|Home | About | Journals | Submit | Contact Us | Français|
Diabetic neuropathy is a major complication of diabetes that affects the sensory and autonomic nervous systems and leads to significant morbidity and impact on quality of life of patients. Mitochondrial stress has been proposed as a major mediator of neurodegeneration in diabetes. This review briefly summarizes the nature of sensory and autonomic nerve dysfunction and presents these findings in the context of diabetes-induced nerve degeneration mediated by alterations in mitochondrial ultrastructure, physiology and trafficking. Diabetes-induced dysfunction in calcium homeostasis is discussed at length and causative associations with sub-optimal mitochondrial physiology are developed. It is clear that across a range of complications of diabetes that mitochondrial physiology is impaired, in general a reduction in electron transport chain capability is apparent. This abnormal activity may predispose mitochondria to generate elevated reactive oxygen species (ROS), although experimental proof remains lacking, but more importantly will deleteriously alter the bioenergetic status of neurons. It is proposed that the next five years of research should focus on identifying changes in mitochondrial phenotype and associated cellular impact, identifying sources of ROS in neurons and analyzing mitochondrial trafficking under diabetic conditions.
The World Health Organization (WHO) predicts that by 2025 there will be 300 million people with diabetes. In North America 19 million people currently have diabetes that has an incidence of 6% and rising. Of those, about 90–95% have non-insulin-dependent diabetes mellitus (type 2 diabetes) and 5–10% have insulin-dependent diabetes mellitus (type 1 diabetes). In the USA approximately $25 billion per annum of health service costs is spent on treatment of diabetic complications that include retinopathy, nephropathy, heart disease and neuropathy. In 1998 approximately $15 billion of heath service expenditure was associated with the neurological complications (sensory and autonomic neuropathy and blindness). Incidence of sensory and autonomic neuropathy in diabetic patients can be as high as 50% and leads to incapacitating pain, digestive abnormalities, erectile dysfunction, heart arrhythmia, sensory loss, foot ulceration (up to 2 million Americans have this complaint), infection, gangrene and poor wound healing. The end result is often lower extremity amputation that accounts for approximately 80,000 cases each year in the USA and in rare cases sudden death, presumably on a cardiac basis. There is no effective therapy, only palliative treatment is available at the present time. These alarming figures are predicted to rise by approximately 5-fold over the next 10 years due to the epidemic in obesity and the associated increase in incidence and earlier time of onset of type 2 diabetes (from American Diabetes Association web site: http://www.diabetes.org).
Diabetic neuropathy in type 1 and 2 diabetes in humans is comprised of symmetrical sensory polyneuropathy, autonomic neuropathy and a variety of rare forms. Sensory neuropathy is associated with reduction of motor and sensory nerve conduction velocity and structural changes in peripheral nerve including endoneurial microangiopathy, abnormal Schwann cell pathology, axonal degeneration, paranodal demyelination and loss of myelinated and unmyelinated fibers - the latter due to a dying-back of distal axons that presents clinically as reduced epidermal nerve fiber density [1–4]. Diabetic patients do not typically present with devastating autonomic failure involving the sympathetic and parasympathetic nervous systems, although the incidence of autonomic nervous system dysfunction is likely to reflect the diligence and sophisticated techniques with which it is sought. The morbidity contributed by complaints involving cardiovascular, genitourinary, sudomotor and alimentary symptoms is significant to individual patients lives, may result in subclinical disease diminishing the safety factor of autonomic function and, in a poorly understood manner, result in increased mortality . The appreciation of the actual burden of autonomic neuropathy in diabetic patients has lagged well behind the characterization of somatic sensory neuropathy.
Neurodegeneration is most profound in the longest axons of neurons, and defective axon regeneration impedes tissue re-innervation . There is no molecular signature or sign of apoptosis-dependent loss of sensory or sympathetic neuron perikarya in diabetic humans or animals [7–11] although loss of small neurons does occur in long-term animal models of diabetes [9,12]. The distal dying-back and formation of axonal dystrophy (with swellings) of axons are critical pathological features [6,13–17] and mimic axonal pruning and degeneration observed in the CNS and PNS in other pathological states .
It is generally believed that oxidative stress is the key pathological process inducing nerve damage in diabetes [19,20]. Oxidative stress, possibly triggered by vascular abnormalities and associated microangiopathy in the nerve [1,21,22], is a key pathological process inducing nerve damage in diabetes in humans and experimental models [19,20,22]. Diabetes-induced oxidative stress in animal models of in type 1, type 2 and pre-diabetes in sensory neurons and peripheral nerve is demonstrated by increased production of reactive oxygen species (ROS) [23–27], lipid peroxidation [28–30] and protein nitrosylation [29–32], and diminished levels of reduced glutathione [28,33] and ascorbate . Treatment with anti-oxidants such as α-lipoic acid, γ-linolenic acid and aldose reductase inhibitors prevent many indices of neuropathy in STZ-diabetic rats [19,20,25,27]. The neurons and Schwann cells do initiate protective mechanisms involving up-regulation of antioxidant pathways [34–36], however, the neurodegenerative outcome is energy failure in the nerve, observed as a decrease in high energy intermediates (e.g. phosphocreatine) [19,20], impaired axonal transport of proteins  and sub-optimal ion pumping [38–41].
In cultured endothelial cells it has been proposed that high [glucose] drives excessive electron donation to the electron transport chain in mitochondria resulting in mitochondrial hyperpolarization and elevated production of ROS . Brownlee et al. have proposed that this mitochondrial-dependent process is a central mediator of oxidative stress in complications of diabetes [23,42]. The theory suggests that high [glucose] in tissue targets for diabetic complications leads to increased supply of NADH in the mitochondria, and that this increased electron availability and/or saturation may cause partial reduction of oxygen to superoxide radicals in the proximal part of the electron transport chain [23,42]. Subsequent large elevations in ROS then induce degeneration of tissue. Studies in cultured embryonic sensory neurons have shown that high [glucose] induces toxicity through an apoptotic route involving a mitochondrial-dependent pathway . While this work reveals some possible novel pathways for therapy, the interpretation of the results is difficult since in human diabetes and in animal models there is no evidence of apoptosis of neurons [8–10,43–45]. It is the case, however, that high [glucose] can induce neurotoxicity. For example, in STZ-diabetic rats the stress-activated kinase, p38, is enhanced in sensory neurons in lumbar dorsal root ganglia (DRG) and linked to nerve dysfunction but the apoptotic pathway is not triggered [7,46,47].
Our work and studies by Wiley et al. shows that in adult sensory neurons from STZ-diabetic rats the mitochondrial inner membrane potential is depolarized and not hyperpolarized as observed in endothelial cells exposed to high [glucose] [48–50]. Our work showed that mitochondrial depolarization in STZ-diabetes could be prevented by systemic treatment with low dose insulin, that did not impact on hyperglycemia, or neurotrophin-3 (NT-3) thus further questioning a central role for high [glucose] in mitochondrial dysfunction in diabetes [48,49]. Furthermore, insulin and other neurotrophic growth factors could directly modulate mitochondrial polarization through a phosphoinositide 3-kinase dependent pathway . Adult sensory neurons cultured from normal rats and treated with high [glucose] did not develop any indices of oxidative stress (e.g. elevated caspase-3 activation or enhanced ROS) . Furthermore, normal or diabetic neurons exposed to high [glucose], in fact as high as 50 mM, do not undergo apoptosis or any form of cell death [35,48,52]. In addition, embryonic neurons permitted to mature or adult mouse ganglia in vitro do not undergo apoptosis when exposed to high [glucose] [53,54]. Clearly, the phenotype of matured or adult sensory neurons and subsequent response of these cells to the diabetic state differs from that of endothelial cells and embryonic neurons where, for example, high [glucose] can induce apoptosis.
We and others have proposed that mitochondrial dysfunction in sensory neurons in diabetes could be associated with impaired Ca2+ homeostasis [38–40,48,55–61] and recently reviewed by Verkhratsky and Fernyhough . This includes increased steady-state intracellular Ca2+ concentration ([Ca2+]i), increased frequency of high threshold Ca2+ currents and decreased depolarization-induced Ca2+ signals. Importantly, diabetes leads to a significant decrease of caffeine-induced Ca2+ release from intracellular stores [39,40,57,62], suggesting altered endoplasmic reticulum (ER) Ca2+ ([Ca2+]ER) homeostasis. In fact, we have shown that impaired Ca2+ ion homeostasis is most profound in lumbar DRG neurons, which have the longest axons  and are initially targeted in the human condition [2,3]. Sensory neurons of STZ-diabetic rats simultaneously develop depolarization of the mitochondrial inner membrane. Importantly, a rise in [Ca2+]i of 200 nM (which is seen in diabetes) can trigger elevated intramitochondrial Ca2+ concentration ([Ca2+]m) by entry of Ca2+ through the Ca2+ uniporter and other routes [63–67], which causes partial or complete inner mitochondrial membrane depolarization . Plasma membrane depolarization-induced Ca2+ transients are prolonged in diabetic neurons and blockade of mitochondrial uptake of Ca2+ using the mitochondrial uncoupler, carbonyl cyanide m-chlorophenylhydrazone (CCCP), prevented these abnormalities . This implies that mitochondrial buffering of Ca2+ plays a role in shaping Ca2+ transients in diabetic neurons. Finally, increased [Ca2+]m can stimulate ROS levels leading to oxidative stress [65,69–72].
Extensive studies in a range of cell types have shown that a rise in [Ca2+]m can elevate the rate of NADH production by enzymes of the Krebs cycle including pyruvate dehydrogenase, isocitrate dehydrogenase, and α-ketoglutarate dehydrogenase . Further studies have shown that Ca2+ modulates the function of the F1F0 ATP synthase . Furthermore, a rise in [Ca2+]m has been suggested to enhance electron transport and cytosolic Ca2+ to activate the adenine nucleotide translocase (ANT) . The literature suggests that ATP production can be augmented quickly with a modest increase in [Ca2+]m. Rapid measurements of [Ca2+]m and NADH oxidation rate showed that Ca2+ uptake was rapid (less than 100 ms). The issue of how Ca2+ modulates mitochondrial ROS generation is at the heart of understanding how Ca2+ can be both a physiological and a pathological effector of mitochondrial function. Because the reduced ubiquinone (QH·) intermediate in the Q cycle is a significant source of ROS, two related parameters that can regulate ROS generation are 1) the effective concentration of QH·, which is increased when the distal respiratory chain is inhibited, and 2) the frequency of QH· occurrence, which is increased when the rate of electron transport is enhanced. Thus both stimulation and inhibition of mitochondria can result in enhanced ROS generation [65,72]. Stimulation of the tricarboxylic acid (TCA) cycle and oxidative phosphorylation by Ca2+ would enhance ROS output by making the mitochondrion work faster and consume more O2. Indeed, mitochondrial ROS generation correlates well with metabolic rate [65,76], suggesting that a faster metabolism simply results in more respiratory chain electron leakage (approximately 1–2% of transported electrons leak to generate ROS) [77,78]. It should be noted, however, that at this juncture there are no published studies demonstrating that abnormal Ca2+ buffering within the mitochondria acts as a source of ROS under diabetic conditions in adult neurons.
Studies on mitochondrial physiology in diabetic neuropathy lag behind those in other diabetic complications, such as cardiomyopathy and nephropathy. Table 1 summarizes findings in an array of animal models, and in humans, in type 1 and 2 diabetes with respect to analysis of functioning of the electron transport chain (ETC). In general, mitochondrial-based electron transport functions are diminished. The diabetes-induced factors causing reduced rates of ETC complex activities remain poorly understood, however, investigators have begun to use proteomic and gene array techniques to identify alterations in gene expression that presumably underpin such changes in mitochondrial physiology (for example, see Bugger et al (2009) ). Our preliminary results on mitochondrial function in DRG of age matched versus 22wk STZ-diabetic rats are shown in Table 2 and and33 and are in general agreement with the findings in heart, muscle and kidney. Mitochondrial respiration of the full ETC, coupled and uncoupled, and of complex IV (ascorbate/TMPD assay) are significantly decreased by 30–40% in STZ-diabetic rats (with correction by insulin therapy; Table 2). Table 3 shows that the complex I and IV enzyme activities are also significantly inhibited. This collection of results showing reduced complex I activity in diabetic tissues do appear to contrast with findings in epineurial arterioles serving the sciatic nerve in STZ-diabetic rats where elevated superoxide formation could be blocked using inhibitors of complex I , again highlighting the need to consider the role of mitochondrial function in diabetes on a per tissue and animal model basis.
During sensory nerve degeneration in diabetes in humans and animal models subtle changes of mitochondrial number and size have been described in Schwann cells of myelinated and non-myelinated axons. Glycogen accumulation in the outer compartment of the mitochondrion were described in axons  but the ultrastructure of mitochondria in neuronal cell bodies is reportedly unremarkable . Although mitochondria, vesicles, and 10–20 nm tubules have been described in intraepidermal axons, intraepidermal axonal swellings have not been characterized in purely diabetic patients [14,15]. The most detailed study to date  involved swellings in patients with HIV neuropathy (4 of which were also diabetic) and non-HIV idiopathic sensory neuropathy. Intraepidermal axonal swellings in that study contained numerous accumulated mitochondria, vesicles and neurofilaments and dermal nerves also contained intraxonal mildly enlarged mitochondria. Schwann cells with abnormal appearing mitochondria were also encountered  and confirmed previous work in sural nerve biopsies .
Abnormal mitochondrial ultrastructure is an important part of sympathetic autonomic neuropathology involving pre- and postsynaptic elements in prevertebral sympathetic ganglia in man and a variety of rodent models and appears to be part of an emerging story in type 1 and 2 diabetic mouse models [80–83]. Mitochondria are a normal part of presynaptic boutons, likely reflecting their metabolic activity, but have been recently found in increased numbers in autophagic vacuoles admixed with synaptic vesicles in murine diabetic sympathetic ganglia . More impressive are the accumulated mitochondria, tightly aggregated without a significant amount of intervening cytoplasm, in post-synaptic dendrites in which they may produce nearly pure aggregates (arrows, Figure 1A) often composed of smaller mitochondria than those in adjacent cell bodies (arrows, Fig 1B). In some cases mitochondria are admixed with multivesicular bodies or autophagosomes (arrows, Figure 1C) or with delicate tubulovesicular elements (Figure 1D). It is not typical for dystrophic pre- and postsynaptic elements to occur together as part of single synapses, rather dystrophic pre- and postsynaptic elements are coupled with ultrastructurally unremarkable post- or presynaptic elements, respectively. Recent studies of prevertebral sympathetic ganglia of NOD, STZ-treated and STZ-treated NOD/SCID mice  and in the spontaneously genetically diabetic Akita mouse , have demonstrated the presence of striking abnormalities in mitochondrial ultrastructure. In particular, the accumulation of small, hyperchromatic dense mitochondria that were significantly smaller than those in nondiabetic control mouse ganglion or those in the superior cervical ganglia (SCG) in the same diabetic mice in which superior mesenteric ganglia (SMG) and celiac ganglia (CG) pathology routinely occurs. Rarely such changes accompany loss of ribosomes resulting in neuronal pallor or degenerative changes in neuronal nuclei not culminating in a classical apoptotic appearance (Figure 1E). These axonopathic and neuronopathic changes are accompanied by neuronal loss in the prevertebral SMG/CG of the Akita mouse. It should be remembered that early studies of human sympathetic ganglia in autopsied human subjects uncovered a mild, poorly characterized decrease in neuronal number expressed as decreased neuronal density/mm2 of ganglionic surface area . Future studies of human ganglia using non-biased morphometric techniques may indeed identify both dystrophic terminals and neuronopathy underlying the development of autonomic neuropathy in diabetes, further complicating its pathology and possible pathogenesis. Nerve terminal damage is likely to dis- or misconnect ganglionic neurons and, particularly for prevertebral ganglia serving the viscera, contribute to the loss of integrated reflexes, perhaps complicated by the loss of selected subpopulations of neurons.
Determination of the functional significance of the alterations in the number and size of the mitochondria in diabetic sympathetic ganglia is difficult but insights have recently been provided by the analysis of a variety of human and experimental neuropathic conditions. Mitochondria are dynamic organelles which undergo fusion and fission which are regulated by a number of genes which have a role in control of normal energy supply and participate in the pathogenesis of a variety of neurodegenerative conditions [85–88]. Increased energy demand with accompanying changes in ATP/ADP gradients may induce fusion and division of mitochondria often resulting in the accumulation of small mitochondria in perikaryal and dendritic sites. Small mitochondria may more readily translocate from one dendritic and nerve terminal site of increased energy demand and help in the local regulation of energy demand and balance. Mitochondrial pathology is most prominent in dendrites, sites in which mitochondria are more highly charged and metabolically active than in axons  and, under normal conditions, they are more threadlike in shape than in axons and presynaptic terminals . Significant functional differences (e.g., susceptibility to Ca2+ overload) have also been described between synaptic and nonsynaptic mitochondria . Aggregates of mitochondria may also contribute to the development of local pathology as the result of increased production of ROS, loss of normal energy production or abnormality of Ca2+ handling ability.
There is a complex relationship between mitochondrial fragmentation and bioenergetics, as a decrease in ATP can also stimulate fragmentation . Mitochondrial fission may initially help evenly distribute energy throughout the cell body and axon. Mitochondria also undergo fission or fragmentation in response to increased levels of nitric oxide, i.e. nitrosative stress, which may secondarily result in ATP decline, further synthesis of ROS, the overexpression of Drp1 and neuronal injury . A local increase in NO resulting from the deposition of β-amyloid protein in patients with Alzheimer’s disease (AD) and in the CNS of mutant mouse models of AD has been proposed to produce S-nitrosylation of dynamin-related protein, mitochondrial fission, loss of dendritic spines and synaptic loss [94,95].
Genetic alterations in the fusion inducing protein mitofusin-2 (MFN2) of cultured DRG produced defects in axonal transport of mitochondria and their accumulation in abnormal clusters of small fragmented mitochondria in both neuronal cell bodies and proximal axons [85,96]. Proteotoxic stress in mitochondria resulting in the sumoylation of Drp1 also activates mitochondrial fission and induction of autophagy to eliminate oxidized proteins or protein quality control, preserving organelles . In this setting, either inhibiting fission or promoting fusion prevents increase in ROS . The impact of high glucose concentration on mitochondrial trafficking and associated functions remains to be fully elucidated. Studies of cultured rat liver cells or rat myoblasts demonstrated that mitochondrial fragmentation is necessary for high glucose-induced increase in respiration with concomitant overproduction of ROS . Treatment of pancreatic beta cells, neurons and endothelial cells with high [glucose] results in fragmentation/fission of mitochondria [100–103]. In several settings mitochondrial fragmentation is eventually followed by cell death and/or apoptosis , an outcome which can be prevented or modulated by shifting the dynamic balance to favor mitochondrial fusion . In cultured DRG initial mitochondrial fission has been proposed to result from up-regulation of the mitochondrial fission protein Drp1 resulting in protective or metabolic fission followed later by activation of Bim and Bax resulting in apoptosis, the latter phase resulting from regulation of Bim/Bax activation . Although initial fragmentation and clumping of mitochondria has been demonstrated in vivo in diabetic animals, apoptosis of DRG neurons in diabetic animals remains a contentious arena. In cultured endothelial cells the balance between the metabolic stimulation of mitochondria by elevated glucose levels and free radical production has been proposed to represent the turning point of either adaptation for cell survival or initiation of fatal pathways resulting in cell death in diabetes .
Mitochondrial dysfunction occurs in a range of diabetic complications and given its central role in controlling the bioenergetic status of the cell must be considered a prime trigger of degeneration. It remains unclear how such impaired mitochondrial function triggers cell damage in the nervous system and enhanced ROS production derived from aberrant mitochondrial physiology remains an unproven causal factor. Failure to synthesize adequate ATP for high energy requiring axonal functions such as excitation, ion flux, axonal transport and growth cone motility would seem more attractive options, particularly in neurons where impaired bioenergetic status may enhance oxidative stress induced by high [glucose] or diminish ability to scavenge ROS. These issues are briefly outlined and discussed in the model displayed in Figure 2. In diabetic neuropathy studies have begun to outline the array of impairments in mitochondrial physiology, although studies still lag behind those in nephropathy and cardiomyopathy. Diabetes-induced changes in mitochondrial phenotype would seem key factors leading to altered activity of the electron transport chain and Krebs cycle components.
Aberrant growth factor-dependent signaling, especially through the insulin pathway could be involved, however, hyperglycemia and associated raising of intracellular glucose concentration maybe a central trigger of altered mitochondrial proteome expression. In many cell types high intracellular glucose causes inhibition of oxidative phosphorylation with enhancement of anaerobic glucose metabolism through glycolysis, known as the Crabtree effect . Under such conditions the need for a fully functional mitochondrial proteome would be lessened and would explain recent findings and our preliminary data. It is critical in the next 5 years that the glucose-dependent signal transduction pathway that is proposed to become active in the diabetic state and that initiates altered cell bioenergetics in neurons is dissected and manipulated in vitro and in vivo. Inspection of the data from studies in non-neurons would suggest that the AMP-activated protein kinase (AMPK) and/or peroxisome proliferator-activated receptor-gamma coactivator 1 alpha (PGC-1a) pathways are worthy of investigation. There is a plausible role for the polyol pathway in such a paradigm, however, it must be remembered that neurons do not express aldose reductase and so the activity of such a pathway in neurons in diabetes remains unclear . It should be noted, however, that polyol levels do accumulate in DRG, sympathetic ganglia and glomerular mesangial cells under hyperglycemic conditions [107–109] and treatment with aldose reductase inhibitors can modulate sensory and autonomic neuron phenotype in lumbar DRG of STZ-diabetic rats [46,109].
Dr. Chowdhury was supported by grants to PF from CIHR (grant # MOP-84214), to PF and RES from Juvenile Diabetes Research Foundation (grant # 1-2008-193) and to RES from the NIH (DK19645). This work was also funded by the St Boniface General Hospital and Research Foundation and the Manitoba Medical Services Foundation.
Paul Fernyhough, Division of Neurodegenerative Disorders, St Boniface Hospital Research Centre, R4046 – 351 Taché Avenue, Winnipeg, MB R2H 2A6, Canada and Department of Pharmacology & Therapeutics, University of Manitoba, Winnipeg, MB, Canada, Tel: (204) 235 3692, Fax: (204) 237 4092.
Subir K. Roy Chowdhury, Division of Neurodegenerative Disorders, St Boniface Hospital Research Centre, R4046 – 351 Taché Avenue, Winnipeg, MB R2H 2A6, Canada, Tel: (204) 235 3692, Fax: (204) 237 4092.
Robert E. Schmidt, Dept of Pathology and Immunology, 3720 West Building, Washington University School of Medicine, St. Louis, MO 63110, USA, Tel: (314) 362-7429, Fax: (314) 362-4096.