Pronociceptive changes in response properties of rostroventromedial medullary neurons in a rat model of peripheral neuropathy

Techniques for producing neuropathy

The unilateral axotomy and ligation of the tibial and common peroneal nerves was performed under pentobarbitone anaesthesia (50 mg/kg i.p.) as described in detail earlier (Decosterd & Woolf, 2000). Briefly, the skin of the lateral surface of the thigh was incised and a section made directly through the biceps femoris muscle exposing the sciatic nerve and its three terminal branches. Following ligation and removal of 2–4 mm of the distal nerve stumps of the tibial and common peroneal nerves, muscle and skin were closed in two layers. In sham-operated animals, the surgical procedure was identical, except that the tibial and common peroneal nerves were not ligated or sectioned.

Behavioural verification of neuropathy

Development of hypersensitivity was verified behaviourally in animals habituated to the experimental conditions 1–2 h daily for 2–3 days. For assessment of tactile allodynia, the hind limb withdrawal threshold was determined by stimulating the sural nerve area in the hind paw of the operated limb with monofilaments. The calibrated series of monofilaments used in this study produced forces ranging from 0.16 to 15 g (North Coast Medical, Inc. Morgan Hill, CA, USA). The monofilaments were applied to the foot pad with increasing force until the rat withdrew its hind limb. The lowest force producing a withdrawal response was considered the threshold. The threshold for each hind paw of each rat was based on three separate measurements and the median of these values was considered to represent the threshold. Threshold values ≤1 g were considered to represent hypersensitivity. It should be noted that the currently used strain of rats delivered by Harlan (Horst, The Netherlands) has an exceptionally low withdrawal threshold to monofilament stimulation in the baseline (unoperated) condition: in 10 unoperated control animals, the lowest withdrawal threshold was only 4 g, and therefore the criterion for hypersensitivity was set to as low as ≤1 g in this study.

Electrophysiological recordings

For electrophysiological recordings, the anaesthesia was induced by pentobarbitone at a dose of 50 mg/kg i.p. and the animal was placed in a standard stereotaxic frame according to the atlas of Paxinos & Watson (1998). Anaesthesia was maintained by infusing pentobarbitone (15–20 mg/kg/h). The level of anaesthesia was frequently monitored by observing the size of the pupils and by assessing withdrawal responses to noxious stimulation. When necessary, the infusion rate of pentobarbitone was increased. The rats were spontaneously breathing. A warming blanket was used to maintain body temperature within the physiological range. Peripheral perfusion was checked by evaluating the colour of ears and extremities. The skull was exposed and a hole drilled for placement of a recording electrode in the RVM. The desired recording site in the RVM was 1.8–2.3 mm posterior from the ear bar, 0.0–0.9 mm lateral from the midline, and 8.9–10.7 mm ventral from the dura mater.

Single neuron activity was recorded extracellularly with lacquer-coated tungsten electrodes (tip impedance 3–10 MΩ at 1 kHz) and then amplified and filtered using standard techniques. Data sampling was performed with a computer connected to a CED Micro 1401 interface and using Spike 2 software (Cambridge Electronic Design, Cambridge, UK).

Actual recordings did not start until the animal was under light anaesthesia; that is, the animals gave a brief withdrawal response to a noxious pinch, but the pinch did not produce any longer-lasting motor activity, and nor did the animals have spontaneous limb movements. Neurons were classified on the basis of their response to noxious pinch of the tail with a haemostatic clamp. This stimulus was painful when applied to the finger of the experimenters. Neurons giving excitatory responses to pinch were considered ON-cells, those giving inhibitory responses were considered OFF-cells and neurons showing no or only a negligible (<10%) change in their discharge rates as a response to pinch were considered NEUTRAL-cells. This classification scheme of medullary neurons was modified from that described by Fields et al. (2006). A noteworthy difference is that we did not verify whether pinch-evoked responses of RVM neurons were associated with spinal reflex responses as in the original classification scheme (Fields et al., 2006). Therefore, the populations of ON- and OFF-cells in this study may not be identical with those in a study in which cells are classified strictly according to the classification scheme of Fields et al. (2006). Our previous results suggest, however, that there is only little difference in the classification of RVM neurons whether or not spinal reflex responses are concurrently measured in lightly anaesthetized animals (Pertovaara et al., 2001).

Peripheral test stimuli

In electrophysiological experiments, cold stimuli (peak stimulus temperature, 4 °C; baseline temperature, 35 °C; rate of stimulus temperature decrease, 4 °C/s; duration of the peak temperature of 4 °C, 10 s) were applied with a feedback-controlled Peltier device (LTS-3 Stimulator; Thermal Devices Inc., Golden Valley, MN, USA) to the lateral side of the hind paw innervated by the sural nerve. Whereas some models of neuropathy may be associated with significant skin temperature changes that provide a confounding factor to assessment of thermal responses, particularly when using radiant heat for stimulation (Luukko et al., 1994), applying thermal stimuli with a contact thermostimulator and starting from a standard adapting temperature reduces the possibility that a neuropathy-associated change in skin temperature will influence the results.

ON- and OFF-neurons in the RVM typically have very large receptive fields covering almost the whole body and allowing testing of not only responses evoked by stimulating the operated limb but also responses evoked by the nonoperated region within the same neurons. In this study, responses evoked by stimulating the tail and the viscera were assessed to observe potential nerve injury-induced changes outside of the injured region. Mechanical stimulation of the skin consisted of applying a haemostatic clamp to the tail for 5 s. When applied to the experimenter's hand, this stimulus produced a painful pinch sensation. Noxious visceral stimulation consisted of colorectal distension (CRD) at a noxious intensity (80 mmHg) (Ness et al., 1991). CRD was applied for 10 s by inflating with air a 7–8-cm flexible latex balloon inserted transanally into the descending colon and rectum. The pressure in the balloon was controlled by an electronic device (Anderson et al., 1987).

When the stimulus-evoked responses were analysed, the baseline discharge frequency recorded during a corresponding period just before the stimulation was subtracted from the discharge frequencies determined during stimulation; that is, positive values represent excitatory responses evoked by peripheral stimulation and negative ones inhibitory responses.

Course of the study

Four groups of animals were included in the electrophysiological study: (i) sham group tested 1 week after operation; (ii) sham group tested 8 weeks after operation; (iii) SNI group tested 1 week after operation; and (iv) SNI group tested 8 weeks after operation. In each of these groups, behavioural assessment of sensitivity to monofilament stimulation was performed before the start of the electrophysiological experiment.

After induction of anaesthesia, the microelectrode was lowered to the RVM. After a single cell had been found, its spontaneous activity was recorded for 2–3 min. Then, cold stimulation was applied to the sural nerve area in the operated limb followed by pinch of the tail and CRD at 1-min intervals. The testing procedure, including the order of testing different submodalities of nociception, was the same in all experimental groups. To minimize serial effects, every other animal tested in this series belonged to the sham group and every other to the SNI group.

At the completion of the study, an electrolytic lesion was made in the recording site, the animals were given a lethal dose of pentobarbitone and the brains were removed for verification of recording and microinjection sites.

Statistics

Data are presented as mean ± SEM. For assessment of differences in incidence of various types of neurons in different experimental conditions, the chi-squared test was used. Two-way anova followed by the Student–Newman–Keuls test was used for assessing differences in neuronal responses between the experimental conditions. P < 0.05 was considered to represent a significant difference.

Spontaneous activity

The spontaneous discharge rate of ON-cells was significantly changed from 1 week to 8 weeks postoperatively (F_1,69 = 8.9, P < 0.005; Fig. 2A), and the change in the discharge rate varied with the experimental group (Sham vs. SNI animals; F_1,69 = 4.8, P < 0.04). One week following the operation, the spontaneous discharge rate of ON-cells was significantly higher in the SNI than in the Sham group. By the 8-week postoperative time point, the spontaneous discharge rate of ON-cells was significantly decreased in the SNI group only. Consequently, the spontaneous discharge rate of ON-cells 8 weeks following the operation was no longer higher in the SNI than in the Sham group, but was, if anything, the opposite.

The spontaneous discharge of the OFF-cells tended to increase from the 1-week to the 8-week postoperative time point (F_1,69 = 14.8, P < 0.0005; Fig. 2B). The spontaneous discharge rate of OFF-cells was significantly lower in the SNI group than in the Sham group (F_1,69 = 8.5, P < 0.005; Fig. 2B), independent of the postoperative time point (1 week vs. 8 weeks; F_1,69 = 14.8, P < 0.0005).

Peripherally evoked responses

Application of cold (4 °C) to the sural nerve area in the operated hind limb evoked markedly stronger excitatory responses in ON-cells of the SNI than of the Sham group (F_1,87 = 116, P < 0.0001; 3, 4), independent of the postoperative time point (F_1,87 = 0.67). Cold-evoked responses of ON-cells did not vary with elapsed time from the surgery (1 week vs. 8 weeks; F_1,87 = 0.9). Cold stimulation induced stronger inhibitory responses in the OFF-cells of the SNI than in the Sham group (F_1,53 = 64.6, P < 0.0001; Fig. 4B). Postoperative time (1 week vs. 8 weeks) did not influence the magnitude of the cold-evoked inhibitory response in OFF-cells (F_1,53 = 0.8).

Noxious tail pinch produced significantly stronger excitatory responses in ON-cells of the SNI than of the sham group (F_1,87 = 54, P < 0.0001; 5, 6), and this difference in the response magnitude was significantly larger at 8 weeks than at 1 week after the operation (F_1,87 = 17.4, P < 0.0001). There were no significant differences in pinch-evoked inhibitory responses of OFF-cells between the SNI and the Sham groups (F_1,53 = 3.2; Fig. 6B). Pinch-evoked inhibitory responses in OFF-cells, however, were stronger at 8 weeks than at 1 week after the operation (F_1,53 = 6.7, P < 0.02), independent of nerve injury (F_1,53 = 0.4).

CRD at a noxious intensity (80 mmHg) evoked significantly stronger excitatory responses in ON-cells of the SNI than of the Sham group (F_1,87 = 7.2, P < 0.01; 6, 7), and this difference in the response magnitudes was significantly larger at 8 weeks than at 1 week following the operation (F_1,87 = 10.7, P < 0.005). One week following the operation, ON-cell responses evoked by CRD were minor ones and not different between the experimental groups, whereas at 8 weeks following operation, the CRD-evoked ON-cell response was significantly increased in the SNI group only (see Fig. 6C for detailed results of post hoc tests). In OFF-cells, the inhibitory response evoked by CRD was not significantly different between the SNI and Sham groups (F_1,53 = 2.4; Fig. 6D), and the magnitude of the response evoked by CRD in OFF-cells was not significantly different when assessed 1 week vs. 8 weeks postoperatively (F_1,53 = 1.8).

Recording sites

Figure 8 shows the recording sites in the medulla.

Spontaneous activity and peripherally evoked responses

Spontaneous discharge rate of RVM ON-cells, which are known to have an excitatory effect on nociceptive transmission (Fields et al., 2006), was significantly increased 1 week but not 8 weeks after nerve injury. The spontaneous discharge rate of RVM OFF-cells, which have an inhibitory effect on nociceptive transmission (Fields et al., 2006), was decreased both 1 week and 8 weeks following nerve injury. Taken together, these findings suggest that both ON- and OFF-cell activities in the RVM promote neuropathic hypersensitivity in a tonic fashion during the first week following nerve injury. At a later phase of neuropathy, however, only the OFF-cell activity has a tonic pronociceptive effect, as a result of its decreased activity.

When assessing responses of RVM cells to peripheral stimulation, we focused on mechanical and cold stimulation, because clinical studies indicate that hypersensitivity to mechanical stimulation and cold are frequent and prominent symptoms after traumatic nerve injuries, whereas hyperalgesia to heat occurs only occasionally under neuropathic conditions (Scadding & Koltzenburg, 2006). Importantly, hypersensitivity to cold and mechanical stimulation is also a prominent finding in rats with the SNI model of neuropathy (Decosterd & Woolf, 2000). Because very little is known about the possible influence of somatic neuropathy on visceral sensations, we also determined responses of RVM cells to CRD.

The results indicate that SNI produced a significant hypersensitivity to cold in both ON- and OFF-cells throughout the observation period of 8 weeks. It should be noted, however, that the increase of excitatory responses to cold was considerably stronger in ON-cells than the increase of inhibitory ones in OFF-cells. Possibly, the SNI-induced reduction in the spontaneous activity of OFF-cells limited the amount of a further stimulus-evoked inhibition of OFF-cells. Responses to cutaneous pinch and visceral stimulation were markedly enhanced in ON-cells 8 weeks but not 1 week following nerve injury, whereas OFF-cell responses to pinch or visceral stimulation were not influenced by SNI. The present finding that hypersensitivity to peripheral stimulation was observed predominantly in presumably pronociceptive ON-cells of the RVM is in line with the hypothesis that a positive feedback loop involving ON-cells in the RVM is involved in promoting neuropathic hypersensitivity to peripheral stimulation. This hypothesis is further supported by the earlier behavioural findings indicating that nerve injury-induced hypersensitivity to mechanical stimulation (Pertovaara, 2000; Porreca et al., 2002) and cold (Urban et al., 2003) is dependent on descending facilitatory influence from the RVM. Moreover, it is noteworthy that although behaviourally assessed mechanical hypersensitivity develops within 24 h in the SNI model of neuropathy, it may not be fully developed until the second postoperative week (Decosterd & Woolf, 2000), whereas hypersensitivity to cold is fully developed within a week (Allchorne et al., 2005). These behavioural findings parallel the present neurophysiological results with ON-cells. However, when considering the potential behavioural significance of SNI-induced changes in stimulus-evoked responses of RVM neurons in the present study, one should note that assessments of neuronal responses to noxious peripheral stimulation were not accompanied by simultaneous assessments of corresponding behavioural responses. Moreover, behavioural responses of SNI animals to two of the currently used stimuli (pinch of an uninjured tail and CRD) have not been assessed in any previous study.

Comparison with previous studies

In line with the present findings, it has been shown earlier that ON-cells in the RVM have a pronociceptive role in acute inflammation as indicated by the association of their discharge rate with heat hypersensitivity induced by mustard oil (Kincaid et al., 2006; Xu et al., 2007). Previous studies addressing the contribution of RVM cells to neuropathic hypersensitivity have given partly contradictory results. It has been reported that chemical ablation of RVM cells expressing the µ-opioid receptor both prevents and reverses spinal nerve ligation-induced experimental neuropathy (Porreca et al., 2001). Because ON-cells are the only RVM neurons responding directly to µ-opioid agonists (Heinricher et al., 1992), the selective reversal of neuropathic hypersensitivity by ablation of medullary cells expressing µ-opioid receptors suggests, in line with the present results, that ON-cells promote neuropathic hypersensitivity. On the other hand, the results of our previous electrophysiological study (Pertovaara et al., 2001) suggested that spinal nerve ligation-induced neuropathy of 2–3 weeks' duration may not produce as marked pronociceptive changes in response properties of RVM cells as the SNI model of neuropathy in the present study or acute inflammation in other studies (Kincaid et al., 2006; Xu et al., 2007). Among possible explanations of the difference between the results of our present and earlier study (Pertovaara et al., 2001) are the differences in the genetic background of the animals, model of neuropathy, and postoperative time point of study. Interestingly, the sciatic constriction-induced model of neuropathy had no significant influence on response properties of RVM cells giving excitatory responses to noxious stimulation (Luukko & Pertovaara, 1993), whereas complete denervation of the sciatic area, as opposed to incomplete denervation in the present study, produced an enhancement of excitatory responses of RVM cells to peripheral stimulation of the adjacent intact skin area (Pertovaara & Kauppila, 1989). Together, these findings support the hypothesis that the magnitude or time course of the change in the response properties of RVM cells may vary with the model of peripheral neuropathy, with models involving sectioning of the peripheral nervous system resulting in strong neuropathic changes of the supraspinal pain control system. As anatomical feedback loops implicated in the crosstalk between the brainstem and spinal cord are essential for pain modulation (Almeida et al., 2006), the drastic changes associated with the afferent barrage of nociceptive messages after nerve fibre sectioning may have a role in the differences observed between models. In the specific case of the RVM, a clear anatomical basis for reciprocal interaction exists, as it receives direct projections from the spinal cord (Chaouch et al., 1983) and ON- and OFF-cells are known to project directly to dorsal horn laminae I–II and V (Fields et al., 1995) and modulate spinal nociceptive transmission (Fields et al., 2006).

Various models of peripheral neuropathy may increase spontaneous discharge rates and responses to somatic stimulation in nociceptive spinal dorsal horn neurons (e.g. Palecek et al., 1992; Laird & Bennett, 1993; Takaishi et al., 1996; Pertovaara et al., 1997), whereas their responses to visceral stimulation were not influenced 2–3 weeks after spinal nerve injury (Kalmari et al., 2001). It remains to be studied whether the SNI-induced changes in response properties of RVM cells observed in the present study reflect corresponding changes in response properties of spinal dorsal horn neurons, changes in supraspinal processing of nociceptive signals, or both.