Early Peripheral Nerve Responses to Elevated Blood Glucose in the Diabetic Mouse
Summer Scholarship 2003/2004
Peripheral neuropathy is a major consequence of long-term diabetes. Previous studies have focused on primary damage to the neurons as the cause of this condition, but we propose an alternate hypothesis, namely that Schwann cells (glial support cells) are the primary targets of hyperglycemic damage and that neuronal loss is a secondary consequence. In a diabetic mouse model, we found evidence of physiological stress (activation of cell death pathways) amongst Schwann cells before we were able to detect signs of neuropathological change. This supports our view that Schwann cells are the primary targets of damage in diabetic peripheral neuropathy, and motivates further study.
Diabetes is an ever-increasing problem in westernised countries, including New Zealand. The disease is fast approaching epidemic proportions, placing strain on the health system. A common result of long-term diabetes is a loss of sensation in the extremities, with pain and even amputation. This condition is called diabetic polyneuropathy, and is due to the damage and death of peripheral motor and sensory nerves. After 25 years of the disease, up to 50% of patients report clinical symptoms of neuropathy, as a consequence of hyperglycemia (Pirart, 1978). There are several theories for the cause of diabetic peripheral neuropathy, all focusing on primary insult to the neuron itself. These include: hypoxia, where vascular changes as a result of hyperglycemia leave the nerve hypoxic; oxidative damage, where elevated blood glucose leads to increased reactive-oxygen-species (ROS) production by mitochondria, which leads to DNA damage. Also metabolic insult has been postulated, with hyperglycemia leading to changes in metabolites, growth factors, receptors and other cellular components which may result in apoptosis; osmotic insult may be another possibility (Feldman, 2003). These hypotheses in combination and in isolation seem unable to account for all experimental findings, so other mechanism(s) seem likely to be involved.
Clinically, the first sign of diabetic neuropathy is slowing conduction velocity of the peripheral nerves, without loss of sensation. This would imply that at early stages there is no loss of axons, but action potentials traversing them are slower. How is this possible? Myelin is needed for fast action potential conduction velocity, so we would assume myelin is being lost from the peripheral axons. Schwann cells myelinate peripheral axons. Schwann cells have been shown to be affected in diabetic neuropathy (Eckersley, 2002), possibly by the same mechanisms that damage neurons. We also know, from recent research, that Schwann cells are essential for neuron survival via the neuregulin-erbB pathway (Reithmacher et al, 1997). ErbB3 is a receptor tyrosine kinase on the Schwann cell, which binds neuregulins. Neuregulins are factors produced by developing and mature neurons, and are suggested as being survival factors (Bao et al, 2003). In ErbB3 knockout mice, Schwann cells or their precursors are almost totally absent, and there is around an 80% loss of sensory and motor neurons.
So, bringing this information together: diabetes causes Schwann cells to die, meaning the ErbB receptor they produce is not present. Neuregulins produced by neurons cannot bind, and there is no reciprocal survival signalling pathway back to the neuron, so the neuron dies. This then leads to a plausible, alternative hypothesis: the cause of diabetic peripheral neuropathy is a primary insult to the Schwann cell, and once these cells have de-differentiated or died, the axons are no longer supported via neuregulin-erbB, and die as a consequence of the Schwann cell loss.
If this was the case, we would expect to see pathological changes in the Schwann cells before neurons show signs of damage, and before measurable neuropathic symptoms, such as slowing conduction velocity, reduced force generation and longer duration of action potential. Previous studies report severe presentation of peripheral neuropathy within 8 weeks after diabetes induction in the mouse, with significant conduction velocity reduction and self-mutilation (Fahim et al, 1998). Here, we test our hypothesis by inducing diabetes in mice, then by testing peripheral nerve function at 1, 2, 4 and 8 weeks post induction. In addition, we look for an upregulation of markers of activation of cell death and de-differentiation amongst the myelinating Schwann cells of peripheral nerves.
In this study, 22 mice were used, with four diabetic mice and one or two control mice in each time point: 1, 2, 4 and 8 weeks post injection. Diabetic animals were injected i.p.with 200 mg/kg streptozotocin in 0.35 mL citrate buffer pH 4.2, whereas controls were injected with 0.35 mL vehicle (citrate buffer). Streptozotocin induces an inflammatory response in the β cells of the pancreas. β-cells secrete insulin, so death of these cells renders the animal diabetic. Previous evidence shows diabetic symptoms include a high water consumption and loss or lack of weight gain. All animals were monitored by measurement of daily water consumption and bi-weekly measurement of body weight.
Surgery and Data Gathering:
This protocol was approved by the University of Otago Animal Ethics Committee, No. 80/03. Mice were anaesthetised with 0.2 mL intra-muscular Saffan, and monitored by frequent toe-pinch and corneal reflex, with 0.1 mL anaesthetic "top-ups" every 30 minutes. The left hind leg was dissected to expose the sciatic nerve and the gastrocnemius muscle. A Harvard force transducer (ADI Instruments) was connected via a fine straight tungsten wire to the achilles tendon, whilst stimulating electrodes were placed in contact with the sciatic nerve in the thigh. Using "MacLab" and an electrical stimulator (Devices Instruments) muscle twitches were initiated using a 0.1 ms duration stimulation at 0.75-1.5 V, or until a maximal response could be obtained, with the force recorded in mN on the program "Scope". Sustained contractions or tetani were initiated using a 500 ms train of stimuli at 90 Hz with voltage between 0.75 and 1.5 V, or until maximal force could be seen. Sensory nerve data was collected using a fine tungsten pin pushed through the skin of the upper foot, passing close to sensory axons in that area. The recording electrode in the foot was coupled with a reference electrode under the skin of the belly, and an earth electrode on the tail. Nerve stimulation was as for muscle twitches. For each condition, ten samples were taken, and comparisons analysed using students t-test. After data gathering had been completed, measurements of distance from stimulation point to muscle mid-belly and to the sensory electrode in the foot were taken for calculation of conduction velocity. Sciatic nerve sections of around 1 cm were taken for immunohistochemical analysis, as well as the gastrocnemius muscle from both legs of the animal. After sacrifice, blood glucose was measured, and only streptozotocin-injected animals with blood glucose levels of above 16 mmol/L were considered diabetic.
Nerve tissue from each animal was sectioned longitudinally (10 ~m section, -20°C, Leica cryostat) and placed on coated slides. These were then lightly fixed in 2% paraformaldehyde for 10 minutes followed by washes of PBS, and 15 minutes in 0.1 M glycine in 0.01 M PBS. Tissue was stained using primary antibodies for S100 (Dako, rabbit anti-cow, 1:500), a widely used Schwann cell marker. Caspase-3 (goat anti-human, an enzyme associated with the apoptosis cascade), and p75 (indicates a de-differentiation response by the cells).
Antibodies were diluted in 1 % BSA in PBS, and 1 00 ~L of the primary antibody solution was pipetted over the slides. These were left in a dark, humidified chamber overnight at 4°C. Negative control slides were left with only 1 % BSA in PBS (no primary antibody), otherwise processed in the same way as stained slides. Slides were washed twice in 1xPBS, then twice in 2xPBS, followed by a final double wash in 1xPBS. Each wash lasted 5 minutes. Secondary antibodies (green: Alexa 488, red Alexa 594) were left in a dark, humidified chamber for 3 hours at room temperature. Again, serial 5 minute PBS washes were performed: twice in 1xPBS, twice in 2xPBS, then twice again in 1xPBS. The slides were coverslipped using 90% glycerol with 10% PBS. Slides were viewed using an Olympus fluorescence microscope and digital camera (Diagnostic Instruments, SPOT-RT Slider), and digital images were processed using Adobe Photoshop.
Measures of nerve and muscle health were taken from Scope recordings. For motor nerve and muscle: the latency from stimulation to first sign of contractile response was used to calculate motor nerve conduction velocity; the time from first sign of contractile response to the peak of contraction (Time-To-Peak); and contraction amplitudes of twitch and tetanus, which were normalized by muscle weight. For sensory neurons: latency from end of stimulation to first change in action potential was used to calculate action potential conduction velocity; amplitude of action potential, and duration of action potential., Sample traces from "Scope" are shown in Figure 1. We would expect diabetic neuropathy to alter these recorded parameters, with slowing conduction velocities, longer Times- To-Peak, smaller amplitudes of contraction and action potential, and increases in action potential duration. Diabetic animal data was compared with their age-matched controls using student t-tests.
Figure 1: Sample traces from Scope recordings. (A) gastrocnemius muscle twitch, (B) tetanic contraction, (C) sensory nerve action potential.
Confirmation of Diabetes
Body weight was monitored as an indication of diabetes. Figure 2 shows controls in graph A, and diabetics in graph B, followed from time of injection to protocol end point: sacrifice. While the weight changes are not significantly different (P>0.05), there is, as expected, a trend towards increasing body weight for controls, and decreasing body weight for diabetics.
Figure 2: Control and Diabetic body weight changes from beginning to end of protocol.
What was significant (P≤0.05), was water consumption between controls and diabetics:
Controls: 4.45 (±0.979) mL/day
Diabetic: 9.82 ( ±4.15) mL/day.
Finally, to confirm diabetes, blood glucose was significantly different between controls and diabetics (P≤0.001):
Controls: 9.17 (±1.18) mmol/L
Diabetics: 24.69 (±6.38) mmol/L
Data on muscle, motor and sensory nerves were taken and graphed to follow the onset of diabetic neuropathy. No clear change from normal as a consequence of onset of diabetes was observed for any of the neurophysiological parameters examined (Figures 3 to 9). This was somewhat unexpected as published work indicated "severe neuropathy" 8 weeks after diabetic induction.
Figure 3: Motor nerve conduction velocity
Figure 4: Twitch amplitudes
Figure 5: Twitch Time-To-Peak Contraction
Figure 6: Tetanus Amplitudes
Figure 7: Tetanus Time-To-Peak
Figure 8: Sensory action potential conduction velocity
Figure 8: Sensory action potential amplitude
Figure 9: Sensory AP duration
Longitudinal sections were stained for proteins that would indicate dying (caspase-3) or de-differentiating (P75) cells, and for a Schwann cell marker (S100). At the 1, 2 and 4 week post diabetic induction time points there were no obvious differences in expression from their age matched controls (not shown). In 8-week diabetic animals, there was caspase-3 activation (arrows, Figure 10B) beyond the level seen in age-matched controls (Figure 10A). The sections were double labeled with S100 antibody (Figure 10C, D). The labeled cells indicated by arrows in Fig 10B and 10D are the same cells, indicating that caspase-3 activation is co-localised to S100 positive cells: Schwann cells. In contrast, for the control animal, strongly labeled S100 positive Schwann cells in 10C have no caspase3 counterpart in 10A.
Figure 10: Immunoflourescence in 8-week post injection time points: control (a, c) and diabetics (b, d). A and B are caspase-3 stained sections, arrows in B indication positive-staining cells. C and D are S100 stained sections of the same view as A and B. Arrows indicate well-stained Schwann cells. Scale = 100μm
Co-localisation of caspase-3 and S100 in diabetics is further supported by a cross-sectional view (Figure IIA). This image shows axons wrapped in Schwann cells, with a Schwann cell cytoplasm clearly caspase-3 positive. No p75 activation was seen in any animal, diabetic or control, at any time point (not shown). A negative control is included (Figure 11B), to show without any primary antibody, secondary antibody does not bind tissue.
Figure 11: A: cross-sectional view of an 8-week post induction diabetic animal showing axons enveloped in S100 stained Schwann cells, with a caspase-3 positive cytoplasm, scale = 10Ám. B: negative control, showing no immunoreactivity without primary antibody, scale = 100Ám.
Our hypothesis for these experiments was that primary damage in diabetic neuropathy was occurring not to neurons, but to Schwann cells, with neurons dying as a consequence of Schwann cell loss. The hypothesis predicted we would see signs of Schwann cell death or de-differentiation before measurable neuropathy or neuron death.
Our experiments show that while there may be significant differences between individual sets of diabetic animals in the different electrophysiological conditions, there are no obvious general trends that would indicate diabetic neuropathy was occurring in motor or sensory nerves within 8 weeks post diabetic induction in these mice. We had expected the 1, 2 and 4-week animals to be similar to age matched controls. In most cases they are not. Indeed, the controls are not similar to each other in each case. This would indicate a greater variation in normal muscle; motor and sensory nerve properties than was expected. In future experiments, using multiple controls could possibly reduce this variation. Without confirming diabetic neuropathy, we cannot confirm onset of damage within the nerve. However, we can postulate why we did not see neuropathy. The study we based our protocol on (Fahim et al, 1998), which reported presentation of slowing conduction velocity and other signs of diabetic neuropathy within 8 weeks, used male TO strain mice. Our study used female, C-I29 mice, suggesting genetic and/or gender variations may account for a difference between our work and that previously published.
Though we did not see the expected signs of neuropathy electrophysiologically, we did see some signs of nerve damage. Our immunofluorescence data showed no apparent differences in staining pattern of caspase-3, S100 or p75 between diabetics and agematched controls in the 1,2 and 4 week time points. However, in 8-week animals we did see signs of increased caspase-3 activation in diabetic animals compared to 8-week control animals. As shown in Figures 10 and II A, caspase-3 activation was co-localised to S I 00 positive Schwann cells, and that this pattern and intensity of expression in diabetics was different to that in their age-matched controls. Therefore, even though we did not see measurable neuropathy, we propose that the caspase-3 activation seen in 8week diabetic is the beginning of nerve damage. Of course, without confirming neuropathic symptoms we cannot be sure, but we would expect from this result, that neuropathy would be seen soon after the 8-week time point. Assuming this is correct, caspase-3 is active in Schwann cells before signs of nerve damage, which supports our hypothesis of primary damage to Schwann cells. In other results, no p75 activation was seen at any time point. This would further indicate, taken together with the caspase-3 results, that hyperglycemia induces a death response in Schwann cells, rather than a dedifferentiation response.
To further reveal effects of diabetic neuropathy on the Schwann cell, more work needs to be done, primarily to confirm our results by continuation of the protocol until definite measurable signs of diabetic neuropathy are seen in our mice. It is the ultimate aim of this work to discover how exactly the Schwann cells are being damaged in diabetes, and lead on to the development of new strategies and therapies to preserve nerve function in diabetics and others with peripheral nerve damage.
This research has been generously funded by the Physiology Department, and the Otago Diabetes Research Trust in conjunction with the Otago Medical Research Foundation. Thanks must also go to Dr Phil Sheard and Dr Marilyn Duxon for their help and guidance.
- Bao, J., Wolpowitz, D., Role, L. W. and Talmage, D. A. Back signaling by the Nrg-l intracelhilar domain. Journal o/Cellular Biology, 161: 1133-41, 2003.
- Eckersley L. Role otthe Schwann cell in diabetic neuropathy. International Review o/Neurobioiogy. 50:293-321, 2002
- Fahim, M. A., EI-Sabban, F. and Davidson, N. Muscle contractility decrement and correlated morphology during the pathogenesis of streptozotocin-diabetic mice. The Anatomical Record, 251, 240-244, 1998
- Feldman EL. Oxidative stress and diabetic neuropathy: a new understanding of an old problem. Journal o/Clinical Investigation. 111(4): 431-3,2003 Feb
- Pirart, J. Diabetes mellitus and its degenerative complications: a prospective study of 4,400 patients observed between 1947 and 1973. Diabetes Care, 1: 168-252, 1978
- Riethmacher, D., Sonnenberg-Riethmachet, E., Brinkmann, V., Yamaai, T., Lewin, G. R., and Birchmeier, c. Severe neuropathies in mice with targeted mutations in the ErbB3 receptor. Nature, 389, 725-30, 1997