Spontaneous Pain, Both Neuropathic and Inflammatory, Is Related to Frequency of Spontaneous Firing in Intact C-Fiber Nociceptors

Behavioral signs of spontaneous pain

None of the animals showed SFL of the contralateral foot. SFL occurred 1 d PO in both SNA and mSNA rats.

In SNA rats, SFL was much reduced by day 2, whereas in mSNA rats, it continued to increase and was much greater at day 2 than in SNA rats and was sustained at this high level until day 7 PO (S. Koutsikou and S. N. Lawson, unpublished observations). At 7 d PO, a greater percentage of mSNA than of SNA rats displayed SFL for >1 s/10 min [80% (20 of 25) compared with 12% (2 of 17); p < 0.001, Fisher’s exact test) (Fig. 2A). Figure 2B shows a significantly (p < 0.001) greater median SFL duration in mSNA compared with SNA rats. In a separate series of experiments (S. Koutsikou and S. N. Lawson, unpublished observations), SFL was not seen in rats (n = 6) that had L4 chromic gut loose ligation only (i.e., no L5 axotomy). Thus, it seems that in this model the combination of axotomy of one SN (L5) and chromic gut loose ligation of the adjacent SN (L4) in mSNA rats is critical for SFL development.

All CFA-treated rats (n = 11) (100%) showed SFL on day 1, whereas only one of six rats showed SFL on day 2 and none on days 4 and 7 after treatment (highly significant; χ² for trend) (Fig. 2C). Furthermore, the duration of SFL was significantly greater at 1 d than at 2, 4, and 7 d after CFA (Fig. 2D). Note that only SFL durations in the five rats in which electrophysiological recordings were made are included in Figure 2, to compare this with spontaneous neural activity in the same group of rats (see below). Inclusion of the data from the other six rats would increase the significance levels seen in Figure 2C.

Relationships between allodynia, hyperalgesia, and SFL after nerve injury

To ensure that SFL after nerve injury is not an evoked response attributable, for example, to allodynia, linear regression analysis was performed between three different types of behavior (Fig. 3; for details of analysis, see legend). In SNA rats, although the maximum SFL was only 1.5 s, there was a correlation between SFL and hyperalgesia (Fig. 3B) and a similar but not significant (p = 0.056) relationship between SFL and allodynia (Fig. 3A). In mSNA rats (SFL range from 4.5 to 22 s), there was no correlation of SFL with either allodynia or hyperalgesia (Fig. 3A, B). In contrast, there was a positive correlation between the extents of allodynic and hyperalgesic behaviors (Fig. 3C) with almost identical regression lines regardless of whether or not the mSNA values were included, suggesting some similarity in mechanisms contributing to these two behaviors. The presence of substantial SFL only in the mSNA rats (Fig. 2A, B), and the lack of correlation between SFL and the evoked behaviors (allodynia and hyperalgesia) in these rats, suggests different mechanisms for SFL and these two evoked behaviors.

Effects of the extent of L4 SN injury

The extent of L4 SN injury was determined with ATF3 immunostaining (see lack of positive nuclei in Fig. 4A and two clearly positive nuclei in Fig. 4B). The relationship of pain behaviors to this injury after L5 axotomy was examined by plotting measures of the different behaviors against the percentages of L4 DRG neuronal nuclei expressing ATF3. In all of the mSNA rats, >35% of L4 DRG neurons were ATF3 positive. These rats consistently showed SFL, the duration of which was variable and not correlated with the extent of L4 injury (Fig. 4C, filled circles). In SNA rats, there was little or no SFL even in those rats with levels of injury in L4 DRGs that were similar to those in mSNA rats (>35%) and again no correlation between SFL duration and L4 ATF3 percentage (Fig. 4C, open circles). Thus, in the presence of L5 axotomy, it is the L4 chromic gut loose ligation, not merely the extent of L4 fiber damage, that is required for generation of SFL.

In contrast, the extents of evoked behaviors (allodynia and hyperalgesia) were both positively correlated with the percentage of L4 DRG neurons with ATF3 staining (Fig. 4D, E), indicating an influence on both behaviors of the number of damaged L4 fibers. Linear regression lines between the extent of behavior and ATF3 percentages (Fig. 4D, E, continuous line) for SNA rats were almost perfectly superimposed on those for SNA plus mSNA rats for both allodynia and hyperalgesia (Fig. 4B, C, dotted line) (that is, adding the mSNA data did not alter significantly either the slope or elevation of these lines). Thus, the few SNA rats with more extreme inadvertent mechanical damage (>35% ATF3) had evoked behaviors that were as great as those in mSNA rats with similar amounts of L4 damage caused by the chromic gut. This is in clear contrast to SFL. We interpret this to mean that, after L5 axotomy, the degree of L4 injury is a major determinant of the extents of allodynia and hyperalgesia, regardless of whether L4 injury is caused directly by loose ligation of the L4 SN, or inadvertently as a consequence of L5 axotomy alone. If this is correct, then the greater extents of these two types of evoked behavior in mSNA compared with SNA rats would have resulted primarily from the consistently greater mechanical damage to L4 SN caused by the chromic gut ligation. Thus, it appears that SFL depended on the presence of chromic gut, most likely the inflammatory effects (see Discussion), whereas the evoked behaviors depended heavily on the extent of L4 SN damage.

Electrophysiological records of SA

Recordings were made in 191 identified LTMs. In control DRGs, none of 2 Aδ LTMs or of 31 Aα/β LTMs showed SA. After nerve injury, in L4 DRGs, no Aδ LTMs were recorded in either SNA or mSNA models. There was a slight, but not significant, increase in the percentage of Aα/β LTM neurons showing SA (13% of 30 units in SNA, and 10% of 30 units in mSNA). The firing rates of these Aα/β LTMs were low and similar in SNA (range, 0.01–0.03 Hz) and in mSNA (range, 0.01–0.02 Hz) rats. One day after CFA, only 1 Aδ LTM was recorded, and this showed SA with a firing rate of 0.4 Hz, and none of the 2 Aα/β LTMs showed SA. Four days after CFA, none of the 8 Aδ LTMs and only 1 of the 87 Aα/β LTMs showed SA (rate, 0.01 Hz). Because the SA was similar in magnitude in Aα/β LTMs in SNA and mSNA, it was unlikely to contribute greatly to SFL seen only in mSNA rats. Therefore, for the subsequent (main) analyses of intact neurons, we focus exclusively on nociceptors.

Because the only difference between the SNA and mSNA models was the L4 loose ligation, the SFL in mSNA must result from changes in firing properties of L4 neurons in mSNA compared with SNA rats. We therefore examined electrophysiologically the SA in L4 neurons in these two models, and to enable our data to be compared with those from other studies (see below), we also examined SA in axotomized L5 neurons of mSNA rats (Fig. 5). Finally, to enable comparison with a model of inflammatory pain, we recorded SA in L4/L5 DRG neurons in CFA1 and CFA4 rats. In addition to the above LTMs, recordings were made in a total of 719 DRG neurons classified by their dorsal root CVs and, where possible, sensory receptive properties. The breakdown of the numbers in each group is shown in Figure 5. A substantial number of C-fiber neurons was recorded in the treatment groups (n = 102) and control groups (n = 149) despite the difficulty of making intracellular recordings in such neurons. This 102 is made up of n = 21 in L5 in mSNA, n = 17 in L4 in SNA, n = 23 in L4 mSNA, n = 13 in CFA1, and n = 28 in CFA4. Because only a proportion of these neurons in any treatment group (for example, about one-third in SNA and mSNA rats) fired spontaneously, in any one treatment group, the numbers of C-fiber neurons recorded with SA was <10 (between 5 and 8). The numbers of rats used for electrophysiology were as follows: 14 mSNA, 7 SNA, 5 CFA (1 d), 26 CFA (4 d), plus 106 control, untreated rats. Many of the rats including most of the control rats provided data for other studies as well.

Percentages of neurons with spontaneous activity

The percentages of C-, Aδ-, and Aα/β-fiber neurons with SA in DRGs of different treatment groups are shown in Figure 5. Because the L5 SN is treated the same (tight ligation with silk followed by axotomy) in both SNA and mSNA rats, and because chromic gut ligature on the L4 SN in mSNA rats is several millimeters from the central stump of the L5 SN, it is unlikely that firing properties of axotomized L5 DRG neurons differ between the two models. The following data show clear differences in the L4 DRGs that could account for the SFL in mSNA but not SNA rats.

Axotomized neurons in L5 DRGs

Axotomized L5 C-fiber neurons showed no SA (Fig. 5A) consistent with previous studies (Liu et al., 2000). Any changes in SA in L5 neurons are unlikely to contribute to the much greater SFL in mSNA rats; because the L5 SN was axotomized in both models, the SA is likely to be similar in the two models. Compared with untreated controls, axotomy of L5 SN (mSNA rats) resulted in no significant change in percentage of axotomized Aδ-fiber or Aα/β-fiber neurons with SA (Fig. 5B, C, first two columns); overall, there was an increase to 6% (not significant) for all A-fiber neurons (data not shown). This is similar to the 8% value reported previously in an SNA model (Ma et al., 2003) and suggests that a similar percentage of axotomized L5 A-fiber neurons have SA in SNA and mSNA rats. Note that muscle spindle afferents were excluded from this control group, and axotomized Aα/β-fiber units firing at >10 Hz were also excluded. We do not exclude a role in neuropathic pain for fast conducting muscle afferent units that develop SA after nerve injury (Michaelis et al., 2000). However, because the L5 SN was axotomized in both SNA and mSNA rats, any SA in L5 neurons cannot account for the main behavioral difference between the two groups, namely, the SFL.

Neurons in ipsilateral L4 DRGs

Figure 5A shows that, compared with control rats, there was a similar large and significant increase in the percentage of nociceptive L4 C-fiber neurons with SA in both SNA (gray column) and mSNA (black column) rats (from 7 to 35% in both cases; Fisher’s test, p < 0.01). Thus, the percentage/number of these C-fiber neurons with SA could not, per se, account for the clear SFL seen in mSNA rats. Of the 17 C-fiber neurons in SNA rats, 9 were nociceptive, 5 were unresponsive, and 3 were unidentified (not fully tested) units. Of the 23 C-fiber neurons in mSNA rats, 15 were nociceptive, 3 were unresponsive, and 5 were unidentified units. For Aδ-fiber neurons, 10 of the 14 units in SNA rats were nociceptive, 2 were unresponsive, and 2 were unidentified, whereas 9 of 13 units in mSNA rats were nociceptive and the remaining 4 were unidentified units.

Of the eight C-fiber nociceptive-type neurons with SA in mSNA rats (Fig. 5A, black column, L4LL), five were nociceptive and three were unidentified (not fully tested). The latter three neurons had electrophysiological properties (action potential variables) that were very similar to those previously reported by Fang et al. (2005b) for nociceptive or unresponsive neurons and different from those of C low-threshold mechanoreceptors. They are unlikely to have been axotomized, because no L5 C-fiber neurons showed SA after axotomy (Fig. 5A, column 2). They were therefore probably nociceptive units with inaccessible receptive fields or very high-threshold nociceptors (see Materials and Methods). These eight C-fiber neurons with SA were recorded in five of eight mSNA rats from which recordings were made. Of the six units with SA in SNA rats (Fig. 5A, gray column, L4Spar), four were nociceptive and two were unresponsive. These units were recorded from all six SNA rats in which recordings were made.

The absence of SA in axotomized L5 C-fibers (Fig. 5A) (all recorded in mSNA rats) suggests that even the unidentified L4 C-fiber units with SA in SNA and mSNA rats are likely to have intact fibers.

L4/L5 neurons after CFA inflammation

A potential cause of the C-fiber SA in L4 DRGs was inflammation attributable to Wallerian degeneration of axotomized L5 peripheral fibers within the same peripheral nerve (sciatic) (see Discussion). We therefore examined whether inflammation alone, albeit applied to the periphery and not the nerve trunk, could have a similar effect. Because we found SFL 1 d but not 4 d after CFA, we examined whether C-fiber SA was greater on day 1 than day 4.

Both 1 and 4 d after CFA treatment, the percentage of C-fiber nociceptors with SA was significantly increased compared with untreated control rats from 7% (control) to 38% (1 d) and 25% (4 d) (Fig. 5A). These values were of a similar magnitude to those in L4 C nociceptors in nerve-injured rats 7 d PO (Fig. 5A, L4Spar and L4LL). This is consistent with inflammatory processes being involved in generation of SA in C-fiber neurons after both nerve injury and inflammation.

However, after CFA treatment, unlike 7 d after nerve injury, there was also a significant increase in percentage of Aδ-fiber nociceptors with SA (from 6% in untreated to 60% at 1 d and to 33% at 4 d) (Fig. 5B). Aβ-fiber nociceptors were not evaluated in CFA1 animals, but there was no increase in CFA4 rats. The percentage of Aδ nociceptors with SA was thus greater 1 and 4 d after CFA than 7 d after nerve injury. One explanation could be similar accessibility of nociceptive terminals of both Aδ- and C-fibers to inflammatory mediators in the periphery, but lower accessibility of myelinated than unmyelinated fibers to such mediators within the nerve.

Membrane potentials of C-fiber neurons with SA

Fluctuations in membrane potential in somata of axotomized A-fiber but not C-fiber neurons have been suggested as a cause of SA (Amir et al., 1999). We did not see such fluctuations in L4 C-fiber neurons, suggesting that the SA in these neurons may originate peripheral to the soma. Furthermore, there were no significant differences in median membrane potentials of C-fiber neurons between (1) units with SA (−50 mV) and without SA (−48 mV) in mSNA rats (p > 0.05; Mann–Whitney U test) or between (2) units recorded in L4/L5 DRGs from untreated rats (−46 mV), L5 DRGs in SNA rats (−50 mV), L4 DRGs in either mSNA (−50 mV) and SNA rats (−51 mV) (Kruskal–Wallis test followed by Dunn’s post test). These are all similar to the median value (−50 mV) reported previously for normal rats (Fang et al., 2005b).

Rate of SA in L4 nociceptive-type neurons in SNA and mSNA rats

Comparison of SA rate in L4 C- and A-fiber nociceptors in mSNA rats

From the above, in mSNA rats, the spontaneous firing rate of L4 C-fiber nociceptors was higher than that of L4 A-fiber nociceptors by 180-fold if medians are compared (C, 1.8 Hz, vs A, 0.01 Hz; p < 0.01) or by >50-fold if means are compared (C, 1.9 Hz, vs A, 0.035 Hz) (Fig. 6A, B; Table 1).

SA in CFA-treated rats

The median firing rates of C-fiber nociceptors with SA were significantly greater at 1 d than at 4 d post-CFA (Fig. 6A), consistent with the possibility that a certain firing rate in these neurons, or total level of input along the SN, might be required for SFL to occur. An ongoing/spontaneous firing rate of 0.59 Hz (Fig. 6, Table 1) in approximately one-third of C-fiber nociceptors in L4 and L5 DRGs 1 d after CFA appears to be enough to cause a small amount of SFL, whereas 0.15 Hz 4 d after CFA was not. Thus, a common feature between the rats with SFL in CFA1 and mSNA models is a high rate of spontaneous firing in C-fiber nociceptors.

Estimate of total CNS input from L4 C- and A-fiber SA

Relationships between SFL and rate of spontaneous firing in C-nociceptors

If faster firing in intact C-nociceptors is necessary for SFL, then the extent of SFL may be related to the firing rate. First, we compared these in individual rats. We had SFL records for only 13 rats in which successful C-fiber recordings were made. (CFA4 data are not included, because SFL was not measured in the same rats. N.B., No foot lifting was ever seen in CFA4 rats, and SA firing rates can be seen in Fig. 6.) In Figure 7A, we compare the SFL and SA firing rates for each of these 13 rats. Note that the firing rate was low or 0 in rats with no SFL even for an mSNA rat (rat 5) and higher firing rates were seen in rats with SFL, in both mSNA and in CFA1 rats. The mean firing rate for each rat was plotted against the SFL duration (Fig. 7B). A highly significant linear correlation indicates that, for these 13 rats, the SFL was related to the mean C-neuron firing rate (including neurons with and without SA).

We also determined, for each of the four groups, CFA1, CFA4, SNA, and mSNA, the mean and median amount of SFL (in seconds per 10 min) and the mean rate of spontaneous firing in C-fiber neurons with SA for that group, providing us with one point summarizing the values for that whole group. For this, we used all the SFL data in Figure 2 and the values for firing rate given in Figure 6 and Table 1. Interestingly, linear regression analysis shows that SFL is highly significantly related to both the mean and median L4 C-fiber firing rate (Fig. 7C). Extrapolation from small numbers of DRG neurons should be interpreted with caution. However, we believe these neurons are randomly sampled and thus representative of the total population of neurons within the DRGs examined. The patterns that emerge from all the evidence together are persuasive and can be used to generate hypotheses for future testing.

SFL as a behavioral sign of spontaneous pain

The lack of correlation between SFL and evoked pain behaviors (allodynia or hyperalgesia) in mSNA rats suggests that SFL in these rats results not from allodynia attributable to contact but from spontaneous pain.

It is not known how much SFL (how many seconds per 10 min) is required to indicate spontaneous pain. An afferent input strong enough to override normal locomotory/exploratory behavior would be needed for an animal to lift/examine its foot; even a few seconds of SFL probably indicates substantial spontaneous pain as would be the case in humans. In mSNA rats, the consistent presence of SFL, the relatively short SFL duration, and generally normal behavior indicate this to be a model reliably causing spontaneous pain behavior but with only mild–moderate discomfort. However, distractions should be limited during SFL measurement, because they probably cause greater variability in responses to ongoing rather than evoked, synchronized, afferent input.

Our findings of increased percentage of L4 C-fiber neurons with SA 1 week after L5 SN injury agrees with previous studies in rat (Wu et al., 2001) and monkey (Ali et al., 1999). The lower values in our study (35%) than that of Wu et al. (>50%) may relate to strain differences and/or shorter initial SA observation periods in our study (2 vs 5 min) that could exclude units with very low firing rates (<1 in 2 min).

Two additional studies (Boucher et al. 2000; Ma et al. 2003) failed to demonstrate SA in spared L4 C-fibers after L5 SN injury, perhaps because (1) Boucher et al. (2000) transected the L4 SN just before dorsal root fiber recordings, thus preventing SA originating in C-fibers distal to the transection reaching the dorsal roots; (2) Ma et al. (2003) used postoperative antibiotics. These would reduce inflammatory responses to bacterial infection (if present despite aseptic operating conditions), or may directly influence inflammatory responses (Hamilton-Miller, 2001) such as that to chromic gut.

The similar proportion (one-third) of L4 C nociceptive-type neurons showing SA in both SNA and mSNA rats suggests involvement of the same C-fiber neuron population in both models. However, the large increase in frequency of spontaneous firing in L4 C-fiber nociceptive-type neurons in mSNA compared with SNA rats suggests that effects of chromic gut add to those of L5 axotomy, resulting in increased firing rates in this same group of neurons.

This study does not address the role of central sensitization. The higher rate of firing (∼2 Hz) in L4 C-fiber neurons in mSNA rats, which translates into 5 kHz along the L4 dorsal root, may well be enough to trigger SFL in the absence of central sensitization. However, this barrage may also cause central sensitization (Woolf and Wall, 1986), which could enhance its effectiveness in generating SFL/spontaneous pain. It seems unlikely that A-fiber nociceptors make a significant contribution to SFL in mSNA rats, because the frequency of firing is very low and the proportion with SA remains low. However, we cannot exclude an effect of even a small A-fiber input, especially if enhanced by central sensitization.

Inflammation as a cause of SA in intact L4 C-fibers

There are similarities in the L4 intact C-nociceptors after nerve injury and CFA treatment that suggest a common inflammatory mechanism in both cases. These include the following: (1) the proportion with SA and (2) an increase in action potential height (Djouhri, Fang, McMullan, and Lawson, unpublished observations). The positive correlation between SA firing rate in C-nociceptors and SFL, in both inflammatory and nerve injury models, is strongly suggestive of a link between these variables in both types of model.

Our findings lend support to the growing evidence for inflammatory mediators influencing sensory neurons and having a role in generation of neuropathic pain (for review, see Li et al., 2000; Watkins and Maier, 2002; Lawson, 2005). Wallerian degeneration of peripheral nerve fibers causes an inflammatory response (Shamash et al., 2002), which is thought to be the cause of SA in intact C-fiber afferents running in the same peripheral nerve (Li et al. 2000; Wu et al. 2001). Evidence of a role for inflammatory mediators includes the following. After CCI, SA in DRG neurons is markedly increased by inflammatory mediators (Song et al., 2003). In another neuropathic pain model, TNFα (tumor necrosis factor α), a proinflammatory cytokine, causes increased dorsal root SA in C-fibers and increased SFL when applied to the DRG (Schafers et al., 2003). Nerve growth factor (NGF) plays a key role in the response of sensory neurons to inflammation; inflamed peripheral tissues upregulate NGF (Woolf et al. 1994), which leads to NGF-dependent SA in C-fiber DRG neurons (Djouhri et al. 2001), probably because NGF influences neuronal expression of ion channels or receptors that contribute to SA generation (Lawson, 2005). NGF levels are raised in adjacent spared L4 DRG neuronal somata after L5 axotomy, possibly because of raised NGF in peripheral nerve during Wallerian degeneration (Shamash et al., 2002) and reduced competition for target-derived NGF by axotomized L5 fibers allowing increased NGF uptake by L4 fibers (Obata et al., 2004). Nociceptive DRG neurons express high levels of trkA (the high affinity NGF receptor) (Fang et al. 2005a) and can therefore take up and be influenced by NGF.

Possible causes of inflammatory influences on intact L4 neurons in SNA and mSNA nerve injury models are summarized in Figure 1. Inflammation 1 and increased NGF availability in skin are thought to result from degenerating L5 fibers (see above) and inflammation 2 is chromic gut induced (see Introduction). Inflammation 1 alone (SNA) results in a lower rate of SA than inflammations 1 plus 2 (mSNA). Alone neither inflammation 1 (SNA) nor inflammation 2 (after L4 chromic-gut loose ligation only) (S. Koutsikou and S. N. Lawson, unpublished observations) causes significantly increased SFL. Cumulative effects of inflammation 1 (perhaps with increased NGF availability in the periphery) plus inflammation 2 thus appear to be required for both the SFL and the faster SA rate in intact L4 C-fibers in mSNA rats.

Sensations that may result from C-fiber SA

The different types of spontaneous pain sensation probably depend on the central projections of spontaneously active afferent fibers. Because activation of C- and A-fiber nociceptors generate unpleasant/unbearable burning and sharp pain, respectively (for review, see Djouhri and Lawson, 2004), SA in these two fiber groups may evoke, respectively, the burning and stabbing spontaneous pain reported in human patients. Thus, SA in intact C-fiber neurons reported here may cause burning spontaneous pain. It is not known whether A-fiber nociceptors are responsible for stabbing spontaneous pain, and, if so, whether these have axotomized or intact fibers. However, because SA in Aδ-nociceptors was induced by peripheral inflammation but not nerve injury or inflammation applied to the nerve (chromic gut), the myelin sheath may normally protect such fibers from inflammatory mediators except in the periphery or possibly after demyelination.

In summary, we present evidence linking high spontaneous firing rates in intact C-fiber nociceptive neurons with SFL in both neuropathic and inflammatory pain models. In neuropathic models, a low spontaneous firing rate in these neurons, possibly resulting from inflammatory effects of degenerating axotomized L5 fibers, is not associated with significant SFL. Significant SFL is accompanied by the higher firing rate that appears to depend on added inflammatory influences provided in this mSNA model by the chromic gut L4 SN ligation. This plus the overall correlations between SFL and C-nociceptor firing rate after nerve injury and CFA treatment lead us to propose that spontaneous neuropathic pain may result from cumulative neuroinflammatory influences not necessarily all at the same site on the neuron, causing relatively rapid spontaneous firing of intact C-nociceptors. Our findings therefore support the view, based on the high percentage (50%) of clinical neuropathic pain cases associated with infection/inflammation of peripheral nerves (Said and Hontebeyrie-Joskowicz, 1992), that inflammation is important in neuropathic pain (Devor and Seltzer, 1999; Watkins and Maier, 2002). Importantly, we also suggest an underlying neuronal mechanism linking inflammation, rate of spontaneous firing, and spontaneous pain. Elimination of SA in these intact C-fiber nociceptors may therefore relieve spontaneous neuropathic pain. Additional studies are needed to elucidate the altered electrophysiological properties and/or chemical influences causing this C-fiber SA.