Our study show that the extrinsic renal nerve is affected by experimental diabetes, with the main morphological, ultrastructural findings, associated to the morphometric data, pointing to a small fiber loss. These alterations begin early, after 15 days of diabetes induction and are not prevented by conventional insulin treatment.
It is well known that the pathology of diabetic neuropathy is characterized by primary progressive nerve fiber loss [20, 21] but somatic and automomic C fibers are not usually studied in the same region of the body  making the neuropathologic basis of the small fiber loss difficult to understand. Several investigators accessed the small fiber neuropathy in diabetes, focusing on the intraepidermal nerve fibers [23, 24] but transmission electron microscopy studies are still scanty, particularly at early stages of diabetes. The extrinsic renal nerve provided an unique opportunity to investigate the involvement of small and large nerve fibers at once, since it has not only a large number of unmyelinated fibers, but also two distinct groups of myelinated fibers, as shown by the diameter distributions (Figure 4). Despite that advances on the pathogenesis of the glomerular lesions in diabetic nephropathy have been made, there is still a lot to understand. Our results suggest that lesions of extrinsic renal nerves might play a role on the glomerular functional alterations that occur very early in this diabetes model and these alterations could be involved in some degree to the development of the diabetic nephropathy.
More recently, altered function of mitochondria in metabolic diseases have been correlated to their aberrant morphology and growing evidence suggests that mitochondrial dynamics play an important role in diabetes establishment and progression . Mitochondrial phenotype is abnormal in sensory neurons, Schwann cells and sympathetic ganglia in diabetes  but little information on nerve mitochondria is available. We show severe ultrastructural alterations on the mitochondria of the renal nerve axons from diabetic rats, as early as 15 days, and these alterations were delayed but not prevented by insulin treatment. An early work reveals that hyperglycemia in diabetes triggers nutrient excess in neurons, that mediates the phenotype changes in mitochondria, and the development of axon loss in sensory neuropathy is linked to this nutrient excess [26, 27]. We are demonstrating, for the first time, the presence of mitochondrial damage on myelinated and unmyelinated fibers of the renal nerves in diabetes that might be contributing to the fiber loss observed in these nerves since early stages of the disease.
In somatic nerves, different types of sensations travel along fiber population with different average diameters. The renal nerves contain efferent post-ganglionic sympathetic fibers, thought to be exclusively unmyelinated. The myelinated fibers might be the afferent ones that would transmit sensory information from the kidneys to the central nervous system . We suggest that the small myelinated fibers could be those afferent fibers related to baro- and chemoreflex, based on information from the aortic depressor nerve myelinated fibers sizes . The chronic diabetic groups, treated and untreated, presented larger average values of myelinated fiber and axon area and diameter compared to controls. These data suggest a loss of thin myelinated fibers, shifting the average to higher values and also affecting the histogram distributions. A shift of the size distribution towards larger fibers observed in both, acute and chronic diabetic groups, associated with the shift to the right observed for the unmyelinated fiber size also in acute and chronic diabetic animals demonstrates thin fiber loss. This data is corroborated by the observation of smaller myelinated and unmyelinated fibers density in chronic diabetic animals, compared to controls, indicating the increasing severity of the fiber loss at chronic stages. This thin fiber loss could cause kidney denervation and also influence on the autonomic imbalance described in diabetes. Also, the shift observed on the unmyelinated fiber distribution, although less pronounced than the myelinated distributions, suggest the possible atrophy of thin axons in chronic diabetes, which might also contribute to the nephropathy, since the majority of the afferent renal nerves are unmyelinated. Patel and Zhang  observed that the volume reflex is reduced in STZ-induced diabetic rats after 15 days. They also showed that the natriuresis due to renal sympathoinhibion is blunted in response to volume expansion and the restoring the glucose levels to normal by insulin treatment in diabetic rats reversed the volume receptor reflex back to normal. We showed that insulin treatment delayed fiber loss in the short term diabetes but was not able to prevent it in long term. Yagihashi and Sima  discussed that the characteristic of the autonomic neuropathy involving both sympathetic and parasympathetic peripheral nerves is a structural change, consisting in dystrophic axonal alterations. These authors  affirm that the important morphometric changes such as fiber atrophy and loss of unmyelinated fibers was progressive during the course of diabetes, which is compatible with the results from the present study.
Despite the fact that the literature describes preferably a reduction of the myelinated fiber and axon sizes in somatic nerves of STZ-induced diabetic rats [31–33] our results show, in an essentially autonomic nerve, a small myelinated fibers loss, accompanied by degeneration and atrophy of the unmyelinated fibers, in this model of diabetes. This is suggestive that somatic and autonomic nerves react in distinct ways to diabetes. Another interpretation is because somatic nerves have a large number of myelinated fibers, the small fiber loss observed in autonomic nerves, might not be easily or readily identified in somatic ones, particularly because most studies used light microscopy, that has no resolution for the unmyelinated fibers investigation.
It is well known that Schwann cells are important in the regeneration processes that follow nerve injury. Nerves that are experiencing regeneration will have a larger number of Schwann cells due to their augmented duplication rate under these circumstances . We showed that density of Schwann cell is larger in chronic diabetes, treated and untreated, indicating possible regeneration of the large myelinated fibers. In fact, this result, together with the larger average G ratio in chronic diabetes and the shift to the right on the G ratio distribution, is a morphological indication that the remaining myelinated fibers on the chronic diabetic nerves have thinner myelin sheath, compatible with multiplication of the Schwann cells in order to accomplish the regeneration process. Regeneration of fibers is a slow process and this is probably the reason that these alterations were not present in acute diabetic animals.
There has been increasing evidence that microvascular pathological abnormality and ischemia may be involved in the pathogenesis of diabetic neuropathy . In fact, Fazan et al.  demonstrated severe damage of the endoneural vessels present on both STZ groups, besides the insulin treatment, using a similar chronic diabetes model from this study. We showed that at early stages of diabetes, vascular damage occurs in the endoneural space of the renal nerves that increase in severity on chronic diabetes regardless insulin treatment.
It is important to mention that the insulin treatment is demonstrated to prevent or correct the axonal atrophy caused by STZ-diabetes in the large myelinated fibers [6, 35–37] as well as the axonal diameter and their distributions . The STZ-diabetes model is widely used to investigate experimental diabetic peripheral neuropathies [6, 36, 38, 39], but few studies have performed a detailed assessment of either unmyelinated fibers or capillary morphology in this animal model, particularly on renal nerves. Thus, the present study adds useful information for further investigations on the ultrastructural basis of nerve function in diabetes and also provides support for the role of the renal nerve neuropathy on the development of the diabetic nephropathy.