Combined PTEN Knockdown and Local Insulin in Chronic Experimental Diabetic Neuropathy Free

Diabetic polyneuropathy (DPN) is a common, chronic, irreversible peripheral neurodegenerative complication of persons with type 1 and type 2 diabetes. Sensory neurons are especially vulnerable, but the reasons are unknown (1–7). DPN development in type 1 diabetes also correlates with inadequate replacement of physiological levels of insulin, whereas in type 2 diabetes, there is resistance to circulating insulin (8). In vivo, low-dose subhypoglycemic insulin applied intrathecally, intranasally (to access cerebrospinal fluid or dorsal root ganglia [DRG] root sleeves), or near nerve can each selectively reverse features of experimental DPN (9–12). By acting on insulin receptors (IRs) (12–15), insulin acts through canonical growth factor signaling cascades that emphasizes phosphoinositide 3-kinase (PI3K)-phosphorylated Akt (pAkt) intrinsic support of growth (16,17) but are blocked by the gateway molecule phosphatase and tensin homolog deleted on chromosome 10 (PTEN) (18). PTEN inhibition or knockdown (KD) enhances the outgrowth of neurites from adult sensory neurons in vitro and axon regrowth in vivo (18). Moreover, levels of PTEN in sensory neurons are elevated in both experimental types 1 and 2 diabetes (19), and local KD of PTEN improves regeneration in experimental diabetes (19). The objective of this work was to identify whether PTEN inhibition or KD and local insulin had additive impacts on the plasticity of sensory neurons, including that of their skin branches, and their partnering Schwann cells (SCs) (20,21).

Cohorts of adult male CD1 mice (8–12 weeks of age, 30–50 g) and rats were used in compliance with University of Alberta Health Sciences Laboratory Animal Services and Canadian Council on Animal Care guidelines. Diabetes was induced by three consecutive daily injections of intraperitoneal streptozotocin (85, 70, and 55 mg/kg; Sigma-Aldrich, St. Louis, MO) dissolved in citrate buffer (pH 4.5), and control littermates received citrate buffer alone. Diabetes was confirmed at a fasting glucose level >16 mmol/L (OneTouch Ultra 2 glucometer) and reconfirmed at end point. Interventions were performed after a duration of diabetes of 4 months, with treatments over 28 days. Embryonic cell isolation was approved by the institutional animal care and use committee at the National University of Singapore.

E15 rat embryos were obtained from pregnant C02 euthanized Sprague-Dawley rats (InVivos Pte Ltd, Singapore) and placed in a cold L15 medium (Gibco by Life Technologies, Singapore). DRGs and spinal cords were stripped of glia then placed in L15 medium supplemented with 1% penicillin-streptomycin and dissociated with 0.05% trypsin/EDTA (Thermo Fisher Scientific, Waltham, MA). Next, cells were distributed into 96-well plates coated with poly-D-lysine (50 μg/mL; Thermo Fisher Scientific) and laminin (10 μg/mL; Invitrogen, Carlsbad, CA). Seeding for neurons was at a density of 10,000 cells/well in single culture and at 5,000 cells/well in coculture with mouse SCs (ScienCell Research Laboratories, Carlsbad, CA), which were plated at 3,000 cells/well (five sectors each imaged randomly in two fields). Postnatal day 9 mouse SCs isolated from sciatic nerves were purchased from ScienCell Research Laboratories.

Media were prepared by adding glucose (Invitrogen) to complete SC medium (ScienCell Research Laboratories) and complete Neurobasal medium (Gibco, Grand Island, NY). For single SC cultures, elevated glucose concentrations in the media were at 10, 30, and 60 mmol/L. For single embryonic neuron cultures, elevated glucose concentrations were at 35, 45, and 60 mmol/L. Baseline concentrations were 25 mmol/L for neurons and 5.5 mmol/L for SCs. For embryonic neurons, the tested concentrations did not include lower values that are not considered optimal for growth. The choice of concentrations was based on published work (22–24) and preliminary studies of their impact. SCs were first seeded into a 96-well plate at the density of 3,000 cells/well and induced to maturity in basal SC media supplemented with vitamin C (50 μg/mL) for 4 days followed by neuron seeding. In coculture of SCs and neurons, complete Neurobasal medium with the glucose concentration at 60 mmol/L was used as a maximal supraphysiological stress.

Neurons were incubated for 2 days in complete Neurobasal medium supplemented with insulin (Sigma-Aldrich) at 0.01–2.0 μmol/L. SCs were incubated in complete SC medium supplemented with insulin at 0.1–5.0 μmol/L for 2 days. For combined experiments, SCs were cultured in the absence or presence of bpV(pic) (0.1–5 μmol/L; Sigma-Aldrich) and/or insulin (1 μmol/L) for 2 days. SCs and neurons were cocultured in Neurobasal medium (adjusted to 60 mmol/L glucose) for 7 and 14 days, maintained by changing media every 2 days.

After removing media, the cells were rinsed with PBS three times for 3 min, fixed by 4% paraformaldehyde (Sigma Singapore) for 30 min at room temperature, and then followed by PBS rinsing and blocking in 3% BSA and 0.25% Triton X-100 for 1 h at room temperature. They were then incubated overnight at 4°C with the following primary antibodies: rabbit anti-insulin receptor (RRID:AB_2296149, 1:250, cat. no. ab5500; Abcam) for SCs and mouse anti-NeuN (RRID:AB_10711040, 1:1,000, cat. no. ab104224; Abcam) for embryonic neurons, rabbit anti-insulin receptor (1:250), and rabbit anti-caspase-3 (RRID:AB_443014, 1:200, cat. no. ab13847; Abcam). For coculture of SCs and neurons, mouse anti-NF160 (RRID:AB_306083, 1:500, cat. no. ab7794; Abcam) and rabbit anti-MBP (RRID:AB_1141521, 1:200, cat. no. ab40390; Abcam) antibodies were used. After washing with PBS, cells were incubated for 1 h at room temperature with Alexa Fluor 488 goat anti-mouse (RRID:AB_143160, 1:1,000, cat. no. A11017; Molecular Probes, Eugene, OR) and Alexa Fluor 594 goat anti-rabbit (RRID:AB_142057, 1:500, cat. no. A-11072; Molecular Probes) as the secondary antibodies. They were then rinsed with PBS three times, each time for 3 min, and imaged on a confocal Zeiss LSM 800 microscope with Airyscan (RRID:SCR_015963; Carl Zeiss, Jena, Germany). IR area, mean fluorescence, and several adjacent background readings of IR and DAPI or NeuN in single channels of each image were first measured using Fiji software (RRID:SCR_002285; https://fiji.sc). The total corrected cellular fluorescence was calculated (25) as integrated density – (area of selected cell × mean fluorescence of background readings). The ratio of caspase-3 cells to the total number of NeuN neurons was used to report apoptosis. Neurites stained with both NF-160 and MBP antibodies were considered myelinated. At 7 and 14 days, the length of unmyelinated and myelinated neurites were measured by Simple Neurite Tracer, an open-source plugin of Fiji, as previously described (26).

For adult neuron immunocytochemistry, L4, L5, and L6 DRGs from adult rats were harvested and maintained in DMEM F12 medium with added 1:100 N2 (Gibco), penicillin-streptomycin 100 units/mL (Invitrogen), 100 ng nerve growth factor (Invitrogen), and 10 μmol/L cytosine β-arabinose furanoside (Sigma) and at 24 h, were stained with anti-NF200 antibody (1:600, cat. no. N4142; Sigma-Aldrich) with secondary goat anti-rabbit antibody, Alexa Fluor 546 (cat. no. A-11035; Thermo Fisher Scientific). Cells were imaged with a fluorescence microscope (Axioscope, Carl Zeiss) at 20× resolution. Quantification of neurite outgrowth was evaluated by WIS-NeuroMath software.

Adult SCs were harvested from mouse DRGs and separated by centrifugation in 15% BSA, then plated on poly-L-lysine and laminin precoated slides. The cells were grown in DMEM (Gibco) supplemented with 10% FBS (Gibco), forskolin (6 μmol/L; Tocris Bioscience), NRG-1 (50 ng/mL; R&D Systems), gentamicin (25 μg/mL; Gibco), and penicillin-streptomycin (50 units/mL, each) for 5–7 days, then fixed with 4% paraformaldehyde. SCs were stained with mouse anti-PTEN antibody (1:100, cat. no. sc-7974; Santa Cruz Biotechnology) and either rabbit anti-Sox10 (1:200, cat. no. ab227680; Abcam) or rabbit anti-p75 (1:200, cat. no. 8874; Abcam) overnight at 4°C then incubated with secondary goat anti-mouse Alexa Fluor 546 (1:200; Invitrogen) and goat anti-rabbit Alexa Fluor 488 (1:200; Invitrogen) for 1 h at room temperature. SCs were visualized on a Leica SP5 confocal microscope.

Antibodies were rabbit anti-insulin receptor (cat. no. ab5500; Abcam), mouse anti-NeuN (cat. no. ab104224; Abcam), rabbit anti-caspase-3 (cat. no. ab13847; Abcam), mouse anti-NF160 (cat. no. ab7794; Abcam), rabbit anti-MBP (cat. no. ab40390; Abcam), Alexa Fluor 488 goat anti-mouse (cat. no. A-11017; Molecular Probes), Alexa Fluor 594 goat anti-rabbit (cat. no. A-11072; Molecular Probes), goat anti-rabbit, Alexa Fluor 546 (cat. no. A-11035; Molecular Probes), anti-PTEN antibody (cat. no. sc-7974; Santa Cruz Biotechnology), rabbit anti-Sox10 (cat. no. ab227680; Abcam); rabbit anti-p75 (cat. no. 8874; Abcam), and mouse NF200 antibody (cat. no. N4142; Sigma-Aldrich).

The survival of postnatal SCs was assayed at 2 days by the LIVE/DEAD Viability/Cytotoxicity Kit (Molecular Probes) according to the manufacturer’s instructions. Briefly, the medium was removed and the cells rinsed gently with PBS. The working solution was prepared by diluting calcein AM and EthD-1 in SC media to final concentrations of 2 and 4 μmol/L, respectively, with 100 μL added into wells, and transferred to the incubator for 30 min. Cells were rinsed twice with PBS, and images were obtained by laser scanning confocal microscopy (Zeiss LSM 800 microscope with Airyscan).

The proliferation rate of postnatal SCs was evaluated by alamarBlue (AB) (Bio-Rad, Hercules, CA) staining at 2 days of culture. SCs were seeded into a 96-well plate at a density of 1,000 cells/well, grown in media supplemented with 10% FBS, then washed three times in PBS. AB was then diluted by FBS-free SC media (1:10) and added (250 μL) to each well, and the plate was incubated for 2 h. Incubated AB solution of each well was then transferred to four separate wells (50 μL per well) of a new 96-well plate for measurement of the fluorescence absorbance using a microplate reader at the excitation and emission wavelengths of 570 nm and 600 nm, respectively. The number of viable cells correlated with the percentage of AB reduction calculated according to the manufacturer’s instructions.

Intact uninjured and 3-day sciatic nerve axotomized DRGs at L4, L5, and L6 DRGs from adult rats were used for in vitro sensory neuron outgrowth studies. DRGs were isolated and cultured in DMEM F12 medium with added 1:100 N2 (Gibco), penicillin-streptomycin 100 units/mL (Invitrogen), 100 ng NGF (Invitrogen), and 10 μmol/L cytosine β-arabinose furanoside (Sigma-Aldrich). These primary cultured cells were treated with insulin at 10 nmol/L or scrambled sequenced siRNA or PTEN siRNA at 20 nmol/L or in combination for 24 h. siRNAs were mixed with HiPerFect Transfection Reagent (QIAGEN, Germantown, MD) diluted in DMEM F12 medium and incubated for 20 min. After 24 h, the cells were fixed with 2% PFA for 10 min and then underwent immunohistochemistry. Rat siRNA sequences for in vitro studies were 5′-ATCGATAGCATTTGTAGTATA-3′ and scrambled control siRNA 5′-AATTCTCCGAACGTGTCACGT-3′.

PTEN or scrambled siRNA (QIAGEN) was mixed in HiPerFect Transfection Reagent for 20 min at room temperature, then with saline. Sequences were PTEN siRNA 5′ATCGATAGCATTTGCAGTATA-3′ and scrambled control siRNA 5′-AATTCTCCGAACGTGTCACGT-3′. While under isoflurane anesthesia, 20 μL of PTEN siRNA solution was subcutaneously injected into the plantar skin of the right hindleg and near the sciatic nerve (with insulin 0.1 IU in 10 μL of saline or saline carrier 1 h prior to siRNA injection into the plantar skin). Scrambled control siRNA solution was subcutaneously injected into the right plantar skin and near the sciatic nerve (with the saline carrier 1 h prior to siRNA injection into the plantar skin). The dosing was 1.56 μL of PTEN siRNA (20 μmol/L) or scrambled siRNA reconstituted in 5 μL of HiPerFect and saline (13.44 μL) for a total volume of 20 μL subcutaneously at the paw and sciatic nerve for 3 days weekly for 4 weeks (total of 12 doses). The PTEN siRNA–alone mice were treated identically without insulin. Immediately following injection, Hanks’ balanced salt solution was pipetted onto the surface of the plantar skin and sciatic nerve to facilitate conduction. Electrodes were then held in contact with the skin, with the negative electrode positioned near the site of injection and the positive electrode positioned on the proximal end of the plantar skin. Five 25-V pulses at 1 Hz and 50 ms duration were given with an ECM 830 Electro Square Porator unit.

Motor and sensory electrophysiological recordings (compound muscle action potentials [CMAPs], sensory nerve action potentials [SNAPs], motor conduction velocities [MCVs], and sensory conduction velocities [SCVs]) were performed percutaneously as in previous work (10) under isoflurane anesthesia using platinum subdermal needle electrodes inserted at the stimulation and recording sites, maintaining a near-nerve temperature of 37.0 ± 1.0°C. Briefly CMAPs were recorded over the tibial innervated interossei on the dorsum of the foot, with supramaximal stimulation at the sciatic notch and knee. SNAPs were recorded from notch and knee with stimulation of intraplantar sensory nerves. Distance measurements were made three times, and the average was taken.

Mice were placed in compartments atop a plexiglass platform, and a thermal conducting lamp was positioned directly below the hind paw. After acclimatization, thermal sensitivity was tested by applying a radiant heat source to the middle of the hind paw and then timing the latency (in seconds) to withdrawal using a plantar test device (Ugo Basile) and measured in triplicate. To prevent damage to the animal’s paw, exposure to the thermal lamp was limited to a maximum of 30 s. Triplicate measurements were performed to give an average withdrawal latency. To assess mechanical sensitivity, mice were placed on a metal mesh platform within individual compartments accessed to apply Von Frey filaments of increasing force to the animal’s hind paw until the filament bent. Filaments used in these experiments ranged from 0.07 to 4 gauge. A positive response was recorded if the filament incited a withdrawal response 60% of the time over five trials (three withdrawals of five trials).

Bilateral footpads were harvested with a 3-mm punch and placed in paraformaldehyde, L-lysine, and sodium periodate solution. After 24 h, samples were rinsed with 0.1 mol/L Sorenson phosphate buffer, cryoprotected overnight at 4°C in a 20% glycerol solution, then frozen in optimal cutting temperature compound and stored at −80°C. Footpad sections of 25 μm thick underwent antigen retrieval in 65°C Tris-EDTA buffer solution for 1.5 h, then were blocked and stained using a 1:500 solution of rabbit anti–protein gene product 9.5 (EnCor Biotech Inc., Gainesville, FL) as the primary antibody and goat anti-rabbit antibody, Alexa Fluor 546 (1:200; Thermo Fisher Scientific) as the secondary antibody. Analysis of footpad sections used a Leica SP5 confocal microscope to obtain Z-stack images using a step size of 0.5 μm and 63×/1.2 objective lenses. For each skin section, five consecutive frames were imaged using ImageJ software in three randomly chosen sections per animal counted (blinded). The number of both horizontal (<45° angle from dermal epidermal junction) and vertical (>45° angle) nerve fibers crossing from the dermis into the epidermis were counted, averaged, and expressed per millimeter of epidermis or as an axon density by a blinded experimentalist. All fibers analyzed were in the epidermis.

L4–L6 DRGs were immersed in Zamboni fixative at 4°C overnight, washed in PBS, and kept in 20% phosphate-buffered sucrose for 12 h, then embedded in Tissue-Tek optimal cutting temperature compound. DRGs were cryosectioned at 2 μm, then stained with anti-PTEN antibody (1:500, cat. no. sc-7974; Santa Cruz Biotechnology) and colocalized with anti-NF200 antibody (1:600, cat. no. N4142; Sigma-Aldrich) or rabbit monoclonal anti-pAkt [1:200, Phospho-Akt (Ser473) (193H12) antibody, cat. no. 4058; Cell Signaling Technology) colocalized with mouse monoclonal anti-β3 tubulin antibody (1:600, cat. no. T8328; Sigma-Aldrich). Sections were washed with PBS containing 0.05% Tween 20 and blocked with 1% BSA and 5% goat serum for 30 min, then incubated with antibody in a humid chamber at room temperature for 2 h. The secondary antibodies were Alexa Fluor 488 goat anti-mouse (1:200; Invitrogen) and Alexa Fluor 546 goat anti-rabbit (1:200; Invitrogen) for 90 min at room temperature. Sections were washed and mounted with DAPI to stain nuclei (VECTASHIELD, Vector Laboratories, Newark, CA). Using ImageJ software, five pAkt or PTEN images per animal (n = 4) had mean gray values measured for integrated fluorescence intensity per area and the percentage of pAkt-positive cells were counted (five images per animal).

Samples were fixed in 2.5% glutaraldehyde buffered by 0.025 mol/L cacodylate. They were then postfixed in 2% osmium tetroxide, dehydrated, and embedded in epon resin. Sections 1-μm thick from the distal end of the nerve were then sectioned using an ultramicrotome, mounted onto slides, and stained with 0.5% toluidine blue. Using bright-field oil immersion microscopy images at 100× (Zeiss Axioscope), the numbers and density of myelinated axons per area in four sections per animal (n = 3–4 mice) were counted using ImageJ software.

Results were calculated as means ± SEM. Groups were compared with one-way (occasionally two-way, as appropriate) ANOVAs with Tukey post hoc tests (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 in the figures and legends). Two subanalyses used paired t tests from subgroups of data derived from same-mouse comparison of sides (#P ≤ 0.05).

Primary data are available through the corresponding author.

Dissociated embryonic sensory and motor neurons and SCs from 9-day postnatal mice were analyzed at 2 days during 25–60 mmol/L glucose concentrations (Fig. 1A–C) with labeling for IR and the neuronal marker NeuN. No alteration in fluorescence intensity of IR expression in neurons or SCs was identified (Fig. 1D and E). Adult SCs (Fig. 1F) and sensory neurons (Fig. 1G) expressed PTEN.

The percentage of activated caspase-3, an index of cellular stress or apoptosis (27) in embryonic neurons, also reported as caspase-3/NeuN, was lessened by insulin (Fig. 2A–D), and the number of live neurons cells increased with insulin.

At 2 days, AB staining identified rises in neonatal SC proliferation in response to insulin, which were greatest at 1 μmol/L (Fig. 2E and F). Concentrations of the PTEN inhibitor bpV(pic) <2 μmol/L combined with insulin showed the highest proliferation rates (Fig. 2G). At higher doses of insulin, the proliferation impact was reduced.

We confirmed impacts of insulin or PTEN KD (18,28) but noted that the combination was associated with greater evidence of outgrowth: mean neurite outgrowth and branching (Fig. 3) in both uninjured and previously injured neurons. The impact of neurite outgrowth was greater than that of brain-derived neurotrophic factor, which did not significantly improve overall growth but increased sprouting.

Trends toward greater myelination of sensory neurons observed by 7 days with insulin alone were significant when combined with the PTEN inhibitor bpV(pic) (Fig. 4A and B). The length of myelinated neurites treated by insulin combined with bpV(pic) was greater than that of the untreated or insulin-alone groups (Fig. 4C). By 14 days, all groups had near-complete myelination (Fig. 4D and E), but the mean myelination length by combined insulin and bpV(pic) was greater than the untreated or bpV(pic)-alone groups (Fig. 4F).

In 4-month diabetic mice, we compared MCVs and SCVs in hindlimbs given scrambled control siRNA contralateral and PTEN siRNA in the ipsilateral limb or, in separate mice, scrambled control siRNA with saline contralateral and PTEN siRNA with insulin 10 nmol/L ipsilaterally. CMAP amplitudes were comparable at baseline, but there were declines in CMAPs bilaterally by day 28 unrelated to our experimental end points (Fig. 5A and B). CMAPs are technically highly sensitive to edema, a consequence of repeated local injections and repeated recordings in small mouse hind paws. CMAP configuration and latencies were otherwise well preserved. To confirm this impact, a parallel analysis in nondiabetic mice using an identical injection and testing protocol confirmed declines in CMAP amplitudes with preserved conduction velocities (Fig. 5C–E), allowing confidence in interpreting MCV changes in diabetes.

In diabetic animals at all time points, there were expected reductions in MCVs not altered by scrambled siRNA injections, scrambled siRNA injections + saline carrier, or PTEN siRNA administration. However, at day 28, mouse limbs treated with PTEN siRNA + insulin had improvements in MCV rendering values comparable to control nondiabetic mice. Final MCVs were higher than those of contralateral paws treated with scrambled siRNA + saline (Fig. 5F and G). SNAPs were more variable and, as expected, were not altered by diabetes or by any of the interventions (Fig. 5H and I). SCVs had expected reductions secondary to diabetes. At day 28, there was a mild improvement in nerves treated with PTEN siRNA + insulin, albeit less improvement than MCV and remaining below the range of nondiabetic controls (Fig. 5J and K).

Thermal sensation was diminished in mice with diabetes at onset, and at day 14 of treatment, there was no impact of scrambled siRNA, scrambled siRNA + saline, or PTEN siRNA. However, there was an improvement in mice treated with PTEN siRNA + insulin. At day 28, there was no change in thermal sensation of mice treated with scrambled siRNA, but there was mild improvement when given scrambled siRNA + saline and greater improvement when given PTEN siRNA or especially PTEN siRNA + insulin. In the latter group, final values were comparable to control nondiabetic mice (Fig. 6A and B). Mechanical sensation was diminished in mice with diabetes at the outset, and on day 14 of treatment, there was no change in hindlimbs treated with scrambled siRNA, scrambled siRNA + saline, or PTEN siRNA. However, there was improvement with PTEN siRNA + insulin. At day 28 of treatment, treatment with PTEN siRNA alone or PTEN siRNA + insulin had improved mechanical sensation but not to levels of nondiabetic controls (Fig. 6C and D). Injections of PTEN siRNA or PTEN siRNA + insulin had no impact on glycemic levels (Fig. 6E). Overall, combined treatment hastened early (14-day) and final (28-day) sensory improvement.

Diabetes (scrambled siRNA + saline) alone was associated with a decline in tibial but not sural myelinated axon density and a rise in the g-ratio in both sural and tibial axons. PTEN siRNA with insulin improved tibial myelinated fiber density and both sural and tibial axon g-ratios (Fig. 7A–F). In both sural and tibial nerves, mean axon diameters of myelinated fibers were not different among the groups, indicating that differences in myelin thickness rather than axon caliber were associated with g-ratio rises in untreated diabetes and improvements with treatment. Hind paw epidermal innervation in chronic diabetic mice was reduced and had limited improvement with PTEN siRNA alone that was greater following PTEN siRNA + insulin (Fig. 7G–K).

As a readout of PTEN KD and insulin signaling, pAkt expression in DRG neurons by DRG fluorescence intensity and the percentage of neurons expressing pAkt had rises in mice given PTEN siRNA that were greater in mice given PTEN siRNA + insulin (Fig. 8A–C). To confirm in vivo siRNA KD of PTEN (18,19,29), we noted an ∼50% decline in its DRG protein expression fluorescence intensity (Fig. 8D and E).

In this work, the major findings were that 1) IR expression in neurons and SCs is unaltered by high glucose levels; 2) insulin and PTEN inhibition had additive impacts on improving SC proliferation, in vitro neurite outgrowth, and myelination; and 3) insulin and PTEN KD had additive impacts on improving several features of chronic experimental DPN in mice, including MCV slowing, sensory loss, myelin thinning, and epidermal denervation. As a direct neurotrophic factor, impaired insulin signaling contributes to neuropathic abnormalities in diabetes (2,6,12,30,31). The impact of several neurotoxic metabopathic mechanisms on neuron growth and plasticity is less clear. For example, advanced glycation end product/receptor for advanced glycation end product ligation enhances sensory neuron outgrowth rather than suppresses it (32). Similarly, work targeting polyol flux or protein SUMOylation have had circumscribed impacts (7). While neurons are not dependent on insulin to take up glucose (31), they express IRs as do SCs (13–15,33,34). Here, we confirmed that IR expression is not altered in high-glucose conditions. Schwannopathy from diabetes (35,36) includes not only SC expression of aldose reductase and inappropriate polyol flux (37) but also glucose-induced changes in p75-caveolin signaling (38) and specific deficits in SC migration (39). During diabetes, SCs may lose their ability to supply energy substrates and instead provide toxic lipid species to axons (40). Importantly, peripheral axons and SCs depend on each other for support and maintenance of the differentiated cellular phenotype and for growth signaling (29). In this work, insulin supported proliferative activity in neonatal SCs. Insulin alone or with PTEN inhibition also enhanced the myelination of neurons in coculture. At higher doses of insulin, however, this impact was lost. While not further investigated here, insulin resistance, as described in neurons exposed to higher insulin doses in vitro, may be a mechanism of this decline (8). Demyelination is an independent feature of human DPN, particularly in severe disease (41), and in our work, there were improvements of myelin thinning in both the sural, a largely sensory nerve, and tibial, a mixed nerve. Epidermal axon counts also identified loss of distal sensory axons in the model, as well as their recovery. Insulin mediates its role through two main pathways, including PI3K/pAkt, also enhanced by PTEN KD, and extracellular signal–regulated kinase/mitogen-activated protein kinase (ERK/MAPK), although the former is regarded as more important (17,42).

The studies of outgrowth and branching in adult sensory neurons confirm a key role to support plasticity. This was observed not only in naive adult neurons without a prior axotomy injury but also with prior axotomy injury. While these experiments did not demonstrate a preconditioning impact associated with greater regeneration from injury alone, we nonetheless considered it important to demonstrate an impact of combined intervention in neurons with a preexisting axotomy lesion (43).

MCV and SCV slowing of human and experimental DPN can develop before myelin thinning or within unmyelinated axons, suggesting alterations in nodal ion channel function perhaps also linked to changes such as resistance to ischemic conduction failure (44–46). Hind paw injections in this work reduced CMAP amplitudes, measures already inherentlyvariable, from injectate swelling in the small mouse hind paw. These technical problems do not preclude studies of local delivery since the changes are less likely in larger animals or humans. Conduction velocity was improved but more apparent in motor than sensory fibers and appeared later in the treatment period. We also noted loss of both mechanical and thermal sensitivity, a consistent finding in this model, in contrast to hyperalgesia sometimes detected in shorter or differing models. Given the lack of change in nondiabetic animals with injections, we believe that the results are uninfluenced by paw swelling. Loss of thermal sensitivity can be associated with primary loss of epidermal axons, although it can be argued that more subtle and earlier dysfunction of axon terminal signaling in DPN develops in the absence of structural change. Dermal axon loss may contribute to the nonnoxious mechanical sensory loss as tested in this work. Our combined strategy reversed thermal loss completely, and similar impacts were noted on mechanical sensation, both of which were associated with improved epidermal innervation, albeit not necessarily the full explanation for the behavior change. Changes in behavior need not require strict parallel alterations in axon numbers (47,48). An additional limitation of the study was that a separate diabetic with insulin group was not added given the impacts already identified in previous work (9).

Despite these caveats, taken together, our findings identify remarkable improvements in combining two interventions that converge on PI3K-pAkt signaling, applied to two different, but connected cellular constituents, neurons and SCs, of the peripheral nervous system. These findings support the concept that strategies that support regenerative and growth programs, both in neurons and SCs, reverse the key hallmarks of experimental diabetes irrespective of specific metabolic alterations that may target them. However, longer-term treatment, more intensive intervention, or further combinations that target disease mechanisms may be required. Finally, we draw attention toward the elegance of a local strategy to reverse DPN using a molecular approach and mobile siRNA that does not require a viral carrier.

Acknowledgments. The authors thank Dr. Nguyet T. M. Nguyen, Dr. Son L. Tran (Agency for Science, Technology, and Research), and Dr. Kien V. Vu (National University of Singapore) for sharing some experimental materials. Dr. Ngoc T. Nguyen (Industrial University of Ho Chi Minh City) initially extracted raw data. Dr. Ambika Chandrasekhar (University of Alberta) provided technical assistance.

Funding. The work was supported by the Canadian Institutes of Health Research (operating grant FRN15686).

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

Author Contributions. V.M.P. and P.K. completed the majority of the experimental work. V.M.P., P.K., and A.A. analyzed results and wrote and edited drafts of the manuscript. A.A. and T.P. completed experimental work and edited drafts of the manuscript. N.T. supervised the work in Singapore and reviewed and edited final drafts of the manuscript. D.W.Z. supervised the overall project and the experiments in Alberta, analyzed and designed experimental work and data from Alberta, and contributed to writing early and final drafts of the manuscript. All the authors offered collaborative intellectual contributions to the manuscript. D.W.Z. 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 data and the accuracy of data analysis.