Background
Oxaliplatin is a platinum-based chemotherapeutic agent approved for the treatment of colorectal cancer [
1]. Although particularly effective for treating colorectal cancer, oxaliplatin causes neurotoxicity in a high percentage of patients [
2] that is dose-limiting and can only be prevented by reducing or stopping the drug. Oxaliplatin causes acute and chronic forms of neurotoxicity in the clinic. Acute oxaliplatin neurotoxicity presents with neuro-sensory symptoms that develop during or soon after each drug infusion then recover within a few days or weeks [
2,
3]. These symptoms are exacerbated by cold exposure and associated with electrophysiological signs of peripheral nerve hyperexcitability [
4]. With repeated treatment, oxaliplatin causes a chronic sensory neuropathy with distal paraesthesiae and dyesthesiae, loss of tendon reflexes, vibration sense and proprioception, and sensory ataxia in severe cases [
2,
3]. The chronic neurotoxicity of oxaliplatin is cumulative and less reversible than its acute syndrome.
There have been previous studies of oxaliplatin-induced neurotoxicity in rodent models. Single doses of oxaliplatin have been reported to acutely disturb nucleolar morphology in DRG neurons [
5] and alter behavioural responses indicating sensory allodynia and hyperalgesia [
6,
7]. Chronic oxaliplatin treatment causes reduced sensory nerve conduction in the tail or hind-limb of treated rodents [
8,
9], altered sensory responses [
10,
11] and changes in the size profiles of DRG neurons [
8,
9,
12] suggestive of neuronal atrophy or the loss of DRG neurons. The doses of oxaliplatin employed in these previous rodent studies have varied widely but were often lower than those used clinically, when expressed as per unit of body surface area or considered on the basis of relative systemic exposure achieved in rats [
13] and humans [
14].
In the current study, we investigated the effect of oxaliplatin on neuronal size profiles and neurofilament immunoreactivity in DRG tissue from adult Wistar rats following multiple treatments to a cumulative dose of approximately 180 mg/m
2. This dose was broadly comparable to those that are achieved clinically [
1] and induce changes in sensory nerve conduction and DRG morphology in the rat model [
8]. Immunohistochemistry was used to identify subpopulations of DRG neurons and assess their relative susceptibilities to oxaliplatin-induced neurotoxicity, as in recent studies [
8,
15]. The RT97 primary antibody employed in these studies recognises phosphorylated KSP repeats in the tail domain of phosphorylated neurofilament heavy subunit (pNF-H) [
16]. The epitopes of the RT97 antibody are strongly expressed in rat DRG tissue within the cell bodies of subpopulations of large DRG neurons and large-diameter myelinated nerve fibres [
17]. Phosphorylated neurofilaments are major cytoskeletal proteins of large myelinated sensory neurons [
18]. Disturbance of neurofilament phosphorylation has been implicated in a wide range of neurodegenerative diseases [
19] but its role in oxaliplatin-induced neurotoxicity is unknown.
In this paper, we report that neuronal pNF-H expression, as determined by RT97 immunohistochemistry of rat DRG tissue, was significantly reduced after oxaliplatin treatment. This loss of pNF-H immunoreactivity was shown to correspond with the relative neurotoxicity of oxaliplatin, cisplatin and carboplatin, but was not associated with the loss of DRG cells, generalised reduction of neuronal marker or neurofilament expression, or with paclitaxel-induced neurotoxicity.
Discussion
In this paper we report, for the first time, an effect of oxaliplatin on neurofilament expression in rat DRG tissue as determined by immunohistochemistry using the RT97 primary anti-neurofilament antibody. The epitopes of the RT97 antibody are phosphorylated KSP repeats within the tail domain of phosphorylated neurofilament heavy subunit [
16], which are expressed in a specific pattern in DRG tissue of healthy adult rats [
17]. In DRG these epitopes are expressed within the cell bodies of a subpopulation of large neurons, presumably during early post-translational modification of neurofilament subunits, and in their axons, where neurofilament subunits become hyper-phosphorylated during their polymerisation into stabilised polymeric complexes [
26]. Aberrant neurofilament phosphorylation, as indicated by altered tissue immunoreactivity for phospho-specific anti-neurofilament primary antibodies, such as the RT97 antibody, has been associated with other disorders of DRG neurons [
27,
28], but a role in oxaliplatin neurotoxicity has not been considered.
We demonstrated that chronic oxaliplatin treatment was associated with a specific pattern of loss of pNF-H immunoreactivity in rat DRG tissue in the current study. The loss of pNF-H immunoreactivity was evident visually from qualitative changes in the intensity of neuronal cell body immunostaining in DRG sections and by statistically significant reductions in the numbers and size of pNF-H-immunoreactive neurons in oxaliplatin-treated animals, confirmed in two independent experiments. Strong pNF-H immunoreactivity appeared to remain in DRG nerve fibres after oxaliplatin treatment indicating that its loss was specific for the neuronal cell bodies, and that the treatment caused no nonspecific masking of pNF-H epitopes, under these experimental conditions. In control DRG, the neuronal immunoreactivity for pNF-H overlapped or colocalised with parvalbumin, non-phospho-specific-NF-M and non-phospho-specific-NF-H, but the number of DRG neurons displaying immunoreactivity for these primary antibodies was not changed by oxaliplatin. Therefore, the loss of neuronal pNF-H immunoreactivity was not associated with the loss of DRG cells or any generalised reduction of neuronal marker or neurofilament expression induced by oxaliplatin.
The present study also demonstrated that the extent of loss of pNF-H immunoreactivity corresponded to the relative neurotoxicity of oxaliplatin, cisplatin and carboplatin. When these platinum agents were ranked according to their effect on the number of pNF-H-immunoreactive DRG neurons, oxaliplatin had the greatest effect, followed by cisplatin and then carboplatin had the least. Their ranking corresponded with the relative cumulative dose-potencies of oxaliplatin, cisplatin and carboplatin for reducing sensory nerve conduction velocity in rats, which occurs after cumulative doses of 15, 46.7 and 302 μmol/kg, respectively [
24]. In addition, this ranking corresponded with the proportion of patients developing peripheral neurotoxicity of any severity grade after treatment with these platinum drugs, which is reported to occur in ~90% [
29], ~50% [
30] and ~6% [
31] of patients treated with oxaliplatin, cisplatin and carboplatin, respectively.
These findings link the loss of pNF-H with the neurotoxicity of oxaliplatin, although its exact role in this toxicity remains to be elucidated. Phosphorylated neurofilaments have important physiological roles in maintaining axonal calibre and fast conduction velocity of large myelinated nerve fibres [
18,
26], and their loss causes neuronal and axonal atrophy, and reduced sensory nerve conduction velocity, of DRG neurons [
32‐
34]. Therefore, the loss of neuronal pNF-H expression demonstrated in the current study may be causally linked to the decreased size profiles of DRG neurons and reduced sensory nerve conduction velocity, which are induced by chronic oxaliplatin treatment in rodent models [
8,
9,
12].
The molecular mechanisms responsible for loss of pNF-H immunoreactivity induced by oxaliplatin are unclear and require further study. Defects in early neurofilament phosphorylation could account for the loss of RT97 cell body staining without changes in its nerve fibre immunoreactivity or altered immunoreactivity of non-phospho-specific antineurofilament primary antibodies. The main pharmacological mechanism of platinum-based drugs is the formation of platinum-DNA adducts that inhibit DNA replication and transcription [
35]. After exposure to platinum drugs, DNA-platinum adducts have been detected in DRG neurons [
36,
37] and their level is correlated with the severity of neurotoxicity [
38,
39]. Therefore, the loss of RT97 immunoreactivity could occur due to inhibited transcription of neurofilament kinase genes. However, confirmation of a mechanism involving defective neurofilament phosphorylation or inhibited transcription would be technically difficult in DRG tissue because of the confounding effects of persisting RT97 immunoreactivity of the ganglionic nerve fibres and non-specific inhibition of DNA transcription by platinum drugs [
40]. Whatever the mechanism, it was evident from this study that pNF-H is a specific marker of DRG neuronal subpopulations particularly susceptible to damage from chronic oxaliplatin exposure, and changes in numbers of pNF-H immunoreactive neurons, are readily measureable endpoints of oxaliplatin neurotoxicity in the rat. Similarly, the current study confirmed our previous observations [
8] of parvalbumin being a specific marker of DRG neurons susceptible to oxaliplatin toxicity and significant changes in size profiles of parvalbumin immunoreactive neurons during this neurotoxicity. Detecting oxaliplatin-induced neurotoxicity in the rat model using pNF-H or parvalbumin immunohistochemistry is statistically more powerful and utilises fewer animals than nerve conduction studies. However, unlike immunochemical endpoints, nerve conduction measurements can be repeated at different times in the same animal and provide functional information.
Paclitaxel causes peripheral neurotoxicity in a high proportion of treated patients [
41] and reductions in sensory nerve conduction velocity in the rat [
25,
42‐
44], but had no effect on the number or size of pNF-H immunoreactive neurons in this study. The mechanism of paclitaxel neurotoxicity may involve microtubule binding and disturbance of microtubule polymerisation with resulting axonal damage [
43,
45] and secondary reactive changes in DRG cell bodies [
25,
46,
47]. In contrast, the mechanism of oxaliplatin neurotoxicity may involve a loss of phosphorylated neurofilaments at the level of DRG cell bodies with secondary changes in axonal calibre and conduction velocity. In this way, disturbance of major neuronal cytoskeletal proteins, such as microtubules and neurofilaments, may be a common mechanistic theme whereby different anticancer drugs from various classes damage the peripheral nervous system.
Methods
Animals and Drugs
Age-matched 10-week old female Wistar rats were used for experiments that weighed approximately 270 g at the commencement of the study. All animals were housed in a temperature and humidity-controlled environment with uninhibited access to food and water. Oxaliplatin (Sigma-Aldrich, St. Louis, MO, USA and Sanofi-Synthelabo NZ Ltd, Auckland, NZ) and carboplatin (Mayne Pharma, Vic, Australia) were diluted for injection in 5% dextrose (Baxter Healthcare, Old Toongabbie, Australia) for intraperitoneal injection at 15 ml/kg. Cisplatin (Sigma) was diluted in 0.9% sodium chloride (Baxter Healthcare) for intraperitoneal injection at 15 ml/kg. Paclitaxel (Phytogen Life Sciences Inc., Delta, BC, Canada) was solubilised in a 1:1 solution of Cremophor EL (Sigma-Aldrich) and ethanol to make a stock solution of 6 mg/ml, then further diluted with 0.9% NaCl (Baxter Healthcare) for administration by intraperitoneal injection at an injection volume of 12.5 ml/kg. Animals were treated twice per week either with oxaliplatin (1.85 mg/kg), carboplatin (8 mg/kg) or their control drug vehicle of 5% dextrose, or with cisplatin (1 mg/kg) or its control vehicle of 0.9% sodium chloride, an injection volume of 15 ml/kg for 8 weeks. Paclitaxel-treated animals received 12.5 mg/kg of drug once weekly for a total of 9 weeks, and control animals were treated with the Cremophor EL/ethanol/0.9% NaCl solution at the same dosing frequency and injection volume. To prevent time-dependent variation in pharmacokinetics and pharmacodynamics all injections were performed between 1 and 3 p.m. The Animal Ethics Committee of the University of Auckland approved all animal procedures.
Single-label Immunohistochemistry
One week after the conclusion of treatment, terminal anaesthesia was induced by administering 0.9 ml of 3 mg/ml pentobarbitone (Chemstock Animal Health Ltd, Christchurch, New Zealand). Subsequently, transcardiac perfusion with 60 ml of 0.9% NaCl (Baxter Healthcare) followed by 60 ml of 4% paraformaldehyde in 0.1 M phosphate buffer was carried out. L5 DRGs were carefully dissected from each animal, post-fixed in 4% paraformaldehyde for 2-6 hours and cryoprotected in a 30% sucrose solution until the tissues sunk. Following cryoprotection, the DRG were placed in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA, USA), snap frozen in liquid nitrogen and stored at -80°C. Each dorsal root ganglion was sectioned on a cryostat (Leica CM 3050) at a thickness of 10 μm onto polylysine-coated slides that were then stored at -80°C. For immunostaining, frozen tissue slides were warmed to room temperature, washed in PBS containing 0.2% Triton X-100 and incubated in 1% H2O2 in 50% methanol for 10 minutes. To prevent non-specific binding, the slides were blocked for 1 hour in PBS containing 0.2% Triton X-100 with 3% normal goat serum (Sigma-Aldrich) and 20 mg/ml bovine serum albumin (Sigma-Aldrich). Next, the slides were incubated overnight in a humidity chamber with either the mouse monoclonal antibody to the phospho-specific NF-H subunit (RT-97 clone; 1:100; CBL212, Chemicon International, Temecula, CA, USA), rabbit polyclonal anti-parvalbumin (PVA3) primary antibody (1:1000; P. Emson, Cambridge, UK), rabbit polyclonal non-phospho-specific antineurofilament 200 IgG fraction (1:1000; Sigma N4142) or mouse monoclonal non-phospho-specific antineurofilament medium subunit (1:1000; Sigma N5264). The slides were rewashed and incubated with either an anti-mouse or anti-rabbit biotinylated secondary antibody (1:500; Sigma-Aldrich) for 2.5 hours. After further washes, the slides were incubated for 3 hours in an extravidin-peroxidase conjugate (1:500; Sigma-Aldrich). Staining was visualised with 0.5 mg/ml 3,3'-diaminobenzidine tetrahydrochloride (AppliChem, Darmstadt, Germany) and 0.01% H2O2 in 0.4 M phosphate buffer for 10 minutes. Finally, the slides were washed, dehydrated through a series of alcohols, cleared in xylene and coverslipped. The DRG sections were analysed by light microscopy with digital images generated by an Axiocam camera (Carl Zeiss Vision, Hallbergmoos, Germany) and quantitative analysis performed using AxioVision 3.0 (Carl Zeiss Software) software. The cross-sectional area was measured for each immunoreactive neuron and the frequency of expression was generated by counting every cell and expressing the count of immunoreactive neurons as a percentage of the total cell count. Immunoreactive DRG neurons were also categorised on the basis of size into small (cross-sectional area <750 μm2), medium (750-1750 μm2) and large (>1750 μm2) sized cells.
Fluorescent Double Labelling Immunohistochemistry
Frozen DRG slides were defrosted, washed, incubated in H2O2 with methanol and blocked as described previously for single-label immunohistochemistry. Each slide was incubated overnight in a humidity chamber in both mouse anti-pNF-H (1:100) and rabbit anti-parvalbumin (1:1000) primary antibodies. The slides were then washed and incubated in the dark for 4 hours in both anti-rabbit cy3 (1:200; Jackson Laboratories, West Grove, PA, USA) and biotinylated anti-mouse secondary antibodies (1:200, Sigma). The slides were rewashed and incubated in the dark for 3 hours in FITC tertiary antibody (1:200; Sigma-Aldrich). Finally, the slides were washed, cover slipped with Citifluor (Agar Scientific, Essex, UK) and stored overnight at 4°C to prevent bleaching. Fluorescent analysis was performed with a Zeiss Axioplan 2 epifluorescence microscope (Carl Zeiss Microscopy) equipped with fluorescent rhodamine and FITC filters with excitation wavelength ranges of 534-558 nm and 450-490 nm, respectively. Monochrome images were captured by a Dage video camera (Newvicom, Wiesbaden, Germany) and were converted to pseudo-coloured images by Metamorph 6.1 software (Universal Imaging Corporation, Downington, PA, USA).
Statistics
The statistical significance of differences in means between treatment and control groups were assessed using unpaired t-tests and analysis of variance (ANOVA). P values < 0.05 indicated statistical significance.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SJ carried out the oxaliplatin and paclitaxel studies and drafted the manuscript. JS carried out the comparison of oxaliplatin and carboplatin. NJ carried out the cisplatin studies. VI provided technical support. BC participated in the design of the study and its coordination. JL and MM conceived of the study, and participated in its design and coordination and drafted the final manuscript. All authors read and approved the final manuscript.