DW14006 as a Direct AMPKα Activator Ameliorates Diabetic Peripheral Neuropathy in Mice

Diabetes is a chronic metabolic disease, and long-term hyperglycemia injures the tissues and organs such as large vessels, microvessels, heart, brain, kidney, eyes, and feet (1). Microvascular complications including nephropathy, retinopathy, and neuropathy, which are major predictors of type 2 diabetes macrovascular complications, contribute highly to morbidity and mortality (2). Diabetic peripheral neuropathy (DPN) is a common microvascular complication of diabetes affecting up to 66% of patients with diabetes (2–4). Clinical characteristics of DPN include pain, paraesthesia, and sensory loss, which are the risk factors for burns, injuries, and foot ulceration (5,6). Unfortunately, there has not yet been an effective treatment for DPN due to its complex pathogenesis (7,8), and it is imperative to design efficient anti-DPN drugs based on new therapeutic strategies.

Multiple etiological factors are tightly linked to the development of DPN, and there is a growing recognition that physicians and DPN patients must aim to resist multiple risk factors to improve both daily care and clinical outcome (9,10). Pathologically, impaired mitochondrial bioenergetics (mtBE) is one of the main pathogenetic features of DPN. In diabetic rodents, bioenergetics profile and mitochondrial membrane potential (MMP) are aberrant in dorsal root ganglion (DRG) neurons (11), with twofold rise of reactive oxygen species (ROS) in axons (12,13). Inflammation is another vitally pathogenetic feature of DPN, as indicated by high levels of proinflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) in patients and mice with DPN (3,14). In addition, vascular dysfunction is also responsible for diabetic neuropathy, as demonstrated by restrained sciatic nerve blood flow and blood perfusion in diabetic rodents (15).

AMP-activated protein kinase (AMPK) monitors and modulates cellar energy expenditure and energy homeostasis. It regulates anabolic and catabolic pathways involving lipid metabolism and mitochondria-mediated regulation (7). Levels of phosphorylated AMPK (p-AMPK) and peroxisome proliferator–activated receptor γ coactivator-1α (PGC-1α) expressions are decreased in diabetic mice (8,11). Recently, indirect AMPK activators have been reported to activate AMPK by stimulating the AMP/ADP:ATP ratio leading to the inhibition of mitochondrial respiration and further the reduction of mitochondrial ATP synthesis. For example, metformin and resveratrol exhibited abilities in attenuating DPN symptoms, although they both activated AMPK by inhibiting mtBE and showed potential side effects (7,11,16,17). Thus, given the potency of AMPK and the potential defects of indirect AMPK activators, we aimed to study direct AMPK activators against DPN.

Here, we report that small molecule DW14006 (2-[3-(7-chloro-6-[2′-hydroxy-(1,1′-biphenyl)-4-yl]-2-oxo-1,2-dihydroquinolin-3-yl)phenyl]acetic acid) (Fig. 1A) functioned as a direct AMPKα activator and ameliorated peripheral sensory neuropathy and neurovascular dysfunction in both streptozotocin (STZ)-induced type 1 and BKS db/db type 2 diabetic mice by counteracting multiple risk factors including mitochondrial dysfunction, oxidative stress, and inflammation. The related mechanisms were intensively investigated. In addition, the results in adeno-associated virus (AAV) 8–AMPKα–RNAi–injected diabetic mice further verified that DW14006 ameliorated DPN by activating AMPKα. Our work is expected to help better understand the mechanisms underlying the amelioration of direct AMPKα activation on DPN by counteracting multiple risk factors, and highlight the potential of DW14006 in the treatment of DPN.

We designed direct AMPKα activators by the scaffold hopping and bioisosteric replacement principles. A-769662, the first small molecule AMPK activator disclosed by Abbott Laboratories, belongs to thienopyridone family (18), while compound 991 is a typical benzimidazole derivative developed by Merck and Metabasis (19,20). We designed and synthesized new quinolinone-based AMPKα activators retaining the key pharmacophores of A-769662 and compound 991 (Supplementary Fig. 1A). Compound synthesis is described in the Supplementary Material.

Activities of the synthetic compounds in activating recombinant AMPKα1β1γ1 and AMPKα2β1γ1 enzymes were assayed by traditional nonradioactive methods (21). Among the active compounds, DW14006 was finally selected for its highly agonistic activity against AMPKα (Supplementary Fig. 1B and C).

Adult mouse DRG neurons were isolated from normal or diabetic mice as described (11). Neurons from normal mice were grown in defined Hams F12 medium with B27 additives including 10 mmol/L d-glucose and 10 nmol/L insulin, and neurons from diabetic mice were maintained in medium containing 25 mmol/L d-glucose (11).

All animals were received humane care, and animal-related protocols were approved by the Institutional Animal Care and Use Committees at Nanjing University of Chinese Medicine. Male C57BL/6 mice at 7 weeks of age were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd., and male BKS Cg-m^+/+Lepr^db/J (db/db) mice aged 15 weeks were purchased from Model Animal Research Center of Nanjing University. Considering that male mice developed a greater extent of diabetes-induced cognition deficits and peripheral neurovascular dysfunction compared with female mice (22), male mice were here used in animal-based assays.

Type 1 diabetic mice with DPN (STZ mice) were prepared by treating C57BL/6 mice with STZ (23). Briefly, 8-week-old C57BL/6 mice were treated with a single intraperitoneal (i.p.) injection of STZ (Sigma, St. Louis, MO) at 180 mg/kg body weight. Nondiabetic mice (control, age-matched C57BL/6 mice as control in STZ mice-related assay. n = 12) were injected with the same volume of vehicle buffer (Na-Citrate Buffer, pH 4.5). A type 1 diabetic model mouse was defined as a blood glucose level >16 mmol/L after 1-week injection of STZ, and STZ mice were defined at 12 weeks of age (4 weeks after STZ injection) through behavioral tests.

db/db mice aged 16 weeks were used as type 2 diabetic mice that developed features of human DPN (2). Age-matched heterozygotes mice (db/m) were used as control mice in db/db mice–related assay. The criteria used for db/db mice were mainly according to DPN phenotypes (abnormal neurological functions), including but not limited to the blood glucose level.

AAV-induced AMPK knockdown–related assay in DRG tissue was performed to verify the amelioration of DW14006 on DPN through targeting AMPKα. AAV8-AMPKα-RNAi virus and its negative control (NC) vector (AAV8-NC) were purchased from Shanghai Genechem Co., Ltd. The titer of AAV8-AMPKα-RNAi was 9.27 × 10¹² vector genomes/mL. After being injected with STZ for 4 weeks, the STZ-injected C57BL/6 mice were injected with AAV8-AMPKα-RNAi vector or AAV8-NC to tibialis anterior and gastrocnemius muscles (1.7 × 10¹¹ vector genomes/mouse) using 5-μL Hamilton injector (24).

DW14006 was dissolved in physiological saline with 3.5% DMSO and 3.5% tween 80 (vehicle buffer). Nondiabetic C57BL/6 mice, STZ mice, db/db mice or AAV8-AMPKα-RNAi–injected STZ mice (n = 12/group) were daily administrated with 15 or 30 mg/kg DW14006 by i.p. injection for 4 weeks. Control and db/m mice were administrated with the same volume of vehicle buffer as that for diabetic mice. AMPKα expression level was repressed in DRG tissue from STZ mice 2 weeks after being injected with AAV8-AMPKα-RNAi (6 weeks after i.p. STZ) (Supplementary Fig. 4G and H). Administration of DW14006 against STZ mice started after 6 weeks of STZ injection and hyperglycemic status lasted 6 weeks before administration. For db/db mice, DW14006 administration started at 16 weeks of age (2) and hyperglycemic status lasted more than 8 weeks before administration. Animals were killed after the last administration of DW14006 under sodium pentobarbital (5 mg/100 g) anesthesia. The experimental schedule was shown in Fig. 2A and B. LD₅₀ of DW14006 (262.1682 mg/kg) against C57BL/6 mice (i.p.) was obtained by Karber’s method (25). The detail of LD₅₀ calculation was listed in Supplementary Table 5.

Weight and blood glucose levels of mice were weekly measured by published approaches (11). The glucometer used is ROCHE ACCU-CHEK Performa (CAT/TYP:04680464008, SN: 55407954627).

Mice stood on a metal grid with hind paws irritated with a series of Von Frey filaments (Ugo Basile, Comerio, Italy), and a paw withdrawal responding to each irritation was recorded (24,26).

Briefly, a movable radiant heat source was placed directly under the mice plantar surface of the hind paw (Ugo Basile). The average withdrawal time of the left and right hind paws was used for statistical analysis (27).

The motor nerve conduction velocity (MNCV) of the sciatic nerve from the ankle to the sciatic notch was measured via bipolar electrodes with a supramaximal stimulus (3V) of 0.05 ms duration by published description (28).

After anesthesia with isoflurane (RWD, Shenzhen, China), the real-time regional velocity and distribution of blood flow and perfused blood vessel of the sciatic nerve and foot pads were detected by Laser Speckle Contrast Imaging/LSCI (RFLSI Pro; RWD). Blood perfusion areas of the sciatic nerve and foot pads were analyzed by Image J (29).

DRG tissues of the mice were used for paraffin-embedded immunohistochemistry assay (IHC-P) by published approaches (11). The antibodies used for IHC-P were provided in Supplementary Table 1. DAPI was used to stain the nucleus. Morphometric analysis was performed by IHC-Toolbox in Image J (29).

Neurite outgrowth was detected by the approach described in the literature (11). DRG neurons from mice were incubated with primary antibody against neuron-specific β-tubulin isotype III (1:1,000; Sigma Aldrich), followed by incubation with fluorescein 488-conjugated secondary antibody (1:300; Proteintech). Leica fluorescence microscope equipped was used to image. Total length of axon outgrowth was quantified using Neuron J plug-in components in ImageJ software (30).

The integrity of the mitochondrial membrane was measured by staining with JC-1, an ideal fluorescent probe widely used to detect mitochondrial membrane potential (Beyotime Biotechnology, Shanghai, China) and tetramethylrhodamine (Invitrogen) by the published approach (11,31). Fluorescent data were analyzed with Microplate Reader (Molecular Devices, San Jose, CA).

An XF96 Analyzer (Seahorse Biosciences) was used to measure the basal mitochondrial oxygen consumption, maximal respiration, spare respiratory capacity and ATP production of DRG neuron from mice by the published methods (11).

Western blot assay against the homogenates of DRG and spleen tissues was performed as previously described (8) and the details of antibodies were provided in Supplementary Table 2. The blots were imaged using a ChemiDoc MP (Bio-Rad). ELISA assay was performed by commercial kits (Nanjing Jiancheng Bioengineering Institute).

Total superoxidase dismutase (SOD) activity and glutathione (GSH) and malondialdehyde (MDA) levels in serum were determined using commercial kits by manufacturer’s protocols (Nanjing Jiancheng Bioengineering Institute).

Superoxide level was measured by the published approach (32). DRG neurons from mice were seeded in a 12-well plate for 24 h, MitoSOX red mitochondrial superoxide indicator (0.5 μmol/L) and Mito Tracker Green (90 nmol/L) (Invitrogen-Molecular Probes) were added into each well and incubated for 10 min. Fluorescence intensity was quantified with ImageJ software (30).

Considering the poor oral pharmacokinetic properties of DW14006, we also optimized DW14006 by using prodrug strategy. The determined ethyl ester prodrug DW15139 with better pharmacokinetic properties (Supplementary Tables 3 and 4) has well provided leading structural information for DW14006-based anti-DPN drug discovery. In the assay, half male and half female C57BL/6 mice were administered with DW15139 by oral gavage or intravenous injection. Plasma and snap-frozen DRG were collected at eight time points for analysis of DW15139 and its hydrolysate DW14006 levels by liquid chromatography–tandem mass spectrometry assay.

DRG neurons were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocols, and nontargeting siRNA was used as a NC. Interference efficiency was detected by Western blot.

Data were presented as mean ± SE/SD. Unpaired two-tailed Student t test was used for two-group comparison. Statistical comparisons were made by one-way ANOVA or two-way ANOVA with between-group differences identified using Tukey or Dunnett post hoc tests for at least three groups’ comparisons. Significance was defined as P < 0.05.

The data sets generated during the current study are available from the corresponding author upon reasonable request.

As axonal loss contributes largely to DPN induction (33,34), we examined the potential of DW14006 in promoting neurite outgrowth of DRG neuron from mice.

As expected, DW14006 enhanced neurite outgrowth in cultured adult DRG neurons from normal (Supplementary Fig. 3A and B), STZ (Fig. 1B and C), and db/db (Fig. 1D and E) diabetic mice. As indicated in Supplementary Fig. 3C and D, treatment of AMPKα-specific inhibitor compound C (CC, 3 μmol/L, MedChemExpress) (11) obviously blocked the stimulation of DW14006 on neurite outgrowth in cultured DRG neurons from normal mice. Moreover, AMPKα knockdown by AAV8-AMPKα-RNAi deprived DW14006 of the capability in promoting neurite growth of DRG neurons from AAV8-AMPKα-RNAi–injected STZ (STZ+AAV8-AMPKα-RNAi) mice (Fig. 1F and G).

As indicated in Fig. 2, STZ and db/db mice both exhibited increases in 50% paw withdrawal threshold (Fig. 2C and D) and thermal response latencies (Fig. 2E and F), while decreases in MNCV (Fig. 2G and H), indicative of the typical pathological features of DPN. Obviously, DW14006 treatment (15, 30 mg/kg) ameliorated all above-mentioned DPN pathologies (Fig. 2C–H). Notably, AMPKα knockdown by AAV8-AMPKα-RNAi could block such capabilities of DW14006 in STZ+AAV8-AMPKα-RNAi mice, additionally, AMPKα knockdown alone failed to affect the sensory sensitivity and MNCV in STZ mice (Fig. 2I–K). Moreover, DW14006 had no effects on sensory sensitivity and MNCV in nondiabetic mice (Supplementary Fig. 9).

As indicated in Fig. 3A–D, both STZ and db/db mice showed decreases, and DW14006-treated (15, 30 mg/kg) diabetic mice showed increases, in blood flow velocity and blood perfusion ratio of sciatic nerve and foot pad tissue. Notably, AMPKα knockdown by AAV8-AMPKα-RNAi blocked such capabilities of DW14006 in STZ+AAV8-AMPKα-RNAi mice (Supplementary Fig. 6).

Since oxygen consumption rate (OCR) is vital for mtBEs and its impairment is highly related to DPN development (11), OCR was measured in DRG neurons from mice by the Seahorse Biosciences XF96 analyzer.

OCR in DRG neurons from STZ (Fig. 4A) and db/db (Fig. 4B) mice gave low levels of basal respiration (Fig. 4C and D), ATP production (Fig. 4E and F), maximal respiration capacity (Fig. 4G and H), and spare respiratory capacity (Fig. 4I and J) compared with those from age-matched control mice. DW14006 (30 mg/kg) treatment ameliorated these effects (Fig. 4C–J), and AMPKα knockdown by AAV8-AMPKα-RNAi deprived DW14006 of such amelioration against DRG tissue from STZ+AAV8-AMPKα-RNAi mice (Fig. 4K–O).

Next, the potential of DW14006 in ameliorating MMP of DRG neurons was detected by using JC-1 and tetramethylrhodamine dyes (11). As indicated in Fig. 4P–Q and Supplementary Fig. 7, MMP levels were decreased in DRG neurons from diabetic mice and high glucose (HG)-treated (45 mmol/L) primary DRG neurons but increased by DW14006 treatment, implying that DW14006 protected mitochondria from apoptosis in DRG neurons.

Considering the potency of AMPK/SIRT1/PGC-1α (SIRT1, silent information regulator factor 2-related enzyme 1) signaling in sensing oxidative stress and mitochondrial function (8), we detected the potential regulation of DW14006 against this pathway. As indicated in Fig. 5A–D, compared with the results from control mice (control, db/m), the levels of p-AMPKα and PGC-1α in DRG homogenate were decreased from diabetic mice and increased from DW14006-treated (15, 30 mg/kg) diabetic mice, while SIRT1 expression level in DRG homogenate was unaltered from STZ mice but downregulated from db/db mice, which was in agreement with the published report (35). Notably, DW14006 (15, 30 mg/kg) treatment increased SIRT1 expression in DRG homogenate from either type of diabetic mice compared with that from vehicle-treated diabetic mice. In DRG homogenate from STZ+AAV8-AMPKα-RNAi mice, the protein expression levels of AMPKα, p-AMPKα, and SIRT1 were downregulated compared with the results from AAV8-NC–injected STZ mice (STZ + AAV8-NC), and DW14006 administration gave no influence on these protein expressions in STZ+AAV8-AMPKα-RNAi mice (Fig. 5E and F). In addition, mitochondrial complexes I (NDUFS3) and IV (COX IV) were commonly used to evaluate mtBE dysfunction (36), and our results indicated that DW14006 treatment stimulated expressions of NDUFS3 and COX IV, and AMPKα knockdown by AAV8-AMPKα-RNAi blocked the activity of DW14006 in stimulating these two complexes of DRG tissue from STZ+AAV8-AMPKα-RNAi mice (Supplementary Fig. 4).

Additionally, it was noted that DW14006 rendered no effects on the expressions of phosphorylated nuclear factor-κB (p-NF-κB) and PGC-1α after treatment with SIRT1 siRNA in HG-cultured (45 mmol/L) DRG neurons from normal mice (Fig. 6).

Given that mammalian target of rapamycin (mTOR) is a target of AMPK and participates in protein synthesis by a high energy-consuming process (37), we detected the potential of DW14006 in regulating mTOR signaling.

As indicated in Supplementary Fig. 8, levels of phosphorylated Akt (p-Akt) (Thr308) and phosphorylated mTOR (p-mTOR) (Ser2448) expressions were increased in diabetic mice compared with those in control mice, and DW14006 treatment decreased p-Akt and p-mTOR expressions in DRG homogenate from diabetic mice.

Given that oxidative stress can be initiated by high concentrations of glucose in diabetic mice (12,13), we examined the potential effect of DW14006 treatment on oxidative stress in DRG tissue and serum of diabetic mice.

Given that superoxide is a vital component of ROS (12), we detected superoxide level (MitoSOX) in DRG neurons from mice. As indicated in Fig. 7A, B, and G, the superoxide level was upregulated in DRG neurons from diabetic mice, and DW14006 treatment downregulated such a level. Besides, 8-hydroxy-2deoxyguanosine is a biomarker of mtDNA oxidative damage (38), and our data also suggested that DW14006 reduced 8-hydroxy-2deoxyguanosine fluorescence intensity in DRG neurons from diabetic mice by activating AMPKα, which was verified by AMPKα knockdown assay in STZ+AAV8-AMPKα-RNAi mice (Supplementary Fig. 5A–E). Similarly, SOD, GSH, and MDA (product of lipid peroxidation) were also assayed to evaluate the level of oxidative stress (39). As indicated in Supplementary Fig. 5F–H, DW14006 increased SOD activity and GSH level, while it decreased MDA levels in serum from diabetic mice by activating AMPKα as verified by an AMPKα knockdown assay in STZ+AAV8-AMPKα-RNAi mice.

In oxidative stress, nuclear factor erythroid-2–related factor 2 (Nrf2) is released and translocated into the nucleus, thus triggering the expressions of antioxidant response element–mediated antioxidant enzymes and the cystine/glutamate transporters related to glutathione biosynthesis (40). Thus, we investigated the potential of DW14006 in regulating Nrf2 nuclear translocation in response to its antioxidation effect in DRG neurons. As shown in Fig. 7C, D, and H, Nrf2 fluorescence intensity in the nucleus of DRG neurons was decreased from diabetic mice but increased from DW14006-treated diabetic mice. Notably, DW14006 administration failed to affect Nrf2 fluorescence intensity in the nucleus of DRG tissues from DW14006-treated STZ+AAV8-AMPKα-RNAi mice (Fig. 7E and F).

Given the close linkage of inflammation to DPN, we detected the potential regulation of DW14006 against inflammation by ELISA, Western blot, and immunofluorescence assays.

Western blot results indicated that the protein levels of TNF-α and inducible nitric oxide synthase (iNOS) in splenic tissues were increased from diabetic mice and decreased from DW14006-treated diabetic mice (Fig. 8A and C). Besides, ELISA results also demonstrated that DW14006 treatment obviously suppressed the levels of TNF-α and IL-6 (Fig. 8E). Notably, DW14006 administration rendered no influence on TNF-α levels in serum from DW14006-treated STZ+AAV8-AMPKα-RNAi mice (Fig. 8K).

Since activation of inhibitor of κB (IκB) α via phosphorylation results in the release and nuclear translocation of active NF-κB, leading to upregulation of proinflammatory genes (41), we detected IκBα phosphorylation in DRG homogenate from mice. As indicated in Fig. 8B and D, phosphorylated IκBα (p-IκBα) levels were increased in diabetic mice and decreased in DW14006-treated diabetic mice, thereby suggesting that DW14006 treatment suppressed IκBα phosphorylation.

Next, we inspected the nuclear translocation of NF-κB in DRG tissues from mice. As indicated in Fig. 8F–H, the red fluorescence intensity (NF-κB antibody) in the nucleus was obviously increased in diabetic mice and decreased in DW14006-treated diabetic mice. Notably, DW14006 treatment gave no influence on the nucleus NF-κB fluorescence intensity in DRG tissue from DW14006-treated STZ+AAV8-AMPKα-RNAi mice (Fig. 8I and J).

DPN is a long-term complication of diabetes with a complicated pathogenesis. Several indirect AMPK activators were recently reported to improve DPN, but they activated AMPK involving inhibition of mitochondrial respiration and further the reduction of mitochondrial ATP synthesis (7,16). ATP is necessary for neurite outgrowth of DRG neurons, and its abnormality in mtBEs is highly responsible for DPN etiology (12,18). DW14006 as a direct AMPKα activator improved mitochondrial function and increased mitochondrial ATP generation, implying that DW14006 might overcome the potential deficiency from indirect AMPK activators.

In addition, we determined that DW14006 promoted neurite outgrowth of DRG by regulating AMPKα/SIRT1/PGC-1α signaling axis. AMPK activation enhances SIRT1 activity by elevation of NAD⁺ (42), and SIRT1 catalyzes PGC-1α deacetylation (43) that is highly related to mtBE. These facts have thus highlighted the potency of AMPKα-mediated signaling in neurite outgrowth of DRGs. Moreover, diabetic complications are highly correlated with vascular disease, which is also a major cause of diabetic neuropathy (44). DW14006 treatment efficiently improved peripheral nerve tissue perfusion and blood flow velocity accompanied by nerve function improvement in diabetic mice. All results have strongly supported that activating AMPKα should be a promising therapeutic strategy for DPN.

Hyperglycemia-induced mitochondrial superoxide overproduction leads to oxidative stress, and the vast majority of cellular ROS can be traced back to the mitochondria. SIRT1 as a protein deacetylase is recognized as an effective regulator for preventing cells from oxidative stress damage. SIRT1 activation increases Nrf2 expression and its downstream genes (45) and inhibits NF-κB signaling accompanied by enhancing oxidative metabolism (46). DW14006 suppressed antioxidative stress by enhancing mitochondrial respiration and the AMPKα/SIRT1/Nrf2 signaling axis, thus providing a new strategy for designing reagents against DPN. Additionally, we detected the effects of DW14006 in HG-cultured DRG neurons treated with SIRT1 siRNA (Fig. 6). The results indicated that SIRT1 knockdown phenocopied the effects of AMPK knockdown and has highlighted the central node of SIRT1 in connecting AMPK to improving mitochondrial function and decreasing inflammation.

Notably, DW14006 treatment rendered no impact on the body weight of STZ mice but decreased the body weight of db/db mice. Here, we tentatively ascribed this finding to the lower level of total cholesterol in DW14006-treated db/db mice (Supplementary Fig. 2), and we suggested that some of the improvements of DW14006 in db/db mice may benefit from the reduction of lipid accumulation. In addition, the result that DW14006 treatment had no effects on blood glucose levels in diabetic mice highly verified the benefit of DW14006 on DPN were independent of blood glucose regulation. Interestingly, we also determined that STZ mice showed a hypersensitivity in mechanical stimulus at early stage (first 2 weeks) of diabetes but exhibited a decreased sensitivity in mechanical stimulus after 4 weeks of STZ injection. On the basis of the published reports (17,47,48), we speculated that hypersensitivity seemed to be an early feature and sensory repression and loss were advanced features of diabetic mice. Moreover, DW14006 was also determined to suppress protein synthesis involving downregulation of mTOR signaling to maintain energy homeostasis during cellular energy stress in diabetic status (Supplementary Fig. 8). To our knowledge, our work might be the first report on the cross talk between direct AMPKα activation and DPN regulation.

Finally, DPN improvement was evaluated by MNCV and sensory sensitivity assays in the current work, although these assays are very potent for DPN study (28,49) according to the Diabetes Complications Consortium guidelines, sensory nerve conduction velocity (SNVC)-related assays should be of high complementation to our work.

As summarized in Fig. 8L, DW14006 as a direct AMPKα activator ameliorated DPN-like pathology in diabetic mice by counteracting multiple risk factors with SIRT1 as a central node in connecting AMPKα to regulate mitochondrial function. It enhanced mitochondrial function through the AMPKα/SIRT1/PGC-1α pathway, repressed inflammation via AMPKα/NF-κB signaling, and suppressed oxidative stress via the AMPKα/Nrf2 pathway.

Acknowledgments. The authors thank Linnan Qian (Nanjing University of Chinese Medicine) for expert technical assistance. The authors thank Jiaxun Nie, Xue Lu, and Rui Guo (Nanjing University of Chinese Medicine) for offering guides on immunohistochemistry staining and MNCV test.

Funding. This work was supported by National Science & Technology Major Project “Key New Drug Creation and Manufacturing Program” of China (2018ZX09711002), the National Natural Science Foundation for Young Scientists of China (81703806), Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX18_1600), the Open Project Program of Jiangsu Key Laboratory for Pharmacology and Safety Evaluation of Chinese Materia Medica (JKLPSE201801), and the Project of the Priority Academic Program Development of Jiangsu Higher Education Institutions(Integration of Chinese and Western Medicine).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Xu Xu and X.S. designed the study. W.W., Z.W., W.D., and J.Z. synthetized DW14006. Xu Xu, J.L., Xi. Xu, J.X., J.Y., X.Z., Y.L., X.H., and J.W. performed the animal experiments. Xu Xu analyzed and interpreted data. Xu Xu wrote the manuscript. X.S. reviewed the manuscript. All authors approved the manuscript. J.Z. and X.S. are the guarantors of this work and, as such, had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.