Spinal microglia and neuropathic pain in young rats

1 Introduction

Neuropathic pain occurs in children but with lower incidence than in adults. Brachial plexus injury, which often causes devastating chronic neuropathic pain in adults, has minimal impact on neonates, with little evidence of chronic pain behaviour (^{Anand and Birch, 2002}). Similarly, facial neuralgias are rare in children (^{Grazzi et al., 2005}) and adults whose symptoms first present in childhood do not have the same therapeutic requirement as patients with later onset symptoms (^{Resnick et al., 1998}).

Considerable postnatal development occurs within nociceptive pathways over the first postnatal weeks (^{Fitzgerald, 2005}) and central mechanisms that underlie neuropathic pain in adults are unlikely to be fully functional in juvenile animals (^{Fitzgerald and Macdermott, 2005}). A recent study of peripheral nerve injury revealed that mechanical allodynia did not develop if injury was performed before three weeks of age. Importantly, pain symptoms were never detected in these rats even when they reached the age at which injury would normally evoke mechanical allodynia (^{Howard et al., 2005}).

Recent research has focused on proliferation and activation of spinal microglia in the pathogenesis of neuropathic pain following peripheral nerve injury (^{Colburn et al., 1997; Stuesse et al., 2000; Coull et al., 2005; Tsuda et al., 2005; Zhuang et al., 2005}). Microglia are the primary immunocompetent cell type within the CNS serving a major role in the immune response to tissue injury or infection and subsequent removal of cellular debris (^{Kreutzberg, 1996; Watkins et al., 2001}). Often described as existing in two states; ‘resting’ and ‘active’, microglia are able to switch from a surveillance role to a stimulated state mediating neuropathic conditions (^{Coull et al., 2005; Tsuda et al., 2005}). It is evident that conditions leading to activation of spinal microglia, be they physiological, as a consequence of peripheral nerve injury, or chemical, via exogenous application of endotoxins are capable of inducing neuropathic-like pain. Furthermore a transient neuropathic state in naïve rats can be induced by intrathecal injection of ATP-stimulated microglia (^{Tsuda et al., 2003; Coull et al., 2005}).

One possibility, therefore, is that the lack of allodynia in rat pup nerve injury arises from an immature microglial response and the aim of this study is to examine this hypothesis. Postnatal developmental changes in microglial activation have been reported following experimental status epilepticus (^{Rizzi et al., 2003}) and LPS-induced intracerebral inflammation (^{Lawson and Perry, 1995}). However, treatment of juvenile rats with LPS produces an immune challenge that alters the neuroimmune response (^{Boisse et al., 2004}) and affects baseline nociceptive responses as an adult (^{Boisse et al., 2005}) suggesting that the immune system has the capacity to respond to an immune challenge at early postnatal stages (^{Streit, 2001}).

We examined dorsal horn microglia, following peripheral nerve injury and direct intraspinal immune and excitotoxic challenges in young rat pups. The ability of ATP-stimulated microglia to induce mechanical allodynia in naive juvenile rats has also been examined. The results provide novel insights into the developmental profile of CNS neuronal/microglial interactions evoked by nerve damage and evidence for their critical importance in the postnatal development of neuropathic pain.

2 Methods

Experiments were performed on male Sprague–Dawley rats. All experimental procedures were specifically licensed and approved by the UK Home Office and conformed to the guidelines of the International Association for the Study of Pain (www.iasp-pain.org).

3 Spared nerve injury (SNI) surgery

Adult and 10-day-old (P10) rats were anaesthetised with a mixture of halothane/O₂ (3% for induction 2% for maintenance). Spared nerve injury (SNI) surgery was carried out as described elsewhere (^{Decosterd and Woolf, 2000}). Briefly, the left sciatic nerve was exposed at the mid-thigh level distal to the trifurcation and the tibial and common peroneal branches were tightly ligated with 5/0 silk and axotomised, leaving only the sural nerve intact. In sham-operated controls, the sciatic nerve was exposed as above but was not ligated or cut. Muscle and skin flaps were closed with suture and the animals recovered uneventfully. SNI and sham-operated pups were returned to their litter containing uninjured littermates on recovery.

4 Immunohistochemistry

Adult and P10 rats were given an overdose of Euthatal [100mg/ml] and perfused with 4% paraformaldehyde. The lumbar spinal cord was removed, post-fixed and stored in 30% sucrose in 0.1M phosphate buffer/0.02% sodium azide at 4°C. Immunohistochemical staining was performed on 40μm free-floating cryosections of L4/L5 spinal cord. The sections were blocked for 1h in TTBS (0.05M Tris saline, pH 7.4/0.3% Triton X-100) containing 5% normal goat serum (NGS) at room temperature. Sections were then incubated at 4°C (RT) for 72h with rabbit α-ionised calcium binding adapter molecule-1, Iba-1 (Wako, Japan) diluted 1:2000 in TTBS. Three 10min washes in 0.1M phosphate buffer were followed by a 2h incubation at room temperature with biotinylated goat anti-rabbit secondary antibody (Vector Laboratories, Burlingame, CA) diluted 1:500 in TTBS. A further three 10min washes in 0.1M phosphate buffer were followed by a 45min incubation at room temperature with Cy3-conjugated streptavidin (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) diluted 1:4000 in TTBS.

5 Real-time quantitative PCR

Ipsilateral and contralateral L4/L5 dorsal horn tissue was dissected and RNA extracted using Trizol reagent (Invitrogen) as per manufacturer's instructions. Quantitative real-time PCR was performed using the Sybr green detection system with primer sets designed on Primer Express. Specific PCR product amplification was confirmed using dissociation protocol. Transcript regulation was determined using the relative standard curve method as per manufacturer's instructions (Applied Biosystems). Relative loading was determined prior to RT with RNA spectrophotometry followed by gel electrophoresis and post-RT by amplification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). For each time point 4 samples of pooled tissue from 2 rats were analyzed.

6 Intraspinal injection of NMDA and LPS

Rats (adult and P10) were anaesthetised with a mixture of halothane/O₂ (3% for induction 2% for maintenance). An L4/5 laminectomy was performed and intraspinal injections were made into the dorsal horn on one side using a 10μl Hamilton syringe with a fixed 26S gauge needle (model 801 RN, Hamilton Bonaduz AG, Switzerland). Injections were either 20nM N-methyl-d-aspartate (NMDA) in 0.25μl of sterile saline (n=4 rats) or 0.5μl of 100μg/ml lipopolysaccharide (LPS, Sigma 0111:B4 from Escherichia coli) in sterile saline. Control animals were injected with the same volume of sterile saline. After the procedure, the wounds were closed, and the pups returned to their litter. Animals were left for three days in the NMDA study and for 6, 24 and 72h in the LPS study before being terminally anaesthetised with Euthatal [100mg/ml] and tissue prepared for immunohistochemistry as above and qPCR.

7 Microglial culture and activation

A mixed primary culture was prepared from neonatal Sprague–Dawley rat cortex under standard conditions as described previously (^{Nakajima et al., 1992}). Cortices were removed and minced. The tissue was then digested with trypsin for 10min at 37°C. DNase was added and the cells separated by centrifugation. The resulting pellet was triturated in DMEM and the cells plated on 75cm² culture flasks. Cultures were maintained for 10–14 days in DMEM with 10% foetal bovine serum with regular medium changes. Microglia were separated from the primary culture by gentle shaking of the flask and re-plated on plastic Petri dishes. The cells were removed from the dish surface using a cell scraper and collected in 100μl PBS. Subsequently, the cell density of microglia was measured using a cell counter and the volume of PBS adjusted to give a final density of 1000 cells/10μl. This method produces microglia cultures of >95% purity (^{Nakajima et al., 1992}). The cells were subsequently incubated with ATP (50μM) or an equivalent volume of PBS for one hour prior to intrathecal injection (^{Tsuda et al., 2003}).

8 Intrathecal injection of LPS and microglial cultures

Adult and P10–P21 rats were anaesthetised with a mixture of halothane/O₂ (3% for induction 2% for maintenance). Intrathecal injections were performed at the L4–5 level in naïve rats using a 10μl Hamilton syringe (26S gauge, model 801 RN), with a fixed needle. The procedure for LPS was as described elsewhere (^{Cahill et al., 2003}). Briefly, rats were given an intrathecal “priming dose” of LPS (2.0μg) (Adult 30μl; P10 5μl) allowed to recover and then, 24h later, anaesthetised again and given an intrathecal injection of 20μg LPS as a “challenge dose”. The procedure for intrathecal injection of cultured microglia was a single injection in 5μl in P10–18 rats and 7μl in P19–P21 rats.

9 Behavioural testing

Behavioural testing was performed in those animals that were intrathecally injected with LPS or microglial cultures. Flexion withdrawal reflex thresholds were established in both groups to punctate mechanical stimulation of the dorsal surface of the hindpaw using calibrated von Frey nylon filaments (Stoelting, Woodvale, Il). Briefly, filaments with a pre-calibrated bending force were applied sequentially to the dorsal surface of the hind paw, 10 times at intervals of one second. Response threshold was defined as the von Frey hair which produced reflex paw withdrawal in 5 out of 10 applications. Rats were tested for at least one day prior to intrathecal injections to habituate them to the procedure and to establish a baseline. Thresholds were tested again immediately before anaesthetising for intrathecal injections (both ‘priming’ and ‘challenge’ injections, in the case of LPS and then once every hour for five hours following intrathecal injection).

10 Results

10.1 Nerve injury evokes significantly less microglial activation in young compared to adult rats

Microglia were labelled using a microglial marker IBA-1, a microglia specific calcium binding protein. Immunostaining was carried out in L4/5 spinal cord sections 7 days after spared nerve injury (SNI) in postnatal day 10 (P10) and adult rats and examined using confocal microscopy. In adult rats (n=4), a significant increase in Iba-1 immunoreactivity was consistently observed within the ipsilateral spinal cord 7 days following nerve injury (Fig. 1A). Increased expression was observed throughout the L4/5 dorsal horn and motoneurone pool but was concentrated in the somatotopic region of termination of the injured sciatic nerves and in the more superficial laminae of the dorsal horn. In contrast, only a small increase in Iba-1 immunoreactivity was observed in the ipsilateral spinal cord 7 days following SNI at P10 (n=4) (Fig. 1A). Quantitative PCR was used to examine the expression of integrin-α M, Iba-1, MHC-IIDMα and MHC-IIDMβ in the L4/L5 dorsal horn. All four microglial markers were significantly increased in the ipsilateral dorsal horn of adult and P10 rat pups 7 days after SNI (Fig. 1B; n=8 rats per experimental group). However, the magnitude of the increase observed in the adult ipsilateral dorsal horn was significantly greater than in the P10 in all cases.

10.2 Excitotoxic injury activates microglia in both young and adult rat dorsal horn

To examine microglial responsiveness in the young rat spinal cord, the effect of intraspinal injection of an excitotoxic concentration of N-methyl-d-aspartate (NMDA) in P10 and adult rats upon microglial activation was compared. Three days following injection, Iba-1 immunoreactivity was significantly increased around the site of NMDA injection in the dorsal horn of the spinal cord in both adult and P10 rats (n=4 for each age group) (Fig. 2A). Iba-1 immunoreactivity was unaltered in saline controls (Fig. 2A).

10.3 An immune challenge activates microglia in both young and adult rat dorsal horn

To further test microglia responsiveness in the young spinal cord, we challenged the immune system by injecting intraspinal LPS in P10 and adult rats. Intraspinal injection of LPS in both the adult and P10 dorsal horn led to strong upregulation of Iba-1 immunostaining at 3 days and at both ages with the gross morphological changes associated with reactive microglia being observed throughout the spinal cord (n=4 for both age groups) (Fig. 2A). Control intraspinal injections of saline caused minor changes in microglial morphology due to the needle injury (Fig. 2). QPCR analysis examining the specific time-course of expression of mRNA for microglial markers integrin-α M, Iba-1, MHC-IIDMα and MHC-IIDMβ demonstrated that the increased expression of microglial markers in the adult dorsal horn microglial response following intraspinal LPS was significantly greater, with an earlier onset and longer duration than the neonatal response (Fig. 2B).

10.4 Intrathecal injection of LPS causes mechanical allodynia in young rats

As both NMDA and LPS caused microglial activation at P10, we wished to test whether they were also able to produce mechanical allodynia. Using the previously reported LPS ‘priming’ and ‘challenge’ regime, which leads to clear mechanical allodynia in adult rats (^{Cahill et al., 2003}), we tested the effect of a ‘challenge’ dose of intrathecal LPS upon hindpaw mechanical flexion reflex thresholds in younger rats. Fig. 3A shows that both P10 (n=7) and P21 (n=7) rat pups display significant mechanical allodynia which peaks at 60–90min post-injection and has recovered by 150min. Nevertheless, when expressed as a percentage change from baseline, it is clear that the effect is significantly greater at P21 than at P10.

10.5 Intrathecal injection of ATP-stimulated microglia does not result in tactile allodynia in young rats

Having established the pattern of microglial activation in young spinal cord, we next wished to test whether microglia, when activated, are able to cause the same increase in excitability in young animals as they do in adults. Intrathecal injection of exogenously ATP-activated neonatal microglia is known to cause allodynia in adult animals (^{Tsuda et al., 2003}), We therefore tested the effect of administering these cells upon the hindpaw flexion reflex thresholds of young animals. Intrathecal injection of ATP-stimulated microglia induced significant tactile allodynia in P21 rats (n=8 per group) (Fig. 4C) and P16 rats (n=6 per group) (Fig. 4B), but had no effect at P10 (n=11 per group) (Fig. 4A). The time course and extent of allodynia at P16 and P21 was similar to that described in previous studies in adult rats (^{Tsuda et al., 2003}) and was clear despite a non-significant fall in baseline threshold, presumably arising from repeated testing. Tactile allodynia was apparent at 3h post-injection and was greatest at 5h when behavioural testing was terminated. No significant effect was observed in control experiments involved injecting cultured microglia that had not been previously stimulated with ATP (Fig. 4A–C). Fig. 4Aii, Bii and Cii illustrate the peak (5h post-injection) effect in control and injected animals between at the three ages.

11 Discussion

These results show a number of important differences in dorsal horn microglial responses between young and adult animals. While nerve injury in adults triggers a marked activation of dorsal horn microglia at 7 days. The same injury performed at 10 days of age induces only a weak microglial response. This was observed using both morphologically using immunostaining and by quantitative analysis of mRNA expression of well-established microglial markers. Integrin-α M, Iba-1, MHC-IIDMα and MHC-IIDMβ are all well-known markers of microglia and increased expression of these markers is widely believed to be associated with microglial activation. The spinal microglial response that we observed to peripheral nerve injury in adults is consistent with previous reports (^{Eriksson et al., 1993; Coyle, 1998; Tsuda et al., 2005}) although the extent and time course of the changes varies with the type of injury (^{DeLeo et al., 1997; Colburn et al., 1999}).

One possible reason for the weak microglial response to nerve injury performed at P10 is that the innate immune response is immature and spinal microglia are less able to respond robustly to immune challenges when compared to the adult. When LPS, (a bacterial endotoxin known to cause robust activation of the immune system leading to neuroinflammation, and shown to activate microglia in the adult central nervous system) (^{Zielasek and Hartung, 2006}) was injected into the dorsal horn of P10 and adult rats, Iba-1 immunoreactivity increased significantly when examined at 72h post-injection. This significant microglial response was qualitatively indistinguishable between both P10 and adult animals. Furthermore, spinal injection of an excitotoxic dose of NMDA, also known to cause intense microglial activation in the adult spinal cord (^{Gomes-Leal et al., 2004}), induced an equally strong microglial reaction at P10. When the specific microglial reaction following P10 intraspinal LPS was examined quantitatively using qPCR, the increase in Iba-1 mRNA mirrored the immunohistochemistry results, the three other markers were not significantly changed. Therefore, the neonate have an innate immune response and can respond other stressors such as intraspinal LPS and NMDA, although the specific microglial response is greater in the adult.

The weak microglial response to nerve injury at P10 might also be due to differences in the neural signalling. The spinal cord receives no direct injury when peripheral nerves are damaged, not even the nearby inflammation associated with spinal ligation surgery (^{Djouhri et al., 2006}). Therefore the microglial reaction observed must arise from a signal from the damaged and/or undamaged afferent neurones. It seems likely that, following SNI at P10, the specific and peripherally driven signals for the neuropathic immune response, which may be electrical or chemical or both, are either absent or ineffective in activating microglia in young animals. The absence of microglial regulation by peripheral nerves could be related to normal developmental plasticity. The postnatal period is a time of axonal pruning and apoptosis as well as axonal sprouting and growth and microglia, as phagocytes are likely to be responsible for the clearance of cellular debris during this period. In addition microglia secrete many factors, including neurotrophic factors, that can initiate neurite and axonal growth (^{Prewitt et al., 1997; Nakajima and Khosaka, 2004}) and have been shown to modify the non-permissive environment of the CNS to allow subsequent axonal growth to occur (^{David et al., 1990; Lazarov-Spiegler et al., 1998}). A permissive role of microglia for nerve growth and synaptic reorganisation has been suggested for the postnatal development of visual callosal projections during the period of widespread elimination of exuberant callosal axons (^{Rochefort et al., 2002}). In the spinal cord, nerve injury before P10 causes sprouting of nearby intact afferent terminals at the expense of those that have been damaged (^{Fitzgerald et al., 1990; Shortland and Fitzgerald, 1994}), marked plasticity of peptidergic afferents (^{Reynolds and Fitzgerald, 1992}), reduced dorsal horn cell somatodendritic growth (^{Fitzgerald and Shortland, 1988}), reorganised receptive fields (^{Shortland and Fitzgerald, 1991}) and increased synaptic input from damaged A fibres and undamaged C fibres (^{Shortland and Molander, 2000}). One might speculate that a less reactive neuroimmune response is required for this type of plasticity to take place.

A further possibility is that spinal neurones in the neonatal dorsal horn are less responsive to the excitatory cytokines and other factors that are released by the limited microglial response that does occur. Following peripheral nerve injury in the adult, mechanical allodynia is accompanied by an increased expression of the purinergic P2X₄ receptor on spinal microglia (^{Tsuda et al., 2003}). Ongoing activation of these receptors is necessary for the maintenance of the neuropathic pain state. Intrathecal application of ATP-stimulated microglia to naïve adult rats produces a similar allodynic effect to that induced in the peripheral nerve-injured rat. Upon stimulation with ATP, microglia release BDNF which disrupts the anion balance of dorsal horn neurons, resulting in a state of increased excitability within the spinal cord, resulting in behavioural hypersensitivity (^{Coull et al., 2005}). We tested the sensitivity of neonatal spinal cord neurons directly by introducing exogenous ATP-activated microglia into the spinal cord and found that there was no allodynic response to these activated microglia at P10. Responsiveness only begins at P16 and by P21 the response was similar to that seen in adults. Interestingly, by-passing primary afferent involvement through the direct intrathecal application of ATP-stimulated microglia was effective in inducing an allodynic response approximately two weeks earlier than a peripheral nerve injury showing that by P16 dorsal horn neurones are capable of responding to signals from activated microglia. Before that time responsibility for the lack of behavioural response may lie either with the dorsal horn neurons or with the failure of exogenous ATP stimulated microglia to activate resident microglia at P10. Importantly, intrathecal LPS produces behavioural allodynia at P10, demonstrating the inherent ability of young dorsal horn neurones to become sensitised by immune activation although the response is greater in the adult. LPS acts via the functional CD14-Toll-like receptor 4 complex proteins expressed on several cell types including microglia, neutrophils, monocytes, macrophages and astrocytes in neonatal white and grey matter. Activation of this receptor evokes rapid cell death in neighbouring oligodendroglia via calcium- and cytokine-mediated pathways (^{Sherwin and Fern, 2005}). The effects of intrathecal LPS therefore cannot be ascribed solely to microglial activation.

The lack of microglial activation by peripheral nerve injury is consistent with the absence of behavioural allodynia in young rat pups following nerve injury. In a study where SNI rat pups were tested with a range of mechanical stimuli over several postnatal weeks and compared with sham controls, it was evident that no allodynia developed in rats before 3–4 weeks of age (^{Howard et al., 2005}). This lack of response is unique to nerve injury as animals of the same age can show a robust allodynia and hyperalgesia to a range of peripheral inflammatory stimuli, such as formalin, carageenan, complete Freund's adjuvant and mustard oil (^{Marsh et al., 1999; Torsney and Fitzgerald, 2002; Ren et al., 2004}). It is likely that the lack of microglial response following nerve injury at P10 is linked to this lack of allodynia.

In conclusion it is clear that there are significant developmental changes in the neuroimmune response to potentially damaging stimuli in the postnatal spinal cord. We have shown that, at P10, microglia are not substantially activated by nerve injury and that this is correlated with the previously reported lack of behavioural responses to nerve injury in young animals. While the lack of microglial activation is partly a reflection of the overall weaker immune response to a range of stimuli, it is also likely that the specific neural signals indicating nerve damage are absent in the immature nervous system because a neuropathic pain state can be induced in the younger animals by a direct injection of ATP-stimulated microglia into the intrathecal space of the lumbar spinal cord. Further investigation of the postnatal development of spinal microglial/neuronal interactions will help to unravel the relative roles of primary afferent, spinal microglia and dorsal horn neurons in the ontogeny of neuropathic pain.

Acknowledgements

The authors would like to thank the Medical Research Council & Neuroscience Canada for their support. The authors would also like to thank Professor Clifford Woolf for his kind assistance.