impairs sensory nerve function, but also impacts autonomic and motor nerves (Selvarajah et al., 2019). Approximately 30% of patients with DPN experience painful symptoms; however, only a limited number of patients achieve adequate pain relief with current therapeutic strategies (Jensen et al., 2021). Metformin (Met), a first-line therapy for type 2 DM (T2DM), has been reported to alleviate neuropathic pain in various contexts, including chemotherapy (Hacimuftuoglu et al., 2020), osteoarthritis (Wang et al., 2019), post-surgery (Smith and Tran, 2019), and inflammation (Augusto et al., 2022), possibly via activation of AMP-activated protein kinase (AMPK). Supporting this mechanism, PPARα agonists have been shown to improve DPN through the PPARα-AMPK-PGC1α-eNOS axis in db/db mice (Cho et al., 2014; Kołodziejski et al., 2017). In addition, it has been reported that inhibition of fatty acid-binding protein 5 (FABP5), an intracellular fatty-acid binding protein that regulates lipid signaling and metabolism, may also ameliorate peripheral neuropathy (Warren et al., 2024). Despite these therapeutic potentials, current strategies have not resulted in satisfactory clinical outcomes in the treatment of DPN.

Schwann cells (SCs), glial cells of the peripheral nervous system, play essential roles in supporting the growth, maintenance and repair of peripheral nerves (Bosch-Queralt et al., 2023). Their biological functions have been shown to deteriorate under hyperglycemic conditions, leading to various nerve dysfunctions including reduced conduction velocity and axonal atrophy (Dyck and Giannini, 1996; Eckersley, 2002). Although the precise mechanisms are unclear, metabolic changes in SCs under hyperglycemic and/or hyperlipidemic conditions, such as oxidative stress via overproduction of reactive oxygen species (ROS) as well as activation of the polyol pathway, glycation of lipids and proteins, and modulation of protein kinase C activity, are likely to be involved in the pathogenesis of DPN (Mizukami et al., 2011; Obrosova et al., 2007; Song et al., 2003; Vincent et al., 2009; Zhu et al., 2023). Indeed, recent studies suggest that dyslipidemia associated with DM contributes to the development and progression of DPN (Obrosova et al., 2007; Vincent et al., 2009). Therefore, amelioration of both metabolic dysregulation and ROS overproduction in SCs may be a promising therapeutic strategy for DPN. To investigate DPN pathogenesis and explore therapeutic strategies, in vitro models using cultured SCs under high glucose conditions have been widely utilized (Jiang et al., 2020; Zhang et al., 2021).

Imeglimin (Ime), the first drug in a new tetrahydrotriazine class of oral antidiabetic drugs called “glimins”, is expected to overcome many of the limitations of current therapies for T2DM (Konkwo and Perry, 2021). Recent clinical studies in both Japanese and Caucasian patients with T2DM have demonstrated that Ime provides significant and durable antihyperglycemic effects with favorable safety and tolerability profiles (Dubourg et al., 2021; Reilhac et al., 2022). Although the biological effects of Ime on DPN have not been fully elucidated, Ime, but not Met, has been reported to reduce ROS production, decrease the number of damaged mitochondria, and normalize mitophagic activity in islet β-cells of db/db mice (Lachaux et al., 2020; Nishiyama et al., 2023). Thus, Ime may have therapeutic potentials against DPN via suppression of ROS production in SCs.

In the present study, we aimed to elucidate the effects of Ime on DPN by comparing the effects of Ime and Met on cellular metabolic functions and ROS generation in mouse SCs under high glucose conditions. We also examined whether a PPARα agonist pemafibrate (Pema) or a FABP5 inhibitor MF6 modulates these actions under the same conditions.

To study the effects of antidiabetic or lipid metabolism modulating including Met and Ime, Pema and MF6 on DPN pathogenesis including metabolic derangements and oxidative stress, high glucose-stimulated SCs were used as an in vitro model mimicking pathogenesis of DPN. In this study, we adopted a concentration of 0.5 mM or 2 mM for both Met and Ime, which exceeds their clinically observed plasma levels (Chevalier et al., 2023; Kajbaf et al., 2016). This dosage was selected based on previous reports demonstrating dose-dependent effects on AMPK activity and mitochondrial function at concentrations ranging from 0.25 to 10 mM in various cell types (Hozumi et al., 2023). Notably, 0.5–2 mM is commonly used for evaluating mitochondrial responses in vitro (Hallakou-Bozec et al., 2021; Hozumi et al., 2023; Takahashi et al., 2024).

First, we examined the types of FABPs expressed in mouse SCs and the glucose concentration-dependent ROS levels. As shown in Figure 1A, we confirmed positive expression of FABP5 and FABP7 among FABP3, FABP5 and FABP7, which are known to be related to the pathogenesis of DM and neuronal disorders (Boneva et al., 2011). FABP5 and FABP7 were also detected in SCs at protein levels (Supplementary Figure 1). Moreover, we found that ROS levels were increased in a glucose concentration-dependent manner in SCs (Figure 1B). Therefore, in the subsequent experiments, we adopted MF6, a specific inhibitor for FABP5 and FABP7, to inhibit the role of FABPs and selected 50 mM glucose as the high-glucose condition, considering it suitable for examining the effects of Ime, Met, and other agents in a more severe hyperglycemic state that mimics the pathogenesis of DPN. We also confirmed negligible levels of toxicity in SCs by (1) normal-glucose (5.5 mM) and high glucose (50 mM) conditions, (2) 2 mM Met, (3) 2 mM Ime, (4) 10 μM Pema and (5) 10 μM MF6 (Figure 1C).

Next, to address effects of antidiabetic agents, Met and Ime, on cellular metabolic functions of Schwann cells under a high glucose condition, we conducted extracellular flux analysis of SCs that were treated with Met and Ime under normal-glucose (5.5 mM) and high glucose (50 mM) conditions, respectively (Figures 2A–D). SCs treated with high glucose showed no statistically significant changes in metabolic parameter except for an increase in glycolytic reserve, yet there was a tendency toward enhanced metabolic activity with high glucose treatment (Figures 2E–K). Treatment with Met increased glycolytic functions and decreased baseline OCR/ECAR ratio (Figure 2L), but interestingly, treatment with Ime rather enhanced mitochondrial respiratory functions (Figures 2E–I). Such metabolic changes were more evident in SCs that were treated with normal-glucose condition compared those with high glucose condition (Figures 2E–I). Ime also increased proton leak regardless of glucose concentrations, whereas Met did not affect proton leak (Figure 2G). The findings suggest that Ime, as opposed to Met, enhances most of mitochondrial respiratory functions in SCs (Figures 2M, N).

Since it has been well-recognized that lipid metabolism derangements were also associated with the pathophysiology of the development of DPN, we next examined the effects of Pema (Figures 3A, B) and MF6 (Figures 4A, B) on metabolic functions in SCs that were cultured with normal-glucose and high glucose conditions. Pema tended to increase cellular metabolism toward energetic in the high-glucose condition, but the effects were more evident in glycolytic indices than mitochondrial respiratory indices (Figures 3C–J). In contrast, MF6 did not affect glucose dependent metabolic alterations albeit SCs expressed both FABP5 and FABP7 (Figures 4C–J). These findings suggest that PPARα-related signaling, but not FABP5/7-related signaling, may have influence high glucose-dependent metabolic alteration.

To study the effects of Ime, Met, Pema and/or FABPs on high glucose-induced oxidative stress of SCs, levels of ROS were measured. As shown in Figure 5A, (1) levels of ROS were significantly increased under a high glucose condition compared to those under a normal-glucose condition and (2) such elevated levels of ROS were reduced to varying degrees by 2 mM Met and Ime but not by 0.5 mM of those. The effects of 2 mM Ime were more potent than those of 2 mM Met. We also evaluated the combined effects of 2 mM Met or Ime with Pema and/or MF6 under a high glucose condition. As shown in Figures 5B, C, (1) Pema significantly increased the generation of ROS regardless of the conditions with or without Met or Ime and (2) MF6 substantially reduced the suppressive effects of Met and Ime on generation of ROS. Therefore, taken together, the results indicated that Ime had a more potent inhibitory effect than that of Met on high glucose-induced generation of ROS in SCs and the beneficial effects were greatly influenced by FABPs and PPARα-related signaling.

Finally, to explore the potential mechanisms by which Ime or Met suppresses high glucose-induced ROS production in SCs, we assessed the expression of hypoxia-related genes including HIF1A and VEGFA, which are closely associated with regulation of cellular metabolism and ROS production (Majmundar et al., 2010). Although the expression levels of these genes were decreased in a high glucose condition compared with a normal glucose condition in SCs, there were no statistically significant difference between Ime-treated group and Met-treated group under a high glucose condition (Supplementary Figure 2). These findings suggest that suppression against high glucose-induced ROS production by Ime or Met are independent of hypoxia inducible factor-related signaling.

A schematic summary of the present study is shown in Figure 6.

It has been shown that elevated levels of glucose induce oxidative stress in neuronal cells and Schwann cells (Askwith et al., 2009; Khan et al., 2023; Obrosova et al., 2005; Schmeichel et al., 2003) and that increased levels of ROS harm the lipids in myelinated nerves, leading to the loss of axons and destruction of the microvasculature in the peripheral nervous system (Vinik et al., 2006). As for the possible underlying mechanisms causing the hyperglycemia-induced oxidative stress, it was suggested that the persistent hyperglycemic state in T2DM could modulate mitochondrial DNA (mtDNA) and nuclear DNA (nDNA) (Fernyhough et al., 2003), thereby inducing mutations in the mitochondrial genome as well as mitochondrial dysfunction to produce ROS (Hollensworth et al., 2000; Srinivasan et al., 2000). In the present study, a high glucose condition caused significant increases in the levels of ROS and glycolytic reserve in SCs. In addition, the high glucose-induced increase in the levels of ROS was substantially reduced by Met and Ime and the effect of Ime was much more potent than that of Met. Since previous studies showed that Met had beneficial effects on diseases of the peripheral nervous system including DPN by inhibition of the harmful effects of ROS on cell number, cell viability, and migration (Ma et al., 2015), the present study suggested that Ime may have more potent beneficial effect than that of Met on DPN.

Intriguingly, while Met decreased mitochondrial respiratory functions, Ime rather increased indices of mitochondrial respiratory functions in SCs, especially in a normal glucose condition. In addition, Ime increased proton leak irrespective of low- or high-glucose exposure (Figure 2G). Given that Ime suppressed high glucose-induced ROS production in SCs (Figure 4), such increase in proton leak may be associated with ROS suppression via activating of uncoupling proteins (Mailloux and Harper, 2011), although there seems no significant increase in proton leak by Ime in other types of cells including hepatocytes or pancreatic islet β-cells (Hozumi et al., 2023; Li et al., 2022). Indeed, previous reports have shown that Ime increases levels of nicotinamide adenine dinucleotide (NAD⁺) (Hallakou-Bozec et al., 2021; Takahashi et al., 2024), which is a potentially positive regulator of uncoupler proteins through NAD⁺-Sirt1/AMPK-PGC-1α pathway (Nishikiori et al., 2025; Xu et al., 2021). Furthermore, we evaluated the expression levels of representative mitochondrial respiratory chain complexes in SCs in both low- and high-glucose conditions with and without Ime, but there were no remarkable differences in these protein levels (Supplementary Figure 3). Therefore, the changes in cellular metabolism by Ime observed in the present study are likely to be due to the activation of whole metabolic pathways, or changes in the intermittent metabolites, rather than changes in the expression or activities of the mitochondrial respiratory chain complexes themselves in SCs. Future comprehensive analysis may clarify these unresolved molecular mechanisms. Nevertheless, it is noteworthy that Ime activated cellular metabolism in SCs without increasing ROS production, whereas activation of cellular metabolism is generally associated with ROS production. This comprehensive finding supports the idea that Ime may ameliorate disease states in which increased ROS is the main pathophysiology. Although further clinical trials are needed to determine whether Ime significantly reduces the incidence of DPN compared to other anti-diabetes agents, the present findings support the recent notion that the choice of anti-diabetes agent should not focus solely on lowering blood glucose levels, but should also consider its protective effects on various organs (Qaseem et al., 2024).

It has been shown that dyslipidemia causes vascular insufficiency, oxidative stress-induced deterioration of mitochondrial function, and impaired conduction of electrical impulses in neurons, which are involved in the onset and progression of DPN (Davis et al., 2008; Wiggin et al., 2009), and normalization of lipid metabolism is therefore a suitable therapeutic strategy for DPN. A previous study showed that oral administration of the PPARα agonist fenofibrate, which is widely used to treat hyperlipidemia (Wierzbicki et al., 2003), prevented the progression of sciatic neuropathy in diabetic mice by activating of the AMPK-related signaling pathway (Cho et al., 2014). Furthermore, it has been shown that a PPARα agonist stimulated corneal nerve regeneration in patients with T2DM (Teo et al., 2023) and protected the corneal nerve from degeneration in a mouse model of diabetes (Matlock et al., 2020), suggesting that the use of the PPARα agonist may be promising strategy for treating DPN. In the present study, although the PPARα agonist Pema increased both mitochondrial and glycolytic functions under a high glucose condition, only Pema induced a significant increase in the levels of ROS and the Pema-induced elevation in the levels of ROS was synergistically enhanced by Met and Ime. Similar to this, a recent study showed that bezafibrate improved mitochondrial morphology and functions despite causing a mild increase in the production of ROS using fibroblasts obtained from patients with dynamin-1-like protein (DNM1L) mutation (Douiev et al., 2020). In contrast, other studies showed that bezafibrate reduced the levels of ROS significantly in hiPSC-derived neural stem cells (NSC), early neural progenitors (eNP), and neural progenitors (NP) (Augustyniak et al., 2019) and that Pema inhibited angiotensin II-induced production of ROS in human vascular smooth muscle cells by increased catalase activity (Amioka et al., 2022). Furthermore, it was shown that fenofibrate could prevent DPN by protecting endothelial cells through VEGF-independent activation of the PPARα-AMPK-PGC-1α-eNOS-NO pathway (Cho et al., 2014). However, reason why effects of Pema on levels of ROS were different from other fibrates such as bezafibrate and fenofibrate remained to be elucidated. It was speculated that much higher efficacies of PPARα agonist activities and different off-target biological effects of Pema compared to other fibrates may be involved (Yamashita et al., 2020). Taken together, the results suggested that effects of the PPARα agonist Pema on production of ROS may be different among various experimental conditions using different sources of cells.

FABPs, which are known as intracellular lipid chaperones, were found to be expressed in cells of the central and peripheral nervous systems (De León et al., 1996; Furuhashi and Hotamisligil, 2008). Functionally, FABPs interact with various fatty acids and other endogenous lipids including endocannabinoids, thereby playing a pivotal role in the development of the nervous system as well as the mature nervous system to modulate neuronal activity and network functions (Lu and Mackie, 2021). Previous studies have shown that FABP5-deficient mice and mouse models with pharmacological inhibition of FABP5 showed an increase in brain levels of Anandamide (AEA) (Kaczocha et al., 2014; Yu et al., 2014). In contrast to FABP5, the contribution of FABP7 remains to be elucidated, although the expression of both FABP5 and FABP7 was observed in various nervous cells (Glaser et al., 2023; Kawahata and Fukunaga, 2023). However, it was shown that elevated levels of AEA were comparable in FABP5/7 KO mice and FABP5 KO mice (Kaczocha et al., 2014; Yu et al., 2014), suggesting that the FABP that is responsible for the regulation of the AEA levels may be FABP5 but not FABP7. In support of this, the expression level of FABP7 in the adult mouse brain is low (Owada et al., 1996). In the current study, we also detected the expression of both FABP5 and FABP7 in SCs and the expression level of FABP7 was lower than that of FABP5. In the present study, pharmacological inhibition of FABP5 and FABP7 by MF6 did not alter high glucose induced metabolic changes in SCs, whereas Ime- and/or Pema-induced effects on the levels of ROS production under high glucose condition, suggesting that FABP5 and FABP7 may stimulate high glucose induced oxidative stress in SCs independently of their metabolic states. In fact, a recent study showed that FABP5 caused mitochondrial dysfunction related to αSyn oligomerization/aggregation in mitochondria, thereby inducing oxidative stress in neurons (Wang et al., 2021).

The present study has several limitations. First, the precise molecular mechanisms by which Ime reduces oxidative stress in high glucose-treated SCs remain unclear. Although Ime alleviated oxidative stress under high-glucose conditions, this antioxidant effect occurred without concurrent improvement in mitochondrial respiration, suggesting the involvement of alternative pathways. Specifically, NAD⁺-related metabolism, including the role of nicotinamide phosphoribosyltransferase (NAMPT), is presumed to be the central molecular mechanism of action of Ime (Hallakou-Bozec et al., 2021). Elucidating these pathways through metabolomics, multiomics analyses including transcriptomics, and stable isotope tracer approaches across various cell types may further clarify the therapeutic potential of Ime. Second, while ROS levels were evaluated at both 0.5 mM and 2 mM concentrations of Met and Ime, the relevance of these concentrations to physiological or therapeutic levels remains uncertain. Further investigations are needed to validate whether similar effects occur under in vivo conditions. Third, we did not perform osmolarity equalization using agents such as sucrose or mannitol, as our aim was to mimic the pathophysiological hyperglycemic conditions observed in DM, where elevated glucose levels impose both metabolic and osmotic stress on cells. Finally, the findings in the present study do not directly demonstrate the efficacy of Ime in ameliorating DPN. In addition, the effects of Ime and Met on mitochondrial function have not been characterized in detail. Further studies are needed to clarify the additive roles of FABP and PPARα in Ime-induced responses by elucidating their downstream signaling and additional mitochondrial parameters, such as mitochondrial membrane potential in SCs. Moreover, in vivo studies using diabetic animal models are required to evaluate the effects of Ime on sensory disturbances and nerve conduction velocity.

In conclusion, the antidiabetic agent Ime had a more potent suppressive effect than that of Met on a high glucose-induced in vitro model mimicking pathogenesis of DPN using mouse SCs and that their effects were exclusively and synergistically modulated by FABP or PPARα-related signaling. However, since all findings are obtained from in vitro models, the therapeutic potential of Ime for DPN should be interpreted with caution and requires further validation in in vivo studies.