Impaired Nociception in the Diabetic Ins2^+/Akita Mouse

Diabetic sensory neuropathy is a major cause of chronic pain and paresthesias, but the most common symptom is sensory loss, and insensate neuropathy is a dominant risk factor for foot ulcers and amputations (1–3). Rodent models of diabetic neuropathy recapitulate many anatomical and sensory abnormalities observed in patients, and thus appear suitable for translational mechanistic studies (4,5). The sensory abnormalities reported from different rodent models differ qualitatively, with diabetic rats commonly displaying tactile and thermal hypersensitivity, whereas mice usually exhibit loss of sensitivity to mechanical or thermal stimulation similar to that seen in patients (3–5).

Patients with diabetic neuropathy typically display a reduced tactile sensitivity and a reduced ability to detect skin cooling and heating (6–8). A compromised ability to detect stimulation with von Frey filaments and to sense vibration are thus important and simple diagnostic tools for early signs of diabetic neuropathy (9,10). In addition to the loss of sensation and the development of pain and paresthesias, diabetic neuropathy is characterized by reduced nerve conduction velocity, distal fiber loss and reduced axon diameters (11,12). The relative importance of abnormalities in sensory transduction, conduction, and anatomical structure for painful or insensate neuropathy in vivo is currently unclear, but the presence of sensory abnormalities cannot be used to discriminate between patients with painless and painful diabetic neuropathy (12).

Here, we present a detailed characterization of sensory neuropathy in the Ins2^+/Akita mouse using behavioral, cellular, and neurophysiological approaches. Mice that are heterozygous for the Ins2^C96Y (Ins2^+/Akita) mutation develop β-cell endoplasmic reticulum stress, produce little insulin, and become diabetic soon after weaning (13,14). The phenotype is more serious in male mice than in female mice (13,14). Since the Ins2^C96Y mutation leads to the spontaneous development of diabetes, the Akita strain provides a convenient translational model suitable for investigations of diabetic complications and islet transplantations (15). Relatively little information is available about the sensory phenotype of Ins2^+/Akita mice, whereas autonomic neuropathy has been examined in more detail previously (16). The results reported here show that Ins2^+/Akita mice rapidly develop impaired mechanosensation after the onset of hyperglycemia, later followed by impaired thermal (hot and cold) nociception. Investigations of skin-saphenous nerve preparations demonstrate a reduced rate of action potential discharge in mechanosensitive A- and C-nociceptors at high stimulus intensities. The reduced behavioral sensitivity to noxious heat was associated with a reduced proportion of heat-sensitive dorsal root ganglion (DRG) neurons in Ins2^+/Akita mice, as well as a reduced [Ca²⁺]_i response amplitude in heat-sensitive neurons.

Behavioral experiments were carried out according to the U.K. Home Office Animal Procedures (1986) Act. All procedures were approved by the King’s College London Animal Welfare and Ethical Review Body and were conducted under the U.K. Home Office Project License PPL 70/7510. Ins2^+/Akita mice were obtained from The Jackson Laboratory (stock #003548, mouse genome informatics #1857572; Bar Harbor, ME) and maintained on the C57BL/6J strain. Blood glucose was monitored routinely by Contour XT blood glucose meter (Bayer Healthcare, Reading, U.K.) or Stat Strip Xpress meter and strips (Nova Biomedical, Runcorn, U.K.).

The Randall-Selitto paw-pressure test was performed using an Analgesy-Meter (Ugo Basile, Gemonio, Italy). Mice were kept in their holding cages to acclimatize (10–15 min) to the experimental room. The experimenter lightly restrained the mouse and applied a constantly increasing pressure stimulus to the dorsal surface of the hind paw using a blunt conical probe. The nociceptive threshold was defined as the force in grams at which the mouse withdrew its paw. A force cutoff value of 150 g was used to avoid tissue injury.

Tactile sensitivity was assessed using von Frey filaments (0.008–2 g) according to the up-down method of Chaplan et al. (17). Animals were placed in a Perspex chamber with a metal grid floor allowing access to their plantar surface and were allowed to acclimatize prior to the start of the experiment. The von Frey hairs were applied to the plantar surface of the hind paw with enough force to allow the filament to bend and were held static for ∼2–3 s. The stimulus was repeated up to five times at intervals of several seconds. The stimulus interval was adapted to allow for the resolution of any behavioral responses to previous stimuli. A positive response was noted if the paw was sharply withdrawn or if the mouse flinched upon removal of the hair. Any movement of the mouse, such as walking or grooming, was deemed an unclear response, and in such cases the stimulus was repeated. If no response was noted, a higher force hair was tested and the filament producing a positive response was recorded as the threshold.

Thermal nociception was examined by lightly restraining the animal and placing one of the hind paws onto a cold or hot plate maintained at 10°C or 50°C (18,19). The paw withdrawal or escape latency was measured using 30 s as a cutoff to avoid tissue injury.

Single-fiber recordings were carried out on preparations from 18- to 20-week-old adult male littermate Ins2^+/+ and Ins2^+/Akita mice. The saphenous nerve and hind-paw hairy skin were dissected free and immediately placed in warm (32°C) oxygenated synthetic interstitial fluid (SIF) containing the following (in mmol/L): 108 NaCl, 3.5 KCl, 0.7 MgSO₄, 26.2 NaHCO₃, 1.65 NaH₂PO₄, 9.6 Na-gluconate, 5.55 glucose, and 1.53 CaCl₂, buffered to pH 7.4 by bubbling carbogen (95% O₂ and 5% CO₂). The skin was pinned down, corium side up, and the saphenous nerve was placed on a mirrored platform covered in paraffin oil in a separate chamber containing an SIF-paraffin oil interface. The nerve was desheathed and teased into thin filaments, which were placed onto a gold electrode. Electrical signals were recorded using the DAM80 differential amplifier (World Precision Instruments), band-pass filtered at 300 Hz and 10 kHz, and digitized and stored on a computer using the Micro1401 data acquisition unit and SPIKE2 version 8 software (Cambridge Electronic Design). Action potentials were also visualized using an oscilloscope (Gould 420 series).

Receptive fields of mechanosensitive primary afferents located in the hairy skin of the hind paw and lower hind limb were identified using a blunt glass rod and thereafter stimulated with brief electrical pulses (1 ms; DS2; Digitimer) to determine the conduction velocity of the fiber. Fibers with conduction velocity <1.2 m/s were classified as C fibers, those between 1.2 and 10 m/s as Aδ fibers, and those with conduction velocity >10 m/s as Aβ fibers (20).

The mechanical threshold of single fibers was assessed by identifying the minimum force (delivered as a 2-s step by a computer-driven, feedback-controlled stimulator) required to evoke two action potentials (21). Adaptation properties were characterized by applying 15 s of sustained mechanical force of 0.5, 1, 2, 4, 5, 10, 15, and 20 g with 2-min interstimulus intervals to reduce tachyphylaxis. The encoding properties of mechanosensitive Aδ and C fibers were examined by applying ramp-shaped mechanical stimuli.

The thermal sensitivity of some C fibers was investigated. A metal ring was used to isolate the receptive field and a 60-s cold stimulus of ∼31–4°C was applied by superfusing precooled SIF. Thereafter, a 20-s heat stimulus of ∼32–48°C was applied by prewarming the SIF through a heated element. The rate of temperature increase was determined using a custom-made feedback-controlled device. For both the cold and heat stimuli, peristaltic pumps were used to control the SIF flow rate. A digital temperature probe (GIA 2000; Greisinger) was placed inside the ring to record the temperature at the skin. A thermally elicited response was considered if a minimum of two action potentials occurred and the temperature at which the second action potential fired was recorded as the thermal threshold.

Data were analyzed offline using the SPIKE2 software (Cambridge Electronic Design). Action potentials were discriminated using templates based on their amplitude and waveform.

DRG neurons were dissociated enzymatically (using collagenase and trypsin) from ganglia dissected from 12-week-old (one of each phenotype) or 20-week-old (two to three of each genotype) male littermate Ins2^+/+ and Ins2^+/Akita mice using methods described previously (22). Isolated neurons were plated on poly-d-lysine–coated coverslips and maintained at 37°C in 95% air and 5% CO₂. Wild-type (WT; Ins2^+/+) neurons were maintained in MEM AQ containing 5.6 mmol/L glucose (Sigma-Aldrich, Poole, U.K.) and Ins2^+/Akita neurons in DMEM GlutaMAX containing 25 mmol/L glucose (Thermo Fisher Scientific, Paisley, U.K.), each supplemented with 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, 10 μmol/L cytosine arabinoside, and 50 ng/mL nerve growth factor (Promega, Southampton, U.K.) for up to 24 h before experimentation. The serum content of insulin makes the final concentration in medium 4 pmol/L, which is far below that required to support trophic effects on isolated DRG neurons (23).

DRG neurons were loaded with 2.5 μmol/L Fura-2 AM (Invitrogen) in the presence of 1 mmol/L probenecid and 0.01% pluronic acid for ∼1 h at 37°C. In all experiments, the cells were constantly superfused with physiological saline solution containing the following (in mmol/L): 140 NaCl, 5 KCl, 10 glucose, 10 HEPES, 2 CaCl₂, and 1 MgCl₂, pH 7.4 (NaOH). Fura-2 was excited at wavelengths of 340 and 380 nm, and light emission was measured at wavelengths >520 nm. Solutions were applied through a tube placed in close proximity to the cells and ratio images were captured every 2 s. Data were analyzed using ImageMaster software (Media Cybernetics, Rockville, MD) to obtain emission intensity ratios (R_(340/380)) in individual cells.

Hind paw skin was dissected and fixed in 4% paraformaldehyde in PBS and kept at room temperature until embedded in paraffin wax. The 8-µm-thick sections were cut and stained with rabbit polyclonal anti-PGP9.5 (Ultraclone Limited, Isle of Wight, U.K.) overnight at 4°C. Staining was visualized using a biotinylated goat anti-rabbit IgG (Vector Laboratories), followed by Streptavidin-Alexa Fluor 568 (Thermo Fisher Scientific). Skin sections were additionally stained with DAPI to visualize nuclei. The high density of nuclei in the epidermis provides a clear indication of the location of the basement membrane. We have adapted the general method described for human tissue (24,25) to suit thin sections prepared from mouse skin (26,27). The number of intraepidermal nerve fibers (IENFs) extending intact from the basement membrane per millimeter was quantified using a Zeiss Apotome Microscope. The observer was blinded to the identity of the animal.

Islets were isolated from 10- to 12-week-old C57BL/6J mice (Charles River, Margate, U.K.) using collagenase digestion (1 mg/mL, type XI; Sigma-Aldrich, Gillingham, U.K.) and density gradient separation (Histopaque-1077; Sigma Aldrich), as previously described (28). Islets were washed in culture medium (RPMI 1640 medium supplemented with 10% FBS and 1% Penicillin/Streptomycin) and divided into groups of 500 for same-day transplantation.

Male 14- to 20-week-old Ins2^+/Akita mice were transplanted with 500 freshly isolated islets into the renal subcapsular space. Mice were anesthetized with 1–5% isoflurane and 95% oxygen (1 L/min). The kidney was exposed by a lumbar incision, and a small perforation in the kidney capsule was made using a 23-gauge needle. Islets were pelleted in PE50 tubing before being injected under the renal capsule using a Hamilton syringe (Hamilton Robotics, Reno, NV). The capsule was then cauterized and the mouse peritoneum and skin were sutured. Mice were given carprofen (4 mg/kg s.c.; Caprieve; Norbrook Laboratories Limited, Newry, Northern Ireland, U.K.) and bupivacaine (0.5% solution, 2 mg/kg s.c.; Marcaine; Aspen Medical, London, U.K.) for postoperative analgesia.

Quantitative data are presented as the individual raw data points or the mean ± SEM, and the number of observations (animals or cells indicated by n). Data were analyzed by t test, Mann-Whitney U test, or ANOVA followed by Tukey post hoc test as appropriate. Categorical data were analyzed by Fisher exact test. Boxplots show the median and interquartile range, with whiskers indicating the 5th and 95th percentile. Statistical analysis was performed using Statistica or IBM SPSS (version 20 or higher). Differences were considered to be significant at P < 0.05.

Icilin was purchased from Tocris Bioscience (Bristol, U.K.), salts and all other reagents from Sigma-Aldrich, unless stated otherwise.

Ins2^+/Akita mice developed hyperglycemia soon after weaning and became diabetic at ∼4–5 weeks of age (Fig. 1A). Hyperglycemia progressively worsened in male Ins2^+/Akita mice, and at ∼12 weeks of age, their blood glucose values reached or exceeded the upper limit of most blood glucose meters (∼33 mmol/L). In contrast, blood glucose peaked at ∼6 weeks of age in female Ins2^+/Akita mice and thereafter declined toward normoglycemia. Male, but not female, Ins2^+/Akita mice gained body weight at a reduced rate compared with WT littermates (Fig. 1B).

We assessed the behavioral sensitivity of male and female WT and Ins2^+/Akita mice to noxious mechanical and thermal stimulation, starting soon after weaning and continuing until ∼20 weeks of age. Mechanical nociception was assessed using the Randall-Selitto paw-pressure test and punctate stimulation with von Frey filaments, and the sensitivity to noxious thermal stimulation was determined using a modified cold- and hot-plate assay (18,19). Both male and female Ins2^+/Akita mice developed a reduced sensitivity in the paw pressure test, almost immediately after the onset of hyperglycemia (Fig. 1C). Furthermore, female Ins2^+/Akita mice progressively recovered from mechanical insensitivity from ∼10–12 weeks, when their blood glucose concentration started to decline. From 14 weeks of age, female Ins2^+/Akita mice were no longer distinguishable from WT littermates in the paw-pressure test, suggesting that the glycemic profile has a dynamic influence on mechanical nociception in mice (Fig. 1C). The paw withdrawal threshold in response to stimulation with calibrated von Frey filaments did not differ between Ins2^+/Akita and WT littermates, apart from a transiently reduced sensitivity in male Ins2^+/Akita mice at ∼10 weeks of age (Fig. 1D).

The paw withdrawal latency of male Ins2^+/Akita mice in the cold-plate (10°C) and hot-plate (50°C) tests increased at ∼9–10 weeks of age and thereafter remained elevated until at least 19–20 weeks of age (Fig. 1E and F). In contrast, female Ins2^+/Akita mice remained indistinguishable from WT mice in both thermal tests, apart from a transiently reduced cold sensitivity at ∼10 weeks of age (Fig. 1E and F). The paw pressure thus appears to be a particularly sensitive parameter for assessing sensory abnormalities in diabetic mice, and our results indicate that the degree of hyperglycemia experienced by female Ins2^+/Akita mice is not sufficient to influence the behavioral sensitivity to noxious hot or cold stimulation.

The conduction velocities and the mechanical response thresholds of rapidly adapting (RA), slowly adapting (SA) Aβ, and Aδ down hair (DH) fibers were unchanged in preparations from 18- to 20-week-old male diabetic Ins2^+/Akita mice compared with fibers in age-matched male WT littermates (Table 1). Furthermore, RA and DH fibers from diabetic Ins2^+/Akita mice displayed an unchanged action potential firing pattern to sustained force across the range of forces tested (0.5–20 g) compared with fibers in preparations from WT littermates (Fig. 2A, B, E, and F). We did note a significantly reduced firing rate in Ins2^+/Akita SA fibers in response to the lowest force (0.5 g; P = 0.01, Mann-Whitney U test) (Fig. 2C). This observation suggests that SA fibers in Ins2^+/Akita mice display an accelerated adaptation at low stimulus strengths, compared with WT mice (Fig. 2D).

The conduction velocities of Ins2^+/Akita nociceptive Aδ-mechanoreceptors (AMs) (29) were unchanged compared with WT fibers in preparations from 20-week-old male mice. Unmyelinated C-mechanoreceptors (CMs) (30) from Ins2^+/Akita mice, however, displayed a reduced conduction velocity compared with WT fibers (P = 0.007, Mann-Whitney U test) (Table 1), as previously noted in diabetic rats (31). The response threshold in AM fibers from diabetic mice was modestly increased (Table 1), but the most striking differences between the response properties of nociceptors from Ins2^+/Akita and WT mice were revealed by suprathreshold stimulation. WT AM fibers faithfully encoded mechanical force with a linear increase in discharge rate. In contrast, fibers from Ins2^+/Akita mice failed to encode stimuli exceeding 10 g (Fig. 3A) and displayed a significantly reduced impulse rate in response to the highest mechanical forces (Fig. 3A and D). The temporal response profiles evoked by force steps demonstrate that Ins2^+/Akita AM fibers failed to reach the maximal initial impulse rate typical of WT fibers (Fig. 3B and C). The mechanical encoding properties of AMs were examined further using ramp-shaped stimuli. Diabetic AM fibers displayed marginally (but not significantly) reduced average firing frequencies throughout the range of ramp forces used (Fig. 3E and F), an observation that may be consistent with a failure to achieve the initial peak firing rate (Fig. 3C), but a retained ability to encode force in response to continuously increasing ramp stimulation.

Stimulation with force steps produced similar impulse discharge rates in CM fibers from WT and Ins2^+/Akita mice (Fig. 4A and B), but the responses to the maximal ramp force were significantly reduced in Ins2^+/Akita CM fibers (Fig. 4C). Although Ins2^+/Akita CM fibers encoded the intensity of a ramp force stimulus, they appeared to do so at a somewhat lower rate than CM fibers in WT preparations (Fig. 4D and E).

Since diabetic mice displayed behavioral insensitivity to noxious heat and cold, we investigated whether the thermal sensitivity of CMs was altered in Ins2^+/Akita mice. The proportion of C-fiber nociceptors that responded to cold (WT: 48%, n = 14 of 29 vs. Ins2^+/Akita: 48%, n = 12 of 25; P = 0.98, Pearson χ² test) or heat (WT: 51%, n = 15 of 29 vs. Ins2^+/Akita: 45%, n = 9 of 20; P = 0.64, Pearson χ² test) was unchanged in Ins2^+/Akita mice compared with WT littermates. Cold-sensitive nociceptors (CMC fibers) from diabetic mice exhibited a marginally, but not significantly, reduced temperature threshold for activation (Ins2^+/Akita 18.2 ± 2°C; WT 21.4 ± 1.7°C; P = 0.25, Mann-Whitney U test) (Fig. 5A and B). Cold stimulation evoked a significantly reduced number of action potentials in CMC fibers from Ins2^+/Akita mice compared with WT mice (Ins2^+/Akita 18 ± 4 action potentials; WT 37 ± 7 action potentials, respectively; P = 0.018, Mann-Whitney U test) (Fig. 5C–E). Similarly, heat-sensitive nociceptors (CMH fibers) from diabetic mice displayed slightly, but not significantly, higher thresholds for heat activation (Ins2^+/Akita 40.4 ± 1.2°C; WT 38.4 ± 1.0°C; P = 0.14, Mann-Whitney U test) (Fig. 5F and G) and a reduced number of action potentials in CMH fibers from Ins2^+/Akita compared with WT mice (Ins2^+/Akita 16 ± 5 action potentials; WT 25 ± 5 action potentials, respectively; P = 0.18, Mann-Whitney U test) (Fig. 5H–J).

We used [Ca²⁺]_i imaging to determine whether the behavioral insensitivity of Ins2^+/Akita mice to noxious heat and cold was associated with a compromised thermal sensitivity of isolated DRG neurons (Fig. 6A and B). These experiments were performed on neurons isolated from 12- to 20-week-old male WT and Ins2^+/Akita mice. Stimulation with heat ramps (from 25 to 48°C) evoked [Ca²⁺]_i responses in a significantly reduced proportion of DRG neurons from Ins2^+/Akita mice compared with neurons isolated from WT mice (Ins2^+/Akita mice: 19.3%, 428 of 2,222, n = 4 animals; WT mice: 28.5%, 300 of 1,054, n = 3 animals; P < 0.001, Fisher exact test). Our measurements revealed a modestly, but significantly, reduced temperature threshold for heat activation in neurons from diabetic mice (Ins2^+/Akita 41.5 ± 0.2°C; WT 42.9 ± 0.2°C) (Fig. 6C) (P < 0.001, t test). The heat-evoked [Ca²⁺]_i response amplitude was reduced by 40% in neurons from Ins2^+/Akita mice compared with WT mice (Ins2^+/Akita 0.77 ± 0.02 R_(340/380); WT mice 1.2 ± 0.04 R_(340/380) (Fig. 6D) (P < 0.001, t test). These results strongly suggest that an impaired neuronal transduction of noxious heat may contribute to the behavioral insensitivity observed in vivo. The proportion of neurons responding to the transient receptor potential vanilloid 1 (TRPV1) agonist 1 μmol/L capsaicin (Ins2^+/Akita neurons: 44.3%, 759 of 1,715 neurons; WT neurons: 42.4%, 447 of 1,054; P > 0.3, Fisher exact test) and the response amplitude evoked by capsaicin were indistinguishable in cultures from WT and Ins2^+/Akita mice, indicating that the reduced heat responsiveness of Ins2^+/Akita DRG neurons is not likely due to an altered expression of the noxious heat transducer TRPV1.

In contrast to the reduced heat sensitivity of Ins2^+/Akita DRG neurons, cold ramps (35–10°C) elicited indistinguishable [Ca²⁺]_i responses in neurons isolated from mice of the two genotypes. Cooling stimulated [Ca²⁺]_i responses in a small proportion of DRG neurons from 20-week-old Ins2^+/Akita (2.4%, 55 of 2,318) and WT littermates (2.5%, 40 of 1,604). Cold-evoked [Ca²⁺]_i response amplitudes did not differ between neurons of the two genotypes (Fig. 6F), whereas the cold activation temperature was marginally, but significantly, higher in Ins2^+/Akita neurons (24.5 ± 0.7°C) compared with WT neurons (22.3 ± 0.9°C) (Fig. 6E). These results indicate that the behavioral insensitivity to noxious cold observed in diabetic Ins2^+/Akita mice is not associated with compromised cellular cold transduction mechanisms.

Diabetic neuropathy is accompanied by a loss of IENFs in patients and in animal models of diabetes, a feature that is frequently used as a diagnostic tool in various small-fiber neuropathies (25,32,33). Plantar skin sections prepared from male 20-week-old diabetic Ins2^+/Akita mice and WT littermates were immunohistochemically labeled for the expression of the pan-neuronal marker PGP9.5 (Fig. 7A). We measured the IENF density by counting the number of nerve fibers (in mm⁻¹) that extended intact from the epidermal/dermal junction into the epidermis, in thin (8 µm) paraformaldehyde-fixed paraffin-embedded sections of plantar skin (24,25,32). Our results revealed a significant loss of IENFs in diabetic mice. Compared with paw skin from WT littermates, the density of PGP9.5-positive epidermal fibers was reduced by about a third in diabetic Ins2^+/Akita mice (Fig. 7B).

Finally, we examined whether the impaired nociception and loss of IENFs in Ins2^+/Akita mice could be reversed by improved glycemic control. A group of diabetic Ins2^+/Akita mice underwent the transplantation of pancreatic islets obtained from healthy C57BL/6J mice (500 islets under the kidney capsule). After transplantation, blood glucose levels declined steadily during the first few days after transplantation, followed by a slower phase of decline, and at 3 weeks after islet implantation all mice had glucose levels <15 mmol/L (Fig. 7C). Intriguingly, nociception improved rapidly and significantly after transplantation (Fig. 7D and E). The behavioral sensitivity to noxious mechanical and cold stimulation recovered to the same level as in naive WT mice within a few days of transplantation, and thus occurred before normoglycemia was established. The IENF density in paw skin from transplanted mice had increased significantly at the end of the experiment (5–6 weeks after transplantation) compared with Ins2^+/Akita mice that had not received an islet transplant. The IENF density in transplanted mice was indistinguishable from that in healthy WT mice (Fig. 7B).

Here, we present a detailed characterization of sensory neuropathy in the Ins2^+/Akita mouse, and demonstrate that diabetes leads to an impaired mechanical and thermal (hot and cold) nociception, which is associated with a marked loss of IENFs. We noted a rapid loss of mechanical nociception at the onset of diabetes, which was followed by a reduced sensitivity to noxious cold and heat 2–3 weeks later. In contrast, we did not observe an altered sensitivity to stimulation with von Frey filaments. We used manual stimulation with calibrated von Frey filaments according to the up-down method of Chaplan et al. (17), rather than the widely used automated von Frey assay (5), which uses stimuli of higher force intensity and generates higher paw withdrawal thresholds. Our behavioral and electrophysiological studies suggest that mechanonociceptors are particularly sensitive to diabetes, whereas low-threshold A-fibers are functionally more resilient. Intriguingly, the loss of mechanical nociception appears to be a dynamic process, with rapid onset after the establishment of hyperglycemia, and a similarly rapid recovery following islet transplantation. In addition, the recovery of female Ins2^+/Akita mice from hyperglycemia was accompanied by a normalized sensitivity in the paw-pressure test. The distinct behavioral profiles observed using different test modalities underscores the value of using a comprehensive behavioral testing strategy.

Patients with diabetic neuropathy commonly report numbness, spontaneous pain, and dysesthesias (34), but loss of thermal (cold and heat detection) and mechanical (mechanical and vibration detection) sensation are the dominant sensory symptoms in patients with mild, severe, or no pain (3,35). Hypersensitivity to thermal or mechanical stimulation (evoked pain) can thus not be viewed as translational measures of spontaneous pain or painful diabetic neuropathy. Importantly, we observed only a reduced responsiveness of mechanosensitive and thermosensitive fibers here, in good agreement with an earlier skin-nerve investigation of mice made diabetic by streptozotocin (STZ) treatment (36).

Our electrophysiological studies demonstrate that Ins2^+/Akita AM and, to a lesser extent, CM nociceptors failed to sustain a high action potential discharge rate in response to intense mechanical stimulation. The force activation threshold of Ins2^+/Akita AM fibers, but not CM fibers, was increased somewhat compared with fibers from WT mice, whereas C fibers exhibited a degree of conduction slowing. Together, these results suggest that neuronal excitability and conduction, rather than transduction of mechanical and thermal stimuli, are primarily affected by diabetes in Ins2^+/Akita mice. Our results are consistent with the loss of mechanical nociception observed in mice in vivo and clinically in patients (1,37,38).

Earlier investigations of STZ-induced diabetic neuropathy in mice (36) demonstrated a reduced mechanical sensitivity of RA fibers and a more pronounced loss of CM nociceptor activity than observed here in Ins2^+/Akita mice. It is possible that the longer interval between consecutive mechanical challenges used here may have allowed a fuller recovery of CM fiber activity. Other possible reasons for differences between the two studies include the age of diabetes onset (which is very early in Ins2^+/Akita), the duration of diabetes, and the possibility that STZ affects cells outside the pancreatic islets (39–42).

Diabetic Ins2^+/Akita mice displayed an impaired sensitivity to noxious heat and cold in vivo and in vitro. Studies of temperature-sensitive CM fibers (CMHs and CMCs) in isolated skin-nerve preparations from Ins2^+/Akita mice demonstrated a markedly reduced responsiveness to cold, but not to heat. DRG neurons from Ins2^+/Akita mice showed a markedly reduced heat responsiveness compared with neurons from WT mice, but an unchanged cold sensitivity. The capsaicin receptor TRPV1 is the principal heat transducer in DRG neurons, although other ion channels have been proposed to contribute to the transduction of noxious heat (e.g., during inflammation or in response to very high temperatures) (43–46). A high concentration of the TRPV1 agonist capsaicin evoked [Ca²⁺]_i responses in equal proportions in WT and Ins2^+/Akita DRG neurons, suggesting that an altered expression of TRPV1 is unlikely to explain the reduced heat sensitivity in Ins2^+/Akita neurons. It is possible that a reduced activity produced by, for example, glycation (47) or altered phosphorylation of TRPV1 may contribute to the reduced sensitivity (48). It is likely that heat-evoked [Ca²⁺]_i responses are mediated in part by voltage-gated Ca²⁺ channels, after an initial TRPV1-mediated depolarization. An altered activity of voltage-gated channels or differences in membrane potential that influence calcium entry via voltage-gated calcium channels may thus be responsible for the reduced heat sensitivity observed in DRG neurons isolated from Ins2^+/Akita mice. The reduced heat responsiveness of isolated DRG neurons, but largely unchanged activity of CMH fibers, may indicate that mechanoinsensitive heat nociceptors are affected by diabetes.

In contrast to the reduced heat responsiveness of Ins2^+/Akita DRG neurons, we did not detect any deficits in the neuronal response to cold ramps. We used a modest cooling rate in our experiments, and different rates of cooling are known to stimulate different sensory neuron populations (49,50). It is therefore possible that a subpopulation of cold-sensitive Ins2^+/Akita DRG neurons with altered properties may have been overlooked.

Nociceptive AM and CM fibers, as well as mechanically insensitive thermoreceptors terminate as free nerve endings in the epidermal layer of the skin (51,52). A reduced density of IENFs may thus have contributed to the impaired thermal and mechanical nociceptive function in vivo.

Transplantation of pancreatic islets isolated from healthy WT mice, progressively improved blood glucose levels and achieved normoglycemia after ∼3 weeks. In contrast, the behavioral sensitivity to noxious mechanical and cold stimulation was restored to the WT level by the time of first behavioral test (2–3 days after transplantation). This rapid recovery of sensory function may indicate that hyperglycemia is not directly responsible for maintaining the impaired nociception. One possible explanation for the rapid restoration of nociceptive function is that insulin itself may modulate sensory neuron function and act as a neurotrophic factor (reviewed in the study by Grote and Wright [53]). A rapid neurotrophic or sensitizing effect of insulin on sensory neurons after transplantation is also consistent with the observation that the insulin signal transduction machinery is intact or even upregulated in Zucker diabetic fatty rats with well-established neuropathy (54). Furthermore, local intraplantar administration of insulin in diabetic mice rapidly (within days) increased the density of IENFs, a time course that is consistent with that observed here for the recovery of behavioral sensitivity in vivo after islet transplantation (55). In our experiments on isolated DRG neurons, we did not match the insulin concentrations to the levels that the DRG neurons would experience in vivo. The observed differences between WT and Akita neurons can thus not be explained by any differences in the insulin concentrations in the cell culture medium. Importantly, this also means that the compromised heat sensitivity of Ins2^+/Akita neurons, which is in line with our behavioral observations in vivo and electrophysiological observations in vitro, are due to diabetes, not simply due to differences in insulin signaling after isolation of the neurons. It is likely that sensory neurons exhibit heterogenous insulin signaling mechanisms, since the expression patterns of the insulin receptors IRS1 and IRS2 are not uniform across DRG neurons (56). The concentration of insulin present in the medium here (∼4 pmol/L) is below the concentrations required to exert trophic effects on isolated sensory neurons (23).

Diabetes caused by Ins2^C96Y is associated with impaired nociception, loss of IENFs, a reduced sensory neuronal heat sensitivity, and a reduced action potential discharge rate in mechanosensitive nociceptive afferent fibers in vitro. Functional and structural improvement from established diabetic neuropathy was previously reported from mice that recovered spontaneously from STZ-evoked diabetes (57). Our results highlight the importance of distinguishing sensory abnormalities from the spontaneous pain that may be caused by diabetic neuropathy in patients. Islet transplantation restored IENF density (assessed at the end of the experiment, 5–6 weeks after transplantation). Remarkably, islet transplantation restored nociceptive function before normoglycemia was established, indicating that lack of insulin, rather than hyperglycemia per se, is responsible for the impaired nociception in diabetic mice.

Ins2^+/Akita, and other strains carrying Ins2 mutations (58), are convenient and attractive models of type 1 diabetes. Here we show that diabetic neuropathy in the Ins2^+/Akita mouse is characterized by a loss of evoked sensory responses, consistent with the sensory profile most commonly observed in patients with painful or painless neuropathy (3). Studies of experimental models that rely on disease-relevant mechanisms and faithfully reproduce the clinical condition, such as the Ins2^+/Akita mouse, are likely to generate conclusions of translational predictive validity. Ins2^+/Akita appears to have an advantage over several other commonly used rodent models of diabetic sensory neuropathy that are associated with mechanical hypersensitivity. The hypersensitivity seen in the latter models contrasts with clinical experience in humans where reduced tactile sensitivity is used to detect the onset of diabetic neuropathy. Although patients with painful diabetic neuropathy may display a more marked loss of sensation than those with painless neuropathy, sensory loss is the primary symptom of both groups (3,35,38). Sensory testing of diabetic rodents is thus appropriate as a sensitive experimental measure of sensory neuropathy, but not as a translational measure of spontaneous pain and dysesthesias.

Funding. This study was supported by the Diabetes UK Alec and Beryl Warren Award to A.J.K., S.B., and D.A.A. (BDA 13/0004649). F.G. was supported by the Johannes und Frieda Marohn-Stiftung.

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

Author Contributions. N.V., F.G., C.G., A.L.A., and A.J.K. performed and analyzed experiments and contributed to experimental design and manuscript preparation. S.B. and D.A.A. contributed to experimental design and manuscript preparation. D.A.A. 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.