Cisplatin induced sensory neuropathy is prevented by vascular endothelial growth factor-A

Abstract

Increased patient survival is a mark of modern anti-cancer therapy success. Unfortunately treatment side-effects such as neurotoxicity are a major long term concern. Sensory neuropathy is one of the common toxicities that can arise during platinum based chemotherapy. In many cases the current poor understanding of the neurological degeneration and lack of suitable analgesia has led to high incidences of patient drop out of treatment. VEGF-A is a prominent neuroprotective agent thus it was hypothesised to prevent cisplatin induced neuropathy. Systemic cisplatin treatment (lasting 3 weeks biweekly) resulted in mechanical allodynia and heat hyperalgesia in mice when compared to vehicle control. PGP9.5 sensory nerve fibre innervation was reduced in the plantar skin in the cisplatin treated group versus vehicle control mice. The cisplatin induced sensory neurodegeneration was associated with increased cleaved caspase 3 expression as well as a reduction in Activating Transcription Factor 3 and pan VEGF-A expression in sensory neurons. VEGF-A₁₆₅b expression was unaltered between vehicle and cisplatin treatment. rhVEGF-A₁₆₅a and rhVEGF-A₁₆₅b both prevented cisplatin induced sensory neurodegeneration. Cisplatin exposure blunts the regenerative properties of sensory neurons thus leading to sensory neuropathy. However, here it is identified that administration of VEGF-A isoform subtypes induce regeneration and prevent cell death and are therefore a possible adjunct therapy for chemotherapy induced neuropathy.

Introduction

Chemotherapy is routinely used alone or in combination with other therapies to treat patients following cancer diagnosis. However, chemotherapy-induced peripheral neuropathy (CIPN) is a major dose limiting effect of many chemotherapeutic agents, contributing to the further deterioration of the quality of life for cancer patients. CIPN is typically seen in about 40% [1-3] (reported in up to 95%) of patients treated with platinum based treatments, including cisplatin [1]. These typically lead to severe neurotoxicity which results in sensory and motor neurodegeneration, and is classified as pain associated with touch, ‘pins and needles’ and numbness as well as loss of motor coordination [3]. These symmetrical sensory complications are routinely identified in the extremities e.g. fingers and toes, with symptoms persisting from weeks/months to many years. Moreover, the symptoms of CIPN may remain after discontinuation of treatment with platinum compounds [1,4,5]. Currently, there are limited treatments available for CIPN. Therefore it is of great importance to identify potential therapeutic agents that would help in either preventing or treating the CIPN.

Cisplatin belongs to the first generation of platinum drugs and has been in use for the last 40 years in the treatment of a number of advanced and metastatic cancers. Platinum based compounds lead to DNA cross linkage ultimately preventing mitosis, thus suppressing tumour growth. However, cisplatin also causes damage to other cell types such as sensory neurons. It is now widely regarded that such treatments cause the described sensory symptoms, although the mechanisms by which these cause neuropathy are still undefined. A number of rodent models (including oxaliplatin [6,7] and cisplatin [8,9] administration) have been developed, which display classical hallmarks of human chemotherapy-induced sensory neurodegeneration.

It is widely accepted that upon a nervous insult such as a nerve injury, a number of neurotrophins and growth factors are up-regulated due to their integral role in neuronal survival and regeneration [10-12]. Recent evidence has identified that the vascular endothelial growth factor family-A (VEGF-A) family has a strong role in neuroprotection. This family consists of two sister protein groups that are unique, differing solely in the final 6 amino acids of the C terminus, which give distinct functional roles to each isoform family [13]. The neuroprotective actions of VEGF-A have been demonstrated extensively in motor degenerative disease [14,15]. In addition, VEGF-A_xxxa is strongly associated with sensory neuronal development, protection and regeneration [16-19], with increased expression in dorsal root ganglia (DRG) sensory neurons upon traumatic nerve injury [20]. Recent work now highlights the role of VEGF-A_xxxb in neuroprotection [12].

This study tested the hypothesis that there is a reduction in the regenerative ability of sensory neurons upon chemotherapy exposure, which can be reversed by treatment with neurotrophic VEGF isoforms. The experimental aims were to target the impact of cisplatin on sensory neurons and determine the role of the VEGF-A family in response to such an insult. Cisplatin led to a reduction in regenerative markers including VEGF-A. Treatment with either VEGF-A₁₆₅a or VEGF-A₁₆₅b prevented both cisplatin induced cell death and reduction in neurite growth.

Materials and methods

Studies in mice treated with or without cisplatin

Adult C57bl6 male (~30 g, 10 total) mice were used. All procedures were performed according to UK Home Office legislation in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and associated guidelines (2012), and were approved by the University of Nottingham AWERB. Animals were provided food and water ab libitum.

Nociceptive behavioral assays

All treatments administered were given via intraperitoneal (i.p.) injection. Treatments were biweekly i.p. injections of either vehicle (phosphate buffered saline) or cisplatin (2 mg/kg [21,22]). Animal weight was monitored regularly. The experiment was terminated at the end of week 3. Prior to onset of nociceptive behavioral testing, mice were habituated to the testing procedure which involved transport and handling, and to the testing environment including mesh and glass-floored Perspex boxes, for 10 minutes before the start of testing. Mechanical withdrawal thresholds were defined as previously described [23]. A total of five applications of each von Frey hair (Linton) were applied to the plantar surface of both hind paws. Von Frey hairs were used to obtain data ranging from 0% to 100% nociceptive withdrawal responses, which were then used to construct force response curves to determine 50% withdrawal threshold. Nociceptive responses to heat were also determined using the Hargreaves test [20]. Both hind paws were tested three times to provide a mean withdrawal latency. A rest period between each hind paw stimulation was enforced to prevent sensitization to the heat stimulation. These parameters were recorded at weekly intervals for duration of the study.

Cell culture

An immortalized embryonic sensory neuronal cell line, 50B11s, was provided as a gift from Drs Ahmet Hoke and Damon Lowes. The cells were prepared and cultured to 80% confluence in a T75 flask as previously described [24] in neurobasal media (Gibco) with Fetal bovine serum (Sigma-Aldrich), 2% B27 (Life Technologies), 0.2% L-glutamine (Sigma-Aldrich), 0.2% (11 mM) glucose (Fisher) as supplements. Prior to differentiation and experimentation the cells were left for 24 hrs in the culture plate. For neuronal growth and cleaved caspase III assays cells were plated at low density (5000 cells per well) on ethanol sterilized coverslips in a 24 well plate. 75 μM forskolin (Abcam) was used to differentiate the 50B11s and was applied after 24 hrs along with other experimental agents; vehicle (phosphate buffered saline), 2.5 nM recombinant human (rh) VEGF-A₁₆₅a (R&D systems), 2.5 nM rhVEGF-A₁₆₅b (R&D systems), 1 nM nerve growth factor (NGF), cisplatin (a range of doses as outlined in figure legends given in μg/ml). Following initial experiments to determine dose, 5 μg/ml cisplatin was used. Cells were incubated under experimental conditions for 24 hours, after which they were fixed with 4% paraformaldehyde (PFA) prior to imaging.

Neurite growth

For neurite growth measurements cells grown in 24 well plates and were imaged using a Nikon Eclipse Ti microscope using differential interference contrast. Each well was imaged with an ×20 objective and for each well 5 random fields of view were captured and used for analysis. Image J was used to analyze images to determine neurite growth [25].

Protein expression

Immunofluorescence

Plantar skin from the hind paws of all mice was taken and placed into 4% PFA overnight, cryoprotected in 30% sucrose solution overnight, all at 4°C. Tissue was frozen in OCT and stored at -80°C until processing. Thin sections (20 μm) of plantar skin from hind paws of the mice were cut on a cryostat and placed onto slides (VWR international). Sections were washed once in phosphate-buffered saline (PBS) for 10 minutes. The skin sections were then subjected to a 10 minute wash in PBS-0.2% Triton X-100 after which they were incubated in blocking buffer (10% FBS, 5% BSA, 0.2% PBS-Triton) for 1 hr at room temperature. The primary antibody (anti-rabbit anti-polyclonal protein gene product 9.5 (PGP9.5), 1 in 100, Ultraclone) was added in blocking solution and incubated at 4°C overnight. Three PBS washes were performed followed by incubation in PBS 0.2%-Triton X-100 containing biotinylated anti-rabbit IgG (1 in 500, Jackson) at room temperature for 2 hrs. Three further washes were performed and streptavidin Alexa Fluor 555 anti-rabbit antibody (Invitrogen) at a concentration of 1:1000 in 0.2% PBS-Triton was added for 1 hr at room temperature. Samples were then subjected to three 5 minute washes with PBS, coverslipped with Fluorsave (Calbiochem) and imaged on a Leica TCS SPE confocal microscope using Leica application suite. 5 random images were taken per skin sample (per field of view from ×40 objective/25600 μm²). A mean was taken from the number of PGP9.5+ve nerve terminals identified at the dermal: epidermal border of each image to provide a mean value per animal. These were then used for subsequent analysis for treatment groups.

Immunocytochemistry

Cultured cells were treated with 4% PFA for 15 minutes and as with the above protocol (Immunofluorescence) were prepared in block solution. Cells were incubated in primary antibody (rabbit anti-cleaved caspase 3, 1:500, Cell Signalling Technology) at 4°C overnight, and with secondary antibody (Alexa Fluor 555 conjugated anti-rabbit antibody, 1:1000, (Invitrogen) in 0.2% PBS-Triton X-100 for 1 h at RT. Cell nuclei were labeled with DAPI (1:2000, Invitrogen) for five minutes at room temperature. Washes and cell imaging were performed as described above. The slides were then imaged using a fluorescent microscope (Nikon Eclipse Ti). Five random images of immunofluorescence were taken per well.

Protein extraction and quantification

50B11 cells were seeded in 10 cm petri dishes and grown to 70% confluence after which they were differentiated with 75 μM Forskolin ± 5 μg/ml cisplatin. After 24 hours of treatment, protein was extracted from the cells. The media from each dish was removed and centrifuged. The supernatant was discarded and the resulting pellet was kept. 150 µl lysis buffer (Radioimmuno Precipitation Assay (RIPA) buffer supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF), 20 μg/ml Protease Inhibitor cocktail, 50 mM sodium fluoride and 1mM sodium orthovanodate (Na₃VO₄), all reagents from Sigma-Aldrich, UK), was added to each dish, and the cells were collected. The lysis buffer was collected in Eppendorf tubes and left on ice for 30 minutes with occasional agitation. The samples were sonicated on ice for 15 minutes and the protein concentration was subsequently quantified using a Bradford assay (Bioad).

Western blotting

100 μg of each protein was loaded in a 4% stacking/12% separating gel and separated by SDS-PAGE at 70 V until the ladder reached the resolving gel, then at 120 V. The proteins were then transferred onto polyvinyldifluoride (PVDF, Biorad) membrane by wet transfer at 90V for 1 hour 30 minutes. The membrane was incubated in 5% milk powder in Tris buffered saline (TBS)-Tween 0.1% for 30 minutes at room temperature with agitation. Primary antibodies, rabbit anti-cleaved caspase 3 (1:1000 Cell Signalling Technology; #9664), rabbit anti-total caspase III (Cell Signaling Technology; #9662), rabbit anti-Pan VEGF (A20, 1 μg/ml, Santa Cruz Bio Technology; sc-152), mouse anti-VEGF165b (2 μg/ml, Abcam; ab-14994), rabbit anti-Activating Transcription factor 3 (ATF3; 2 μg/ml, Santa Cruz Bio Technology; sc-128) and rabbit anti-α/β tubulin (1:500 Cell Signalling; #2148) antibodies were diluted in blocking solution. The membrane was incubated in the primary antibodies overnight at 4°C with agitation. The membranes were washed three times in TBST-0.1% before incubation with secondary antibodies (Licor donkey anti-rabbit and anti-mouse antibodies 1:10000) in TBST-0.1% 1% BSA. After 90 minutes incubation in the secondary antibodies, the membranes were washed thrice with TBST-0.1% before visualizing on the Licor Odessey.

Statistical analysis

Data were acquired and analysed using Microsoft Excel, Image J and GraphPad Prism 6. The results obtained are presented as means with standard error of the mean. Behavioural assays were recorded and analysed as outlined in the methods. No significant difference was observed between the hind paws within groups (P = 0.1724, two way ANOVA between hind paws and groups), therefore in all instances both hind paws per animal were used as replicates in the analysis. A two way ANOVA with post-hoc Sidak’s multiple comparison tests was used to determine any statistical difference of treatment over time. For quantification of CC3+ve cells the total number of cleaved caspase 3 positive cells (CC3+ve) was counted based upon colocalision with DAPI stain. For neurite length analysis, Image J was calibrated to a microscope graticule and neurite length was manually determined using Image J measuring tools. For both in vitro cell based assays and in vivo IENF PGP9.5+ve nerve innervation (per field of view from ×40 objective/25600 μm²), values were determined from 5 random images acquired per sample (i.e. well or tissue sample) and a mean determined. Western blot densitometry was measured using Image J gel plugin. Differences in protein loading were corrected using densitometry of the control protein (i.e. tubulin or actin). Data obtained in in vitro assays, PGP9.5+ve nerve fibre innervation and western blot densitometry were analysed using Mann Whitney and Kruskal Wallis tests for two and multiple group comparisons respectively.

Results

Biweekly administration of cisplatin in mice led to a neuropathic pain phenotype (Figure 1). Compared to gender/age matched sham (vehicle treatment) injected mice, those treated with biweekly i.p. 2 mg/kg cisplatin showed a significant drop in mechanical withdrawal threshold (Figure 1A; Week 3 Sham = 1.59 g ± 0.09 vs. Cisplatin = 1.23 g ± 0.08) and a reduction in heat nociceptive response latencies (Figure 1B; Week 3 Sham = 9.46 s ± 0.17 vs. Cisplatin = 6.11 s ± 0.63). Cisplatin did not induce weight loss in the treatment group compared to the sham group (Figure 1C; Week 3 Sham = 28.08 g ± 0.79 vs. Cisplatin = 25.42 g ± 0.6). Furthermore, there was a significant reduction in intra-epidermal sensory nerve innervation (Figure 2A; Week 3 Sham = 6.14 ± 0.39 PGP9.5+ve terminals/field of view vs. Cisplatin = 2.38 ± 0.8 PGP9.5+ve terminals/field of view) in the cisplatin group (Figure 2C) compared to the vehicle injected mice (Figure 2B).

Figure 1.

Cisplatin induces neuropathic pain behaviour. (A) Intraperitoneal administration of cisplatin (biweekly 2 mg/kg) to adult male mice led to a significant reduction in mechanical nociceptive withdrawal threshold indicative of mechanical allodynia and (B) decrease in withdrawal latency to heat (heat hyperalgesia) versus the control group (i.p. vehicle) (*P < 0.05, **P < 0.01 two way ANOVA with post-hoc Bonferroni tests, n = 5 per group). (C) Cisplatin treatment did not lead to a significant reduction in body weight when compared to the age/gender matched sham group.

Figure 2.

Intraepidermal nerve fibre innervation is lost as a result of cisplatin treatment. (A) Three weeks of cisplatin treatment resulted in a significant reduction in intra-epidermal nerve fibre innervation in the plantar skin/per field of view (PGP9.5 = red, *P < 0.05 Mann Whitney U test). (B, C) Representative examples of PGP9.5 positive nerve terminal staining in (B) control and (C) cisplatin groups (scale bar = 20 μm).

To isolate the actions of cisplatin on sensory neurons, an immortalised nociceptive sensory neuronal cell line, 50B11s, was used [24]. To determine in vitro markers of sensory neurodegeneration, the degree of cell stress/death (cleaved caspase 3) and neurite length, 50B11s were treated with cisplatin in increasing concentrations (0 µg/ml-10 µg/ml; Figure 3) for 24 hrs. Increasing concentrations of cisplatin led to a dose dependent increase in cleaved caspase 3 expression (Figure 3A, 3E, 3F; 0 µg/ml = 0.0% ± 0.0, 10 µg/ml = 72.06% ± 3.09). This was accompanied by a reduction in neurite growth, with decreased total neurite growth per cell (Figure 3B; 0 µg/ml = 234.4 μm ± 20.47 vs. 10 µg/ml = 30.57 μm ± 12.3), a reduction in mean neurite length (Figure 3C; 0 µg/ml = 108.9 μm ± 3.65 vs. 10 µg/ml = 20.85 μm ± 5.86) and the percentage of cells with neurites (Figure 3D; 0 µg/ml = 63.37 ± 2.8 vs. 10 µg/ml = 2.96 ± 1.25). All subsequent experiments were carried at sub-maximal dose of 5 µg/ml as there was also a significant loss of neuronal growth and high degree of neuronal compromise.

Figure 3.

Cisplatin treatment inhibits neurite outgrowth and increases cleaved caspase in immortalised sensory neurons. (A) Increasing doses of cisplatin (given in µg/ml) induced a significant increase in cleaved caspase 3 expression. (B) Cisplatin prevented neurite growth leading to a reduction in total neurite growth per cell, (C) reduced average neurite length per cell and (D) reduced percentage of cells with neurites. (E) Representative examples of cleaved caspase 3 expression (CC3 = Red, DAPI=Blue) in control conditions (0 μg/ml) and (F) the presence of 10 μg/ml cisplatin. (**P < 0.01, ***P < 0.001 one way ANOVA with post-hoc Bonferroni tests, scale bar = 100 μm).

In protein extracted from cells treated for 24 hrs with cisplatin (sub-maximal dose 5 µg/ml) there were also elevated levels of cleaved caspase 3/total caspase (Figure 4A, 4B). Additionally, cisplatin treatment significantly reduced ATF3 expression compared to the vehicle group (Figure 4C, 4D), and was associated with the suppression of regenerative capacity as shown by reduced neurite outgrowth (Figure 3). The expression of pan-VEGF-A_xxx (total VEGF-A isoforms) was significantly reduced compared to the vehicle treated group (Figure 4E, 4F; VEGF-A165) in the cisplatin group (5 µg/ml), but expression of the alternatively spliced variant VEGF-A₁₆₅b was unaltered by cisplatin treatment (Figure 4E, 4F).

Figure 4.

Sensory neurons have altered expression of pro- and anti-degenerative molecules following cisplatin treatment. (A) The immortalised sensory neuronal cell line (50B11) showed increased expression of cleaved caspase 3 (corrected for total caspase 3 expression) following cisplatin (5 µg/ml) treatment. (B) Representative Western blot of cleaved caspase expression. (C) 50B11 cells showed reduced expression of the neuroregenerative marker Activating Transcription Factor 3 (ATF3) following cisplatin (corrected for loading against actin expression). (D) Representative Western blot of ATF3 expression. (E) Total (pan) Vascular Endothelial Growth factor-A expression was decreased by cisplatin treatment, but VEGF-A₁₆₅b levels did not change (corrected for loading against tubulin. (F) Representative Western blots of pan-VEGF and VEGF-A₁₆₅b expression. (A, C and D all *P < 0.05 Mann Whitney test, n = 4).

Treatment of 50B11 cells with exogenous rhVEGF-A₁₆₅a alone (without cisplatin) for 24 hrs led to robust neurite growth compared to vehicle treated cells (Figure 5A; Vehicle = 117.1 μm ± 4.88 vs. rhVEGF-A₁₆₅a = 211 μm ± 12.1; representative images C & E), as did the archetypal sensory neuronal growth factor NGF (Figure 5A; positive control NGF = 157.4 μm ± 10.8; representative images D). Furthermore, rhVEGF-A₁₆₅a led to a significant increase in total neurite outgrowth compared to NGF (Figure 5A). Treatment with rhVEGF-A₁₆₅b did not induce increased total neurite growth compared to vehicle treated cells (Figure 5B; vehicle = 145.1 μm ± 4.72 vs. rhVEGF-A₁₆₅b = 158.2 μm ± 6.48; representative images F). Incubation with either rhVEGF-A₁₆₅a or rhVEGF-A₁₆₅b did not affect cleaved caspase 3 expression (Figure 6; vehicle = 5.94 ± 1.55% vs. rhVEGF-A₁₆₅a = 7.64 ± 1.48% vs. rhVEGF-A₁₆₅b = 3.65 ± 1.08%) versus vehicle treated cells. Thus VEGF165a acted as a growth factor for 50B11 cells, whereas VEGF165b did not.

Figure 5.

Vascular Endothelial Growth Factor-A isoforms induce sensory neuronal neurite growth. (A) VEGF-A₁₆₅a administration resulted in a greater total neurite outgrowth per cell than either the archetypal trophic factor, NGF or vehicle. (B) In contrast, VEGF-A₁₆₅b did not lead to increased neurite growth in 50B11 neurons. (C-F) Representative images of 50B11 neurons grown in the presence of (C) vehicle, (D) NGF, (E) VEGF-A₁₆₅a and (F) VEGF-A₁₆₅b. (**P < 0.01, ***P < 0.001, one way ANOVA with post-hoc Bonferroni tests, scale bar = 50 μm).

Figure 6.

Vascular endothelial growth factor-A isoforms does not affect cleaved caspase 3 expression in immortalised sensory neurons. (A) There was no effect of VEGF-A isoforms on cleaved caspase 3 expression in 50B11 neurons. Representative images of 50B11 neurons cultured with either (B) vehicle, (C) VEGF-A₁₆₅a and (D) VEGF-A₁₆₅b. (CC3 = Red, DAPI = Blue, scale bar = 100 μm).

To determine whether exogenous VEGF-A (either VEGF-A₁₆₅a or VEGF-A₁₆₅b) could prevent cisplatin-induced neuronal damage in vitro, rhVEGF-A₁₆₅a or rhVEGF-A₁₆₅b were co-administered with cisplatin for 24 hrs. Cisplatin treatment (5 µg/ml) again led to a significant reduction in neurite outgrowth (Figure 7A; control (untreated normal media) = 181.6 μm ± 7.92 vs. Cisplatin 5 µg/ml + Vehicle = 97.72 μm ± 6.79) and increased caspase 3 expression (Figure 7B; control (untreated normal media) = 9.29% ± 2.7 vs. Cisplatin 5 µg/ml + Vehicle = 76.22% ± 4.15). Administration of rhVEGF-A₁₆₅a or rhVEGF-A₁₆₅b to cisplatin treated cells led to pronounced neuroprotection, preventing both the cisplatin-induced reduction in total nevurite length (Figure 7A; Cisplatin 5 µg/ml + rhVEGF-A₁₆₅a = 180.7μm ± 9.74 vs. Cisplatin 5 µg/ml + rh VEGF-A₁₆₅b = 163.6 μm ± 9.58) and cleaved caspase expression (Figure 7B; Cisplatin 5 µg/ml + rhVEGF-A₁₆₅a = 31.34% ± 5.89 vs. Cisplatin 5 µg/ml + rhVEGF-A₁₆₅b = 36.21% ± 5.92). However, neither rhVEGF-A₁₆₅a (Figure 7E) or rhVEGF-A₁₆₅b (Figure 7F) treatment completely prevented cisplatin induced cleaved caspase 3 expression (Figure 7B), with slight increases above control conditions.

Figure 7.

VEGF-A isoforms reverse the cisplatin-induced changes in neurite outgrowth and cleaved caspase expression. 50B11 neurons were cultured in 5 μg/ml cisplatin for 24hrs, which led to a reduction in (A) total neurite growth per cell and (B) increased cleaved caspase 3 (CC3 = Red, DAPI = Blue) expression (representative images (C) normal media and (D) cisplatin 5 μg/ml). Administration of either VEGF-A₁₆₅a or VEGF-A₁₆₅b prevented cisplatin induced (5 μg/ml) (A) reduction in total neurite growth. In addition, (E = representative image VEGF-A₁₆₅a) or (F=representative image VEGF-A₁₆₅b) treatment attenuated cisplatin induced cleaved caspase 3 expression (B) (***P < 0.001 one way ANOVA with post-hoc Bonferroni test). However, neither (E) VEGF-A₁₆₅a nor (F) VEGF-A₁₆₅b completely prevented cisplatin induced cell death compared to normal media alone (one way ANOVA with post-hoc Bonferroni test, *P < 0.05, **P < 0.01, ***P < 0.001, scale bar = 100 μm).

Discussion

Chemotherapy induced neuropathy is highly prevalent in cancer patients and survivors [3] and is a dose limiting side effect that can impact upon the continuation of treatment regimes, ultimately on patient survival and on cancer survivor quality of life. We show here that cisplatin treatment results in chronic pain phenotypes in mice, with evidence of mechanical allodynia and heat hyperalgesia. Cisplatin-induced neuropathic pain may occur due to suppression of the endogenous capacity for neuroregeneration and neuroprotection of the peripheral sensory neuron, as endogenous growth factor (VEGF-A_xxxa) and regeneration marker (ATF3) expression are reduced by cisplatin in sensory neurons in vitro. Administration of VEGF-A isoforms can provide neuroprotection and neurotrophic support to these neurons, inhibiting the effects of cisplatin.

Sensory neuropathy is induced by a host of anti-cancer treatments e.g. platinum derived agents, vinca alkaloids. Symptoms range from exacerbated evoked painful sensations, and pain resulting from touch or non-noxious temperature change, through to pins and needles and numbness (paresthesia). The onset of neuropathy can occur immediately following the start of chemotherapy treatment, and most unfortunately, the symptoms can persist not only for the duration of treatment but also past it’s termination, which impacts greatly on the survivors quality of life. Critically, despite not being directly related to the disease, chemotherapy-induced neuropathy weighs heavily on the patient, carers and family. The additional and on-going burden of an iatrogenic medical complication can lead to emotional disturbances arising as a result of reduced quality of life often resulting from limited mobility and reduced sleep quality [26,27]. These contribute to the well-documented increased incidence of depression and anxiety in cancer patients, with patient catastrophization, leading to enhanced perceived pain [28,29]. Furthermore, there are very few available analgesic drugs that can relieve cisplatin-induced pain in the majority of patients [30].

Typically the sensations associated with peripheral sensory neurodegeneration are as described above, but the degeneration itself can also be demonstrated by loss of intra-epidermal sensory nerve fibre (IENF) innervation in the skin [31] and by alterations in sensory nerve conduction velocity [32]. These hallmarks of peripheral sensory neuropathy are apparent in humans but to allow further investigation of the mechanisms of these changes, a number of rodent models are now being characterised [8]. Cisplatin administration in rodents leads to prominent sensory alterations including mechanical and cold hypersensitivity [33,34] with progression to hypoalgesia as neurodegeneration occurs. Our findings demonstrate that this regime of cisplatin leads to a pronounced sensory neuropathy in adult mice as previously reported [21,22], including IENF terminal loss and altered nociceptive behaviour, comparable to reports from previous studies [8] providing a suitable model for the study of novel therapies.

Platinum-based chemotherapy treatment results in significant neuronal damage and degeneration in the peripheral sensory nervous system, with many reports of sensory nerve regression and axonal dysfunction with accompanying neuropathic pain [35]. However, the underlying mechanisms and causes of neuropathy are still unclear. Platinum based treatments induce neuropathy in rodent models but there are contrasting data showing that DRG sensory neurons do not necessarily express all common markers of sensory neuronal damage or regeneration (e.g. ATF3) in response to this cellular stress or insult. Oxaliplatin induces cleaved caspase 3 expression in DRG sensory neurons [12] and regression of neurite outgrowth [36], and pharmacological inhibition of caspase activity blunted pain behaviours induced by oxaliplatin and cisplatin [7]. We demonstrate a pronounced reduction in the expression of the widely accepted neuronal damage marker and inducer of sensory neuronal regeneration, ATF3 [10,37] following cisplatin treatment. In contrast, ATF3 expression in DRG neurons is increased in response to cisplatin treatment [9] or unchanged following exposure to paclitaxel and oxaliplatin [6,38], in animals that displayed on-going neuropathic pain behaviours and alterations in sensory neuronal function typified by increased ongoing neuronal activity and alterations in conduction velocity [39]. In general, previous in vitro data are consistent with our in vivo observations of epidermal sensory nerve regression indicating that sensory neurons lose their normal neuroregenerative capability in response to cisplatin induced peripheral nerve damage.

Here we demonstrate that cisplatin treatment induced a significant neurodegeneration in vivo and in vitro. In vitro, this was accompanied by a reduction of total VEGF-A expression, a potent neuro-protective and regenerative agent [12,18,40]. Studies to date have demonstrated that VEGF-A has prominent actions on neurons, promoting survival and protection. In a number of rodent models of neuropathic pain (traumatic [18] and diabetic neuropathy [40]) VEGF-A can prevent sensory neuropathy. Interestingly, in vitro, total expression of VEGF-A was reduced, but expression of VEGF-A_xxxb was unchanged following cisplatin treatment, which suggests a reduction in the relative expression of VEGF-A_xxxa isoforms. Administration of a specific VEGF-A_xxxb neutralising antibody exacerbated oxaliplatin-induced neuronal toxicity in vitro [12] indicating that endogenous VEGF-A_xxxb levels may be maintained during chemotherapy exposure, and may offer a degree of neuroprotection, as supported by our finding that exogenous VEGF-A₁₆₅b was neuroprotective. Consideration of the independent neuroprotective functions of alternatively spliced isoforms is of great importance, as many well-established anti-cancer drugs are targeted against all VEGF-A isoforms or the functional receptor, VEGF receptor 2. These drugs were developed to inhibit VEGF-A₁₆₅a-mediated angiogenesis and tumour growth [41], but also block the actions of VEGF-A_xxxb isoforms, which have different physiological and pathological actions [20]. Blockade of VEGF-A signalling leads to the suppression of tumour growth and improved survival in some cancers, however these patients also demonstrate pronounced neuropathy [42,43]. This implies that the effects of these treatments on important VEGF-A-mediated neural signalling pathways may contribute to the advancing neuropathy that ultimately results in termination of treatment [44]. Consideration of the function and expression of known and emerging neuroprotective and regenerative growth factors such as VEGF-A₁₆₅b, is now crucial when designing and using such anti-cancer therapy, especially in combination with platinum-based chemotherapy. Understanding the functions that loss of endogenous growth factor support has in the adverse effects of such treatments may provide avenues to nullify those treatment-limiting side effects.

Disclosure of conflict of interest

None.