In contrast to the unsuccessful axonal regeneration in the central nervous system (CNS) in adult mammals, neurons in the peripheral nervous system (PNS) have a potential to grow new axons after injury. Successful nerve regeneration with functional recovery in the PNS can be attributed to both intrinsic and extrinsic factors, such as the ability of neurons and/or Schwann cells to regenerate neurites, the distal environment supportive of axon regrowth, and the target tissues receptive to reinnervation (Figure 1) (1). Although growth inhibitory molecules, such as chondroitin sulfate proteoglycans and myelin-associated glycoprotein, are present in both adult CNS and PNS, Wallerian degeneration of the peripheral nerves distal to the injury provides an optimal environment for axonal sprouting from the injured neurons (1, 2). The sequence of cellular events during Wallerian degeneration includes Schwann cell activation and proliferation, macrophage recruitment, elimination of axonal and myelin debris, and synthesis of neurotrophic and chemotactic factors. However, axonal regeneration and reinnervation depends on appropriate contact of regenerating axonal sprouts with Schwann cell basal laminae in the distal nerve segment. Currently, clinical approaches to repair axonal injury are still far from satisfactory (3).

Diabetic peripheral neuropathy is one of the most common and intractable complications of diabetes mellitus, and its prevalence correlates closely with the degree and duration of the disease (4). Impaired axonal regeneration is one of the salient features of diabetic neuropathy; however, its pathogenesis remains to be elucidated (5). In this paper, we review the findings from recent studies that focus on how diabetes affects the intrinsic and extrinsic factors of peripheral nerve regeneration. In addition, we discuss our own results from in vivo and in vitro studies, which indicate that defective neurite outgrowth after axonal injury occurs under diabetic conditions (6–8).

Impaired axonal regeneration following peripheral nerve injury is a characteristic feature of diabetic neuropathy and has been documented in experimental diabetic animals [e.g., streptozotocin (STZ)-diabetic rats (9) and mice (10), biobreeding Worcester (BB/Wor) rats (11), and Goto-Kakizaki rats (12)]. We classify these animal studies into the three categories indicated in italics in Section “Introduction” and briefly introduce recent studies focusing on the abnormalities in each category (Figure 2).

In 1991, we introduced cell culture system of dissociated DRG neurons from experimental diabetic animals (Figure 3) (65, 66). Since then, this system has been recognized as a useful tool to explore the changes of neurons that are exposed to hyperglycemia for a significant period of time and is now widely employed for the study of diabetic neuropathy (13, 67). However, it remains controversial whether the neurite outgrowth of diabetic DRG neurons is impaired, and the cell culture system of dissociated DRG neurons has the following problems for evaluation of axonal regeneration. First, there is a possibility that severely damaged neurons in diabetic animals are eliminated during the procedure of culture. Second, there is a topographic difference in the site of neurite outgrowth where neurites sprout from neuronal cell bodies in vitro, whereas axonal regeneration occurs at the sites of injury in vivo. Third, neurons are separated from other neurons and non-neuronal cells. It is thus likely that the effects of cellular interplay on axonal regeneration cannot be evaluated by this system. To overcome these issues, we have established a three-dimensional culture system of ganglion explants in which adult peripheral ganglia with nerve fibers are embedded in collagen gel or Matrigel^®, and neurite outgrowth from transected nerve stump has been evaluated (Figure 4) (13, 68). Because of the maintenance of cell-to-cell interactions, it may be reasonable to consider that the explant culture mimics axonal regeneration in vivo better than the use of dissociated cells. We introduced this culture system to vagal nodose ganglia and DRG of STZ-diabetic mice and found that the neurite outgrowth from transected nerve stump in diabetic mice was not diminished but rather enhanced (6, 7). These findings were contrary to our expectations and in stark contrast to those in previous studies in vivo. Since it is well accepted that hyperglycemia impairs neural function, enhanced neurite outgrowth in vitro in our study does not seem to imply the acceleration of proper axonal regeneration after injury in diabetes. Considering that the explant culture lacks the process of Wallerian degeneration and subsequent regenerative events at the site distal to injury, it seems more likely that the excess of regenerating fibers in diabetes may impede appropriate target reinnervation and functional recovery. This hypothesis was supported by the findings of in vivo studies (5, 8, 69), which will be discussed in the next section. Gumy et al. (36) applied the explant culture system to normal adult mice and observed that exposure to high-glucose (60 mM) culture conditions resulted in retardation of neurite outgrowth and Schwann cell migration from DRG explants. These findings are in disagreement with those in our studies (6, 7) and may be accounted for by the differential effects of long-term (>3 weeks) hyperglycemia in vivo in our study versus short-term (<1 week) high-glucose environments in vitro on neurite outgrowth in their study. In neither study, however, the hyperglycemic insults induced apoptotic cell death of DRG neurons. These findings indicate that apoptosis (70, 71) is unlikely to be a major contributor to the impaired axonal regeneration in diabetic neuropathy.

Using electron microscopy, Yasuda et al. (38, 69) evaluated the regenerative capacity of myelinated nerve fibers (MNFs) and unmyelinated nerve fibers (UNFs) in STZ-diabetic and normal control rats at 5 and 24 weeks after sciatic nerve crush. The number and size of regenerating MNFs after crush injury were significantly decreased in diabetic rats relative to those in control rats. The structural changes were consistent with the reduced motor nerve conduction velocity detected in diabetic rats with nerve crush (38). In contrast, the density and number of UNFs were markedly increased in both normal and diabetic groups at 5 weeks after injury, and these parameters returned to their baseline levels in normal control rats but not in diabetic rats at 24 weeks after injury (69). Although the precise mechanisms accounting for the differential effects of hyperglycemia on the regenerative capacity of MNFs versus that of UNFs remain unclear, the authors inferred that the long-lasting impairment of regenerative capacity of UNFs in diabetes might disturb the proper connection between the nerve endings and target organs. This inference seems to be well fitted with our hypothesis based on the in vitro assays (6, 7). Moreover, they found amelioration of the reduced motor nerve conduction velocity and number and size of regenerating MNFs in the crushed nerves in diabetic rats treated with tolrestat, an ARI (38). There were no significant effects, however, on the regenerative capacity of UNFs in the ARI-treated diabetic rats (69). ARI may therefore be more effective for regeneration of MNFs with functional repair than UNFs under diabetic conditions, and the effects may be dependent on the high expression of AR in the Schwann cells (39, 72).

By light microscopic and ultrastructural morphometric analyses, we recently evaluated nerve fiber regeneration from axotomized sciatic nerves of STZ-diabetic rats and non-diabetic control rats (8). Four weeks after axotomy, there was a significant increase in myelinated fiber density and total fiber number at the cut end in diabetic rats relative to those in control rats. In contrast, the average number of myelin lamellae per fiber, measured by electron microscopy equipped with a computer-assisted image analyzing system, was significantly reduced in diabetic rats relative to that in control rats. These findings indicate enhanced regeneration of thinly MNFs after axotomy in rats with experimental diabetes. Impaired myelination of regenerated fibers may be attributable to metabolic disorders of Schwann cells and increased vulnerability of the myelin sheath to oxidative stress under diabetic conditions (73–75). Increased regeneration of small nerve fibers in this study may be associated with the improper target reinnervation described above and painful phenotypes of diabetic neuropathy (76), but further evidence is required to support this contention.

Although the purpose of this paper is to overview altered axonal regeneration in experimental diabetes, several findings illustrated here are closely associated with those from clinical studies. For instance, the restorative effects of ARIs on MNF regeneration in diabetic rats (11, 38) were also indicated in the specimens of sural nerve biopsies from patients with diabetic neuropathy (77, 78). Exogenously applied neurotrophic factors, growth factors, and other biological molecules improved axonal regeneration and other neurological functions in animal models of diabetes (5, 17, 18, 79); however, the translation of these molecules (e.g., NGF, BDNF, and VEGF) to humans has failed in clinical trials (80–82). These failures may be attributable to the irrelevance of animal models and outcome measures to the pathology and pathophysiology of human diabetic neuropathy (83).

Because the loss of corneal nerves has been shown to correlate well with the severity of peripheral nerve damage, corneal confocal microscopy (CCM) is recognized as a novel non-invasive diagnostic modality for patients with diabetic neuropathy (84). CCM has also been applied to animal models of diabetes, and Yorek et al. (85) reported that glycemic control by continuous administration of insulin in STZ-diabetic mice significantly restored corneal nerve innervation. In agreement with this study, long-term normoglycemia by simultaneous pancreas and kidney transplantation in patients with type 1 diabetes improved corneal nerve fiber density, branch density, and length (86). In contrast, the transplantation failed to ameliorate the loss of epidermal nerve fiber density that was evaluated by skin biopsies (87). These findings imply that CCM can be a more sensitive means than skin biopsies for the assessment of early nerve regeneration and repair after therapeutic interventions.

Diabetes mellitus has been shown to affect peripheral nerve regeneration in various ways. In the first part of this paper, we reviewed the topics of axonal regeneration in animal models of diabetes from the viewpoint of intrinsic and extrinsic factors, in particular, focusing on the events that occurred at the sites distal to the injury. Delayed Wallerian degeneration, alterations in the structure and function of the ECM components, and enhanced AGE–RAGE signaling axis caused by hyperglycemia can be major obstacles to axonal regrowth after injury. In the second part, we illustrated the excessive regenerative capacity of small nerve fibers under diabetic conditions in vivo and in vitro (6–8, 69). There is room for further investigation if these findings imply dysfunctional neurite outgrowth that disturbs appropriate target reinnervation and functional recovery. In addition, much more attention should be paid to the diabetes-induced abnormalities of target tissues receptive to reinnervation.

All the authors designed and performed the experiments described in this paper in accordance with the ethical guidelines in their institutions. KS and SY wrote the paper.

The research was conducted in the absence of any commercial or financial relationship that could be construed as a potential conflict of interest.