GPR40 Activation Abolishes Diabetes-Induced Painful Neuropathy by Suppressing VEGF-A Expression

More than half of the patients diagnosed with diabetes develop some form of peripheral neuropathy (1–3). Diabetes-induced painful neuropathies are associated with hyperglycemia-induced increased permeability of the blood-nerve barrier of exo- and endoneurial vessels (2,4,5), affecting sensory afferents and autonomic and motor nerves and leading to altered proprioception and nociception (6). The earliest changes of peripheral neuropathies occur at the level of the unmyelinated C-fibers, with initial changes in ion channel expression and sensitization, which cause, depending on the neuronal subtype, hyper- or hypoexcitability (7–9). Because conduction is more energy consuming in unmyelinated nerves, C-fibers may have increased dependence on metabolic support from Schwann cells. At later stages, axonal demyelination and degeneration of peripheral sensory nerves occur, causing a decrease in intraepidermal nerve fiber density (2,10).

Although the pathomechanisms are not completely understood, excess glucose and fatty acids are metabolized, decidedly increasing the synthesis of reactive oxygen species (ROS) and reactive metabolites, such as oxidative or glycosylated byproducts (2,11,12). These hyperglycemia-induced reactive metabolites support vascular remodeling, causing impaired vasodilation, increased capillary thickening, and endothelial hyperplasia. As a consequence, the oxygen supply decreases, further damaging neuronal cells and evoking exaggerated responses to sensory stimuli (13,14). The decreased oxygen supply and the increased synthesis of ROS can induce the expression of vascular endothelial growth factor-A (VEGF-A), which mediates interruptions of tight junction formation of endothelial cells and vasodilatation and increases blood-nerve barrier permeability, making VEGF-A inhibitors a first-line treatment in diabetic retinopathy (2,4,5,15–18).

The G-protein–coupled free fatty acid receptor GPR40 (FFA1) is considered a promising target for the treatment of type 2 diabetes, causing an intensive search for potent and safe GPR40 agonists (19–21). Activation of GPR40 by endogenous agonists (medium- and long-chain fatty acids, such as docosahexaenoic acid, eicosapentaenoic acid, linoleic acid, and 20-hydroxyeicosatetraenoic acid) enhances the glucose-induced insulin secretion from pancreatic β-cells, preventing hypoglycemia, similar to sulfonylureas (22,23). GPR40 is expressed most abundantly in pancreatic β-cells, but also in several other cell types (e.g., endothelial cells, adipocytes, macrophages, and certain neuronal subpopulations) of the central nervous system and the vascular system, where its function is largely unknown (24–27). Interestingly, activation of GPR40 attenuates mechanical and thermal hypersensitivities in models for inflammatory pain (carrageenan and complete Freund adjuvant) and trauma-induced neuropathy (27,28). We investigated the effects of GPR40 activation on diabetes-induced vascular permeability and peripheral neuropathy using the streptozotocin (STZ) model for type 1 diabetes, in which pancreatic β-cells are destroyed, allowing investigation of β-cell–independent GPR40 functions separately from GPR40 effects on insulin release.

For all animal experiments, male C57BL/6NRj (8–12 weeks old) mice (Janvier, Le Genest, France) or GPR40^−/− mice with C57BL/6NRj background were used (23,29). All mice were kept in a temperature- (22 ± 0.5°C), humidity- (55%), and light-controlled environment. In all experiments, the ethics guidelines for investigations in conscious animals were obeyed, and the procedures were approved by the local ethics committee (Regierungspräsidium Darmstadt, Darmstadt, Germany; permission FU/1198).

Adult nonfasted mice were injected i.p. with 90 mg/kg STZ (Cayman Chemical, Ann Arbor, MI) on 2 consecutive days (3). Body weight and blood glucose levels were determined before and every 7 days after STZ injection using a blood glucose monitor (Omnitest 5; Braun, Melsungen, Germany). Mice were considered hyperglycemic and diabetic with blood glucose levels ≥350 mg/dL. Where indicated, the VEGF receptor inhibitor axitinib (TargetMol) (30 mg/kg in 33% PEG400 in acidified water) was injected daily i.p. starting 4 days after STZ injection. GW9508 (Hycultec, Beutelsbach, Germany) (50 mg/kg in DMSO) and VEGF-A₁₆₄ protein (R&D Systems, Minneapolis, MN) (8 μg/kg in PBS) were injected into naïve mice or at days 10 and 11 after STZ injection.

Blood was collected from mice by cardiac puncture with a syringe, prefilled with citrate buffer 9NC (Sarstedt, Nümbrecht, Germany) and diluted with Dulbecco’s PBS (without Ca²⁺/Mg²⁺)/10% ACD-A buffer (66 mmol/L citric acid monohydrate, 85 mmol/L trisodium citrate dihydrate, and 111 mmol/L d-glucose).

Liquid chromatography–quadrupole time-of-flight mass spectrometry analysis was performed to analyze free fatty acids and triglycerides as previously described (30) on a TripleTOF-6600 (SCIEX, Darmstadt, Germany) coupled to a Nexera-X2 (Shimadzu Corporation, Kyoto, Japan) using positive and negative electrospray ionization (30). Quantitative Liquid chromatography–tandem mass spectrometry was performed to analyze free fatty acids using a hybrid triple quadrupole ion trap mass spectrometer (QTrap-5500; SCIEX) (31,32). The Supplementary Material provides a detailed description of the method.

Frozen tissue was homogenized by sonication (SONOPULS, Bandelin, Germany). MHEC5-T cells (ACC 336; DSMZ GmbH, Braunschweig, Germany) were grown in DMEM medium with low glucose supplemented with GlutaMAX supplement, 10% FCS, and 1% penicillin-streptomycin (all Gibco, Carlsbad, CA; Thermo Fisher Scientific, Waltham, MA) and 50 μg/mL gentamycin (Lonza Pharma & Biotech, Basel, Switzerland) at 37°C and 10% CO₂. Human umbilical vein endothelial cells (HUVECs) were grown in EndoGRO medium (SCME001; Merck, Darmstadt, Germany) at 37°C and 5% CO₂. MHEC5-T cells or HUVECs (10,000 cells/cm²) were incubated overnight before glucose (30 mmol/L) was added for 5 or 24 h. Where indicated, the cells were treated afterward for 19 h with 0.5 μmol/L GW9508 (Hycultec), 0.5 μmol/L AMG837, 5 μmol/L DC260126 (both Cayman Chemical), 1 μmol/L GF109203X or 10 μmol/L U0126 (both Tocris Bioscience, Bristol, U.K.). RNA was isolated using the mirVana miRNA Isolation Kit, and quantitative RT-PCR was performed in triplicate with 10 ng cDNA using the First Strand cDNA Synthesis kit and Gene Expression Assay Kit (both Thermo Fisher Scientific). Primers used were GPR40 (Mm00809442_s1), Vegf-a (Mm00437306_m1), and Gapdh (Mm99999915_g1) (all Thermo Fisher Scientific). Analysis was performed by using the ΔΔ C(T) method, whereby each replicate of the controls was normalized to the mean of those controls (33).

Cryosections (10 μm) from sciatic nerves or aortas were fixed using ice-cold methanol for 10 min and blocked for 60 min (3% bovine serum albumin and 10% goat serum in PBS) followed by incubation with antibodies (Supplementary Table 1). Images were taken (Observer.Z1; ZEISS, Oberkochen, Germany) and analyzed using Image J (version 1.52i; National Institutes of Health, Bethesda, MD) by determining the signal intensity in a 250-pixel range with line plot profiles. Values <50% of the maximal intensity were defined as absent.

Mice were injected i.v. with 5 mL/kg 1% Evans Blue in PBS. After 1 h, they were euthanized, and sciatic nerves were prepared as described above. Sciatic nerve cryosections were analyzed by microscopy (Observer.Z1) using Image J software (version 1.52i). Fluorescent signal intensity (RawIntDen) of sciatic nerve tissue was divided by the signal in blood vessels using five different nerve sections per mouse.

After euthanasia of the mice, the sciatic nerves were exposed and incubated with nanoparticles (0.06 μm SkyBlue fluorescent particles; Kisker-Biotech, Steinfurt, Germany) diluted 1:200 in PBS for 10 min. Sciatic nerves were washed, dissected, crushed, and digested with 3 mg/mL collagenase IA (Sigma-Aldrich, Deisenhofen, Germany) in DMEM (Gibco) for 45 min at 37°C. After filtration through 70-μm nylon mesh (BD Biosciences, Franklin Lakes, NJ), the flow-through was analyzed by FACS (MACSQuant Analyzer-10; Miltenyi Biotech) and FlowJo software (version 10; Treestar, Ashland, OR).

For transendothelial electrical resistance (TEER) experiments, HUVECs (15,000 cells/100 μL) (SCCE001) were added apically to fibronectin-coated thincerts (0.4 μm) (Greiner Bio-One, Kremsmünster, Austria), incubated for 2 h at 37°C, and then inserted into 600 μL medium (basolateral). After overnight incubation, inserts were transferred into CellZScope (Nanoanalytics, Münster, Germany). After 4 days, the compounds were added apically and basolaterally. TEER values 1 h before compound addition were set as 100%.

To determine expression of angiogenic proteins, the Proteome Profile Mouse Angiogenesis Array Kit (R&D Systems) was used. Proteins (300 μg) were extracted from aortas of naïve mice and incubated with the membranes according to the manufacturer’s protocol. Chemiluminescent signals were detected with various exposure times to determine mediators with high or low expression. Analysis was performed with Image J (version 1.52i). Data <10,000 RawIndDen (pixel) were considered to be under the detection level.

For Western blot analysis, membranes were blocked with Intercept Blocking Buffer (LI-COR Biosciences, Lincoln, NE), incubated with antibodies (Supplementary Table 1), detected with the Odyssey Imaging System, and analyzed using Gel Analyzer software (LI-COR Biosciences).

MHEC5-T cells were grown in six-well plates overnight at 37°C and 10% CO₂. The next day, glucose (30 mmol/L) was added for 5 or 24 h. Where indicated, the cells were treated afterward for 19 h with the different compounds. VEGF-A release was determined in culture supernatants in duplicate using the Mouse VEGF Quantikine ELISA Kit (R&D Systems).

Mechanical hypersensitivity was determined using a Dynamic Plantar Aesthesiometer (Ugo Basile, Gemonio, Italy) as described previously (34). The time until paw withdrawal was determined by increasing force at the same rate for each trial (0–5 g over 10 s) with a cutoff time of 20 s. Thermal hypersensitivity was determined by using the Hargreaves test as described previously (35). Paw withdrawal was determined by increasing heat at the midplantar region of a paw at the same rate for each trial (up to 12% starting at 32°C) using an IITC Plantar Analgesia Meter (Model 336G; IITC Life Science) (Hargreaves test) with a cutoff time of 20 s. Baseline measurements were performed on 2 consecutive days before STZ injection.

The proximal sciatic nerve was prepared at the plexus, divided into specimens 2 mm wide, fixed in indium tin oxide for 24 h, and embedded in EPON for transmission electron microscopy. Ultrathin cross sections were cut with a microtome (Ultracut; Leica, Jena, Germany) and analyzed with a JEM-1400Plus (JEOL, Tokyo, Japan). For quantitative evaluation of enlargement of neurons and distance measurements, SightX-Viewer (version 1.2.3.537; JEOL) was used. Measurements were performed in axons in the middle region of the nerve. Only exactly cross-sectioned axons, not obliquely sectioned axons, were measured.

Statistical significance was determined by using the Student t test, one-way ANOVA by using the Bonferroni post hoc test, and two-way ANOVA by using the Bonferroni post hoc test through Graph Pad Prism 8 (Graph Pad Software, San Diego, CA). For all panels, data are shown as mean ± SEM.

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

Diabetes was induced by injecting wild-type and GPR40^−/− mice with 90 mg/kg STZ on 2 consecutive days. The STZ model was used, because STZ induces hyperglycemia through destruction of pancreatic β-cells (36,37). This model has the advantage that β-cell–independent GPR40 functions can be studied without altering insulin release. Both genotypes showed no significant change in bodyweight (±5%) throughout the experiment (Fig. 1A) and had comparable development of blood glucose levels over the course of 3 weeks (Fig. 1B). Increased triglyceride plasma levels are a hallmark of diabetes and are one source for free fatty acids, including the GPR40 ligands. Accordingly, triglyceride plasma levels were elevated in wild-type mice 14 days after STZ injection. Notably, naïve GPR40^−/− mice already had increased baseline triglyceride levels, which were similar to the concentrations seen in diabetic wild-type mice (Fig. 1C and Supplementary Table 2). Plasma levels of free fatty acids known to activate GPR40 were not increased in wild-type mice after diabetes induction (Supplementary Table 3). However, basal plasma levels of docosahexaenoic acid were moderately increased only in naïve GPR40^−/− mice compared with naïve wild-type mice but not after diabetes induction (Fig. 1D and Supplementary Table 3). Next, we investigated the effect of hyperglycemia on GPR40 expression in the vascular system and the peripheral nerve system using aortas and sciatic nerves as examples. Gpr40 mRNA decreased in the pancreases of wild-type mice after STZ injection, mirroring the STZ-induced destruction of β-cells (Fig. 1E). Interestingly, aortas and sciatic nerves of wild-type mice expressed Gpr40 mRNA, and this expression increased following the induction of diabetes (Fig. 1F and G).

Because vascular permeability is increased in diabetic mice and GPR40 is expressed by endothelial and epithelial cells (24–27), we investigated whether or not GPR40 deletion affects endothelial tight junction formation. Aortas from naïve GPR40-deficient mice showed significant reduction in the expression of tight junction proteins ZO-1 and claudin-1 compared with wild-type mice (Fig. 2A and B). Similarly, in the epineuriums of sciatic nerves in GPR40^−/− mice, reduced expression of ZO-1 and claudin-1 was observed (Fig. 2C and D). Moreover, 14 days after STZ injection, ZO-1 and claudin-1 signals within the epineurium of the sciatic nerves decreased in wild-type mice to the levels observed in naïve GPR40^−/− mice (Fig. 2C and D). Notably, Western blots showed no significant reduction of ZO-1 or claudin-1 in either tissue (Supplementary Fig. 1), suggesting that at this stage, tight junction dysfunction is based on a shift in the steady state between formation and disassembly of tight junctions.

Because GPR40 depletion reduces endothelial tight junction formation, we determined whether or not vascular permeability is also altered in GPR40^−/− mice. As reported previously (38), the Evans Blue staining was significantly increased in sciatic nerve tissue of diabetic wild-type mice, demonstrating an increased vascular permeability (Fig. 3A). Interestingly, vascular permeability was also increased in naïve GPR40-deficient mice (Fig. 3A), which is in accordance with the low tight junction protein expression in sciatic nerves in these mice. To study whether or not the integrity of the perineuriums and epineuriums of sciatic nerves is compromised in GPR40^−/− mice, we determined the ability of fluorescent nanoparticles with an average size of 0.06 μm to diffuse into exposed sciatic nerves (31). In wild-type mice, the number of nanoparticles in sciatic nerves significantly increased 21 days after STZ injection, demonstrating diabetes-induced microlesions (Fig. 3B and Supplementary Fig. 2A). In sciatic nerves from naïve GPR40-deficient mice, the number of fluorescent nanoparticles was fourfold higher than that in wild-type mice and did not significantly increase after diabetes induction (Fig. 3B), demonstrating the reduced integrity of perineurial and epineurial barrier function.

Because the data showed that the integrity of endothelial barriers was decreased in GPR40-deficient mice, we studied whether or not GPR40 directly influences endothelial barrier permeability. The GPR40 antagonist DC260126 significantly decreased the barrier permeability of HUVECs as determined by TEER measurements and was as efficient as the positive control EGTA (5 mmol/L). In contrast, the GPR40 agonist AMG837 had no effect on barrier permeability (Fig. 3C). This increased permeability was not due to cytotoxic effects, because for concentrations up to 5 μmol/L DC260126 or AMG837, no cytotoxicity was observed (Supplementary Fig. 2B). Taken together, the loss of GPR40 negatively affects tight junction formation of endothelial cells, increases endothelial barrier permeability, and causes microlesions in the epineuriums in sciatic nerves.

To investigate in more detail the mechanism underlying the GPR40-mediated regulation of vascular permeability and endothelial tight junction formation, we determined the differential expression of 52 angiogenic factors in aortas from naïve wild-type and GPR40^−/− mice using an angiogenesis array (Fig. 4A–C). VEGF-A isoforms were investigated independently by Western blot analysis, because it was not addressed in the angiogenesis array. Strikingly, VEGF-A was the only factor known to increase endothelial permeability, which was elevated in GPR40^−/− mice (Fig. 4D–E). VEGF-A isoforms VEGF-A₁₆₄ and VEGF-A₁₈₈ were detectable in aortas, whereas only VEGF-A₁₆₄ was significantly increased in naïve GPR40^−/− mice (Fig. 4D–F).

To examine whether or not GPR40 regulates VEGF-A expression, we first used mouse endothelial cells (MHEC5-T), which express Gpr40 mRNA at levels similar to aortas or sciatic nerves (Supplementary Fig. 3A and B). Because GPR40 is upregulated during hyperglycemia, MHEC5-T cells were incubated with high glucose concentrations (30 mmol/L), which induced Gpr40 mRNA expression in MHEC5-T cells by twofold (5 h) to threefold (24 h) (Fig. 5A). Vegf-a mRNA showed only an intermittent increase at 5 h before returning to baseline (Fig. 5B). Importantly, 24-h incubation with glucose together with the GPR40 antagonist DC260126 increased Vegf-a mRNA expression, whereas the GPR40 agonist AMG837 did not affect Vegf-a mRNA expression (Fig. 5C), suggesting that GRP40 signaling is already fully activated under the culture conditions. Next, we tested the involvement of the main signaling pathway activated by GPR40 (protein kinase C [PKC] signaling) and, as control, ERK1/2 signaling as a known inducer of VEGF expression in Vegf-a mRNA expression and VEGF-A release. As expected, Vegf-a mRNA expression decreased after ERK1/2 inhibition and, in accordance with the involvement of GPR40, increased after PKC inhibition (Fig. 5D).

Concerning the release of VEGF-A from MHEC5-T cells, the GPR40 antagonist DC260126 increased VEGF-A release, while the GPR40 agonists AMG837 and GW9508 suppressed VEGF-A release (Fig. 5E). Interestingly, the suppressive effect of GPR40 activation was reversed by inhibition of ERK1/2 signaling but not by inhibition of PKC (Fig. 5F). Thus, the data suggest that GPR40-mediated suppression of Vegf-a mRNA expression and VEGF-A release is mediated by different signaling pathways, including PKC-mediated suppression of Vegf-a mRNA expression and ERK1/2-mediated suppression of VEGF-A release.

Treatment of HUVECs with high glucose concentrations for 24 h did not affect Gpr40 mRNA expression but induced a moderate increase in Vegf-a mRNA expression (Fig. 5G and H). Importantly, the GPR40 antagonist increased Vegf-a mRNA expression, while GPR40 agonists had no effect on VEGF-A expression (Fig. 5I). Taken together, the data show that GPR40 inhibition increases VEGF-A expression and its release from endothelial cells, while GPR40 agonists decrease VEGF-A release from endothelial cells

Next, we studied the role of GPR40 in the development of diabetes-induced hypersensitivity. STZ-induced hyperglycemia caused in wild-type and GPR40^−/− mice a robust mechanical hypersensitivity starting 7 days after STZ injection (Fig. 6A). Also, wild-type mice showed mild thermal hypersensitivity, which reverted at later time points to thermal hyposensitivity (Fig. 6B). In contrast to the wild-type mice, GPR40^−/− mice did not show thermal hypersensitivity but instead showed similar thermal hyposensitivity to wild-type mice (Fig. 6B). The thermal hyposensitivity marks a functional loss of function of thermosensitive neurons. Here, gross morphology of the myelinated nerve fibers showed no differences between diabetic mice and controls, whereas some of the unmyelinated axons were swollen and enlarged (Fig. 6C and D and Supplementary Fig. 4A). The distances between adjacent axon membranes and between axon membranes and cell membranes of the Schwann cells were enlarged, and the cytoplasm of nonmyelinating Schwann cells ensheathing the unmyelinated axons were slightly swollen (Fig. 6C and E and Supplementary Fig. 4B). Thus, we observed alterations in unmyelinated neurons and their microenvironment but no axon degeneration. Fittingly, immunohistochemistry did not show immune cells within the sciatic nerves, except quiescent resident macrophages (F4 80⁺/Iba1⁻) (Supplementary Fig. 5).

To test the role of VEGF signaling in the development of the observed mechanical and thermal hypersensitivities, wild-type mice were treated daily with axitinib (30 mg/kg i.p.), which inhibits VEGF receptors 1/2 (IC₅₀ 0.1–0.2 mmol/L) and, with lesser potency, platelet-derived growth factor receptor and cKIT (IC₅₀ 1.6–5 mmol/L) (39). The axitinib treatment completely prevented development of STZ-induced mechanical hypersensitivity (Fig. 6F) and thermal hyper- and hyposensitivities (Fig. 6G) as well as the diabetes-induced decrease of ZO-1 and claudin-1 expression in the epineurium of the sciatic nerve (Fig. 6H and I). Also, axitinib administration reverted the altered tight junction formation in naïve GPR40^−/− mice to levels similar to those in wild-type mice (Supplementary Fig. 6).

Next, we determined whether or not GPR40 agonists could decrease STZ-induced hypersensitivities in a VEGF-A–dependent manner. Therefore, the GPR40 agonist GW9508 (50 mg/kg i.p.), and VEGF-A (8 μg/kg i.p.) were injected alone or together on 2 consecutive days (days 10 and 11 after STZ application) into wild-type mice. GW9508 completely reversed diabetes-induced mechanical and thermal hypersensitivities (Fig. 7A and B). In contrast, injection of VEGF-A alone had no effect on mechanical hypersensitivity but produced a strong thermal hyposensitivity (Fig. 7A and B), which is normally observed only at later stages of the disease. Coinjection of VEGF-A with GW9508 completely reversed the analgesic effects of GW9508 on mechanical hypersensitivities (Fig. 7A). Likewise, GW9508 was not able to improve thermal hyposensitivity induced by VEGF-A (Fig. 7B), supporting the notion that VEGF-A acts downstream of GPR40. The GPR40 agonist GW9508 did not affect mechanical or thermal paw withdrawal latencies in STZ-treated GPR40^−/− mice, demonstrating the GPR40-specific effect of GW9508 (Supplementary Fig. 7). Concerning its downstream mediators, we found that GW9508 treatment inhibited Vegf-a expression in sciatic nerves (Fig. 7C) and reversed the diabetes-induced reduction of ZO-1 and claudin-1 expression in the epineuriums of the sciatic nerve proteins (Fig. 7D and E). Notably, GW9508 injection did not affect tight junction formation in naïve wild-type mice (Supplementary Fig. 8). VEGF application was able to suppress the effect of GW9508 on both tight junction proteins (Fig. 7D and E) and induced microlesions in the epineuriums of sciatic nerves (Fig. 7F). Importantly, VEGF-A administration decreased tight junction formation in naïve wild-type mice without affecting mechanical or thermal sensitivity (Supplementary Fig. 9), underlining the notion that increased vascular permeability in the absence of increased glucose concentrations is not sufficient to induce neuropathies. Taken together, the data show that GPR40 activation suppresses glucose-induced VEGF-A expression in endothelial cells and decreases diabetes-induced vascular permeability and hypersensitivities.

GPR40 is an interesting target for the treatment of type 2 diabetes, because it supports the glucose-induced insulin secretion from pancreatic β-cells (22,23,40). GPR40 is also expressed in the human vascular system in endothelial cells and smooth muscle cells, where its function is largely unknown (24,41). Here we show that GPR40 deficiency leads to the downregulation of tight junction proteins in the vascular system, causing increased vascular permeability. This is in accordance with a previous report showing that GPR40 promotes tight junction assembly in airway epithelial cells (26); our data extend the regulatory role of GPR40 for tight junction stability toward vascular endothelial cells as well as epineural cells of peripheral nerves. Toward the mechanism through which GPR40 can regulate endothelial barrier, we observed an upregulation of VEGF-A in GPR40^−/− mice. VEGF-A has a well-established role in diabetic vascular diseases, which includes an increased blood-nerve barrier permeability (2,4,5,42). The key role of VEGF signaling for diabetes-induced painful neuropathy in the STZ type 1 diabetes model is demonstrated by our finding that the VEGF receptor inhibitor axitinib completely prevents the development of mechanical and thermal hypersensitivities. Because axitinib can also inhibit platelet-derived growth factor receptors and cKIT, the involvement of VEGF was shown using a second line of evidence by administration of VEGF-A, which promoted neuronal loss of function as highlighted by the induction of thermal hyposensitivity.

VEGF-A is expressed in endothelial cells and has the ability to alter tight junctions of endothelial cells and increase vascular endothelial permeability (16,18,43). We observed upregulation of Vegf-a mRNA expression in human and murine endothelial cells in the presence of high glucose concentrations. This upregulation was further enhanced in the presence of a GPR40 antagonist. In contrast, cotreatment with a GPR40 agonist prevented upregulation and release of VEGF-A from endothelial cells and preserved tight junction integrity. Interestingly, GPR40 effects on Vegf-a mRNA expression and VEGF-A release were mediated by different signaling pathways. ERK1/2 and PKC mediated suppression of Vegf-a mRNA expression, and independently of PKC, ERK1/2 mediated suppression of VEGF-A release, as previously described for GPR91 (44).

The integrity of the vascular endothelial barrier depends on tight junctions, which restrict paracellular flow and regulate the integrity of the blood-nerve barrier (45–48). Because increased vascular permeability alone was not sufficient to induce mechanical or thermal hypersensitivity in naïve GPR40-deficient mice, additional factors are necessary to initiate neuronal hyperexcitability and cause at later stages the typical neuronal loss of function. Here, the increased vascular permeability in the absence of GPR40 allows easier access for the excess amounts of free fatty acids and glucose under hyperglycemic conditions, thereby inducing metabolic stress through ROS synthesis (13,14,49,50). Thermal stimuli are transduced by small unmyelinated sensory C-fibers, which are believed to be more sensitive to damaging metabolic factors produced during hyperglycemia because they lack the protection and nutrient supplementation provided to mechanosensitive myelinated Aδ-fibers by myelinating Schwann cells (2,8). Notably, the STZ model for type 1 diabetes in the presented form models the early phases of diabetes-induced neuropathy, which only showed swelling of some of the unmyelinated nerves but no axonal degeneration or inflammatory responses within the sciatic nerves.

We found that Gpr40 expression was upregulated after induction of STZ-induced hyperglycemia in aortas and sciatic nerves. Because in parallel tight junction proteins were downregulated and vascular permeability increased, the upregulation of GPR40 could point toward a compensatory mechanism aiming to reestablish the endothelial barrier. However, GPR40 upregulation was not accompanied by an increase of known endogenous GPR40 agonists, whose absence was further underlined by the strong effects of the GPR40 agonist on Vegf-a mRNA expression, tight function formation, and hypersensitivity. Most importantly, treatment of diabetic mice with a GPR40 agonist completely reversed diabetes-induced downregulation of tight junction proteins as well as diabetes-induced hypersensitivities. Because vascular permeability regulates the access of glucose to peripheral nerves and initiates neurotoxic effects, increased permeability becomes one of the key factors in progression of diabetic neuropathy. Therefore, decreasing vascular permeability using GPR40 agonists may cause diabetic neuropathies to revert in the early stages and may delay or even stop progression of neuropathies at later stages. Thus, in addition to supporting insulin release, therapy for type 1 or type 2 diabetes–induced polyneuropathies might present an additional beneficial consequence of GPR40 agonist treatment.

Acknowledgments. The authors thank Professor Stefan Offermanns for contributing the GPR40^−/− mice and providing helpful discussion; Susanne Schiffmann for enabling us to perform the TEER measurements; Thomas Ulshöfer, Karin Schilling, Annette Wilken-Schmitz, Maike Herborn, and Andrea Eichhorn for providing excellent technical assistance; and Carlo Angioni for providing the analytical data.

Funding. This work was supported by DFG (Deutsche Forschungsgemeinschaft) grants SCHO817/3-3, SFB1039 (TP A08 and Z01), and GRK2336 (TP07) and the Fraunhofer Foundation Project: Neuropathic Pain as well as the Fraunhofer Cluster of Excellence for Immune-Mediated Diseases, Frankfurt am Main, Germany.

The funders had no role in the design of the study, collection, analysis, or interpretation of data, writing of the manuscript, or decision to publish the results.

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

Author Contributions. V.K. performed in vivo and in vitro experiments. V.K. and K.S. designed experiments and wrote the paper. S.P. bred and provided GPR40^−/− mice. S.P. and V.R. assisted with in vivo experiments. M.S. and E.L.-D. performed and analyzed TEM measurements. J.W. prepared the material for transmission electron microscopy. L.H. and G.G. performed and analyzed liquid chromatography (LC)–quadrupole time-of-flight mass spectrometry (MS) and LC-MS/MS measurements. All authors read and approved the manuscript. K.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.