Bioenergetics in diabetic neuropathy: what we need to know


Eva L. Feldman, MD, PhD, Department of Neurology, University of Michigan, AATBSRB, 109 Zina Pitcher Place, Ann Arbor, MI 48109-2200, USA. Tel: +1 734-763-7274; Fax: +1 734-763-7275; E-mail:


Progress in developing treatments for diabetic neuropathy is slowed by our limited understanding of how disturbances in metabolic substrates – glucose and fatty acids – produce nerve injury. In this review, we present the current oxidative stress hypothesis and experimental data that support it. We identify weaknesses in our understanding of diabetes-disordered metabolism in the neurovascular unit, that is, in critical cell types of the microvascular endothelium, peripheral sensory neurons, and supporting Schwann cells. Greater understanding of peripheral nervous system bioenergetics may provide insight into new drug therapies or improvements in dietary interventions in diabetes or even pre-diabetes.

Diabetic neuropathy (DN) is a frequent and severe complication of diabetes. It is more common and persistent in patients with type 2 than type 1 diabetes (Young et al., 1993) and, over time, DN affects approximately 60% of all patients with the disease (Vincent and Feldman, 2004). DN is characterized by progressive, length-dependent loss of peripheral nerve axons in a stocking and glove (distal to proximal) pattern (Said, 2007), resulting in pain, decreased sensation, and eventually complete loss of sensation. In the United States, DN is the leading cause of diabetes-related hospital admissions and non-traumatic amputations (Feldman et al., 2003; Boulton et al., 2005; Edwards et al., 2008; Feldman, 2008). DN predominantly occurs in a distal symmetrical pattern, with progressive skin denervation (reduced intraepidermal nerve fiber density) (Said, 2007; Sullivan et al., 2007) over the duration of diabetes (Shun et al., 2004), and a proximal-distal graded loss of myelinated fiber density observed in the peripheral nerves (Johnson et al., 1986).

The condition of diabetes mellitus is characterized by hyperglycemia. For decades, the persistent or recurrent periods of hyperglycemia have been thought to produce the microvascular complications of diabetes including neuropathy, determined through epidemiological studies (Chisholm, 1993; Epidemiology of Diabetes Interventions and Complications, 1999). More recent studies also suggest a link between dyslipidemia, a common risk factor for macro- and micro-vascular disease, and DN (Vincent et al., 2009). Hyperglycemia and dyslipidemia are predicted to produce gross aberrations in global energy balance and metabolism that could lead to cellular injury and neuropathy. Energy deficits are also predicted to underlie the pathophysiological pattern of distal to proximal injury through impaired ability to produce energy at the extremities of long axons far from the cell body. Despite the facts that energy metabolism is generally well understood and that there has been substantial research into bioenergetic failures that produce DN, there is little experimental data to support the hypothesis for the fundamental cellular mechanisms behind nerve injury. This lack of a clear mechanism or mechanisms has prevented the development of effective targeted treatments to prevent or reverse DN. In this article, we consider the oxidative stress theory of diabetes complications and available bioenergetics data obtained in the peripheral nervous system (PNS) to identify areas that may be fruitful for developing greater understanding of DN.

An association between hyperglycemia-derived production of reactive oxygen species (ROS) and the development of DN is well-accepted (Brownlee, 2001; Tomlinson and Gardiner, 2008), with a primary mechanism contributing to the onset and progression of DN suggested to be related to hyperglycemia-induced overproduction of superoxide (Oinline image by the mitochondrial electron transport chain (Nishikawa et al., 2000; Brownlee, 2001) (Fig. 1). In systemic hyperglycemia, elevated glucose content in sciatic nerve (SCN) (Thurston et al., 1995; Kishi et al., 1999; Obrosova et al., 2005), and increased oxidative stress in plasma, SCN and dorsal root ganglia (DRG) (Vincent et al., 2004b), as well as in vitro hyperglycemia-induced DRG neuron mitochondrial injury (Vincent et al., 2004a; 2005) all support the hypothesis that glycolytic flux is increased in the diabetic PNS (Brownlee, 2001; Tomlinson and Gardiner, 2008), contributing to mitochondrial ROS generation. Furthermore, enhanced fatty acid oxidation and lipotoxicity are implicated in the pathophysiology of type 2 diabetic cardiomyopathy (van de Weijer et al., 2011) and nephropathy (Murea et al., 2010). However, experimental evidence for increased metabolic flux through glycolysis or beta-oxidation in diabetic PNS is lacking. Brain glucose utilization is more widely studied. Data suggest that neurons continue to metabolize glucose during hyperglycemia via non-glycolytic mechanisms, and that there is a complex interplay between astroglia and neurons to utilize metabolic substrates that is not well understood (Izawa et al., 2009). Further understanding of the utilization of energy substrates in specific nerve cell types is critical.

Figure 1.

Prevailing view that excessive glycolytic metabolism is responsible for the generation of harmful reactive oxygen species (ROS), mitochondrial injury, and diabetic neuropathy. White boxes, experimental evidence derived from neuronal tissue and cells; black boxes, associations based on hypotheses derived from multiple, non-neuronal tissue, and cell types.

The uptake and utilization of glucose are not insulin dependent in peripheral neurons (Greene and Winegrad, 1979). Glucose transporters (GLUT) are required for facilitated glucose uptake: peripheral neurons are dependent upon GLUT3 (Simpson et al., 2008), and Schwann cells utilize both GLUT3 and GLUT1 for facilitated glucose uptake (Magnani et al., 1996). To our knowledge, the effects of diabetes on this GLUT1 and GLUT3-mediated glucose transport have not been investigated. Glucose accumulates in SCN in alloxan-induced diabetes in rats at 26 weeks (Thurston et al., 1995), and in streptozotocin (STZ)-induced diabetic rats at 4 weeks (Kishi et al., 1999) and 6 weeks (Obrosova et al., 2005), suggesting no impairment of facilitated glucose transport into the nerve in experimental diabetes. However, Kishi et al. (1999) demonstrate that despite this elevated SCN glucose content in STZ-diabetic rats after 1 month of diabetes, both nerve and DRG glucose uptake are reduced compared with controls. Thus, changes in glucose transporter expression and activity warrant consideration and future investigation in the detailed evaluation of peripheral nerve bioenergetics in diabetes. In the face of poor glucose control, if bioenergetics proves to be a key to neuron injury, an ability to target specific GLUT may be one way to protect peripheral neurons against an energy substrate crisis.

In his unifying hypothesis, based primarily on research in endothelial cells, Brownlee (2001) proposed that ROS inhibition of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) activity (Du et al., 2000) leads to accumulation of glyceraldehyde-3-phosphate and upstream glycolytic metabolites, which are then diverted into alternative pathways including the advanced glycation end products pathway, PKC pathway, hexosamine pathway, and polyol pathway (Brownlee, 2001). Specifically, hyperglycemia is associated with overproduction of tricarboxylic acid (TCA) cycle-derived electron donors, subsequent increased mitochondrial Oinline image production (Du et al., 2001) and Oinline image-induced inhibition of GAPDH in cultured endothelial cells (Du et al., 2000). Work by Thurston and colleagues partially support this theory in peripheral nerves: glucose-6-phosphate and fructose-1,6-bisphosphate are elevated following 26 weeks of alloxan-induced diabetes (Thurston et al., 1995). However, in the same study, elevated nerve lactate and ATP lead to the conclusion that there is increased glycolysis and decreased TCA cycle activity, although no TCA cycle intermediates were measured or quantified (Thurston et al., 1995) to better support this conclusion.

Hexokinase directs glucose into the glycolytic pathway under normoglycemic conditions, however, there is a glucose concentration-dependent activation of aldose reductase in nerve (Greene and Lattimer, 1983; Greene et al., 1984; 1987). Thus, activation of the polyol pathway is a prominent metabolic feature of diabetic rat peripheral nerve (Stevens et al., 1994; Obrosova et al., 2005). Early in the course of diabetes, this polyol flux promotes nicotinamide adenine dinucleotide phosphate (NADPH) depletion, decreased ATP production and neuron injury (Stevens et al., 1994). Additionally, aldose reductase is localized to Schwann cells in the peripheral nerve (Kern and Engerman, 1982), and polyol pathway hyperactivity is associated with diminished energy flux in diabetic nerve (Greene and Lattimer, 1984; 1986). Hexokinase saturation and maximal glycolytic flux is one of the proposed mechanisms underlying the accumulation of nerve glucose and its direction into the polyol pathway in diabetes (Tomlinson and Gardiner, 2008); however, there appear to be no data in the literature on the degree of hexokinase activity in diabetic nerve. Gardiner et al., (2007) observed complex effects of STZ-induced diabetes on hexokinase I expression in rat DRG and suggested that metabolic flux through the glycolytic pathway is reduced in diabetes. Studies on excised rat retinas (Ola et al., 2006) concluded that glucose metabolism downstream of hexokinase is not elevated by hyperglycemia or diabetes, but that intermediates of alternative glucose metabolism, such as those of the polyol pathway, are increased. In clinical trials, inhibitors of aldose reductase have generally failed to produce the desired results in decreasing the progression of neuropathy, although study data continue to suggest that improvements in the inhibitors and trial design may ultimately produce a therapeutic benefit (Tsai and Burnakis, 1993).

Work by Tretter and colleagues involving the relationship between key enzymes of the TCA cycle and oxidative stress suggests that in the absence of metabolite flux from glycolysis to the TCA cycle, ROS and subsequent further ROS generation by the TCA cycle itself (Tretter and Adam-Vizi, 2005) may contribute to mitochondrial dysfunction. Aconitase, catalyzing the citrate to isocitrate reaction in the TCA cycle, is inhibited by ROS, including Oinline image(Gardner and Fridovich, 1992; Gardner et al., 1995) and H2O2(Tretter and Adam-Vizi, 2005). When aconitase is fully inhibited by H2O2 in nerve terminals, alpha-ketoglutarate dehydrogenase (α-KGDH; catalyzing the α-KG to succinyl-CoA reaction) remains functional and a segment of the TCA cycle (α-KG to oxaloacetate) is maintained by glutamate, which is converted to α-KG via transamination (see Tretter and Adam-Vizi, 2005 for comprehensive explanation). ROS-mediated inhibition of aconitase may cause this truncated TCA cycle to come into effect in diabetic tissues (Fig. 2). This truncated segment of the TCA cycle has been suggested to function in the absence of glucose (Yudkoff et al., 1994; Erecinska et al., 1996), and, it could be inferred, when glucose is metabolized by non-glycolytic pathways. Additionally, α-KGDH is itself a source of ROS production, the level of which increases when α-KG is utilized as a fuel source over glucose in isolated brain synaptosomes (Tretter and Adam-Vizi, 2004). Thus, in a state of oxidative stress and/or decreased glycolysis, aconitase is completely inhibited, and α-KGDH remains sufficiently active to deliver electron donors to the electron transport chain, but further contributes to the production of ROS (Fig. 2). We have demonstrated loss of aconitase activity in response to hyperglycemia in DRG neurons in vitro (Vincent et al., 2005) but the consequences in regards to ongoing energy metabolism in these cells remain to be determined.

Figure 2.

Proposed relationship between diabetes, ROS, and the tricarboxylic acid (TCA) cycle. In a state of oxidative stress and/or decreased glycolysis, aconitase is inhibited. α-KGDH remains sufficiently active to permit truncated TCA cycling and delivery of electron donors to the electron transport chain, but further contributes to the production of ROS. α-KGDH, alpha-ketoglutarate dehydrogenase; ROS, reactive oxygen species.

Thus, the theory that excessive glycolytic metabolism is responsible for the generation of harmful ROS, mitochondrial injury, and DN (Brownlee, 2001; Tomlinson and Gardiner, 2008) is based on studies that are disparate, incomplete, and performed across multiple tissue and cell types. In addition, to our knowledge, there are no studies that address the interplay between glucose, fatty acids and peripheral nerve energy metabolism in diabetes, nor the relationships between microvascular, Schwann cell, and neuronal metabolism. With the development of quantitative mass spectroscopy techniques, further metabolite studies demand attention. An understanding of diabetic metabolic abnormalities distinct to the PNS may be crucial to developing effective treatments for DN-related peripheral nerve disease. These data, then, can more fully support the multiple mechanisms that are proposed to be involved in cumulative neuron and nerve injury in diabetes that include oxidative stress, loss of neurotrophic support, insulin resistance, myelin dysfunction, and inflammation (Vincent et al., 2011).


The author declares no conflict of interest.