The polyol pathway is a glucose shunt that diverts excess glucose to form fructose. AR is the first and rate-limiting enzyme of the pathway. It reduces glucose to sorbitol, and in the process its cofactor NADPH is oxidized to NADP. Sorbitol dehydrogenase (SDH) then converts sorbitol to fructose, whereas its cofactor NAD+ is converted to NADH. The polyol pathway was first recognized as a key factor in the development of diabetic cataracts, and soon afterwards, it was found to be involved in diabetic neuropathy, retinopathy, and nephropathy. AR also reduces galactose to form galactitol. Because a high galactose diet also produces tissue lesions similar to diabetic lesions, galactosemia has been used as an experimental surrogate for diabetes. It has also been suggested that conversion of sorbitol to fructose by SDH causes reduction of
NAD+/NADH ratio, creating a "pseudohypoxic" state that might contribute to the deleterious effects of hyperglycemia (10). As galactitol is not metabolized by SDH, galactosemia does not simulate some aspects of hyperglycemia. The key evidence for AR's involvement in diabetic neuropathy is that several structurally different AR inhibitors were shown to be effective in preventing the development of this disease in animal models (11), suggesting that the beneficial effects of these drugs is because of the inhibition of AR and not other nonintended target enzymes. However, these drugs did not demonstrate a significant beneficial effect in clinical trials (12,13), raising doubts on the validity of this model. In particular, it was pointed out that the amount of ARI required to normalize nerve blood flow and nerve conduction velocity (NCV) deficit in diabetic rats, exceeds that required to normalize the sorbitol level in the nerves, indicating the lack of correlation between polyol pathway activity and diabetic neuropathy (14). However, this observation does not disprove AR's role in the disease because, as discussed later, sorbitol accumulation is not the cause of diabetic lesions in the nerve. Rather, it is the flux of glucose through the polyol pathway that produces the toxic effect of hyperglycemia. Kinetic calculations demonstrate that when there is a rapid conversion of sorbitol to fructose by SDH, even partial inhibition of AR activity would appear to completely normalize sorbitol levels in diabetic tissue (2). Thus, normalizing nerve sorbitol level does not equate to complete blockage of the polyol pathway activity. However, this theory has not been proven in animal model. Genetic studies to determine AR's role in this disease become essential.
Because it is easier to develop TG mice than KO mice, TG approach was first used to investigate the role of AR in diabetic neuropathy. The first reported mouse model utilized the major histocompatibility complex (MHC) promoter, which is active in all tissues, to drive the expression of human AR (hAR) complementary DNA (cDNA) in TG mice. Indeed all tissues from the MHC-hAR TG mice tested were found to express the hAR mRNA, including liver, skeletal muscle, heart, kidney, brain, and lung. Under normal rearing condition, these mice developed thrombi of the renal vessels, but no abnormality was evident in the brain, lung, heart, thymus, spleen, intestine, liver, muscle, spinal cord, and sciatic nerve when examined under light microscopy (15). Under nondiabetic condition, sorbitol and fructose contents in the sciatic nerve of the TG mice were similar to that of the wild-type (WT). When induced to become diabetic, sciatic nerve sorbitol and fructose levels in the TG mice were twice that of their WT littermates, suggesting that the hAR transgene increased the nerve AR activity by about twofold. Under normoglycemic condition, there was no difference in the motor NCV (MNCV) or the structure of the sciatic nerve, indicating that overexpression of AR did not affect the normal development and function of this tissue. When fed with a diet consisting of 30% galactose, they developed more severe neuropathy than the WT mice as indicated by higher reduction of MNCV and in the mean myelinated fiber size of the sciatic nerve (16). When induced to become diabetic by streptozotocin, the MHC-hAR mice also exhibited a greater reduction of tibial MNCV and more severe myelinated fiber atrophy of the sciatic nerve (17). Further, membrane-associated PKC and Na+/K+-adenosine triphosophatase (ATPase)
activities were significantly reduced in the diabetic MHC-hAR mice sciatic nerves with the perineurial tissues removed.
Reduction of PKC activity in diabetic nerves is particularly intriguing. In other tissues of diabetic animals such as retina and kidney, PKC activity is thought to be activated (18), and an inhibitor of PKC-PII isoform has been shown to prevent the development of diabetic lesions in these tissues (19,20). PKC inhibitors have been shown to prevent the development of diabetic neuropathy in animal models (5,21). To clarify these apparently contradictory observations, PKC activity and PKC isoforms were examined in the perineurial tissues of MHC-hAR and WT mice, and the results revealed a complex response of PKC isoforms to hyperglycemia (22). In the membrane fraction of endoneurial tissues there was a significant reduction of PKC activity in the diabetic MHC-hAR mice but not in the non-TG mice. On the other hand, in the membrane fraction of the epineurial tissues diabetes led to similar increases in PKC activity in the both TG and WT mice, indicating that AR transgene contributed little to the activation of PKC in this tissue. Western blot analysis showed that in the epineurial tissues of WT mice diabetes caused a reduction in PKC-a protein level in the membrane fraction with a concomitant rise in the cytosol fraction, suggesting translocation induced by hyperglycemia. The hyperglycemia-induced redistribution of PKC-a is even more exaggerated in the MHC-hAR mice. On the other hand, hyperglycemia induced the translocation of PKC-PII protein from the cytosol to the membrane, opposite to that of PKC-a. There was slightly more PKC-PII translocated to the membrane in the TG mice than in the WT mice, but the difference was statistically insignificant. Hyperglycemia had no effect on the level or cellular distribution of PKC-PI in this tissue. In the endoneurial tissues PKC-a was not affected by hyperglycemia, whereas PKC-PI and -pII translocation from the cytosol to the membrane was increased. Again, the AR transgene only contributed a small and statistically insignificant increase in their membrane translocation. The hyperglycemia-induced membrane translocation of PKC-PI and -pII suggests activation of these PKC isoforms. As the perineurial tissues contain microves-sels, these findings lend support to the notion that vascular lesions contribute to the pathology in diabetic nerves, and explain the beneficial effects of a PKC-PII inhibitor on diabetic neuropathy. These changes in PKC activity and translocation of different PKC isoforms in the endoneurial and epineurial tissues were all normalized by ARI, indicating that hyperglycemia-induced activation of PKC is mediated by AR activity. The fact that the AR transgene did not make a statistically significant difference in the translocation of some of the PKC isoforms is probably a reflection of the modest (more than twofold) increase in AR activity contributed by the transgene in these tissues.
Another controversial issue in diabetic neuropathy is whether it is because of metabolic dysfunction of the nerve or because of lesions in the vessels resulting in ischemia in the nerve tissues they supply. The fact that administration of different vasodilators prevented the development of diabetic neuropathy strongly supports the vascular theory (23-25). To determine if the nerve tissue also contributes to diabetic neuropathy, TG mice containing the hAR cDNA fused to the rat myelin protein zero (P0) promoter were developed (26). These P0-hAR mice express hAR specifically in the Schwann cells and not in other tissues. To better illustrate increased severity of diabetic neuropathy contributed by the increased level of AR, the hAR transgene was maintained in the F1 (hybrid of CBA mouse strain x C57BL) genetic background that is quite resistant to the development of this disease. Under normal rearing condition, the TG mice did not exhibit any structural or functional abnormality in their sciatic nerve. In both galactosemic and hyperglycemic conditions the WT (F1) control mice showed small and statistically insignificant reduction in MNCV in their sciatic nerve. On the other hand, the P0-hAR mice exhibited a significant decrease in sciatic nerve MNCV, indicating that Schwann cell-specific overexpression of AR led to more severe functional deficits in this tissue. Diabetes did not lead to any drop in the reduced glutathione (GSH) level in the sciatic nerve of the WT mice, but caused a significant drop in the nerve GSH content in the TG mice, indicating that overexpression of AR increases oxidative stress. Interestingly, although there was a twofold increase in the sorbitol and fructose levels in the sciatic nerve of the diabetic TG mice compared with that of the diabetic WT mice, there was no significant difference in the galactitol levels in the sciatic nerve of the galactosemic WT and TG mice. This is probably because the galactitol level in the nerve of the galac-tosemic WT mice was already very high. The nerve galactitol level in the galactosemic WT or TG mice was about 30-fold higher than the sorbitol level in the nerve of diabetic TG mice, a reflection of the fact that galactitol is not metabolized by SDH or other enzymes. Such a high level of polyol might cause severe osmotic stress leading to leakage, reducing the increased accumulation of galactitol contributed by the hAR transgene.
Although the MHC-hAR and P0-hAR TG mice experiments clearly showed that increased AR activity exacerbates diabetic neuropathy, it might be argued that such a high level of AR activity is not found in normal animals and therefore, the transgene created an artificial disease mechanism irrelevant to the pathogenesis of the disease in WT animals. Therefore, AR gene KO mice provide an important animal model to complement ARI and TG studies. AR-deficient mice appear normal in every respect except that they drink and urinate twice as much the WT mice because of impairment in their urine concentrating mechanism (27). This mild polyuric behavior does not affect their serum electrolyte levels. To detect the potential protective effect of AR deficiency on diabetic neuropathy in a better manner, the AR null mutation was maintained in C57BL mice, which developed more severe diabetic neuropathy than the F1 mice. Under normal rearing condition, MNCV and morphology of the sciatic nerve of the KO mice were normal. When induced to become diabetic the WT (C57BL) mice showed a significant decrease in the sciatic nerve MNCV, whereas there was no change in the KO mice. Further, diabetes caused a significant reduction in the sciatic nerve GSH level, but had no effect on the nerve GSH level in the AR null mice (28). These results demonstrated that AR is a key enzyme in the pathogenesis of diabetic neuropathy and that hyperglycemia-induced oxidative stress is primarily the consequence of increased flux of glucose through the polyol pathway.
There were suggestions that the polyol pathway-mediated diabetic lesion is primarily the consequence of the activity of SDH rather than AR. Conversion of sorbitol to fructose by SDH leads to increase in the NADH/NAD+ ratio similar to that experienced by hypoxic cells. This "pseudohypoxic state" is thought to induce PKC, increase prostaglandin synthesis, increase free radical production, and increase nitric oxide production (10). However, attempts to use SDH inhibitors (SDI) to test this hypothesis gave contradictory results. A SDI called S-0773 has been shown to partially attenuate hyperglycemia-induced vascular dysfunction in the eye, sciatic nerve, and aorta in rats (29). However, other SDIs were found to exacerbate diabetic peripheral and autonomic neuropathy (30,31). A spontaneous SDH null mutation in mice had been characterized (32). There is no apparent abnormality in the SDH-deficient mice and their sciatic nerve MNCV and morphology are normal. When these mice were induced to become diabetic there was a slightly higher reduction in their sciatic nerve MNCV than that of the WT mice. However, the difference was not statistically significant (33). Nonetheless, the result indicates that blocking the SDH activity did not improve diabetic neuropathy, demonstrating that pseudohypoxia is unlikely to be the pathogenic mechanism of this disease. This set of experiments also revealed that nerve sorbitol level does not correlate to nerve dysfunction. The sorbitol level in the sciatic nerve of nondiabetic SDH null mice was about four times higher than that of diabetic WT mice. Yet the nondiabetic SDH null mice had normal MNCV, whereas the diabetic WT mice had significant slowing of MNCV. This, and the fact that the sorbitol level in the sciatic nerve of diabetic mice was only in the range of 0.2-0.6 nmol/mg dry weight, whereas myoinositol level was around 20 nmol/mg dry weight, indicate that polyol pathway-induced osmotic stress is unlikely to be a contributing factor in the pathogenesis of diabetic neuropathy in mice. Indeed, the increase in sorbitol in the diabetic nerve did not cause a compensatory decrease in myoinositol. Therefore, myoinositol depletion does not contribute to the development of diabetic neuropathy in mice.
The results of AR, TG, and KO mice experiments confirmed the findings from AR inhibitors studies. Although each of these studies might leave room for alternate interpretations, the complete agreement of these three experimental approaches unequivocally demonstrate that AR plays a key role in the pathogenesis of diabetic neuropathy. The experiments with Schwann cell-specific AR-overexpressing TG mice clearly showed that metabolic dysfunction in the nerve tissue also contribute to the pathogene-sis of the disease. Further, the experiments with SDH null mice showed that blocking this point of the polyol pathway has no beneficial effect on diabetic neuropathy, putting to rest the controversies arising from contradictory findings of the SDI studies. These experiments also demonstrated that accumulation of sorbitol does not contribute to the development of diabetic neuropathy in mice. The pathogenesis of the disease most likely involves oxidative stress generated from the flux of glucose through the polyol pathway. This is in agreement with a number of studies, which showed that antioxidants were able to prevent the development of diabetic neuropathy (34-36).
Was this article helpful?