Characterisation of glyoxalase I in a streptozocin-induced mouse model of diabetes with painful and insensate neuropathy
- First Online:
- Cite this article as:
- Jack, M.M., Ryals, J.M. & Wright, D.E. Diabetologia (2011) 54: 2174. doi:10.1007/s00125-011-2196-3
- 434 Views
Diabetic peripheral neuropathy (DN) is a common complication of diabetes; however, the mechanisms producing positive or negative symptoms are not well understood. The enzyme glyoxalase I (GLO1) detoxifies reactive dicarbonyls that form AGEs and may affect the way sensory neurons respond to heightened AGE levels in DN. We hypothesised that differential GLO1 levels in sensory neurons may lead to differences in AGE formation and modulate the phenotype of DN.
Inbred strains of mice were used to assess the variability of Glo1 expression by quantitative RT-PCR. Non-diabetic C57BL/6 mice were used to characterise the distribution of GLO1 in neural tissues by immunofluorescence. Behavioural assessments were conducted in diabetic A/J and C57BL/6 mice to determine mechanical sensitivity, and GLO1 abundance was determined by western blot.
GLO1immunoreactivity was found throughout the nervous system, but selectively in small, unmyelinated peptidergic dorsal root ganglia (DRG) neurons that are involved in pain transmission. GLO1 protein was present at various levels in DRG from different inbred mice strains. Diabetic A/J and C57BL/6 mice, two mouse strains with different levels of GLO1, displayed dramatically different behavioural responses to mechanical stimuli. Diabetic C57BL/6 mice also had a reduced abundance of GLO1 following diabetes induction.
These findings reveal that the abundance of GLO1 varies between different murine strains and within different sensory neuron populations. These differences could lead to different responses of sensory neurons to the toxic effects of hyperglycaemia and reactive dicarbonyls associated with diabetes.
KeywordsAdvanced glycation end-productsAllodyniaDiabetic neuropathyDorsal root gangliaGlyoxalase IMechanical sensitivityPeripheral nervous system
Diabetic peripheral neuropathy
Dorsal root ganglia
Receptor for AGEs
Single nucleotide polymorphism
Diabetic peripheral neuropathy (DN) is a common secondary complication of long-standing hyperglycaemia as a consequence of diabetes mellitus . Over 50% of diabetic patients suffer from DN with the prevalence increasing with duration of diabetes . The majority of diabetic patients present with insensate neuropathy or negative symptoms, although estimates have varied widely, suggesting that 10–50% of diabetic patients have painful neuropathy or positive symptoms [3–5]. However, a clear understanding of the underlying pathological mechanisms that determine which signs and symptoms of neuropathy, either positive or negative, arise in certain patients has yet to be determined.
One of the mechanisms linked to neuronal dysfunction in DN is the accumulation of AGEs [6, 7]. Non-enzymatic glycation modifies lipids, nucleic acids, and both intra- and extra-cellular proteins, thereby altering normal cellular functions . Signalling pathways activated through the receptor for AGEs (RAGE) also cause sensory neuron dysfunction, although uncertainty remains over the ligands that produce these effects [9, 10]. In clinical studies and rodent models of diabetes, AGEs accumulate in sites affected by diabetic complications including the kidney, retina and peripheral nerves . In diabetic patients, the accumulation of AGEs in the skin precedes clinical neuropathy , correlates with the severity of symptoms  and predicts the progression of DN .
Recently, reactive dicarbonyls or α-oxoaldehydes such as methylglyoxal, glyoxal and 3-deoxyglucose have been found to be important precursors of AGEs and to contribute in their own right to diabetic complications [14, 15]. Reactive dicarbonyls are formed as a normal by-product of many metabolic pathways, including glycolysis, protein breakdown, lipid peroxidation and degradation of glycated proteins, all of which are enhanced in diabetes mellitus [7, 16]. α-Oxoaldehydes are up to 20,000-fold more reactive than glucose, leading to increased carbonyl stress and accelerated production of AGEs . Importantly, α-oxoaldehydes are elevated in diabetic patients . Therefore, reactive dicarbonyls are thought to have an important role in the pathogenesis of DN.
The glyoxalase enzyme system, composed of glyoxalase I (GLO1) and glyoxalase II, is a key cellular pathway that detoxifies reactive dicarbonyls and limits AGE formation. GLO1 levels and activity can be altered in disease states including diabetes [19–21]. Thus, GLO1 has been investigated in diabetic retinopathy, nephropathy, cardiomyopathy and endothelial dysfunction . However, the distribution and functional role of GLO1 have not been investigated in DN. The present study was designed to investigate the distribution and abundance of GLO1 in the peripheral nervous system and to better understand its potential role in modulating the development of DN.
Male C57BL/6 (Charles River, Sulzfeld, Germany), A/J, BALB/cJ, 129P3/J, C3H/HeJ, and DBA/2J mice (Jackson, Bar Harbor, ME, USA) were purchased at 7 weeks of age, 1 week prior to beginning experiments. Mrgd (also known as Mrgprd) mice were generated by D. Anderson (Division of Biology, California Institute of Technology, Pasadena, CA, USA) and maintained as a breeding colony. Genotypes were confirmed as previously described . The animals were housed in 12/12-h light/dark cycle under pathogen free conditions. Mice had free access to standard rodent chow (8,604, 4% energy from fat; Harlan Teklad, Madison, WI, USA) and water. All animal use was in accordance with NIH guidelines and conformed to principles specified by the University of Kansas Medical Center Animal Care and Use Protocol.
Streptozocin induction of diabetes
Diabetes was induced in 8-week-old male C57BL/6 and A/J mice. C57BL/6 mice were injected with a single intraperitoneal injection of streptozocin (180 mg/kg body weight; Sigma, St Louis, MO, USA) dissolved in 10 mmol/l sodium citrate buffer, pH 4.5. A/J mice were injected with 85 mg/kg streptozotocin followed by a second injection with 65 mg/kg streptozotocin 24 h later. Controls received sodium citrate buffer alone. Animals of either strain that failed to develop hyperglycaemia by 1 week after initial injection were re-injected with 85 mg/kg streptozotocin. Food was removed from all cages for 3 h before and after streptozotocin injection. Animals did not receive insulin at any time.
Animal weight and blood glucose levels (glucose diagnostic reagents; Sigma) were measured 1 week after streptozotocin injection and every week thereafter. Mice were considered to have diabetes if their non-fasting blood glucose level, measured from tail vein sampling for intermediate measures and from decapitation pool for the terminal measure, was >12.0 mmol/l at every measurement.
Mechanical behavioural responses to a Semmes Weinstein–von Frey monofilament (1.0 or 1.4 g) (Stoelting, Wood Dale, IL, USA) were assessed at 4, 5 and 6 weeks after streptozotocin injection. Mice underwent a training session on the day prior to the first day of testing. Mice were placed in individual clear plastic cages (11 × 5 × 3.5 cm) on a wire mesh screen elevated 55 cm above the table and allowed to acclimatise for 30 min. The combined mean per cent withdrawal of three applications to each hindpaw was calculated for each animal for each testing session. The group means were calculated. Mice were killed following behavioural testing at 6 weeks post streptozotocin.
Unfixed lumbar dorsal root ganglia (DRG), sciatic nerve and spinal cord were dissected, frozen, sectioned in 16- to 20-μm cross-sectional serial sections, mounted on slides (Superfrost Plus; Fisher Scientific, Chicago, IL, USA) and stored at −20°C. After thawing for 5 min at room temperature, slide-mounted tissue was circled with a Pap Pen (Research Products International, Mt Prospect, IL, USA) to create a hydrophobic ring. Tissue sections were then covered with a blocking solution (0.5% wt/vol. porcine gelatine, 1.5% vol./vol. normal donkey serum, 0.5% vol./vol. Triton-X, Superblock buffer; Pierce, Rockford, IL, USA) for 1 h at room temperature. Primary antibodies were incubated overnight at 4°C. For GLO1 immunohistochemistry, a mouse anti-GLO1 primary antibody (1:100; Abcam, Cambridge, MA, USA) and a donkey anti-mouse secondary antibody (Alexa 488 or 555, 1:2,000; Molecular Probes, Eugene, OR, USA) were used to label positive cells in the L4/5/6 DRG, sciatic nerve and spinal cord. Primary rabbit anti-neurofilament H (1:2,500; Chemicon, Billerica, MA, USA) or rabbit anti-peripherin (1:1,000; Chemicon) with donkey anti-rabbit secondary antibody (Alexa 555, 1:2,000; Molecular Probes) were used to label and counterstain cell populations in lumbar DRG. Sections were washed twice for 5 min with PBS with 0.01% vol./vol. Triton-X followed by incubation for 1 h with fluorochrome-conjugated secondary antibody diluted in PBS with 0.01% vol./vol. Triton-X and Superblock (Pierce). Following three washes with PBS, slides were coverslipped and stored at 4°C until viewing. Staining was blocked by preadsorption of the mouse anti-GLO1 antibody with human GLO1 (R&D Systems, Minneapolis, MN, USA).
Quantitative real-time PCR
Total RNA was isolated from sciatic nerve, DRG and brain using Trizol reagent (Ambion, Austin, TX, USA) and a kit (RNeasy Mini; Qiagen, Valencia, CA, USA). The concentration and purity were determined using an analyser (2100; Agilent Technologies, Santa Clara, CA, USA). Total RNA (0.63 μg) was synthesised directly into cDNA using a kit (iScript cDNA Synthesis Kit; Bio-Rad, Hercules, CA, USA). Quantitative RT-PCR was performed using a kit (iScript One-Step RT-PCR kit with SYBR green; Bio-Rad). The primers are shown in the Electronic supplementary material (ESM) Table 1.
All reactions were performed in triplicate. ΔCt values and fold changes were calculated using the Pfaffl analysis method . The mRNA levels for Glo1 were normalised to Gapdh.
Brain, lumbar spinal cord, sciatic nerve and DRG were rapidly isolated, frozen in liquid nitrogen and stored at −80°C. Tissue was homogenised for 2 min in 50 μl cell extraction buffer (Invitrogen, Carlsbad, CA, USA) with protease inhibitor cocktail (Sigma), 200 mmol/l NaF and 200 mmol/l Na3VO4. The homogenates were incubated on ice for 30 min before centrifugation for 10 min at 4,500×g and 4°C. Protein concentrations were determined using a protein assay based on the Bradford assay (Bio-Rad, Sydney, NSW, Australia).
Samples containing 100 μg of protein were separated by electrophoresis through 4% to 20% SDS-PAGE gels (125 V, 1.5 h, 4°C) and transferred on to nitrocellulose paper (35 mA, overnight, 4°C). Nitrocellulose membranes were blocked with blocking buffer (3% non-fat milk and 0.05% vol./vol. Tween-20 in phosphate buffered saline) for 1 h at room temperature to block non-specific binding sites, followed by overnight incubation with a goat anti-GLO1 primary antibody (R&D Systems) diluted 1:5,000 in blocking buffer at 4°C. The donkey anti-goat IgG-horseradish peroxidase secondary antibody (Santa Cruz, Santa Cruz, CA, USA) was used diluted 1:2,500 in blocking buffer at room temperature for 1 h.
Nitrocellulose membranes were stripped using western blot stripping buffer (Restore Plus; Pierce), followed by 1 h incubation with an actin primary antibody (Millipore, Billerica, MA, USA) diluted 1:100,000 in blocking buffer at room temperature. The donkey anti-mouse IgG-horseradish peroxidase secondary antibody (Santa Cruz) was used diluted 1:2,500 in blocking buffer at room temperature for 1 h. The chemiluminescent signal was acquired using Supersignal West Femto Maximum Sensitivity Substrate (Pierce) and a CCD camera (BioSpectrum Imaging System; UVP, Upland, CA, USA). NIH ImageJ software (http://rsbweb.nih.gov/ij/index.html, Nov. 2010) was used to measure and quantify densitometry readings.
Data are expressed as mean ± SEM. Number of animals per given experimental setting is given as n value, unless otherwise indicated. Differences were analysed by two-tailed Student’s t test for unpaired data and repeated measures ANOVA followed by Bonferroni’s test, as appropriate. Significance was defined as p ≤ 0.05.
Genetic differences in Glo1 mRNA expression
Characterisation of GLO1 abundance in the nervous system
Behavioural sensitivity to mechanical stimuli
Mechanical sensitivity of the hindpaw was tested at 4, 5 and 6 weeks after streptozotocin injection. Consistent with this strain and model, diabetic C57BL/6 mice showed reduced responses to a 1.4 g von Frey monofilament (Fig. 6c) [23, 26]. After 4 weeks of diabetes, diabetic C57BL/6 mice displayed a withdrawal response to mechanical stimulation of 41.2 ± 5.32% (mean ± SEM), while non-diabetic mice responded 64.4 ± 11.7% of the time. This mechanical insensitivity persisted and worsened with progression of the disease, with diabetic mice withdrawing 23.2 ± 2.97% at 6 weeks post streptozotocin (Fig. 6c).
Conversely, as early as 1 week following streptozotocin injection, diabetic A/J mice displayed increased responses to a 1 g von Frey monofilament (data not shown). Mechanical allodynia was present at 4, 5 and 6 weeks following streptozotocin injection, with diabetic A/J mice responding nearly 75% of the time (Fig. 6b). Thus, diabetic C57BL/6 mice developed increased mechanical thresholds and severe mechanical insensitivity, whereas diabetic A/J mice developed reduced mechanical thresholds and mechanical allodynia after 4 weeks of diabetes.
GLO1 distribution and abundance in diabetes
Chronic hyperglycaemia is the main driving force of the development and progression of DN. Elevated intra- and extracellular glucose concentrations lead to accelerated production of AGEs. AGEs alter the function of many proteins, including tubulin, mitochondrial electron transport chain proteins, insulin and neuronal extracellular matrix proteins [8, 27–31]. AGEs are elevated in peripheral nerves and skin of diabetic patients and rodents [11, 12, 32, 33]; indeed, elevated levels have also been measured in plasma of type 1 diabetic patients without complications, suggesting AGEs may have a role in the development and/or progression of diabetic neuropathy . GLO1, a key enzyme in anti-glycation defence, is critical to preventing this accumulation of AGEs and limiting their toxic effects by breaking down the highly reactive precursors, α-oxoaldehydes.
Here, Glo1 mRNA expression in the DRG varied significantly in multiple inbred strains of mice. A previous study using Affymetrix exon arrays also reported high genetic variability of Glo1 in the amygdala of a large panel of inbred strains of mice . Williams et al. discovered that the genomic region encoding Glo1 exists as copy number variant in mice, with certain strains undergoing duplications and others having lost the duplicated region, which could explain expression differences in the DRG . Interestingly, copy number variants have recently been characterised in humans as a substantial source of genetic diversity and disease [36, 37]. While amplification of Glo1 has been described in human cancers, natural variation of a copy number variant encompassing Glo1 has yet to be described in humans . However, many intronic and exonic single nucleotide polymorphisms (SNPs) and SNPs located in regulatory 3′ and 5′ untranslated regions have been described . This suggests that underlying genetic differences, alterations in regulation or modifications in disease may influence the expression and activity of GLO1 [8, 40]. To date, little has been done to define these in humans.
The present study is the first to characterise expression of Glo1 in the peripheral nervous system. Our results reveal that expression of Glo1 is restricted to small, unmyelinated peptidergic neurons in the DRG, which are primarily responsible for transmission of noxious or painful sensory information. This was particularly interesting given the critical function, highly conserved nature and previous reports on the ubiquitous presence of GLO1 . Generally, this suggests that populations of neurons may be vulnerable to different types of damage in diabetes. Specifically, non-peptidergic C-fibres and myelinated neurons may experience increased carbonyl stress and accumulation of AGEs, leading to altered neuronal function. Alternatively, peptidergic neurons may be more protected from the toxic effects of elevated glucose and reactive dicarbonyls due to the abundance of GLO1. This may have significant consequences on the phenotype of altered sensation in DN.
In this study, responses to mechanical stimuli in diabetic mice with different levels of GLO1 varied. Previous studies have also suggested a link between mechanical and thermal sensation, GLO1 and AGEs. Toth et al. and Bierhaus et al. showed that RAGE has a role in sensory neuron dysfunction in diabetes [9, 42]. Diabetic Rage (also known as Ager)−/− mice were protected from pathological, physiological and behavioural signs of DN [6, 9, 42]. Mechanical sensitivity is mediated by unmyelinated and myelinated sensory neurons, which respond to a range of sensory stimulus modalities. Indeed, most C-fibres also respond to mechanical sensitivity . Thus, afferent information encoding mechanical information probably converges in the spinal cord to convey complex mechano-sensory information. It is therefore plausible to suggest that a balance between fibre types is required to convey appropriate sensory information from the periphery. The ability of specific neuronal populations to be more susceptible to and/or protect against the hyperglycaemia-induced damage that occurs in diabetes may alter this balance, leading to abnormal peripheral nociceptive input and producing different signs and symptoms of neuropathy.
Although the present study does not directly address the mechanisms causing positive or negative mechanical sensitivity associated with DN, this issue is clearly important and relevant to humans with DN. Sensory loss, including chronic numbness and insensitivity to pain or touch, develops in the majority of affected human patients. Painful symptoms, including tactile allodynia, are reported in approximately 30% of patients with DN and are often present early in disease progression, the prevalence being slightly higher in type 2 diabetes [44, 45]. This dichotomy in positive vs negative symptoms is reflected in animal models of DN, with some resembling the painful and others the painless aspects of DN [5, 26, 46]. Our current findings in C57BL/6 and A/J mice are consistent with this feature and may point to ways of addressing how genetic differences, including altered GLO1 expression, help shape the phenotype of DN in diabetic patients.
Based on our findings, we hypothesise that an imbalance of fibre type input may underlie the development of positive or negative mechanical sensitivity. In the setting of high GLO1 abundance, as in A/J mice, peptidergic neurons may be protected from extensive damage by reactive dicarbonyls and retain the ability to transmit noxious sensory information. However, A-fibre and non-peptidergic sensory neurons have no detectable production of GLO1 and may be more vulnerable to direct damage by reactive dicarbonyl, which leads to neuronal dysfunction in these fibres. This is supported by studies suggesting that a range of sensory afferents are likely to contribute to overall pain sensation, including A- and C-fibres . Therefore, despite protection from GLO1, it is possible that peptidergic C-fibre function may be affected by damage to other fibres. Conversely, in the setting of low GLO1 abundance, as in C57BL/6 mice, peptidergic neurons may accumulate reactive dicarbonyl-derived AGEs and experience physiological dysfunction similar to that in other fibres. Thus, peptidergic C-fibre mechano-thresholds may be increased due to intrinsic dysfunction stemming from reduced GLO1 levels and accumulation of reactive dicarbonyls.
Finally, our results revealed that GLO1 levels are reduced by diabetes in C57BL/6 mice, which could also play a role in determining the phenotype of DN. Reductions in GLO1 in diabetic C57BL/6 mice could occur through increased proteosomal degradation, particularly if GLO1 itself is modified by methylglyoxal, or also by RAGE-mediated reduction of GLO1 . Both in vivo and in vitro overproduction of GLO1 has been shown to dramatically reduce levels of reactive dicarbonyls and AGEs [48–50]. However, it remains to be seen whether physiological overproduction is robust enough to produce similar reductions. Given the restricted distribution of GLO1 in the DRG and behavioural differences following induction of diabetes, our findings suggest that genetic variation in Glo1 could modify the spectrum of DN symptoms.
In conclusion, we have shown that GLO1 is predominantly produced in peptidergic sensory neurons in the DRG and that its abundance varies in two inbred strains of mice that display different mechanical sensitivity, suggesting each develops a different phenotype of neuropathy. These findings warrant further investigation into the role of GLO1 in DN.
This work was supported by the Juvenile Diabetes Research Foundation and NIH RO1NS43314 to D.E. Wright, and by the Kansas IDDRC, P30 NICHD HD 002528.
M.M.J., J.M.R. and D.E.W. all contributed to the conception and design, analysis and interpretation, writing, and final approval of the published version.
Duality of interest
The authors declare that there is no duality of interested associated with this manuscript.