Brain, Vol. 124, No. 11, 2319-2334,
November 2001
© 2001 Oxford University Press
Does diabetes target ganglion neurones?
Progressive sensory neurone involvement in long-term experimental diabetes
1 Department of Clinical Neurosciences, University of Calgary, Alberta, 2 Department of Anatomy and Cell Biology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
Correspondence to:
Dr D. W. Zochodne, University of Calgary, Department of Clinical Neurosciences, Room 182A, 3330 Hospital Drive, NW Calgary, Alberta, Canada T2N 4N1 E-mail: dzochodn@ucalgary.ca
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Targeting of dorsal root ganglia by diabetes could account for the selective sensory abnormalities that patients with early diabetic polyneuropathy develop. In this work, we addressed survival, phenotype and gene expression in sensory neurones in lumbar dorsal root ganglia in a long-term model of experimental streptozotocin-induced diabetes in rats, designed to reflect human disease. Motor and sensory conduction slowing developed early, by the 2-month time point. At 2 months, sensory neurones had no detectable alterations in their calibre or gene expression, assessed using quantitative in situ hybridization studies for mRNA markers that included
CGRP, ßCGRP, NFM, t1-tubulin, SP, VIP, B50 (GAP43), galanin, somatostatin, PACAP, HSP27, c-jun, SNAP 25, p75, TrkA, TrkB and TrkC. By 12 months, however, diabetics had developed neurone perikaryal and distal axon atrophy, accompanied by generalized downregulation of mRNA expression, particularly of CGRP transcripts, PACAP, SP, NFM, p75, trkA and trkC. With the exception of HSP-27, no elevation in mRNAs that increase after injury, such as VIP, galanin, CCK, PACAP, B50 and t1-tubulin, was observed and constitutive levels, when detectable, trended towards lower rather than increased levels. There was relative preservation of neurone numbers at 12 months; only a non-significant trend towards fewer diabetic neurones was detected using a rigorous and systematic physical dissector counting approach through the entire L5 ganglia. There was no change in the relative populations of CGRP- and SP-immunoreactive neurones. Our findings indicate that even long-term experimental diabetes is associated with relative preservation of sensory neurone populations, but the neurones are atrophic and their gene expression is altered. This pattern of change differs from that following axotomy, implies a degenerative rather than an injury phenotype and has important implications for how such neurones might be rescued. diabetic neuropathy; dorsal root ganglion; sensory neurone; neurofilament
CCK = cholecystokinin; CGRP = calcitonin gene-related peptide; DRG = dorsal root ganglia; GAL = galanin; GAP 43 = growth-associated protein 43; HSP-27 = heat shock protein 27; NFM = neurofilament medium subunit; NGF = nerve growth factor; NPY = neuropeptide Y; OCT = optimum cutting temperature; PACAP = pituitary adenylate cyclase activating polypeptide; SP = substance P; STZ = streptozotocin; Trk = tyrosine kinase; VIP = vasoactive intestinal polypeptide
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
We have postulated that direct targeting of dorsal root ganglia (DRG) by diabetes could account for the apparent selective early sensory abnormalities that patients with early diabetic polyneuropathy develop. DRG may be particularly vulnerable in this disease for several reasons. Sensory neurones in the DRG are not protected from the blood–brain or blood–nerve barrier (Arvidson, 1979
). DRG have higher metabolic requirements than nerve trunks (Greene et al., 1979; Kadekaro et al., 1985) and as a result entrain higher levels of blood flow while demonstrating features of partial autoregulation (Zochodne and Ho, 1991). DRG also appear to have lower ambient oxygen tensions (Zochodne and Ho, 1991). All of these physiological characteristics suggest they may be particularly susceptible to microangiopathy, an important mechanism of disease in diabetes. In previous work we have suggested that diabetic DRG may exhibit selective reductions in local blood flow (Zochodne and Ho, 1994; Zochodne et al., 1994), while changes of blood flow in the mixed (containing motor, sensory and autonomic axons) nerve trunk are controversial (Zochodne, 1996a). Despite their possible involvement, DRG have undergone limited studies in experimental and human work. Careful, unbiased methods for estimating neurone numbers have not been applied to animal models, and the extent of possible sensory neurone cell loss is unknown.
Another possible mechanism of peripheral nerve dysfunction in diabetes that has been considered is defective target synthesis, uptake or retrograde transport of the trophic factors that support sensory neurones (Brewster et al., 1994, 1995; Diemel et al., 1994; Fernyhough et al., 1994, 1995b; Zochodne, 1996b; Tomlinson et al., 1997). Indeed, recombinant human nerve growth factor (NGF) has been studied in two major trials in patients with diabetic neuropathy but the second, larger, phase III trial was negative (Roche, personal communication).
NGF rescues sensory neurones from the changes in gene expression that follow nerve transection or axotomy. These changes include declines in expression of the peptides calcitonin gene-related peptide (CGRP) and substance P (SP), declines in neurofilament expression but increases in the expression of galanin (GAL), vasoactive intestinal peptide (VIP) and B50 [also called growth-associated protein 43 (GAP 43)] (Funakoshi et al., 1993; Hokfelt et al., 1994; Verge et al., 1995, 1996). The rationale for neurotrophin therapy in human diabetic neuropathy has been based, in part, on incomplete evidence that diabetic sensory neurones alter their gene expression in a pattern similar to that occurring after axotomy, as a result of impaired NGF availability (Liuzzi et al., 1998). The actual pattern of gene expression in an appropriate long-term model of experimental diabetic neuropathy, using quantitative in situ hybridization however, has not been explored fully.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Diabetic model: electrophysiology
The protocols were reviewed and approved by the University of Calgary Animal Care Committee using the Canadian Council of Animal Care guidelines. Experimental diabetes was induced in male Sprague–Dawley rats (200–300 g weight at onset) by a single i.p. (intraperitoneal) injection of streptozotocin (STZ) (65 mg/kg). Controls received the STZ citrate buffer carrier alone. Rats were accepted as diabetic if the whole-blood glucose level was
16.0 mmol/l. Glucose was measured with a glucometer in whole blood taken from the proximal ventral tail vein (AccuCheck III; Boehringer Mannheim Canada, Dorval, Quebec, Canada). Hyperglycaemia was confirmed at the end-point using fresh plasma and a glucose oxidase method (Ektachem DT6011 Analyzer; Kodak, Rochester, NY, USA). Insulin was not given to any of the diabetic rats. Rats were maintained on sawdust-covered plastic-bottomed cages to prevent wire cage-related neuropathy (Zochodne et al., 1995) and had access to rat chow and water ad libitum. Conduction velocity was recorded in sciatic–tibial motor fibres and caudal sensory fibres at a near-nerve subcutaneous temperature of 37°C with pentobarbital (65 mg/kg i.p.) anaesthesia. The methods used have been reported elsewhere (Zochodne and Ho, 1994; Zochodne et al., 1995; Singhal et al., 1997). At the end-point, anaesthetized rats underwent resection of the sciatic and sural nerves for morphometry. Bilateral L4 and L5 DRG were removed for morphometric studies, immunohistochemistry and in situ hybridization. The rats were killed with a high dose of pentobarbital. The cohorts of rats used in this study were also used in other work reported separately, addressing nitric oxide in experimental diabetes (Zochodne et al., 2000).
Histological preparation
Nerves and ganglia were fixed in 2.5% glutaraldehyde in 0.025 M cacodylate buffer overnight, then washed in buffer alone (0.15 M cacodylate), postfixed with osmium tetroxide (2% in 0.12 M cacodylate), dehydrated in alcohols and embedded in epon. Sections were made using an ultramicrotome at 1 µm thickness and stained with toluidine blue. For immunohistochemistry, DRG were fixed in Zamboni's fixative (see below). For in situ hybridization, specimens were frozen rapidly in isopentane.
Neurone counting
Number per section
Mean numbers of neurones with nucleoli per transverse section of the DRG were counted at 2 and 12 months. At 2 months, numbers of neurones with nucleoli were counted from four random 1-µm sections through the mid-portion of the L3–L5 ganglion and the mean number of neurones per transverse section was calculated. At 12 months, each rat had six to 12 sections through the mid-portion of the L5 (only) ganglion counted and the mean number of neurones per transverse section was calculated similarly. At 12 months, the mean number of CGRP- and SP-positive neurones per transverse section was calculated from ~24–32 sections per ganglion per rat (the mean was calculated for each experimental rat).
Physical dissector
This approach was reserved for the 12-month time point. L5 ganglia were sectioned along their longitudinal axis starting at one end (the end was defined as the first section that showed 10 neurones) and stopping at the other end. Each section was 1 µm in thickness. We studied six to 12 adjacent pairs of sections, stained with toluidine blue, separated by a distance (K) of 300 µm through the length of the ganglion (the number of pairs studied varied because of the variation in the size of the ganglion). Ganglia were then photographed and scanned and prints made at a magnification of x500. The first section (A) of the dissector pair was made into a transparency and compared with the printed copy of the second section (B). Nuclei of neurones in section A, but not section B, were counted for each pair, and the number was multiplied by 300 to give the total number of DRG neurones. Nuclei were preferred as unequivocal unique features that could be counted easily in each section, in contrast to the `tops' advocated in similar work by others. The choice of nuclei allowed us to use the unbiased approach of using paired sections as a physical dissector (Coggeshall and Lekan, 1996).
Pathology
These studies were carried out at the 12-month time point. A randomly selected transverse section from the mid-portion of the L5 dorsal root ganglion in five diabetic and five non-diabetic control rats was examined under oil-immersion light microscopy. A single examiner (D.W.Z.), who was masked to the identity of the specimens, evaluated ~100–150 neurones from each section and scored the numbers of abnormal neurones that were microvacuolated, small and dark, atrophic or replaced by nests of Nageotte.
Neurone and axon morphometry
For each rat/L5 ganglion at 2 and 12 months, we studied ~200 neurones with visible nucleoli from the mid-portion of the ganglion. Neurone diameter was measured under oil-immersion light microscopy interfaced with an image analysis program (Java; Jandel Scientific, San Rafael, Calif., USA). At 12 months, transverse sections of the sural nerve, stained with toluidine blue, were evaluated under oil-immersion light microscopy for myelinated fibre number and density, fibre and axon diameter and area, myelin thickness and fibre size distribution using a video microscopy image analysis set-up as described by Auer (Auer, 1994).
Immunohistochemistry
For immunohistochemistry, tissues were fixed in modified Zamboni's fixative (2% paraformaldehyde, 0.5% picric acid, 0.2 M phosphate buffer) overnight at 5°C. The tissue was then washed in PBS (phosphate-buffered saline), then dimethyl sulphoxide and again with PBS. It was then covered with PBS and 20% sucrose and left at 5°C overnight, and was then embedded in optimum cutting temperature (OCT) compound (Miles Laboratories, Elkant, Ind., USA), frozen, then sectioned at 16 µm. Sections were placed on poly-D-lysine-coated slides and incubated for 48 h at 4°C with rabbit antiserum to rat CGRP diluted 1 : 1000 (Cambridge Research Biochemicals, Wilmington, Del., USA) or SP diluted 1 : 1000 (Diasorin, Stillwater, Minn., USA). Slides were then washed with PBS and incubated with fluorescein isothiocyanate-conjugated goat and rabbit immunoglobulin G (Incstar, Stillwater, Minn., USA), diluted 1 : 50, for 1 h at room temperature. After further washing in PBS, coverslips were mounted with bicarbonate-buffered glycerol (pH 8.6) and viewed with a fluorescence microscope (Zeiss Axioplan).
In situ hybridization studies
Preparation of probes
Oligonucleotide probes complementary to and selective for the following mRNAs were synthesized (University of Calgary DNA Services, Calgary, Alberta, Canada): the neurotrophin receptors rat tyrosine kinase (Trk) A [the counterpart of bases 1198–1245 of the human TrkA sequence (Meakin et al., 1992; Merlio et al., 1992)], rat TrkB (full length) base pairs 1361–1408, rat TrkB(all) bases 1313–1360 (Middlemas et al., 1991), rat TrkC (full length) bases 1654–1701; rat TrkC(all) bases 1189–1236 (Merlio et al., 1992); the common neurotrophin receptor p75 bases 873–920 (Radeke et al., 1987) [the TrkB(all) and TrkC(all) probes recognized both full-length and truncated transcripts, whereas TrkA, TrkB (full length) and TrkC (full length) were selective for all known full-length transcripts with a Trk domain]; the peptides rat GAL base pairs 152–199 (Vrontakis et al., 1987), rat vasoactive intestinal polypeptide (VIP) base pairs 347–394 (Nishizawa et al., 1985), rat cholecystokinin (CCK) base pairs 298–341 (Deschenes et al., 1984), rat neuropeptide Y (NPY) base pairs 1671–1714 (Larhammar et al., 1987), rat SP base pairs 145–192 (Middlemas et al., 1991), rat CGRP base pairs 664–698 (Amara et al., 1985), rat ßCGRP base pairs 656–690 (Amara et al., 1985), pituitary adenylate cyclase activating polypeptide (PACAP) bases 701–747 (Hurley et al., 1995); the immediate early gene c-jun base pairs 778–825 (Ryseck et al., 1988); the heat shock protein rat HSP-27 bases 408–455 (Uoshima et al., 1993); the cytoskeletal genes for neurofilament medium subunit (NFM) bases 1222–1270 (Napolitano et al., 1987) and rat t-1 tubulin bases 1548–1594 (Lemischka et al., 1981); the synaptosome-associated protein SNAP 25 bases 133–179 (Catsicas et al., 1991); and the regeneration-associated gene rat B50 bases 70–117 (Karns et al., 1987). All probes were checked against the Genbank database (NIH) to ensure that homology to sequences other than the cognate transcript was no greater than 60%. The probes were labelled at the 3' end with -[35S]dATP (New England Nuclear, Boston, Mass., USA) using terminal deoxynucleotidyltransferase (Amersham, Pharmacia Biotech, Amersham, UK) in a buffer containing 10 mM CoCl2, 1 mM DTT (dithiothreitol), 300 mM Tris base and 1.4 M potassium cacodylate (pH 7.2) and purified through Nensorb-20 columns (New England Nuclear), and DTT was added to a final concentration of 10 nM. The specific activities obtained ranged from 2 to 5 x 106 d.p.m./ng oligonucleotide.
In situ hybridization
Deeply anaesthetized animals had their right and left L4 and L5 lumbar DRG rapidly dissected and frozen in OCT compound (Tissue Tek; Miles Laboratories, Elkhart, Ind., USA) in a Cryomold (Tissue Tek). Before sectioning, blocks containing pairs of DRG from animals of the control and experimental groups were fused with OCT compound to ensure processing under identical conditions. Sections were cut at 6 µm on a Micron cryostat (Carl Zeiss, Toronto, Canada), thaw-mounted on to Probe-ON slides (Fisher Scientific, Nepean, Canada) and stored with desiccant at –20°C until hybridization.
Hybridization was carried out according to published procedures (Dagerlind et al., 1992). The sections were brought to room temperature, air-dried and, without any additional treatment, covered with a hybridization buffer containing 50% formamide (Sigma, St Louis, Mo., USA), 4 x SSC (1 x SSC contains 0.15 M NaCl, 0.015 M sodium citrate), 1 x Denhardt's solution (0.02% bovine serum albumin and 0.02% Ficoll), 1% sarcosyl (N-laurylsarcosine), 0.02 M phosphate buffer (pH 7.0), 10% dextran sulphate, 500 mg/ml heat-denatured sheared salmon sperm DNA, 200 mM DTT and 107 d.p.m./ml of probe. The slides were placed in a box humidified with 50% formamide and incubated at 42°C for 18 h. After hybridization, the slides were washed for 60 min with four changes of 1 x SSC at 55°C and then brought to room temperature over 30 min while in the final rinse, then dipped in distilled water and dehydrated in 60 and 95% ethanol.
To generate radioautograms, the incubated slides were dipped in NTB2 nuclear track emulsion (Kodak) diluted 1 : 1 with distilled water and stored in the dark with desiccant at 4°C. The sections were exposed for 2–4 weeks, then developed in Kodak D19 for 3 min, fixed, washed and mounted with glycerol and a coverslip for analysis on a Zeiss microscope equipped with a Darklite dark-field slide-holder or stained with toluidine blue, mounted with Permount (Fisher Scientific) and a coverslip for viewing under bright-field illumination.
In situ hybridization control experiments
The specificity of the hybridization signal for each probe used in the study was ascertained by hybridization of sections of paired L5 ganglia from two rats with a 14-day right sciatic nerve injury (so that markers associated with intact and injured states could be visualized). Series of sections were hybridized with labelled probe, labelled probe with a 1000-fold excess of cold probe or labelled probe with a 1000-fold excess of another, dissimilar cold probe of the same length and similar G–C content.
Quantification
All slides were analysed qualitatively and consistent trends noted. Slides were selected for quantitative analysis on the basis of their representation of consistent trends observed. Image analysis was performed on DRG sections processed for in situ hybridization to CGRP, ßCGRP, PACAP, NFM and HSP-27 mRNAs. Montages of photomicrographs were prepared from 12 sections of diabetic and control tissue (x450). Individual neurones with a visible nucleus were identified and numbered (~128–206 cells per montage). Under x63 oil immersion light microscopy with an interactive computer-assisted image analysis system (Richardson et al., 1989), cross-sectional areas of individual neurones and the percentages of cytoplasmic area covered by silver grains were measured in each ganglion. Volumes were calculated from the larger of the two cross-sectional areas in adjacent sections on the assumption that the neurones were spherical. For each image, the density threshold was adjusted interactively so that the area per grain was constant for all neurones analysed on a given slide. Correction for grain overlap was made to obtain a parameter linearly related to the density of the silver grains (Richardson et al., 1989). Software for the image analysis system was generously provided by W. G. Tatton (Dalhousie University) and supplemented with published programs for data analysis and graphics (Press et al., 1988) (MATLAB; Mathworks, Natick, Mass., USA). Cells were considered labelled if they had more than five times the background level of silver grains, as determined by averaging grain counts over defined areas of the neuropil devoid of positively labelled cell bodies.
Analysis
Results were calculated as mean ± standard error of the mean. Student's t-test (one or two-tailed depending on the expected direction of change, if any) was used to compare diabetic rats and non-diabetic control rats.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Diabetic model: electrophysiology
Diabetic rats had gained less weight than controls at both time points and had hyperglycaemia (Table 1
). At both time points, diabetics had slowing of motor and sensory conduction velocity (Table 1).
|
Sural nerve morphometry
There was a non-significant trend towards fewer sural myelinated fibres in diabetic rats at 12 months. Fibre density was similar in diabetics and controls. Myelinated fibres from diabetic rats had a smaller mean fibre diameter, axon diameter, myelin thickness, fibre area and axon area than fibres from non-diabetic rats (Table 2
). Fibre-size histograms identified a change towards smaller size categories in fibres from diabetic rats (Fig. 1). There were fewer fibres >12 µm in diameter in diabetic rats.
|
|
DRG neurone morphometry, neurone counts and pathology
Mean neuronal diameter and area did not differ between diabetics and controls at 2 months. At 12 months, however, there was a significant reduction in neurone diameter and area in diabetic rats compared with non-diabetic controls and diabetic rats at 2 months, indicating the development of neurone atrophy by the 12-month time point (Table 2
and Figs 2 and 3).
|
|
Although there was a trend towards smaller total numbers of neurones in diabetic animals compared with non-diabetics by 12 months using unbiased counting and a physical dissector, the difference did not achieve statistical significance (Table 2
). Similarly, there was a non-significant trend towards fewer neurones per transverse section at 12 months but not at 2 months of diabetes. We found slightly fewer neurones per transverse section at 2 months than at 12 months but attribute the difference to the inclusion of L3 and L4 ganglia at 2 months in the counts (12-month counts were exclusively L5). Diabetic and control rats had similar numbers of CGRP- and SP-positive neurones per transverse section at 12 months (Table 2). At the light microscope level, no obvious alteration in neurone appearance was observed in DRGs from 12-month diabetic rats when a masked scoring system was used for microvacuolated neurones, shrunken or degenerate-appearing neurones, dark neurones or nests of Nageotte. Lipofuscin granules, small vacuoles, shrunken neurones and nests of Nageotte were observed occasionally in both diabetic rats and controls. There was a non-significant trend towards more shrunken neurones with retraction of their edges in diabetic rats, but the difference did not achieve statistical significance. We did not observe eccentric nuclei, central chromatolysis or dystrophic neurites.
In situ hybridization studies
At 2 months, mRNA expression patterns and levels of the peptides CGRP, ßCGRP, SP, the cytoskeletal proteins NFM and t1-tubulin, the growth-associated protein GAP 43 (B50) and the neurotrophin receptors p75 TrkA, TrkB or TrkC were similar or only slightly different in diabetic rats compared with non-diabetic controls (data not shown). However, by 12 months, alterations in the relative levels of the phenotypic markers expressed were readily noted and the number of markers examined was increased to gain better insight into the extent of the changes. The neuronal and axonal atrophy observed in the 12-month diabetic rats was accompanied by decreased NFM mRNA expression, which was most evident in the large neurones (Fig. 4). Quantification of the hybridization signal over individual neurones revealed a decrease in the average density of label from 34.3 ± 1.8 in non-diabetic control rats to 23.7 ± 0.9 in the diabetics. There was also a change in the percentage of large neurones (diameter >35 µm) with detectable message from 42% in the non-diabetic controls to 33% in the diabetic rats and a corresponding increase in small to medium-sized neurones expressing NFM mRNA from 54% (control rats) to 59% (diabetic rats). Decreases in peptide expression were also evident. Levels of both CGRP transcripts were decreased in the ganglia of diabetic rats (Fig. 5). The mean labelling hybridization signal in neurones from diabetic rats relative to that in neurones from non-diabetic controls decreased from 105.1 ± 16.1 to 92.9 ± 10.9 for CGRP and decreased from 16.6 ± 1.6 to 9.0 ± 0.9 for ßCGRP. There were also decreased percentages of neurones expressing detectable CGRP (22% reduction) or ßCGRP (33% reduction), the former being evident in all size ranges of neurones and the latter in predominantly small to medium-sized neurones (Fig. 5). A slight decrease in SP expression was also detected, with no discernible change in SOM expression (data not shown). The SNAP 25 hybridization signal was qualitatively reduced in neurones from 12-month diabetic rats. Neurotrophin receptor expression was only mildly to moderately altered in neurones from the diabetic rats, the most evident downregulation being observed in the TrkA subpopulation (Fig. 6). The decreases in p75 and TrkC message levels were most apparent in large neurones, which were usually heavily labelled for these markers. No alteration in TrkB mRNA expression was readily detectable.
|
32 x background) hybridization signal.
|
|
In contrast to axotomy, diabetes at either 2 or 12 months was not associated with increases in mRNA expression of GAL, NPY, PACAP, VIP, CCK, t
1-tubulin, B50 or c-jun in neurones. In fact, when GAL, PACAP, t1-tubulin, B50 and c-jun were detectable, the relative constitutive levels of expression of these markers were lower in the 12-month diabetic rats, as seen in Figs 7 and 8 (data for c-jun not shown). Elevation of these markers following peripheral nerve injury was verified (an example is shown for B50 in Fig. 7) to ensure that the absence of elevation in neurones from 12-month diabetic rats was not due to failure of the technique. The decrease in constitutive levels of expression was most evident for PACAP mRNA, the expression of which was confined primarily to small- to medium-size neurones in the intact or diabetic rats (Fig. 8). The percentage of neurones expressing detectable mRNA fell from 37 to 25 and the average neuronal labelling index [ratio of silver grain density over neuronal cytoplasm to grain density over areas of the neuropil devoid of positive hybridization signal (background)] decreased from 21.8 ± 3.0 to 11.6 ± 2.1. The only phenotypic marker of plasticity/stress to exhibit elevated expression in diabetic sensory neurones was HSP-27 (Fig. 7). There was a consistent trend towards elevated numbers of small- to medium-sized neurones expressing detectable levels of this message in 12-month diabetic animals (66%, versus 47% in non-diabetic controls), with a decrease in the percentage of large neurones (20%, versus 29% in non-diabetic controls). The average neuronal labelling index was also greater for DRG neurones from diabetic rats (46.4 ± 3.2) than for those from non-diabetic controls (33.6 ± 2.3).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The major findings of this study were as follows. (i) Experimental diabetes of both short (2 months) and long (12 months) duration was associated with slowing of motor and sensory conduction velocity. (ii) There were trends towards reduced total numbers of sural myelinated fibres and of total L5 sensory neurone numbers in diabetes, but neither trend was significant. (iii) There was evidence of sural myelinated fibre atrophy, with shifts to smaller size categories; a similar shift in fibre size category was observed in sensory neurones. (iv) The relative proportions of CGRP- and SP-containing neurones were not altered. (v) In DRG from 12-month diabetic animals there were declines in the expression of mRNAs for
CGRP, ßCGRP, p75, TrkA, TrkC and NFM in sensory neurones without increases in mRNAs for the injury markers GAP-43, CCK, VIP, GAL, t1-tubulin and c-jun. Indeed, neurones expressed lower constitutive levels of the plasticity markers B50 and t1-tubulin and the peptides GAL and PACAP. (vi) An increase in the stress-associated protein HSP-27 was observed in some neurones from 12-month diabetic animals. We believe that a longer-term experimental model of diabetes mellitus in the rat may provide a more appropriate approach to the understanding of human disease. In this respect, our model differs from most reported in the literature. Since the lifespan of a rat is ~2 years, we believe that our 1-year model reflects how human peripheral nerves and ganglia may respond to several decades of hyperglycaemia. Because we chose an intermediate end-point, the changes we observed were not clouded by the effect of advanced age. Since the density of DRG neurones between 2 and 12 months of diabetes did not change significantly in non-diabetics, it is unlikely that ageing influenced our findings.
There have been no prior studies of experimental diabetes that have applied an unbiased Coggeshall-style approach to the assessment of true neuronal numbers in DRG (Coggeshall and Lekan, 1996). It has been suggested that the use of the dissector approach should be the gold standard for neurone counts in all investigations (Saper, 1996), but this rigorous method has been adopted by only a minority of investigators. As the use of the physical as opposed to an optical dissector approach is time-consuming and technically demanding, we limited its application to the 12-month end-point, when significant changes in neurone numbers were more likely to have occurred. We think it unlikely that such a dissector approach at 2 months would have proved helpful, since all of our measures identified a much milder phenotype at this time. Moreover, at 2 months there was no indication in our measurements of the number of neurones per transverse section (a method of counting neurones used much more frequently in published work) that there was early neurone loss. There have been two reports of DRG neurone morphology in experimental diabetes, and these have included a longer model such as ours, but neither used a similar method to count neurone numbers (Sidenius and Jakobsen, 1980; Sasaki et al., 1997). Finally, our observation of a trend towards a decline in neurone number at 12 months closely matched the trend in the loss of sural fibres that we also observed at this time, an expected correlation.
Our findings support the concept that experimental diabetes targets sensory neurones (Zochodne, 1996a). We believe that combined studies of morphometry and gene expression identify a slowly progressive degenerative condition of sensory neurones associated with perikaryal and axon atrophy. It is likely that neurone loss will be more overt with an even longer model, or perhaps more intense hyperglycaemia of long duration. At earlier time points we did not also observe sural fibre loss, a negative result that prompted us to concentrate on the later time point in this work. Sural nerve axon atrophy, but not loss, has been observed previously at 6 months (but not 2 months) of identical experimental diabetes in our laboratory by Scott and colleagues (Scott et al., 1999). Susceptibility factors are also likely to be an important determinant of neurone survival, and it is possible that other models, such as the STZ mouse model, in which we have observed more obvious sural fibre loss (Kennedy and Zochodne, 2000), might also show more sensory neurone loss. Russell and colleagues (Russell et al., 1999) and Srinivasan and colleagues (Srinivasan et al., 2000) have identified (by terminal dUTP nick end labelling) the labelling of sensory neurones in experimental diabetes within 3 months after its onset. This implies that sensory neurones in diabetes undergo early apoptosis. While these findings may correctly identify early DNA damage in diabetic sensory neurones, clear evidence of loss obtained by rigorous counting is not yet available. If early and widespread neurone loss were a feature of all experimental diabetes models, dramatic distal fibre dropout should accompany it. Morphometric studies of axon numbers in experimental diabetes have been consistent in failing to identify such massive dropout. Since we did not examine very distal sensory terminals, e.g. in footpads, it is possible that the widespread changes in gene expression we observed, and the neurone atrophy, were associated with the retraction of very distal axon terminals. Recent work has shown that the examination of such distal fibre territories as the skin is particularly sensitive in the identification of early fibre dropout in human diabetes (Kennedy et al., 1996, 1999).
We believe that the in situ hybridization results of this study support our previous work identifying the loss of structural proteins (neurofilaments and microtubules) in parallel with declines in their incorporation into axons (Scott et al., 1999). Transgenic mice that lack axonal neurofilaments develop more severe experimental diabetic neuropathy than normal littermates, which also supports the idea that a normal complement of neurofilaments may protect against axon damage in diabetes (Zochodne et al., 1996).
Our findings do not support the concept that changes in diabetic DRG neurones might simply reflect retrograde `axotomy-like' responses to distal axon damage or transection (Liuzzi et al., 1998). While the study of Liuzzi and colleagues showed elevated expression of the injury-associated cytoskeletal element class III ß-tubulin, we were unable to detect an elevation in another injury-associated cytoskeletal protein, t1-tubulin. Indeed, constitutive mRNA levels of this protein and B50, a regeneration-associated protein, were reduced. We also did not observe more overt and significant loss of sural fibres that would be associated with distal axon damage, as occurs in axotomy. We believe that these are important findings because they suggest that an initial target of diabetes is the neurone, not the distal axon. We postulate that the injury response of peripherally axotomized non-diabetic sensory neurones may reflect the combination of two events: (i) loss of constitutive retrograde trophic support from the periphery with declines in trophin-dependent mRNA species, such as CGRP and neurofilament subunits; and (ii) elevation in expression of injury-associated cytokine and growth factors that regulate injury-associated mRNAs positively, such as VIP and galanin (for a review, see Verge et al., 1996). In diabetes, the loss of retrograde support of neurotrophins could conceivably account for the declines in axon calibre, neurone numbers, neurofilament, PACAP, CGRP, ßCGRP, TrkA, TrkC p75, B50 and t1-tubulin that we and others have identified. However, it is unlikely that injury-related cytokines, such as interleukin-6 and LIF (leukocyte migration inhibitory factor), are elevated in diabetes, because markers known to upregulated by these molecules (VIP and GAL) fail to occur. Our one anomalous finding was the elevation in HSP-27 that was observed in some neurones. Elevated HSP-27 expression has been described in injured sensory neurones (Costigan et al., 1998; V.M.K.V., personal observations) and appears to have a role in sensory neurone survival (Lewis et al., 1999). Interestingly, Lewis and colleagues showed that overexpression of this protein increases the survival of neonatal sensory neurones following withdrawal of NGF (Lewis et al., 1999). Thus, elevated HSP-27 expression in trophin-deprived diabetic neurones may represent an attempt to mitigate this state and maximize survival. Defective trophic factor uptake or transport and reduced levels in diabetes are an attractive explanation for sensory neurone dysfunction in diabetes (Brewster et al., 1994; Fernyhough et al., 1995a, 1995b, 1998; Ihara et al., 1996; Tomlinson et al., 1996, 1997). This possibility has been reviewed recently and there is evidence for and against it (Zochodne, 1996b). The state is further exacerbated by reduced levels of neurotrophin receptor expression that we (in a long-term study) and others (short-term studies) have observed (Delcroix et al., 1997; Mohiuddin and Tomlinson, 1997; Fernyhough et al., 1998).
Recombinant human NGF did not improve human disease in a recent clinical trial involving diabetic subjects. As diabetes affects fibres of all size ranges, it is unlikely that single molecules affecting only certain subpopulations of neurones will reverse the range of abnormalities in diabetic neuropathy. Consideration of the delivery of multiple trophic molecules to reverse the deficits may be required. Examples of these include insulin-like growth factor 1, insulin and neurotrophin 3. Insulin-like growth factor 1 rescues sensory neurones from hyperglycaemic apoptotic cell death and is an attractive candidate for a role in diabetic sensory neurone dysfunction (Matthews and Feldman, 1996; Russell et al., 1999). Its levels may be reduced in diabetes and some evidence has suggested that treatment with insulin-like growth factor 1 is of benefit in animal models of diabetes, but the studies were limited to regenerative deficits and pain behaviour (Ishii, 1993, 1995; Ishii and Lupien, 1995). Treatment with low-dose near-nerve insulin unilaterally reversed conduction abnormalities in experimental diabetes and appeared to promote small-fibre sprouting (Singhal et al., 1997). Administration of exogenous neurotrophin 3 reversed established sciatic nerve conduction velocity deficits in a 12-week STZ diabetes model (Mizisin et al., 1999). It would now be of interest to compare specific trophic molecules and combinations of them in a long-term experimental model to determine their effect on the degenerative neuronal phenotype.
We also observed relative preservation of numbers of neurones expressing CGRP and SP using immunohistochemical counting. The results are of significance since diabetic models demonstrate lower thresholds for nociceptive behaviour, a finding that may have suggested acquisition of the SP phenotype by a larger number of neurones in diabetics. Other mechanisms probably underlie hyperalgesia in diabetics (Ahlgren and Levine, 1993, 1994; Sasaki et al., 1998).
Additional abnormalities, discussed above, might account for the primary and progressive neurone dysfunction we have evidence for in this model. DRG may have reductions in local blood flow in diabetes (Zochodne et al., 1994; Zochodne and Ho, 1994; Sasaki et al., 1997). Declines in DRG perfusion might be more relevant than changes in the nerve trunk because of the much higher metabolic demands and vulnerability of neurones and neurites in the ganglia. The absence of a more effective blood–neurone barrier may make the DRG vulnerable to other abnormalities, such as the entry of harmful glycosylated proteins.
We believe that the relative preservation of neurones is important in considering future therapeutic approaches to this condition. Declines in the gene expression of DRG proteins may be useful indices of disease that could be addressed with new targeted forms of intervention.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Brenda Boake provided expert secretarial assistance. D.W.Z. is a Senior Scholar of the Alberta Heritage Foundation for Medical Research. V.M.K.V. is a Medical Research Council of Canada (CIHR) Scholar. The work was supported by the Medical Research Council of Canada.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Ahlgren SC, Levine JD. Mechanical hyperalgesia in streptozotocin-diabetic rats. Neuroscience 1993; 52: 1049–55.[ISI][Medline]
Ahlgren SC, Levine JD. Protein kinase C inhibitors decrease hyperalgesia and C-fiber hyperexcitability in the streptozotocin-diabetic rat. J Neurophysiol 1994; 72: 684–92.
Amara SG, Arriza JL, Leff SE, Swanson LW, Evans RM, Rosenfeld MG. Expression in brain of a messenger RNA encoding a novel neuropeptide homologous to calcitonin gene-related peptide. Science 1985; 229: 1094–7.
Arvidson B. Distribution of protein tracers in peripheral ganglia. A light and electron microscopic study in rodents after various modes of tracer administration. Acta Univ Upps 1979; 344: 1–72.
Auer RN. Automated nerve fibre size and myelin sheath measurement using microcomputer-based digital image analysis: theory, method and results. J Neurosci Methods 1994; 51: 229–38.[ISI][Medline]
Brewster WJ, Fernyhough P, Diemel LT, Mohiuddin L, Tomlinson DR. Diabetic neuropathy, nerve growth factor and other neurotrophic factors. [Review]. Trends Neurosci 1994; 17: 321–5.[ISI][Medline]
Brewster WJ, Fernyhough P, Diemel LT, Mohiuddin L, Tomlinson DR. Changes in nerve growth factor and preprotachykinin messenger RNA levels in the iris and trigeminal ganglion in diabetic rats: effects of treatment with insulin or nerve growth factor. Brain Res Mol Brain Res 1995; 29: 131–9.[Medline]
Catsicas S, Larhammar D, Blomqvist A, Sanna PP, Milner RJ, Wilson MC. Expression of a conserved cell-type-specific protein in nerve terminals coincides with synaptogenesis. Proc Natl Acad Sci USA 1991; 88: 785–9.
Coggeshall RE, Lekan HA. Methods for determining numbers of cells and synapses: a case for more uniform standards of review. [Review]. J Comp Neurol 1996; 364: 6–15.[ISI][Medline]
Costigan M, Mannion RJ, Kendall G, Lewis SE, Campagna JA, Coggeshall RE, et al. Heat shock protein 27: developmental regulation and expression after peripheral nerve injury. J Neurosci 1998; 18: 5891–900.
Dagerlind A, Friberg K, Bean AJ, Hokfelt T. Sensitive mRNA detection using unfixed tissue: combined radioactive and non-radioactive in situ hybridization histochemistry. Histochemistry 1992; 98: 39–49.[ISI][Medline]
Delcroix JD, Tomlinson DR, Fernyhough P. Diabetes and axotomy-induced deficits in retrograde axonal transport of nerve growth factor correlate with decreased levels of p75LNTR protein in lumbar dorsal root ganglia. Brain Res Mol Brain Res 1997; 51: 82–90.[Medline]
Deschenes RJ, Lorenz LJ, Haun RS, Roos BA, Collier KJ, Dixon JE. Cloning and sequence analysis of a cDNA encoding rat preprocholecystokinin. Proc Natl Acad Sci USA 1984; 81: 726–30.
Diemel LT, Brewster WJ, Fernyhough P, Tomlinson DR. Expression of neuropeptides in experimental diabetes: effects of treatment with nerve growth factor or brain-derived neurotrophic factor. Brain Res Mol Brain Res 1994; 21: 171–5.[Medline]
Fernyhough P, Diemel LT, Brewster WJ, Tomlinson DR. Deficits in sciatic nerve neuropeptide content coincide with a reduction in target tissue nerve growth factor messenger RNA in streptozotocin-diabetic rats: effects of insulin treatment. Neuroscience 1994; 62: 337–44.[ISI][Medline]
Fernyhough P, Diemel LT, Hardy J, Brewster WJ, Mohiuddin L, Tomlinson DR. Human recombinant nerve growth factor replaces deficient neurotrophic support in the diabetic rat. Eur J Neurosci 1995a; 7: 1107–10.[ISI][Medline]
Fernyhough P, Diemel LT, Brewster WJ, Tomlinson DR. Altered neurotrophin mRNA levels in peripheral nerve and skeletal muscle of experimentally diabetic rats. J Neurochem 1995b; 64: 1231–7.[ISI][Medline]
Fernyhough P, Diemel LT, Tomlinson DR. Target tissue production and axonal transport of neurotrophin-3 are reduced in streptozotocin-diabetic rats. Diabetologia 1998; 41: 300–6.[ISI][Medline]
Funakoshi H, Frisen J, Barbany G, Timmusk T, Zachrisson O, Verge VM, et al. Differential expression of mRNAs for neurotrophins and their receptors after axotomy of the sciatic nerve. J Cell Biol 1993; 123: 455–65.
Greene DA, Winegrad AI, Carpentier JL, Brown MJ, Fukuma M, Orci L. Rabbit sciatic nerve fascicle and `endoneurial' preparations for in vitro studies of peripheral nerve glucose metabolism. J Neurochem 1979; 33: 1007–18.[ISI][Medline]
Hokfelt T, Zhang X, Wiesenfeld-Hallin Z. Messenger plasticity in primary sensory neurons following axotomy and its functional implications. [Review]. Trends Neurosci 1994; 17: 22–30.[ISI][Medline]
Hurley JD, Gardiner JV, Jones PM, Bloom SR. Cloning and molecular characterization of complementary deoxyribonucleic acid corresponding to a novel form of pituitary adenylate cyclase-activating polypeptide messenger ribonucleic acid in the rat testis. Endocrinology 1995; 136: 550–7.[Abstract]
Ihara C, Shimatsu A, Mizuta H, Murabe H, Nakamura Y, Nakao K. Decreased neurotrophin-3 expression in skeletal muscles of streptozotocin-induced diabetic rats. Neuropeptides 1996; 30: 309–12.[ISI][Medline]
Ishii DN. Insulin and related neurotrophic factors in diabetic neuropathy. [Review]. Diabet Med 1993; 10 Suppl 2: 14S–15S.
Ishii DN. Implication of insulin-like growth factors in the pathogenesis of diabetic neuropathy. [Review]. Brain Res Brain Res Rev 1995; 20: 47–67.[Medline]
Ishii DN, Lupien SB. Insulin-like growth factors protect against diabetic neuropathy: effects on sensory nerve regeneration in rats. J Neurosci Res 1995; 40: 138–44.[ISI][Medline]
Kadekaro M, Crane AM, Sokoloff L. Differential effects of electrical stimulation of sciatic nerve on metabolic activity in spinal cord and dorsal root ganglion in the rat. Proc Natl Acad Sci USA 1985; 82: 6010–13.
Karns LR, Ng SC, Freeman JA, Fishman MC. Cloning of complementary DNA for GAP-43, a neuronal growth-related protein. Science 1987; 236: 597–600.
Kennedy WR, Nolano M, Wendelschafer-Crabb G, Johnson TL, Tamura E. A skin blister method to study epidermal nerves in peripheral nerve disease. Muscle Nerve 1999; 22: 360–71.[ISI][Medline]
Kennedy JM, Zochodne DW. The regenerative deficit of peripheral nerves in experimental diabetes: its extent, timing and possible mechanisms. Brain 2000; 123: 2118–29.
Kennedy WR, Wendelschafer-Crabb G, Johnson T. Quantitation of epidermal nerves in diabetic neuropathy. Neurology 1996; 47: 1042–8.
Larhammar D, Ericsson A, Persson H. Structure and expression of the rat neuropeptide Y gene. Proc Natl Acad Sci USA 1987; 84: 2068–72.
Lemischka IR, Farmer S, Racaniello VR, Sharp PA. Nucleotide sequence and evolution of a mammalian alpha-tubulin messenger RNA. J Mol Biol 1981; 151: 101–20.[ISI][Medline]
Lewis SE, Mannion RJ, White FA, Coggeshall RE, Beggs S, Costigan M, et al. A role for HSP27 in sensory neuron survival. J Neurosci 1999; 19: 8945–53.
Liuzzi FJ, Bufton SM, Vinik AI. Streptozotocin-induced diabetes mellitus causes changes in primary sensory neuronal cytoskeletal mRNA levels that mimic those caused by axotomy. Exp Neurol 1998; 154: 381–8.[ISI][Medline]
Matthews CC, Feldman EL. Insulin-like growth factor I rescues SH-SY5Y human neuroblastoma cells from hyperosmotic induced programmed cell death. J Cell Physiol 1996; 166: 323–31.[ISI][Medline]
Meakin SO, Suter U, Drinkwater CC, Welcher AA, Shooter EM. The rat trk protooncogene product exhibits properties characteristic of the slow nerve growth factor receptor. Proc Natl Acad Sci USA 1992; 89: 2374–8.
Merlio JP, Ernfors P, Jaber M, Persson H. Molecular cloning of rat trkC and distribution of cells expressing messenger RNAs for members of the trk family in the rat central nervous system. Neuroscience 1992; 51: 513–32.[ISI][Medline]
Middlemas DS, Lindberg RA, Hunter T. TrkB, a neural receptor protein-tyrosine kinase: evidence for a full-length and two truncated receptors. Mol Cell Biol 1991; 11: 143–53.
Mizisin AP, Calcutt NA, Tomlinson DR, Gallagher A, Fernyhough P. Neurotrophin-3 reverses nerve conduction velocity deficits in streptozotocin-diabetic rats. J Peripher Nerv Syst 1999; 4: 211–21.[ISI][Medline]
Mohiuddin L, Tomlinson DR. Impaired molecular regenerative responses in sensory neurones of diabetic rats: gene expression changes in dorsal root ganglia after sciatic nerve crush. Diabetes 1997; 46: 2057–62.[Abstract]
Napolitano EW, Chin SS, Colman DR, Liem RK. Complete amino acid sequence and in vitro expression of rat NF-M, the middle molecular weight neurofilament protein. J Neurosci 1987; 7: 2590–9.[Abstract]
Nishizawa M, Hayakawa Y, Yanaihara N, Okamoto H. Nucleotide sequence divergence and functional constraint in VIP precursor mRNA evolution between human and rat. FEBS Lett 1985; 183: 55–9.[ISI][Medline]
Press WH, Flannery BP, Teukolsky SA, Vetterling WT, editors. Numerical recipes in C. Cambridge: Cambridge University Press; 1988.
Radeke MJ, Misko TP, Hsu C, Herzenberg LA, Shooter EM. Gene transfer and molecular cloning of the rat nerve growth factor receptor. Nature 1987; 325: 593–7.[Medline]
Richardson PM, Verge VMK, Riopelle RJ. Quantitative radioautography for NGF receptors. In: Rush RA, editors. Nerve growth factors. Chichester (UK): Wiley; 1989. p. 315–26.
Russell JW, Sullivan KA, Windebank AJ, Herrmann DN, Feldman EL. Neurons undergo apoptosis in animal and cell culture models of diabetes. Neurobiol Dis 1999; 6: 347–63.[ISI][Medline]
Ryseck RP, Hirai SI, Yaniv M, Bravo R. Transcriptional activation of c-jun during the G0/G1 transition in mouse fibroblasts. Nature 1988; 334: 535–7.