The contribution of satellite glial cells to chemotherapy-induced neuropathic pain

Authors


  • Funding sources

    This work was supported by the European Community's 7th Framework Programme through the Marie Curie Initial Training Network Edu-GLIA, the Israel Cancer Association, the Israel Science Foundation (Grant No. 212/08), the US–Israel Binational Science Foundation (Grant No. 2007311) and the Hebrew University Center for Pain Research.

  • Conflicts of interest

    There are no personal or financial conflicts of interest in publishing these data.

Correspondence

Menachem Hanani

E-mail: hananim@cc.huji.ac.il

Abstract

Background

Chemotherapy-induced peripheral neuropathy is a serious side effect in cancer treatment, a major manifestation being neuropathic pain that can be debilitating and can reduce the quality of life of the patient. Oxaliplatin and taxol are common anti-cancer drugs that induce neuropathic pain by an unknown mechanism. We tested the hypothesis that satellite glial cells in dorsal root ganglia (DRGs) are altered in chemotherapy-induced peripheral neuropathy models and contribute to neuropathic pain.

Methods

Mice were injected with either oxaliplatin or taxol and examined at 7–30 days. Glial fibrillary acidic protein (glial activation marker) expression was determined by immunohistochemistry. Satellite glial cells in isolated DRG were injected with the fluorescent dye Lucifer yellow and the incidence of dye coupling among these cells that surround different neurons was quantified.

Results

Taxol or oxaliplatin increased glial fibrillary acidic protein expression in satellite glial cells. Gap junction-mediated coupling between satellite glial cells was increased by up to fivefold after oxaliplatin and by up to twofold after taxol. This is consistent with work on other pain models showing that augmented satellite glial cell coupling contributes to chronic pain. Administration of the gap junction blocker carbenoxolone to chemotherapy-treated mice produced an analgesic-like effect.

Conclusions

We propose that increased coupling by gap junctions is part of satellite glial cell activation, and that augmented coupling contributes to the lowering of pain threshold in oxaliplatin- and taxol-treated mice. We further propose that gap junction blockers may have potential in treating chemotherapy-induced neuropathic pain.

1. Introduction

Around a third of all cancer patients who undergo chemotherapy suffer from chemotherapy-induced peripheral neuropathy (CIPN) (Gutiérrez-Gutiérrez et al., 2010). Multiple classes of chemotherapeutic compounds induce CIPN, and its extent depends on the cumulative dose. CIPN is a major dose-limiting factor that interferes with the treatment schedule and can result in treatment cessation (Quasthoff and Hartung, 2002). A major manifestation of CIPN is neuropathic pain (Windebank and Grisold, 2008). CIPN can interfere with the daily activities of the patient, impairing their quality of life.

What's already known about this topic?

  • Chemotherapeutic agents are known to cause neuropathic pain in patients.
  • Increased gap-junctional coupling between satellite glial cells (SGCs) has been observed in dorsal root ganglia (DRGs) from several pain models and has been proposed to contribute to neuropathic pain.

What does this study add?

  • We found that taxol and oxaliplatin increased gap-junctional coupling among SGCs in mouse DRG.
  • Blocking gap junctions in vivo reduced mechanical hypersensitivity induced by these chemotherapeutic drugs.

Chemotherapeutic drugs are designed to kill rapidly dividing cells. However, it is clear that these drugs also inflict damage on sensory neurons. Some chemotherapeutic agents (e.g., taxanes) are designed to disrupt microtubules in the mitotic spindle of dividing cells, but these drugs may also disrupt microtubule-based transport in the axons of sensory neurons. Other chemotherapeutic drugs designed to block DNA replication in dividing cells (e.g., platinum compounds) have been shown to cause DNA cross-linking in dorsal root ganglion (DRG) neurons (McDonald et al., 2005).

Ectopic neuronal activity in the DRG is thought to be an important driver of pain caused by nerve injury or inflammation (Devor, 2009; Huang et al., 2010). The neuronal cell bodies in the DRG are surrounded by satellite glial cells (SGCs), which outnumber and tightly envelop them. Recently, SGCs have been suggested to be key modulators in chronic pain (Vit et al., 2006; Dublin and Hanani, 2007; Ohara et al., 2009; Huang et al., 2010). However, very little is known on the actions of chemotherapeutic agents on SGCs.

Sensory ganglia lie outside the blood–brain barrier and are densely vascularized by fenestrated capillaries, making the neurons and SGCs easily accessible to compounds in the circulation, including chemotherapeutic drugs (Jimenez-Andrade et al., 2008). Chemotherapeutic drugs show greater accumulation in sensory ganglia than in peripheral nerves (Cavaletti et al., 2000, 2001). It is therefore likely that these drugs would cause damage preferentially on these ganglia.

During pathological conditions, such as nerve injury or inflammation, SGCs demonstrate an altered phenotype similar to that seen in activated astrocytes, which includes increased expression of glial fibrillary acidic protein (GFAP) and synthesis of cytokines (Woodham et al., 1989; Elson et al., 2003; Takeda et al., 2007; Jasmin et al., 2010). SGCs are therefore said to undergo activation due to injury. Increased coupling by gap junctions between SGCs has been observed in several inflammatory pain and axotomy models (Hanani et al., 2002; Ledda et al., 2009; Huang et al., 2010). This is in agreement with reports on augmented expression of the gap junction proteins connexin 43 (Cx43) and Cx26 in the trigeminal ganglia after injury (Garrett and Durham, 2008; Ohara et al., 2008).

Here, SGCs in DRG obtained from taxol or oxaliplatin CIPN mouse models were studied. Taxol is a taxane compound commonly used to treat breast, ovarian and lung cancers. Oxaliplatin is a platinum compound and is one of the few drugs effective at treating metastatic colon cancer.

2. Methods

2.1 Animals and drug treatment

The experiments were approved by the Institutional Animal Care and Use Committee of the Hebrew University-Hadassah Medical School and adhere to the guidelines of the Committee for Research and Ethical Issues of International Association for the Study of Pain. We used Balb/c mice, 2–5 months old, of either sex, weighing 18–27 g. No differences between the sexes were observed in any of the following experiments. Oxaliplatin-treated mice received two intraperitoneal (i.p.) injections of oxaliplatin, 4 mg/kg (Tocris, Bristol, UK), dissolved in saline, given 3 days apart. Taxol-treated mice received two i.p. injections of taxol (Tocris), 18 mg/kg, given 3 days apart. Taxol was dissolved in a 1:1 mixture of Cremophor EL (Sigma, St. Louis, MO, USA) and ethanol and then diluted in saline (1:3) just prior to administration. Control mice were given two injections of the vehicle solution for either oxaliplatin or taxol 3 days apart.

2.2 Behavioural testing

Mice were placed in a clear plastic box on a wire mesh floor and were allowed to accustom to their new environment for at least 20 min before behavioural testing. Pain thresholds were assessed by observing withdrawal responses to mechanical stimulation of the plantar skin of hindpaws using von Frey hairs (Stoelting, Wood Dale, IL, USA). Hairs of 0.07–2 g were applied 10 times at intervals of 5 and 20 s in ascending order. The von Frey hairs were pressed against the plantar skin of hindpaw until the hair buckled and formed a U shape. Sharp retraction of the stimulated hindpaw marked a response. The threshold for withdrawal response (pain threshold) was 6 out of 10 responses. Care was taken not to stimulate the same point on the skin in succession. The right and left hindpaws were averaged. Data for each group were collected from five to six animals.

2.3 Immunohistochemistry

Animals were killed by CO2 inhalation, perfused transcardially with 0.1 mol/L phosphate-buffered saline (PBS, pH 7.3) containing heparin 5 IU/mL. The L4/5 DRGs were removed and placed in 4% paraformaldehyde for 90 min at room temperature. The DRGs were then washed in PBS before incubating them in 20% sucrose in PBS overnight before freezing in Tissue-Tek embedding medium (Sakura Finetek, Torrance, CA, USA). Sections were cut 12-μm-thick using a cryostat (Jung CM3000, Leica Microsystems, Wetzlar, Germany) and thaw mounted on SuperFrostPlus slides (Menzel, Braunschweig, Germany). Sections were washed in PBS and incubated with 50 mmol/L ammonium chloride for 30 min. Sections were washed again in PBS and then blocked in a solution containing 3% bovine serum albumin (BSA) in PBS with 0.3% Triton X-100 for 2 h at room temperature. Primary antibodies against GFAP (rabbit anti-GFAP, Dako, Copenhagen, Denmark) and glutamine synthetase (goat anti-glutamine synthetase, Santa Cruz, CA, USA) were diluted 1:400 and 1:200, respectively, in PBS containing 1% BSA and incubated overnight at 4 °C. Sections were washed in PBS and incubated with secondary antibodies, donkey anti-rabbit conjugated to DyLight 549-TFP ester and donkey anti-goat conjugated to Alexa Fluor 488 (Jackson ImmunoResearch, West Grove, PA, USA) diluted 1:400 in PBS with 10 μmol/L of the fluorescent dye 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) and 1% BSA for 2 h at room temperature. Finally, sections were washed in PBS. Controls omitted the primary antibody. Sections were imaged using a confocal microscope (Olympus FV-1000, Tokyo, Japan), equipped with an IX81 inverted microscope. A 40X (NA 1.3) oil immersion objective was used. DAPI was scanned using a 405-nm laser line for excitation and a 430–470-nm emission filter. DyLight 549 was scanned using a 543-nm laser line for excitation and a 560–620-nm emission filter. Overviews of DRGs were captured using an upright microscope (Axioskop FS2, Zeiss, Jena, Germany), equipped with fluorescent illumination and a digital camera (Pixera penguin 600CL, Los Gatos, CA, USA) connected to a personal computer.

Neurons and SGCs were easily distinguished using DAPI because neuronal nuclei were larger and paler compared with SGC nuclei. Neurons that were surrounded by GFAP-positive SGCs by more than 50% of their circumference were counted and expressed as a percentage of the total number of neurons present in the fields analysed. (This criterion was used because the SGC sheath can be partly very thin, and invisible under light microscopy). Data from each group were collected from three to four animals. Four randomly selected fields were analysed from each animal and averaged.

2.4 Intracellular labelling

Animals were killed with CO2 and DRG L4/5 was removed and attached to the bottom of a silicon rubber-coated dish using fine pins. The dish was placed on the stage of an upright microscope (Axioskop), equipped with fluorescent illumination and a digital camera. The dish was superfused with Krebs solution, which contained (in mmol/L): 118 NaCl, 4.7 KCl, 14.4 NaHCO3, 1.2 MgSO4, 1.2 NaH2PO4, 2.5 CaCl2 and 11.5 glucose; pH 7.3. Individual SGCs in DRGs were injected with the fluorescent dye Lucifer yellow (LY, Sigma), 3% in 0.5 mol/L LiCl solution from sharp glass microelectrodes, connected to an electrometer (model IR 283, Neuro Data Instruments Corp., New York, NY, USA). The dye was passed by hyperpolarizing current pulses, 100 ms in duration; 0.5–1 nA in amplitude at 5 Hz for 5 s to 2 min. The dye injections were made under visual inspection to allow cell identification (Huang et al., 2005). At the end of the injection of each cell, the coupling incidence was determined by the number of glial envelopes belonging to neighbouring neurons that were labelled as the result of the injection of a single SGC.

2.5 Gap junction blocker administration

In the experiments designed to observe the changes in the pain threshold, carbenoxolone (50 mg/kg dissolved in saline) and palmitoleic acid (25 mg/kg dissolved in 50% ethanol, administered in a volume of 50 μL) were injected (i.p.) into the CIPN mouse models, 1 week after the first oxaliplatin/taxol injection, 1 h before behavioural testing. The doses of carbenoxolone and palmitoleic acid were chosen on the basis of previously published work (Hanstein et al., 2010; Huang et al., 2010). Palmitoleic acid and carbenoxolone were purchased from Sigma.

In LY dye injection experiments designed to test whether carbenoxolone could block gap junctions in the DRG, carbenoxolone (100 μmol/L) was added to the Krebs solution and the experiment was carried out as described above.

To test whether the carbenoxolone reached DRG cells in vivo and blocked gap junctions in them, carbenoxolone (50 or 100 mg/kg, i.p.) was injected into the mice that had been treated with oxaliplatin 1 week before. The mice were killed 1 h after carbenoxolone was injected. Their L4/5 DRGs were then removed and LY dye injection experiments were carried out as described above.

2.6 Data analysis

Behavioural data were analysed using two-way analysis of variance (ANOVA). Dye coupling data were pooled for each group from multiple experiments and analysed using Fisher's exact text. This was performed because in different dye coupling experiments, different numbers of cells were injected and relatively small numbers of cells were injected per experiment. Immunohistochemical data were analysed using one-way ANOVA. p < 0.05 was considered as statistically significant. Values are expressed as mean ± standard error of the mean.

3. Results

3.1 Pain threshold

Oxaliplatin- and taxol-treated mice had a healthy and normal appearance and no animal died as a result of the treatment. We measured the pain threshold in oxaliplatin- and taxol-treated mice 1 week, 2 weeks and 1 month after the first injection. The pain threshold was significantly lower in treated mice compared with control mice. Oxaliplatin lowered pain threshold by 70% after 1 week, whereas taxol lowered pain threshold by 88% after 1 week. After 1 month, the pain threshold in oxaliplatin- and taxol-treated mice had risen to levels similar to those of control mice (Fig. 1).

Figure 1.

Changes in pain threshold in oxaliplatin- and taxol-treated mice. (A) Pain threshold in oxaliplatin-treated mice was significantly reduced at 1 week. (B) Pain threshold in taxol-treated mice was significantly reduced at 1 week. Data for each group were obtained from 6 mice. Bars represent the mean ± standard error of the mean; asterisks indicate p < 0.05.

3.2 Expression of GFAP in SGCs

Increased expression of GFAP is used as a criterion for SGC activation. GFAP is expressed by SGCs and Schwann cells in the DRG; however, SGCs can be distinguished from Schwann cells by their morphology and position around the neuron, and also by the presence of glutamine synthetase, which is absent in Schwann cells (Hanani, 2005). Co-staining of DRG from taxol-treated mice for GFAP and glutamine synthetase showed co-localization in SGCs (Fig. 2A–C).

Figure 2.

Glial fibrillary acidic protein (GFAP) expression is increased in satellite glial cells (SGCs) in dorsal root ganglion (DRG) of oxaliplatin- and taxol-treated mice. A DRG section from taxol-treated mouse showing co-localization of GFAP (red) and glutamine synthetase (GS, green) in SGCs (A–C). Arrows indicate examples of SGCs expressing GFAP and GS. 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) (blue) was used to stain nuclei; neuronal nuclei are paler and larger than nuclei of SGCs (A–C, H–K). Overviews of DRGs from mice treated with the vehicle for oxaliplatin (D), oxaliplatin (E), vehicle for taxol (F) and taxol (G) showing GFAP expression after 7 days. Representative images showing GFAP expression in DRG from control mice treated with the vehicle for oxaliplatin (H) and in mice treated with oxaliplatin after 7 days (I). Representative images showing GFAP expression in DRG from control mice treated with the vehicle for taxol (J), and in mice treated with taxol after 7 days (K). A few examples of neurons surrounded by GFAP expressing SGCs are labelled with ‘N’. Scale bar: 30 μm for A–C, 100 μm for D–G and 20 μm for H–K. Quantification of GFAP expression in SGCs in DRG from oxaliplatin-treated mice (L) and taxol-treated mice (M). Data for control groups and 1 week taxol- and oxal-treated groups were collected from 4 mice, data for 2-week- and 1-month-treated groups were collected from three to four mice. The asterisks indicate p < 0.05 compared with control. GFAP-IR, GFAP-immunoreactive.

Immunohistochemical analysis revealed a significant increase in GFAP expression in SGCs in DRG from oxaliplatin- and taxol-treated mice compared with control mice. Neurons that were surrounded by GFAP-immunoreactive (GFAP-IR) SGCs by more than 50% of their circumference were counted, and expressed as a percentage of the total number of neurons analysed. In control mice, about 15% of neurons were surrounded by GFAP-IR SGCs. In oxaliplatin- and taxol-treated mice, the number of neurons surrounded by GFAP-IR SGCs was increased over twofold compared with controls after 1 week. After 1 month, expression of GFAP in oxaliplatin- and taxol-treated mice had returned to levels similar to those in control mice (Fig. 2).

3.3 Gap junction-mediated coupling among SGCs

We injected LY into SGCs to assess the extent of SGC coupling in ganglia from oxaliplatin- and taxol-injected mice compared with control mice. We observed two types of dye coupling between SGCs: between SGCs surrounding a single neuron (see Fig. 3A) and between SGCs surrounding different neurons (see Fig. 3B). We decided to investigate the incidence of SGC coupling around multiple neurons as this has been shown to increase considerably in other mouse pain models (Ledda et al., 2009; Huang et al., 2010).

Figure 3.

Treating mice with oxaliplatin or taxol increased dye coupling between satellite glial cells (SGCs). (A) A LY-injected SGC is coupled to other SGCs only around the same neuron. This type of coupling, which is limited to SGCs surrounding a given neuron, is seen typically in ganglia from control mice. (B) Dye coupling between SGCs around different neurons. The incidence of this type of coupling is higher in ganglia from oxaliplatin- and taxol-treated mice than from control mice. The asterisks indicate the LY-injected cell. Quantification of dye coupling among SGCs in dorsal root ganglion (DRG) from oxaliplatin-treated mice (C) and taxol-treated mice (D). Data for each group were obtained from four to nine mice. The asterisks indicate p < 0.05 compared with control. Scale bars: 20 μm.

In ganglia from control mice the incidence of coupling was low. Both oxaliplatin and taxol increased SGC coupling significantly compared with controls. In oxaliplatin-treated mice, SGC coupling had increased fivefold after 1 week and threefold after 2 weeks (Fig. 3C). In taxol-treated mice, SGC coupling had increased over twofold after 1 week (Fig. 3D). One month after oxaliplatin and taxol treatments, SGC coupling was reduced to levels not significantly different from those of control mice (Fig. 3).

To test whether this coupling was mediated by gap junctions, we added 100 μmol/L of carbenoxolone, a gap junction blocker, to the bath solution and repeated the LY experiments on ganglia from oxaliplatin-treated mice 1 week after the first injection. We observed almost complete blockade of SGC coupling. Only 1 out of 42 injected SGCs was coupled.

3.4 Effect of gap junction blockers on pain threshold

As carbenoxolone blocked gap junctions in vitro, we wanted to test its effect on pain threshold in vivo in oxaliplatin- and taxol-treated mice. Carbenoxolone (50 mg/kg, i.p.) was injected into oxaliplatin- and taxol-treated mice 1 week after the first injection when pain threshold was observed to be at its lowest. Carbenoxolone raised the pain threshold in oxaliplatin- and taxol-treated mice to levels similar to those in controls (Fig. 4). When another gap junction blocker, palmitoleic acid (25 mg/kg, i.p.), was injected into taxol-treated mice 1 week after the first injection, pain thresholds were raised to levels similar to those in controls (Fig. 4).

Figure 4.

Gap junction blockers effect on pain threshold in chemotherapy-induced peripheral neuropathy mouse models. Carbenoxolone (CBX) increased pain threshold in oxaliplatin- (A) and taxol-treated mice (B) to levels similar to those in control mice. Similarly, palmitoleic acid (PA, 25 mg/kg) increased pain threshold in taxol-treated mice (B) to levels similar to those in control mice. Data for each group were obtained from five to six mice. Bars represent the mean ± standard error of the mean; asterisks indicate p < 0.05 compared with control and oxal/taxol + CBX/PA group.

To establish whether carbenoxolone blocks gap junctions in the DRG in vivo, LY dye injection experiments were carried out on DRGs from oxaliplatin-treated mice that had been injected with carbenoxolone (50 or 100 mg/kg, i.p.) 1 h before being killed. The LY dye injection experiments were carried out without carbenoxolone in the solution. Mice injected with 50 mg/kg of carbenoxolone showed 25% SGC coupling, which was not significantly lower than oxaliplatin-treated mice without carbenoxolone treatment. However, mice that were injected with 100 mg/kg carbenoxolone showed only 11% coupling, which was significantly lower than oxaliplatin-treated mice without carbenoxolone (Fig. 5).

Figure 5.

Intraperitoneal injection of carbenoxolone (CBX) to oxaliplatin-treated mice reduced satellite glial cell (SGC) coupling in the dorsal root ganglion (DRG). Quantification of dye coupling among SGCs in DRG from 1-week-oxaliplatin-treated mice without carbenoxolone and with carbenoxolone 50 mg/kg, and 100 mg/kg (i.p.). Data for each group were obtained from three to four mice. The asterisks indicate p < 0.05 compared with oxal group.

4. Discussion

In the current study, we examined SGCs in L4/5 DRG from oxaliplatin and taxol CIPN mouse models. We found that these chemotherapeutic drugs induced functional and biochemical changes in SGCs. An increase in gap junction-mediated SGC coupling was observed in both models, which was accompanied by increased expression of GFAP in SGCs, indicating glial activation. Administration of carbenoxolone, a gap junction blocker, to oxaliplatin- and taxol-treated mice produced an analgesic-like effect.

The conclusion on SGC activation is based on an analogy between SGCs and astrocytes (Watkins and Maier, 2002; Dublin and Hanani, 2007). Astrocytes are known to become ‘activated’ in response to nerve injury, inflammation or ischaemia. Activated astrocytes are characterized by hypertrophy, release of pro-inflammatory cytokines (IL-1, IL-6 and TNF-α), release of nitric oxide and prostaglandins, and up-regulation of the intermediate filaments GFAP and vimentin (Watkins et al., 2001; Pekny and Nilsson, 2005). Likewise, SGCs display increased expression of GFAP after neuronal injury or inflammation and undergo a number of changes similar to those seen in astrocytes, such as synthesis of cytokines (Elson et al., 2003; Takeda et al., 2007; Jasmin et al., 2010).

GFAP expression increased in both oxaliplatin and taxol CIPN models, indicating SGC activation. This is in keeping with previous studies where increased GFAP expression was seen in SGCs from a taxol-based CIPN model in rats (Peters et al., 2007). An increase in SGC coupling has been observed after damage to sensory neurons and was proposed to be part of SGC activation (Hanani et al., 2002; Cherkas et al., 2004; Dublin and Hanani, 2007). Here we showed for the first time that increased expression of GFAP is accompanied by augmented SGC coupling in the CIPN mouse models. This supports our previous suggestion that increased glial coupling is part of the activation process (Huang et al., 2010). Glial coupling was blocked by carbenoxolone, a broad-acting gap junction blocker, confirming that SGC coupling is mediated by gap junctions. This is the first demonstration that systemically administered drugs increase SGC coupling in the DRG.

One of the ways glial cells in the sensory ganglia transmit signals is through intercellular calcium waves (ICWs) via gap junctions and adenosine-5'-triphosphate acting on P2 receptors (Suadicani et al., 2010). This signalling has been shown to be bidirectional between SGCs and neurons (Suadicani et al., 2010). An increase in the number of gap junctions between SGCs, as seen in the CIPN models, would permit a greater spread of ICW in the ganglia. An increase in SGC sensitivity to ATP has been shown in mouse pain models (Kushnir et al., 2011). We hypothesize that an increase in SGC coupling combined with an increase in SGC sensitivity to ATP would result in augmented signalling in the ganglia, which would lead to hyperexcitability of neurons in the sensory ganglia, therefore contributing to chronic pain; see Kushnir et al. (2011) for discussion.

A possible explanation for the increase in SGC coupling seen in the CIPN models may be to enable more effective redistribution of potassium (K+) ions and other harmful substances away from active neurons. It is known that astrocytes in the central nervous system perform ‘spatial buffering’ (regulation of K+ ions) and it is presumed that SGCs also perform the same function (Hanani, 2005). Removing K+ from the perineuronal environment would reduce neuronal excitation and therefore contribute to the lowering of pain. Reducing Cx43 expression in injured rats was shown to diminish pain behaviour, whereas reducing Cx43 expression in non-injured rats was shown to increase pain behaviour (Ohara et al., 2008). This suggests that there is a beneficial effect of SGC coupling as well as an adverse one, and that in a pain state the adverse effect overrides the beneficial one to cause increased neuronal excitability.

When carbenoxolone, a gap junction blocker, was administered in vivo to oxaliplatin- and taxol-treated mice, tactile allodynia was abolished. Carbenoxolone does not cross the blood–brain barrier (Leshchenko et al., 2006), and therefore should act peripherally, blocking gap junctions in the DRG and elsewhere in the periphery. This conclusion was supported by dye coupling experiments on DRG from oxaliplatin-treated mice that had been injected with carbenoxolone (100 mg/kg, i.p.), showing that SGC coupling was significantly reduced (Fig. 5). This confirms that i.p. injected carbenoxolone could reach and block gap junctions in the DRG. These data also suggest that the effect of carbenoxolone is not easily reversed, which is in line with previous reports (Pan et al., 2007).

Carbenoxolone and other gap junction blockers have been shown to reduce pain behaviour in other pain models (Hanstein et al., 2010; Huang et al., 2010), suggesting that their effects are general to various chronic pain states. These results support our hypothesis that augmented SGC coupling contributes to DRG neuronal hyperexcitability and therefore to chronic pain. In a previous report, a variety of gap junction blockers were shown in vitro to reduce excitability of DRG neurons from a mouse inflammatory pain model and also to have analgesic actions (Huang et al., 2010). It should be mentioned that carbenoxolone is thought to have direct effects on neuronal activity. However, while one report claimed that carbenoxolone directly inhibited neuronal excitability (Rouach et al., 2003), another suggested that it increased neuronal activity (Jahromi et al., 2002). If carbenoxolone inhibits neuronal activity directly, the reduced pain behaviour observed with the drug may not be solely the result of gap junction blockade. However, palmitoleic acid, another gap junction blocker structurally unrelated to carbenoxolone, also abolished pain behaviour in taxol-treated mice (Fig. 4), further supporting the idea that the analgesic-like effect of carbenoxolone is the result of gap junction blockade and not due to its other effects (Burt et al., 1991; Huang et al., 2010). Importantly, when carbenoxolone is injected into control animals it has no influence on pain behaviour or on several other behavioural parameters (Hanstein et al., 2010). Therefore, it is likely that carbenoxolone acts largely when gap junctions are augmented, as occurs in several pain models, including CIPN.

Previous work on sensory ganglia in rodent CIPN models mainly focused on the neurons. Oxaliplatin was shown to cause selective atrophy of large CTR1 (a copper transporter) expressing neurons in the DRG (Ip et al., 2010), whereas taxol caused up-regulation of activating transcription factor 3 (a marker of cell injury) in DRG neurons and GFAP in SGCs (Peters et al., 2007). In addition to the multiple changes observed in sensory ganglia after treatment with chemotherapeutic drugs, degeneration of intraepidermal nerve fibres and increased activation of epidermal Langerhans cells were also observed in a CIPN model (Siau et al., 2006). Loss of epidermal innervation mimics sensory nerve axotomy, which is known to induce GFAP up-regulation in SGCs in DRG (Woodham et al., 1989). Therefore, a possible mechanism underlying chemotherapy-induced neuropathies could be via an indirect action on nerve terminals which could lead to SGC activation.

We observed some quantitative differences between the actions of oxaliplatin and taxol, with oxaliplatin being more potent in increasing SGC coupling than taxol. This could be due to the different nature of the two drugs as they are from different classes of compounds. It is not known if SGCs take up oxaliplatin, taxol or their derivatives, but it remains a possibility. The relatively lower increase in SGC coupling observed in the taxol model could be due to impaired trafficking of gap junction proteins (connexins) to the cell membrane, as taxol disrupts microtubule-based transport. In addition to its effects on microtubules, taxol was also shown to stabilize the intermediate filament vimentin (Vilalta et al., 1998), and it can be proposed that it also stabilizes the intermediate filament GFAP. This could explain the more prolonged effect of taxol on GFAP expression compared with that of oxaliplatin.

The doses used in our CIPN models were minimal compared with those reported in previous studies, where higher cumulative doses were used (Ta et al., 2006; Sakurai et al., 2009; Tatsushima et al., 2011). In addition, the doses we used were comparable to those used in human patients. Assuming that a mouse weighs 20 g and has a body surface area of 36 cm2 (Wu et al., 2011), an oxaliplatin dose of 4 mg/kg is equivalent to 22 mg/m2. This is considerably lower than the 85 mg/m2 dose typically given in humans (Maindrault-Goebel et al., 2000). The equivalent taxol dose of 18 mg/kg in mice is 100 mg/m2. This dose is lower than the taxol doses typically given in humans, which range from 135 to 250 mg/m2 (Wiseman and Spencer, 1998). This confirms that we used realistic doses in our models.

In summary, we have shown that both oxaliplatin and taxol cause an increase in SGC coupling, which was previously seen in other models of chronic pain, and that this was accompanied by an increase in SGC activation. Administration of carbenoxolone to oxaliplatin- or taxol-treated mice reduced pain behaviour. We propose that increased SGC coupling is part of SGC activation, and contributes to the lowering of pain threshold in the treated mice. We also propose that gap junction blockers may have potential in CIPN therapy.

Author contributions

R.A.W. was responsible for undertaking the experiments and analysis, while M.H. oversaw the project. The authors discussed the results and both contributed to the writing of the manuscript.

Acknowledgements

We thank Vered Arueti for her help with the immunohistochemistry.

Ancillary