Effects of Polyol Pathway Hyperactivity on Protein Kinase C Activity, Nociceptive Peptide Expression, and Neuronal Structure in Dorsal Root Ganglia in Diabetic Mice

All animal experiments followed the Guidelines for Animal Experimentation of Hirosaki University (approval number M99013). To establish the transgenic mice that overexpress human aldose reductase, C57BL/6 was first mated with DBA2 and the offspring (BDF1) used for transgene integration. BDF1 female mice under hyperovulation were mated with male littermates, and full-length human aldose reductase cDNA with a mouse major histocompatibility antigen class I promoter (H-2Kd) was injected into the fertilized eggs and implanted into the uterus of founder mice (10,11). Integration of the transgene into the offspring (BDF2) was detected by PCR. Female mice transgenic for human aldose reductase were used in this experiment, and littermate mice that did not express human aldose reductase were used for comparison.

Both transgenic and littermate mice at 8 weeks of age were made diabetic by an injection of STZ (160 mg/kg i.p.) (Sigma, St. Louis, MO), and diabetic mice with blood glucose levels >19.4 mmol/l and with constant glycosuria were used. Nondiabetic control (littermate and transgenic) mice were injected with sterile saline alone. Diabetic littermate and transgenic mice were divided into two groups; one group remained untreated, and the other group was treated orally with an ARI (4 mg · kg⁻¹ · day⁻¹) (SNK-860, fidarestat; kindly provided by Sanwa Kagaku Kenkyusho, Nagoya, Japan) for 12 weeks.

During the experimental period, body weight, blood glucose levels, and pain sensation threshold were regularly monitored, and at the end of experiment, motor nerve conduction velocity (MNCV) and sensory nerve conduction velocity (SNCV) were measured. After 12 weeks of diabetes, the animals were killed by an overdose of pentobarbital (Abbot, North Chicago, IL), and the left and right lumbar DRG (L4–6) were taken and processed for biochemical analysis of PKC activity, protein expression of PKC isoforms, and measurement of sorbitol and fructose concentrations. For this purpose, DRG tissues were weighed and frozen by liquid nitrogen until assay. One DRG tissue was fixed in 10% buffered formalin and embedded in paraffin for immunohistochemical analysis of PKC isoforms, nociceptive peptides of CGRP, and substance P and for detection of cell apoptosis. For ganglion cell morphometry, another DRG was fixed in 2.5% glutaraldehyde buffered with 0.05 mmol/l sodium cacodylate (pH 7.3).

All animals were maintained in plastic cages in rooms with a constant temperature of 25°C and a 12-h light-dark cycle. All animals were given free access to food and water during the experimental period. All animals were coded, and examinations of pain sensation threshold as well as nerve conduction velocities were conducted in a double-blinded manner. Tissue samples for morphometric analysis on DRG and immunohistochemistry were also coded and evaluated by the examiners, who were unaware of the identity of the samples.

The mechanical threshold for nociceptive flexion was determined by measuring the foot-withdrawal threshold elicited by stimulation of the left hind paw using analgesimeter (UGP; Basile, Varese, Italy). This device generates a mechanical force that increases linearly with time. The force was applied by a dome-shaped plunger. The nociceptive threshold was defined as the force at which the mouse withdrew its paw. Mice were trained in the paw-withdrawal test for 30 min each day for 3 days. The measurement was performed before induction of diabetes and at 5, 7, and 12 weeks after induction of diabetes. Each time the test was repeated five times, and the mean values represented the threshold of the individuals.

For measurement of MNCV and SNCV, the procedure commonly adopted for measuring M-waves and H-reflexes was used (23). In brief, all mice were anesthetized with isoflurane (Abbot) and placed on a thermostatically controlled heated mat to maintain the body temperature at 37°C. The temperature near the sciatic nerve was also kept constant at 37°C by monitoring with an electronic thermometer (PC-9400 Delta; Sato Keiryoki MFG, Tokyo, Japan) with the aid of a warmed blanket.

For MNCV, the left sciatic nerve was electrically stimulated first at the site of Achilles tendon using a general evoked response stimulator (MS92 electromyogram device; Medelec, London, U.K.) and then at the site of sciatic notch, and the waves were recorded from the second interosseus muscle of the foot. In this case, supramaximal electrical stimulation of 0.1-ms pulses was used to generate M-waves. The point of first negative deflection of M-wave was identified as the period of latency. The latency differences derived from two stimulating sites were divided by the distance between the stimulating sites, yielding the value of MNCV.

For detection of SNCV, the power of electrical stimulation was gradually decreased. With this reduction, the M-wave diminished and instead H-reflex wave appeared. The initial deflection point of H-reflex wave was identified as the latency for SNCV. The identity of the H-reflex was validated by its disappearance on generation of a maximal M- and F-wave. The difference of proximal and distal latency was divided by the distance between the stimulating sites, yielding SNCV. An average of at least five recordings for each was used for measurements.

Tissue levels of sorbitol and fructose in DRG were measured by liquid chromatography with tandem mass spectrometry (LC/MS/MS) method described previously (24). The concentrations were expressed as nanomoles per milligrams of protein.

PKC activities were assayed by the method described previously (12). DRG samples were transferred to a tube containing 1.0 ml homogenization buffer (20 mmol Tris-HCl [pH 7.5], 330 mmol sucrose, 0.5 mmol EGTA, 2 mmol EDTA, 2 μg/ml aprotinin, 25 μg/ml leupeptin, and 1 mmol phenylmethylsulfonyl fluoride) and homogenized with a Polytron. Homogenate was centrifuged at 50,000g for 30 min at 4°C. Supernatant was collected and used as cytosolic fraction. The pellet was resuspended in 0.6 ml homogenization buffer containing 1% Triton X-100 and stored on ice for 1 h. Resuspended solution was centrifuged at 50,000g for 30 min at 4°C, after which supernatant was used as membrane fraction. Phosphorylation assay was carried out in a reaction mixture (20 mmol Tris [pH 7.5], 1 mmol CaCl₂, 10 mmol MgCl₂, 33 μmol octapeptide [RKRTLRRL], 5 mmol EGTA, and 10 μmol γ-³²P-ATP [5–10 × 10⁵ cpm]) (Perkin Elmer Life Sciences, Boston, MA) in the presence or absence of 6.4 μg/ml diorein and 96 μg/ml phosphatidylserine. The reaction was started by the addition of 30 μl cytosol or membrane fraction, incubated at 30°C for 10 min, and terminated by spotting the reaction mixture onto P-81 paper (Whatman, Maidstone, Kent, U.K.). P-81 paper was washed by 75 mmol phosphate four times for 15 min. The radioactivity was counted by liquid scintillation spectrometer (Aloka, Tokyo, Japan).

Western blot analysis was performed using proteins that were extracted as cytosol and membrane fraction for PKC assay. SDS-PAGE was performed using the Xcell SureLock system (Invitrogen, San Diego, CA) in the reducing condition. Aliquots of 100-μg samples of protein were dissolved in the sample buffer (2.5% 2-mercaptoethanol, 62.5 mmol Tris-HCl, 10% glycerol, 2% SDS, 0.0025% bromophenol blue, and 50 mmol reducing agent [dithiothreitol; DTT], pH 6.8) and loaded onto the Novex Tris-glycine Pre-Cast Gel (Invitrogen). After completion of the migration, the proteins were transferred to a polyvinylidene fluoride membrane (Immobilon-P; Millipore, Bedford, MA) in a transfer buffer (25 mmol Tris, 0.2 mol glycine, and 20% methanol) using a wet transfer unit of Xcell SureLock system. For blocking, membranes were incubated with 5% skimmed milk in PBS-T (PBS Triton X-100; 137 mmol NaCl, 2.7 mmol KCl, 1.5 mmol KH₂PO₄, 8.0 mmol Na₂HPO₄, pH 7.4, and 1% Triton X-100) overnight at 4°C. After washing with PBS-T, membrane was incubated with polyclonal anti–PKC-α–, βI–, and βII–specific antibodies (Santa Cruz Bio Tech, Santa Cruz, CA) and β-actin–specific antibody (Santa Cruz) for 1 h at room temperature. The dilution of all antibodies was 1:1,000. A final incubation was carried out with peroxidase-conjugated anti-rabbit or anti-goat IgG (Santa Cruz) for 45 min at room temperature. Immunodetection was performed by enhanced chemiluminescence (Amersham-Pharmacia, Buckinghamshire, U.K.). Quantitative analysis of exposed films was performed using National Institutes of Health image software (Version 1.61).

For immunohistochemical analysis, 4–μm thick sections of formalin-fixed tissues were deparaffinized and pretreated with methanol containing 0.3% H₂O₂ to eliminate endogenous peroxidase activity. Polyclonal anti–PKC-α, -βI, and -βII antibodies (1:200 dilution; Santa Cruz), CGRP (1:1,000 dilution; Affiniti Research Products, Exeter, U.K.), and substance P (1:500 dilution; Affiniti Research Products) were applied to the sections overnight at 4°C. They were then followed by incubation with secondary and tertiary agents using a streptavidin-biotin-peroxidase detection kit (Histofine SAB-PO Kit; Nichirei, Tokyo, Japan). N,N′-diaminobenzidine was used to visualize peroxidase deposition at the antigenic sites, and these sections were counterstained with hematoxylin. The specificity was confirmed by the replacement of the primary antibodies with nonimmune sera or by the omission of the primary antibodies. Using the staining slide, the localization of PKC isoforms on ganglion cells was observed. The number of positively stained nucleated ganglion cells with CGRP- and substance P–specific antibodies was also counted on a single cross section of DRG at a magnification of ×200 in each animal and expressed as a percentage of positive cells to total nucleated ganglion cells counted. For the objective comparison of the staining results among all groups, DRG sections from each group were randomly mounted on a single slide (i.e., six DRG tissues on one slide) and stained under the same condition.

Detection of cell apoptosis was performed by the transferase-mediated dUPT nick-end labeling (TUNEL) method using an apoptosis detection kit (Apop Tag; Chemicon International, Temecula, CA).

The fixed samples of left lumbar DRG (L5) were postfixed in 1% osmium tetroxide and dehydrated through an ascending series of ethanol concentrations and embedded in epoxy resin. One-micron thick sections of DRG were stained with toluidine blue. In morphometric analysis, the cellular area of ganglion cells containing nuclei and their nuclear area were measured at a magnification of ×1,000 by a computer-assisted image analyzing system (National Institutes of Health image version 1.61; National Institutes of Health, Bethesda, MD). The mean values of cellular area, nuclear area, and the ratio of nuclear area to cellular area were obtained from the values of 50 cells on average randomly selected on each cross section of each animal.

Data were expressed as means ± SE. Statistical analysis was carried out on a Macintosh computer (Apple, Cupertino, CA) using a commercially available statistical program (Statview, version 4.11 J; Hulinks, Tokyo, Japan). Comparison of the values among the groups was carried out using one-way ANOVA, followed by Bonferroni’s corrections for multiple comparisons. For analysis of the values of pain sensation threshold, after the comparison among all groups with the above method, specific comparisons among each littermate or transgenic group were made separately. P values <0.05 were considered to be significant.

Transgene expression did not affect the general condition and behavior of the animals. Body weight values in transgenic mice were comparable with those of littermate mice throughout the experimental period (Table 1). At the end of the study, body weight values of diabetic mice were significantly lower than those of nondiabetic mice (P < 0.01), but there were no significant differences between transgenic and littermate mice groups. Blood glucose levels were markedly elevated following the induction of diabetes but were not different between transgenic and littermate mice after diabetes induction. ARI treatment had no effect on body weight or blood glucose level.

There were no significant differences in both sorbitol and fructose contents between nondiabetic transgenic and nondiabetic littermate mice (Table 1). Carbohydrate contents in diabetic groups were remarkably elevated compared with nondiabetic groups and were significantly greater in diabetic transgenic mice than in diabetic littermate mice (P < 0.01). ARI treatment corrected the rise of sorbitol (P < 0.01 for both untreated groups) but only partially of fructose in both diabetic transgenic mice and diabetic littermate mice (P < 0.05 for both).

At the end of the experiment, diabetic transgenic mice showed a significant decrease in MNCV compared with nondiabetic groups (P < 0.01), while the decrease of MNCV in diabetic littermate mice was not significant (Table 1). The difference in MNCV between diabetic transgenic mice and diabetic littermate mice was also significant (P < 0.01). ARI treatment significantly improved MNCV in diabetic transgenic mice (P < 0.01). SNCV was also significantly decreased in diabetic transgenic mice compared with nondiabetic groups (P < 0.01), but the decrease of SNCV in diabetic littermate mice was not significant. The difference of SNCV in two diabetic groups was also significant (P < 0.01). ARI treatment significantly improved SNCV in diabetic transgenic mice (P < 0.01).

At 5 weeks after induction of diabetes there was a trend toward decreased pain sensation threshold in diabetic littermate mice (Fig. 1A). Thereafter, the threshold was conversely increased (P < 0.05 vs. nondiabetic littermate mice at 7 weeks). While the threshold in diabetic transgenic mice was significantly decreased compared with nondiabetic transgenic mice at 5 weeks (P < 0.05), it was thereafter increased (P < 0.05 and P < 0.01 vs. nondiabetic transgenic mice at 7 and 12 weeks, respectively) (Fig. 1B). ARI treatment inhibited these changes in both diabetic groups. At 12 weeks, the threshold was significantly greater in diabetic transgenic mice than in diabetic littermate mice (P < 0.01).

PKC activities in membrane fraction were comparable between the two nondiabetic groups, but they were significantly reduced by diabetes (P < 0.05 for diabetic littermate mice and P < 0.01 for diabetic transgenic mice) (Fig. 2). The activities in diabetic transgenic mice were more severely depressed compared with those of diabetic littermate mice (P < 0.01). ARI treatment significantly improved the activities in diabetic groups (P < 0.05 for diabetic littermate mice and P < 0.01 for diabetic transgenic mice). By contrast, PKC activities in cytosolic fraction were significantly increased in diabetic transgenic mice compared with other groups (P < 0.01), while those in diabetic littermate mice were not significantly altered. ARI treatment corrected this increase.

The relative intensities of the expression bands of PKC-α of the membrane fraction were markedly reduced in diabetic compared with nondiabetic groups, and the reduction in diabetic transgenic mice (57.9%) was more severe than that in diabetic littermate mice (17.6%) (P < 0.01) (Fig. 3). ARI treatment corrected these changes (P < 0.05 for diabetic littermate mice and P < 0.01 for diabetic transgenic mice). In contrast to the changes in membrane fraction, the expression levels of cytosolic PKC-α isoform were significantly elevated by 25.8% in diabetic littermate mice (P < 0.05) and by 36.4% in diabetic transgenic mice (P < 0.05) compared with those in the nondiabetic groups, but there was no significant difference between the two diabetic groups. ARI treatment suppressed the elevation of PKC-α isoform expression in diabetic transgenic mice (P < 0.05) but not in diabetic littermate mice.

The levels of protein expression of PKC-βI isoform in membrane fraction were not significantly different among all of the groups. By contrast, the levels of PKC-βII isoform in membrane fraction were significantly increased in diabetic transgenic mice compared with nondiabetic groups (P < 0.05), but there was no significant difference between the two diabetic groups. ARI treatment inhibited this elevation (P < 0.05). The expression levels of both PKC-βI and -βII isoforms in cytosolic fraction were not different between nondiabetic and diabetic groups, and ARI treatment had no effect on the expression of these isoforms.

In nondiabetic groups, strong positive staining reactions for PKC-α were detected on membrane portions of ganglion cells, predominantly of small to medium sizes (Fig. 4). The membrane reactions in the two diabetic groups appeared to be less apparent compared with the nondiabetic groups and instead showed diffuse weak reactions in the cytosol. Membrane reactions were almost diminished in diabetic transgenic mice. ARI treatment restored the membrane positivity in both diabetic transgenic and diabetic littermate mice. By contrast, reactions for PKC-βI and -βII were mostly located in the cytosol, and diabetic condition did not appear to affect the staining patterns in diabetes (pictures not shown).

Positive reactions of CGRP and substance P were found on small- to medium-sized ganglion cells (Fig. 5). The reactions appeared to be reduced in diabetic groups, markedly so in diabetic transgenic mice. Quantitative evaluations revealed a significant reduction in the population of CGRP-positive cells in diabetic compared with nondiabetic groups (P < 0.05) (Table 2). The reduction of CGRP neurons was most marked in diabetic transgenic mice, but the difference between diabetic transgenic and diabetic littermate mice was not significant. By contrast, reduction of substance P–positive cells was only significant in diabetic transgenic mice compared with the other groups (P < 0.05). ARI treatment partially but significantly corrected the decrease in the population of both CGRP and substance P neurons in diabetic transgenic mice (P < 0.05 for both).

There was no TUNEL-positive reaction in DRG, indicating the absence of the cells undergoing breaks of nuclear DNA strands and hence possibly a lack of ongoing apoptosis in ganglion cells in all experimental groups.

There was no significant difference in the mean cellular area between transgenic mice and littermate mice without diabetes (Table 3). While the cellular area was not altered in diabetic littermate mice, it was significantly reduced in diabetic transgenic mice compared with the nondiabetic groups (P < 0.05). ARI treatment significantly prevented this reduction (P < 0.05). The mean nuclear area was comparable among groups of littermate, transgenic, and diabetic littermate mice, while the area of diabetic transgenic mice was significantly reduced compared with other groups. This reduction was significantly prevented by ARI treatment (P < 0.05).

The density of ganglion cells was comparable among all of the groups; thus, there was no significant loss of ganglion cells in the diabetic groups.

In this study, the effects of aldose reductase overexpression on the neuropathic changes were well exemplified by marked slowing of nerve conduction, augmented changes of altered pain sensation, and reduced nociceptive peptide expressions. Ganglion cell structure and metabolic signals of PKC activity that were severely altered in diabetic transgenic mice may underlie these neuropathic changes, implicating the significance of aldose reductase overexpression in the development of symptomatic sensory neuropathic changes in diabetes. In this study, we clearly demonstrated the decreased PKC activity in sensory ganglion cells in diabetic mice with aldose reductase overexpression. The reduced PKC activity was well accounted for by the reduced membrane expression of PKC-α isoform that was translocated to cytosol. The reduced PKC activity was associated with decreased ganglion cell size, which may possibly be related to distal fiber atrophy in this model, as demonstrated in our previous studies (11). A relationship of the reduced PKC-α expression with the ganglion cell atrophy accords with the recent data that PKC-α is crucial for the regulation of slow axonal transport (25) as well as the maintenance of cell size in an altered osmotic environment, a condition that may occur in hyperglycemic milieu (26). The ganglion cell atrophy is common with other diabetic animal models (27,28) and may be closely related to metabolic imbalance through polyol pathway hyperactivity as well as other metabolic cascade (29). Impaired neurotrophic support, as well as insulin deficiency known to be present in the STZ-induced diabetic rat (30,31), may also lead to such characteristic structural change.

The alterations of PKC activity in diabetic nerve are controversial from depressed (32) to increased (33) or not changed (34). The discordant results may have been derived from the different methods of measurement or the difference in tissue preparations. To support this contention, our previous studies demonstrated that PKC activities are different between endoneurial and epineurial tissues and between membrane and cytosolic fractions (12). Further, the changes in diabetic nerve were different among various isoforms (12). The endoneurial fractions represented the decreased membrane PKC-α isoform with reduced enzyme activity, whereas increased PKC activity due to elevated expression of membrane PKC-β isoform was characteristic in vessel-rich tissues such as epineurium (12). In this study, the immunohistochemical results clearly demonstrated the in situ translocation of PKC-α isoform from the membrane to the cytoplasm, which accords with the results of our Western blot analysis. Contrariwise, PKC-βII isoform in diabetic transgenic mice was slightly elevated, consistent with the data obtained from the STZ rat model (35,36). Interestingly, ARI treatment corrected the alterations of both PKC-α and -βII isoforms. It is thus likely that polyol pathway activation exerts different effects on PKC-α and -β isoforms in the nerve.

The decrease in PKC activity in diabetic nerve was once proposed to be due to reduced synthesis of DAG from substrate of myo-inositol (32). This is unlikely in our model because we could not detect significant reduction of nerve myo-inositol in diabetic transgenic mice in our previous studies (11). Alternatively, excessive oxidative stress related to reduced glutathione as well as impaired nitric oxide production caused by NADPH reduction may be involved in the translocation of PKC-α isoform, as it is claimed to operate in vascular walls (4,9,37). In this setting, the role of mitochondria in the production of oxidative stress through increased glucose flux may be crucial (38). In fact, compared with diabetic wild mice, the levels of reduced glutathione were more severely decreased in diabetic mice overexpressing aldose reductase, specifically in Schwann cells, accompanied by reduced MNCV despite the absence of significant sorbitol accumulation (13).

However, interpretation of aldose reductase overexpression on the neuronal changes of PKC activity must be made with caution because polyol flux should also be exaggerated in satellite cells as well as vascular components of DRG in our model, as expected in human diabetic subjects (12,39). Hence, there is a possibility that the changes may not be a primary event in ganglion cells but could be a secondary event exerted by combined effects of the changes in satellite cells as well as vascular tissues in DRG. It would be interesting to examine the changes of DRG in transgenic mice that overexpress aldose reductase only in Schwann cells in order to clarify whether similar neuronal changes are detected under diabetic condition.

Our study revealed that the pain sensation in diabetic animals was initially hyperalgesic followed by late hypoesthesia in the presence of a progressive delay of both MNCV and SNCV. This result argues against the results obtained from the STZ-induced diabetic rat (36) and the diabetic mice with short duration (40). We only conducted the paw withdrawal test under mechanical pressure for the pain sensation and therefore need further evaluation using different methods such as a test for thermal sensitivities. Nevertheless, our results are still in keeping with the data from recent studies on mice (41) and the neuropathic process in human diabetes (42,43).

Reduced expression of nociceptive peptides of CGRP and substance P detected in this study is not specific in this model but has been reproducibly demonstrated in STZ-induced diabetic rats with 4–8 weeks’ duration (19,44). The expression of CGRP and substance P was well regulated by neurotrophic support, particularly nerve growth factor, neurotrophin-3, and ciliary nerve trophic factor, which are known to be affected in the diabetic state (45) and corrected by insulin or ARI treatment (19,46). These results strongly suggest that sensory neuropathic symptoms develop based on the metabolic aberration occurring in sensory ganglion cells. The results that augmented structural changes of sensory ganglion cells concurrent with reduced PKC activity and the ARI effects in our model provide strong rationale to use ARIs for diabetic patients with mild neuropathy to halt or prevent the development of symptomatic neuropathy in diabetic patients.

This study was supported by a grant-in-aid from the Japanese Ministry of Science, Culture, Education and Sports (#10470054, #40111231), the Japan Applied Medical Research Foundation, and the Japanese Clinical Pharmacology Foundation.

The authors acknowledge the excellent technical assistance of Drs. Noriaki Kato and Moritake Goto in Sanwa Kagaku Kenkyujo for the measurement of nerve polyols.