High-Fat Diet–Induced Neuropathy of Pre-Diabetes and Obesity : Effects of “Healthy” Diet and Aldose Reductase Inhibition

Unless otherwise stated, all chemicals were of reagent-grade quality and purchased from Sigma Chemical (St. Louis, MO). An NEFA C kit for the assessment of nonesterified fatty acids (NEFAs) in serum was purchased from Wako Chemicals (Neuss, Germany), and a rat/mouse insulin enzyme-linked radioimmunosorbent assay kit was purchased from Linco Research (St. Charles, MO). Rabbit polyclonal anti-nitrotyrisine (NT) and anti-hydroxynonenal (HNE) adduct antibodies were purchased from Upstate (Lake Placid, NY) and Calbiochem (San Diego, CA), respectively, and mouse monoclonal anti-poly(ADP-ribose) antibody was purchased from Trevigen (Gaithersburg, MD). Secondary Alexa Fluor 488 and 594 goat anti-rabbit and Alexa Fluor goat anti-mouse antibodies, as well as Prolong Gold Antifade reagent, were purchased from Invitrogen (Eugene, OR). The Avidin/Biotin Blocking kit, M.O.M. Basic kit, Vectastain Elite ABC kit (Standard), DAB Substrate kit, and 3,3′-diaminobenzidine were obtained from Vector Laboratories (Burlinghame, CA). Rabbit polyclonal anti-protein gene product 9.5 (ubiquitin COOH-terminal hydrolase) antibody was purchased from Chemicon International (Temecula, CA). Rabbit polyclonal 12/15-lipoxygenase antibody for Western blot analysis was produced in the laboratory of J.L.N. For immunohistochemistry, 12-lipoxygenase (murine leukocyte) polyclonal antiserum was purchased from Cayman Chemical. Other reagents for immunohistochemistry have been purchased from Dako Laboratories (Santa Barbara, CA).

The experiments were performed in accordance with regulations specified by the National Institutes of Health (Principles of Laboratory Animal Care, 1985 revised version) and Pennington Biomedical Research Center protocol for animal studies. Female C57BL6/J mice (∼20–21 g body wt) had access to water ad libitum and were randomly assigned to receive standard mouse chow or HFD (D12450B, 10 kcal% fat, and D12330, 58 kcal% fat with corn starch, respectively; Research Diets, New Brunswick, NJ), for 16 weeks. The HFD-fed mice were divided into three subgroups. One subgroup was killed for tissue harvest. Two other subgroups were maintained on standard mouse chow with or without the aldose reductase inhibitor fidarestat (16 mg · kg⁻¹· day⁻¹) for another 6 weeks. Blood samples for glucose measurements were taken every 4 weeks and the day before the animals were killed. A glucose tolerance test, assessment of food consumption, serum NEFAs, insulin, sciatic nerve sorbitol pathway intermediate concentrations, NT, and 12/15-lipoxygenase immunofluorescences, as well as all physiological measurements, were performed at the 16-week time point (i.e., before switching to standard mouse chow and at the end of the study). A glucose tolerance test (1.5 mg/kg i.p.) was performed after overnight (12-h) fasting. All other measurements included in the study were performed at one 16-week time point (i.e., before interventions). The physiological and behavioral tests took place in the following order: tactile responses to flexible von Frey filaments (day 1), thermal algesia (day 2), sensory nerve conduction velocity (SNCV), and motor nerve conduction velocity (MNCV) (day 3). Measurements of MNCV and SNCV were taken in mice anesthetized with a mixture of ketamine and xylazine (45 and 15 mg/kg body wt i.p., respectively).

Animals were sedated with CO₂ (12) and immediately killed by cervical dislocation. One sciatic nerve from each mouse (16-week time point) was rapidly dissected and frozen in liquid nitrogen for subsequent assessment of sorbitol pathway intermediate concentrations and 12/15-lipoxygenase expression. Another sciatic nerve, dorsal root ganglia (DRG), and foot pads were fixed in normal buffered 4% formalin. Sciatic nerve and DRG were used for assessment of NT, 4-HNE adducts, poly(ADP-ribose), and 12/15-lipoxygenase by immunofluorescent histochemistry. NT and HNE adducts are generally accepted variables of peroxynitrite-induced injury and lipid peroxidation (a footprint of oxidative stress), respectively. Foot pads were used for assessment of intraepidermal nerve fiber density. Tibial nerves were fixed (see below) and later used for the assessment of myelinated fiber diameter and myelin thickness. Sciatic nerves from 22-week-old rats were treated similarly and used for assessment of variables described above.

Sciatic MNCV and hind limb digital SNCV were measured as previously described elsewhere (13). In all measurements, body temperature was monitored by a rectal probe and maintained at 37°C with a warming pad. Hind limb skin temperature was also monitored by a thermistor and maintained between 36 and 38°C by radiant heat.

Tactile responses were evaluated by quantifying the withdrawal threshold of the hindpaw in response to stimulation with flexible von Frey filaments. Mice were placed in individual plexiglass boxes on stainless steel mesh floor and were allowed to adjust for at least 20 min. A series of calibrated von Frey filaments (IITC Life Science, Woodland Hills, CA) was applied perpendicularly to the plantar surface of a hindpaw with sufficient force to bend the filament for 6 s. Typically, 10–12 mice were stimulated one after another in the same order during one testing procedure (∼2–3 h), and stimulations were repeated five to six times. Brisk withdrawal or paw flinching was considered to be a positive response. In the absence of a response in ≥50% paw stimulations, a filament of the next greater force was applied. Stimulation was stopped at the filament producing a positive response in four of five or six stimulations. The average value for buckling weights of the last and previous filaments (or weight of the last filament if the previous one[s] did not produce any responses) was considered to be tactile response threshold and was recorded for each paw. For example, if the buckling weight of the last filament producing four positive responses was 1.5 g and stimulation with two previous filaments produced two positive responses to the 1.2-g filament and one positive response to the 1.0-g filament, then the tactile threshold would be [(1.5 × 4) + (1.2 × 2) + 1.0]/7 = 1.34 g. The mean value of tactile response thresholds in the left and right paws was taken for statistical analysis. These calculations, although different from those previously used (14,15), produced data in the range of those reported by others in the rat model (compare refs. 15 and 16).

To determine the sensitivity to noxious heat, rats were placed within a plexiglass chamber on a transparent glass surface and allowed to acclimate for at least 20 min. A thermal stimulation meter (IITC model 336 TG Combination Tail Flick & Paw algesia meter) was used. The device was activated after placing the stimulator directly beneath the plantar surface of the hind paw. The paw withdrawal latency in response to the radiant heat (15% intensity that produced a heating rate of ∼1.3°C per s, cut off time 30 s) was recorded. Floor temperature was set at ∼32–33°C (manufacturer's setup). Individual measurements were repeated four to five times and the mean value calculated.

All sections were processed by a single investigator and blindly evaluated. Primary antibodies were omitted in negative controls. Low-power observations of skin sections stained for PGP 9.5 were made using a Zeiss Axioskop microscope. Color images were captured with a Zeiss Axiocam HRc charged-coupled device camera at a 1,300 × 1,030 resolution. Low-power images were generated with a 40× acroplan objective using the automatic capturing feature of the Zeiss Axiovision software (version 3.1.2.1). Low-power observations of sciatic nerve and DRG sections stained for HNE adducts, NT, and poly(ADP-ribose) were made using a Zeiss Axioplan 2 imaging microscope. Color images were captured with a Photometric CoolSNAP_HQ charged-coupled device camera at a 1,392 × 1,040 resolution. Low-power images were generated with a 40× acroplan objective using the RS Image 1.9.2 software.

NT, HNE-adduct, and 12/15-lipoxygenase expressions in the sciatic nerve and DRG neurons were assessed by immunofluorescent histochemistry. In brief, sections (one experimental group per slide) were deparaffinized in xylene, hydrated in decreasing concentrations of ethanol, and washed in water. To exclude the potential contribution of interslide variability to differences among experimental groups, sections of two to three samples in each run were placed on different slides. In the case of a noticeable interslide difference in staining, the whole procedure was repeated. In our experience, when all slides are processed similarly and different antibodies and staining solutions are added in equal amounts and at the same time, the differences in fluorescence (or color staining) among sections from the same group processed on different slides do not substantially exceed differences among sections processed on the same slide. Much higher variability is observed when comparing slides processed in different runs, even with the same lots of antibodies, and a such approach was not used in the present study. For NT, anti-NT antibody was used in a working dilution of 1:400 and secondary Alexa Fluor 488 goat anti-rabbit antibody in a working dilution of 1:200. For HNE adducts, anti-HNE adduct antibody was used in a working dilution of 1:1,600 and secondary Alexa Fluor 488 goat anti-rabbit antibody in a working dilution of 1:200. For 12/15-lipoxygenase, 12-lipoxygenase antiserum was used in a working dilution of 1:3,200 and secondary Alexa Fluor 594 goat anti-rabbit antibody in a working dilution of 1:200. Sections were mounted in Prolong Gold antifade reagent. The intensity of fluorescence was quantified using the the ImageJ 1.32 software (National Institutes of Health, Bethesda, MD) and expressed as means ± SE for each experimental group.

Poly(ADP-ribose) immunoreactivity was assessed as previously described (13) with minor modifications. In brief, sections were deparaffinized in xylene, hydrated in decreasing concentrations of ethanol, and washed in water. Nonspecific binding was blocked with the mouse IG blocking reagent supplied with the Vector M.O.M. Basic Immunodetection kit. Then mouse monoclonal anti-poly(ADP-ribose) antibody was diluted 1:100 in 1% BSA in Tris-buffered saline (TBS) and applied overnight at 4°C in the humidity chamber. Secondary Alexa Fluor 488 goat anti-mouse antibody was diluted 1:200 in TBS and applied for 2 h at room temperature. Sections were mounted in Prolong Gold Antifade Reagent. At least 10 fields of each section were examined to select one representative image. Representative images were microphotographed, and the number of poly(ADP-ribose)-positive nuclei in sciatic nerves and DRG was calculated for each microphotograph. For assessment of poly(ADP-ribose) immunofluorescence in DRG neurons, the intensity of neuronal staining was graded from 1 to 3. The numbers of DRG neurons with weak (grade 1), moderate (grade 2), and intense (grade 3) poly(ADP-ribose) immunofluorescence were calculated as a percentage of the total number of DRG neurons displaying poly(ADP-ribose) immunofluorescence.

Intraepidermal nerve fiber density was assessed as previously described (17) with minor modification. Three randomly chosen 5-μm sections from each mouse were deparaffinized in xylene, hydrated in decreasing concentrations of ethanol, and washed in water. Nonspecific binding was blocked by 10% goat serum containing 1% BSA in TBS (DAKO, Carpinteria, CA) for 2 h and the Avidin/Biotin Blocking kit according to the manufacturer's instructions. Next, rabbit polyclonal anti-protein gene product 9.5 (ubiquitin COOH-terminal hydrolase) antibody was applied in a 1:2,000 dilution. Secondary biotinylated goat anti-rabbit IgG (H plus L) antibody was applied in 1:400 dilution and the staining performed with the Vectastain Elite ABC kit (standard). For visualization of specific binding sites, the DAB Substrate kit containing 3,3-diaminobenzidine was used. Sections were counterstained with Gill's hematoxylin, dehydrated, and mounted in Micromount mounting medium (Surgipath Medical Industries, Richmond, IL). Intraepidermal nerve fiber profiles were blindly counted by three independent investigators under an Olympus BX-41 microscope, and the average values were used. Microphotographs of stained sections were taken on an Axioscop 2 microscope (Zeiss) at 4× magnification, and the length of epidermis was assessed with the ImagePro 3.0 program (Media Cybernetics). An average of 2.8 ± 0.3 mm of the sample length was investigated to calculate a number of nerve fiber profiles per millimeter of epidermis.

Tibial nerve morphometry was performed as previously described (18). Tibial nerves were fixed overnight at 4°C in 2.5% glutaraldehyde buffered with 0.05 mmol/l sodium cacodylate (pH 7.3). The fixed samples were postfixed in 1% osmium tetroxide and dehydrated through an ascending series of ethanol concentrations. Fixed nerves were embedded in epon and polymerized. One-micron–thick semithin transverse nerve sections were stained with toluidine blue. For the morphometric analysis, myelinated nerve fiber diameter and myelin thickness were measured at a magnification of ×1,600 by a computer-assisted image analyzing system (NIH Image and Agfa Arcus scanner connected to Macintosh Quadra 700, Cupertino, CA), as previously described (18).

Glucose, sorbitol, and fructose concentrations in the sciatic nerve were assessed spectrofluorometrically by enzymatic procedures with hexokinase/glucose-6-phosphate dehydrogenase, sorbitol dehydrogenase, and fructose dehydrogenase, as previously described (12).

To assess 12/15-lipoxygenase protein expression by Western blot analysis, sciatic nerve material (∼10 mg) was placed in 200 μl of an extraction buffer containing 50 mmol/l Tris-HCl (pH 7.2), 150 mmol/l NaCl, 0.1% sodium dodecyl sulfate, 1% NP-40, 5 mmol/l EDTA, 1 mmol/l EGTA, 1% sodium deoxycholate, and the protease/phosphatase inhibitors leupeptin (10 μg/ml), aprotinin (20 μg/ml), benzamidine (10 mmol/l), phenylmethylsulfonyl fluoride (1 mmol/l), and sodium orthovanadate (1 mmol/l) and homogenized on ice. The homogenate was sonicated (3 × 5 s) and centrifuged at 14,000g for 20 min. All the aforementioned steps were performed at 4°C. The lysates (20 μg protein) were mixed with equal volume of 2× sample-loading buffer containing 62.5 mmol/l Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate, 5% β-mercaptoethanol, 10% glycerol, and 0.025% bromophenol blue and fractionated in 10% SDS-PAGE in an electrophoresis cell (Mini-Protean III; Bio-Rad Laboratories, Richmond, CA). Electrophoresis was conducted at 15 mA constant current for stacking and at 25 mA for protein separation. Gel contents were electrotransferred (250 mA, 2 h) to nitrocellulose membranes using Mini Trans-Blot cell (Bio-Rad Laboratories) and Western transfer buffer (25 mmol/l Tris-HCl [pH 8.3], 192 mmol/l glycine, and 20% [vol/vol] methanol) (19). Free binding sites were blocked in 2% (wt/vol) BSA in 20 mmol/l Tris-HCl buffer (pH 7.5) containing 150 mmol/l NaCl and 0.05% Tween 20 for 1 h, after which 12/15-lipoxygenase antibody was applied for 2 h for detection of 12/15-lipoxygenase protein expression. The horseradish peroxidase–conjugated secondary antibody was then applied for 1 h. After extensive washing, protein bands detected by the antibodies were visualized with the BM Chemiluminescence Blotting Substrate (POD; Roche, Indianapolis, IN). Membranes were then stripped in the 62.5 mmol/l Tris-HCl (pH 6.7) buffer containing 2% SDS and 100 mmol/l β-mercaptoethanol and reprobed with β-actin antibody to confirm equal protein loading.

Results are expressed as means ± SE. Data were subjected to equality of variance F test and then to log transformation, if necessary, before one-way ANOVA. Where overall significance (P < 0.05) was attained, individual between-group comparisons for multiple groups were made using the Student Newman-Keuls multiple-range test. When between-group variance differences could not be normalized by log transformation (datasets for body weights and plasma glucose), the data were analyzed by the nonparametric Kruskal-Wallis one-way ANOVA, followed by the Bonferroni Dunn test for multiple comparisons. Individual pairwise comparisons between mice fed standard mouse chow and HFD were made using the unpaired two-tailed Student's t test or Mann-Whitney rank-sum test, where appropriate. Significance was defined at P ≤ 0.05.

A 16-week HFD feeding resulted in a 27% increase in body weight compared with mice fed standard mouse chow (Table 1). Daily food consumption was slightly lower in HFD-fed mice (2.58 ± 0.17 vs. 3.03 ± 0.12 g in mice fed standard mouse chow at the 16-week time point). No significant changes in body weight were achieved after switching the HFD-fed mice to standard mouse chow or standard mouse chow containing fidarestat for 6 weeks. Nonfasting blood glucose concentrations were similar among mice fed standard mouse chow, mice fed HFD, and those switched to standard mouse chow with or without fidarestat after HFD feeding. However, glucose tolerance was impaired in mice fed HFD compared with those fed standard mouse chow (Fig. 1A), which is consistent with increased serum insulin concentrations (3.86 ± 0.81 vs. 1.22 ± 0.19 ng/ml in mice fed standard mouse chow, P < 0.01), as well as insulin resistance and impaired glucose utilization previously described in this model (20). Switching to standard mouse chow either with or without fidarestat for 6 weeks improved glucose tolerance (Fig. 1B and C) and reduced serum insulin concentrations to 1.75 ± 0.38 ng/ml in HFD-fed mice switched to standard mouse chow and 2.32 ± 0.63 ng/ml in HFD-fed mice switched to standard mouse chow plus fidarestat (P < 0.01 and P < 0.05 vs. HFD-fed mice and P > 0.45 and P > 0.09 vs. mice fed standard mouse chow, respectively); however, blood glucose concentrations did not return to basal levels in either of these two groups (Fig. 1D and E). Serum NEFA concentrations were 98% higher in mice fed HFD compared with those fed standard mouse chow (1.14 ± 0.275 and 0.576 ± 0.031 meq/l, P < 0.01), and this increase was completely abolished by switching mice to standard mouse chow with or without fidarestat (0.471 ± 0.049 and 0.366 ± 0.044 meq/l, P < 0.01 vs. mice fed HFD for both).

Sciatic nerve glucose concentration increased 33% in mice fed HFD (Table 2), and this increase was completely blunted after switching mice to standard mouse chow with or without fidarestat. Sciatic nerve sorbitol concentration increased 97% and fructose concentration increased 79% in HFD-fed mice compared with mice fed standard mouse chow. An HFD-induced increase in both concentrations was completely blunted after switching mice to standard mouse chow with or without fidarestat.

HFD-fed mice had clearly manifested MNCV and SNCV deficits compared with mice fed the control D12450B diet (Table 3). Additional studies revealed that prolonged (8-week) maintenance of female C57BL6/J mice on D12450B or D12328 diets, typically used as control diets for the HFDs 12451 and 12330, respectively, resulted in quite similar peripheral nerve function (D12450B diet: MNCV 53.1 ± 2.0 m/s, SNCV 36.5 ± 0.3 m/s, thermal response latency 8.5 ± 0.1 s, and tactile response threshold 2.2 ± 0.2 g; and D12328 diet: MNCV 52.9 ± 1.8 m/s, SNCV 36.7 ± 0.6 m/s, thermal response latency 8.4 ± 0.1 s, and tactile response threshold 2.1 ± 0.25 g). Therefore, both control diets are suitable for studying HFD-induced neuropathy. Furthermore, the variables of peripheral nerve function (see above) appeared to be quite similar in C57BL6/J mice after 8 and 16 weeks of the control D12450B diet feeding. Of note, MNCV and SNCV displayed a different response to dietary treatment. MNCV deficit appeared totally nonresponsive to a 6-week intervention with normal chow alone but was essentially normalized by refeeding with normal chow containing fidarestat. In contrast, SNCV deficit was corrected by either standard mouse chow alone or standard mouse chow containing fidarestat. HFD-fed mice had clearly manifest thermal hypoalgesia manifest by a 68% increase in latency of paw withdrawal in response to noxious thermal stimulus (radiant heat). Of interest, no hyperalgesia in HFD-fed mice was registered at earlier time points starting from 1 month after beginning HFD feeding (data not shown). Thermal hypoalgesia was essentially corrected by intervention with standard mouse chow and completely corrected by standard mouse chow containing fidarestat. Another sensory abnormality developing in HFD-fed mice was tactile allodynia. Tactile withdrawal threshold in response to light touch with flexible von Frey filaments was reduced by 46% in HFD-fed mice compared with mice fed standard mouse chow (P < 0.01). Tactile allodynia was essentially corrected by either intervention with standard mouse chow alone or standard mouse chow containing fidarestat.

Sciatic nerve HNE adduct (Fig. 2A and D) and NT (Fig. 2B and E) immunofluorescence increased by 28 and 26%, respectively, in HFD-fed mice compared with mice fed standard mouse chow (P < 0.01 for both compsons). Sciatic nerve NT immunofluorescence was essentially and completely normalized in the HFD-fed mice switched to standard mouse chow and standard mouse chow containing fidarestat, respectively. The number of the sciatic nerve poly(ADP-ribose)-positive nuclei increased by 80% in HFD-fed mice compared with mice fed standard mouse chow (P < 0.01) (Fig. 2, C and F).

In a similar fashion, HNE adduct (Fig. 3A and E) and NT (Fig. 3B and F) immunofluorescence increased by 61 and 20%, respectively, in DRG neurons of HFD-fed mice compared with mice fed standard mouse chow (P < 0.01 for both comparisons). The number of poly(ADP-ribose)-positive nuclei in DRGs was 40% higher in HFD-fed mice compared with mice fed standard mouse chow (P < 0.01) (Fig. 3C and G), with poly(ADP-ribose) immunofluorescence localized primarily in DRG glial cells. The percentage of DRG neurons with weak poly(ADP-ribose) immunofluorescence was lower and of those with intense poly(ADP-ribose) immunofluorescence was higher in HFD-fed mice compared with mice fed standard mouse chow (Fig. 3D and H).

Sciatic nerve 12/15-lipoxygenase expression increased by 42% in HFD-fed mice compared with mice fed standard mouse chow (P < 0.05) (Fig. 4A and B), which is consistent with the corresponding 12/15-lipoxygenase immunofluorescence data (Fig. 4C). 12/15-lipoxygenase immunofluorescence was 20% higher in DRG neurons of HFD-fed mice compared with mice fed standard mouse chow (P < 0.01) (Fig. 4D). Switching from HFD-fed mice to standard mouse chow tended to decrease sciatic nerve 12/15-lipoxygenase immunofluorescence; however, the difference with HFD-fed group before a dietary intervention did not achieve statistical significance. Switching the HFD-fed mice to normal chow plus fidarestat completely normalized sciatic nerve 12/15-lipoxygenase immunofluorescence.

Intraepidermal nerve fiber densities were similar in mice fed HFD and those fed standard mouse chow (Fig. 5). Myelinated nerve fiber diameter and myelin thickness were also similar in the two groups (6.44 ± 0.15 and 6.45 ± 0.17 μ and 2.08 ± 0.06 and 1.94 ± 0.03 μ in mice fed HFD and standard mouse chow, respectively). These findings indicate the lack of small sensory nerve fiber degeneration and axonal atrophy in the mouse model of HFD-induced neuropathy of pre-diabetes and obesity.

The HFD-fed mouse is a new model for studying the pathogenesis of neuropathic changes developing in human subjects with IGT, obesity, and metabolic syndrome that typically includes two aforementioned disorders as well as hyperlipidemia and hypertension (4). The neuropathic changes in HFD-fed mice are exclusively caused by the dietary regiment culminating in pre-diabetes, the most important factor associated with idiopathic neuropathy in nondiabetic human subjects (21). As one can conclude from the presence of thermal hypoalgesia and tactile allodynia, the HFD-fed mice developed small sensory nerve fiber neuropathy, i.e., a condition well documented in human subjects with IGT and the metabolic syndrome (2,4,9,10,22). Moreover, they also developed MNCV and SNCV deficits also reported in human subjects with pre-diabetes, overweight, and obesity (8), although studies of the association between IGT or obesity and nerve conduction changes resulted in contradictory findings (compare refs. 8 and 2). Finally, like human subjects with metabolic syndrome in another study (4), HFD-fed mice did not display any reduction of intraepidermal nerve fiber density. Of note, the presence of intraepidermal nerve fiber loss in human pre-diabetic neuropathy continues to be an area of controversy (compare ref. 4 with 2 and 10). One group (22) even reported a similar severity of intraedipermal nerve fiber loss in human subjects with IGT and overt type 2 diabetes; however, these findings were not reproduced by others (2,4).

Interestingly and excitingly, the key metabolic abnormalities previously thought to be caused primarily by high glucose and shown to contribute to PDN (i.e., increased activity of the sorbitol pathway of glucose metabolism [18,23,24], oxidative-nitrosative stress [25–29], and PARP activation [13,30,31]) clearly manifest in the HFD model of pre-diabetic neuropathy characterized by IGT in the absence of overt hyperglycemia. In particular, the accumulation of both sorbitol and fructose in sciatic nerve of HFD-fed mice was quite comparable to the accumulation in the streptozotocin-induced diabetic mouse model with severe hyperglycemia (18,24, and I.G.O., unpublished observations). Furthermore, sciatic nerve sorbitol concentrations were similar in HFD-fed and leptin-deficient ob/ob mice (32) (the type 2 diabetic mouse model). Of note, an identical sorbitol accumulation in three aforementioned mouse models was present despite the fact that elevation of sciatic nerve glucose concentration, traditionally considered a leading determinant of the sorbitol pathway activity, was very modest in HDF-fed mice (33%) compared with streptozotocin-induced diabetic and ob/ob mice (more than fivefold for both models). These findings indicate that factors other than increased intracellular glucose concentrations may predispose to increased sorbitol pathway activity and support and complement previous observations in models of nondiabetic conditions, i.e., myocardial ischemia (33,34) and aging (35,36). Studies of the effects of HFD on sorbitol pathway enzyme gene and protein expression may help identify an important mechanism contributing to neuropathy and cardiovascular disease in overweight and obese individuals.

The present study is the first to demonstrate that HFD feeding causes oxidative-nitrosative stress manifested by increased HNE adduct and NT immunoreactivities in peripheral nerve and DRG neurons. Oxidative-nitrosative stress is a well-recognized mechanism in PDN, and evidence for the importance of free radicals and oxidants in nerve conduction deficits, neurovascular dysfunction, metabolic and signal transduction changes, and impaired neurotrophic support has been produced by a number of leading experimental groups (24–29,37,38). Furthermore, several studies suggest that oxidative-nitrosative stress plays an important role in sensory neuropathy in animals (28,29,39,40) as well as human subjects (41) with diabetes. Therefore, it is quite plausible that MNCV and SNCV deficits, thermal hypoalgesia, and tactile allodynia in HFD-fed mice are at least partially mediated via oxidative-nitrosative stress. Our results suggest that optimal nutritional management and, in particular, dietary antioxidant supplementation, to minimize oxidative-nitrosative stress, may arrest or retard progression of pre-diabetic neuropathy in human subjects.

Activation of the nuclear enzyme PARP is known to lead to the depletion of NAD and high-energy phosphates, changes in transcriptional regulation and gene expression, exacerbation of oxidative-nitrosative stress, impaired signal transduction, and, in extreme cases, cell necrosis and apoptosis (42,43). Evidence for the important role of PARP activation in pathological conditions associated with oxidative-nitrosative stress, i.e., diabetes complications, is emerging (43–45). We are the first to demonstrate the importance of this mechanism in diabetes-associated MNCV and SNCV deficits, neurovascular dysfunction, and sensory neuropathy (13,16,31). The present study suggests that PARP activation in both sciatic nerve and DRG neurons and glial cells is present at the stage of pre-diabetic neuropathy developing in the absence of overt hyperglycemia. These findings are consistent with a previous report (46) documenting endothelial PARP activation and its association with impaired vascular reactivity in the skin microcirculation of healthy subjects at risk of developing diabetes.

Another important mechanism apparently present at the stage of pre-diabetic neuropathy is the activation of 12/15-lipoxygenase, a nonheme iron–containing dioxygenase that oxidizes esterified arachidonic acid with formation of 12- and 15-hydroxyeicosa-tetraenoic acids and their derivatives (47). These lipid-like substances undergo spontaneous peroxidation, which in turn leads to enhanced oxidative stress, signal transduction changes, and inflammatory response (47,48). Evidence for the important role of 12/15-lipoxygenase in endothelial dysfunction and nephropathy in animal models of diabetes is emerging (47–49). Our recent studies (50) have shown that 12/15-lipoxygenase is abundantly expressed in mouse sciatic nerve, spinal cord and DRG neurons, and human Schwann cells, and its expression is increased by diabetes and high glucose. Furthermore, diabetes-induced 12/15-lipoxygenase upregulation in mouse sciatic nerve was clearly associated with enhanced oxidative-nitrosative stress and PARP activation, and both phenomena were less manifest in diabetic 12/15-lipoxygenase^−/− mice or diabetic wild-type mice treated with a 12/15-lipoxygenase inhibitor than in untreated diabetic wild-type mice (50). Therefore, it is quite plausible that 12/15-lipoxygenase overexpression, at least partially, accounts for enhanced oxidative-nitrosative stress and PARP activation in peripheral nerve and DRG neurons in mice fed HFD.

The present study has also demonstrated that pre-diabetic neuropathy in HFD-fed mice was markedly alleviated by a dietary intervention alone or a combination of a “healthy” diet and pharmacological treatment with the aldose reductase inhibitor fidarestat. Alleviation of PDN by normal-fat diet or normal-fat diet plus fidarestat is a result of dietary or combined dietary and pharmacological suppression of multiple pathogenetic mechanisms including but not limited by increased aldose reductase activity, oxidative-nitrosative stress, and 12/15-lipoxygenase overexpression. Interestingly, MNCV deficit associated with pre-diabetic neuropathy appeared unamenable to a short-term intervention with normal-fat diet but was essentially corrected by a combination of normal-fat diet and fidarestat. In contrast, SNCV was amenable to correction with both interventions. A greater amenability of SNCV compared with MNCV to pharmacological interventions has previously been reported in several experimental neuropathy studies (24,25). Of note, a combination of normal-fat diet and fidarestat was more effective on large motor fiber neuropathy (manifested by MNCV), large sensory fiber neuropathy (manifested by SNCV), small sensory fiber neuropathy (thermal hypoalgesia and tactile allodynia), and some metabolic changes in the peripheral nerve (12/15-lipoxygenase overexpression) than normal-fat diet alone, which suggests that the pathogenetic contribution of sorbitol pathway activation to diabetic neuropathy starts very early at the stage of pre-diabetes. However, it is important that even a short-term intervention with a “healthy” normal-fat diet, introduced at the earliest stage of pre-diabetic neuropathy, before any small sensory nerve fiber loss and axonal atrophy, led to essential disappearance of functional changes characteristic for large and small sensory nerve fiber neuropathy. These findings provide rationale for clinical studies to test whether nutritional approaches cure sensory neuropathy associated with IGT and obesity, as well as metabolic syndrome.

In conclusion, the HFD-fed mouse is an animal model of pre-diabetes and obesity that develops functional, but not structural, changes of large motor and sensory nerve fiber and small sensory nerve fiber PDN and displays evidence of increased sorbitol pathway activity, oxidative-nitrosative stress, PARP activation, and 12/15-lipoxygenase activation developing in the absence of overt hyperglycemia. Like other animal models of human disease, the HFD-fed mouse model may have limitations that can be properly identified only after achieving a consensus on manifestations of neuropathy in human subjects with conditions (IGT, obesity, and metabolic syndrome) that precede and often lead to the development of overt diabetes. A short-term intervention with a healthy normal-fat diet and normal-fat diet plus fidarestat led to essential disappearance of functional changes associated with either sensory (diet alone) or both sensory and motor (diet plus ARI) neuropathy. These findings emphasize the rationale for the development of nutritional approaches for early treatment of neuropathy and other diabetes-associated complications in human subjects with IGT, obesity, and metabolic syndrome that include two of the aforementioned disorders.

The study was supported by American Diabetes Association Research Grant 7-05-RA-102 and a Sanwa Kagaku Kenkyusho grant (both to I.G.O.).

We thank Drs. David A. York and Jianping Ye for valuable recommendations on dietary modeling of obesity and pre-diabetes and Dr. Andrew Mizisin for help with setting assessment of tibial nerve morphometry at Pennington Biomedical Research Center.