We first examined the escape and/or vocalization behavior of rats to transcutaneous stimuli using Neurometer. When 250 and 5 Hz stimuli were applied to the left hind limb, rats escaped from the stimuli or vocalized. The thresholds for 250 and 5 Hz stimuli were quite similar, and were averaged to be 0.74 ± 0.06 mA (T250, n = 12) and 0.84 ± 0.12 mA (T5, n = 12), respectively. Next, 2000 Hz stimulation which is used to evaluate Aβ fibers was applied. Although Aβ fibers were thought to convey innocuous information, escape and/or vocalization behavior was also observed at higher intensity. The threshold for 2000 Hz stimulation (2.14 ± 0.28 mA, T2000, n = 12) was more than two times higher than T250 and T5.

Intracellular recordings were made from 53 DRG neurons. Dorsal root stimulation with a suction electrode elicited antidromic APs with latencies in the range of 0.2 – 15.0 ms. Fast-conducting neurons had a brief action potential duration (APD), whereas slow-conducting ones had a relatively broad APD. Based on the threshold stimulus intensity (TSI), conduction velocity (CV) and APD, DRG neurons were divided into three subgroups, C, Aδ and Aβ fibers (Table 1). These results were consistent with those reported previously 11314151617.

After classification of DRG neurons, C, Aδ and Aβ neurons were stimulated by sine-wave stimuli with gradual increase in intensity using Neurometer.

It is well known that SG neurons in the spinal dorsal horn receive synaptic inputs predominantly from C and Aδ fibers 181920. Therefore, SG is thought to be specifically implicated in nociceptive processing. As shown previously 21222324, SG neurons exhibited miniature and spontaneous excitatory postsynaptic currents (mEPSCs and sEPSC, respectively) under voltage-clamp condition at a holding potential of -70 mV. The large amplitude of EPSCs were reversibly blocked by TTX (1 μM) applied to the surface of the spinal cord leaving mEPSCs intact (data not shown). Therefore, the large amplitude of EPSCs (sEPSCs) were mediated by spontaneous firings of primary afferents or interneurons. Pinch stimulation applied to the skin of the hind limb produced a barrage of EPSCs (defined as evoked EPSCs; eEPSCs) in all 24 SG neurons examined (Figure 4).

When the sine-wave stimuli was applied to the skin of the receptive field, 250 and 5 Hz stimuli evoked a barrage of EPSCs at a stimulus intensity of ~0.8 mA (Figures 5A and 5B). On the other hand, SG neurons showed an increase in frequency of large amplitude of EPSCs with 2000 Hz stimulation at higher intensities (Figure 5C) in spite of few inputs from Aβ fibers to SG, suggesting that the EPSCs elicited by 2000 Hz stimulation were mediated by the activation of Aδ afferents. Figure 5D shows the frequency of eEPSCs (> 50 pA) evoked at the behavioral thresholds of the three stimuli. The average frequency of 5 Hz-evoked EPSCs at the behavioral threshold was increased to 378 ± 30 % of control (1.8 ± 0.3 Hz, n = 7) and that of 250 Hz-evoked EPSCs to 473 ± 69 % (2.9 ± 0.7 Hz, n = 8), and that of 2000 Hz-evoked EPSCs to 602 ± 198 % (2.9 ± 0.8 Hz at the behavioral threshold T2000, n = 6). The large amplitude EPSCs evoked by sine-wave stimuli were blocked by TTX (1 μM) and all EPSCs were abolished by a glutamatergic AMPA receptor antagonist CNQX (20 μM; Figures 6A and 6B).

We observed that the stimulus thresholds producing the escape and/or vocalization behavior in rats with 250 and 5 Hz stimuli were not significantly different, whereas the threshold for 2000 Hz stimulation was two times higher than those of 250 and 5 Hz stimuli. The values of these thresholds are consistent with those of previous study 10. These observations suggest that transcutaneous stimuli at frequency of 250 and 5 Hz induce pain sensation, and the 2000 Hz stimulation may also induce pain at higher intensity. From the analysis in C and Aδ fibers, which predominantly convey noxious information, we found that 5 Hz stimuli activated both C and Aδ fibers, 250 Hz stimulation activated only Aδ at low intensity (~0.8 mA). Interestingly, Aδ fibers were also activated by 2000 Hz stimulation, although 2000 Hz stimulation is thought to activate Aβ fibers. However, the activation of Aδ fibers by 2000 Hz was observed only at high stimulus intensity (> 2 mA). In in vivo analysis of EPSCs evoked in SG neurons, which receive monosynaptic inputs predominantly from C and Aδ fibers 181920, we found that 2000, 250 and 5 Hz stimuli significantly increased the frequency of evoked EPSCs at their behavioral thresholds. Therefore, these results indicate that 250 and 5 Hz stimuli, produce nociceptive responses by activation of Aδ fibers and by activation of C and Aδ fibers, respectively. On the other hand, 2000 Hz stimulation at high intensity could activate Aδ fibers and thus initiate noxious sensation. There was no significant difference in the frequencies of EPSCs evoked by 250, 5 and 2000 Hz stimuli at the behavioral threshold (Figure 5D). In the previous behavioral pharmacological study, topical application of capsaicin to the skin, a specific desensitizer of small fibers 25, increased behavioral thresholds for 250 and 5 Hz but not 2000 Hz stimuli, indicating that 250 and 5 Hz stimuli activate small fibers 10. This observation is consistent with our results.

We observed that C fibers were activated only by 5 Hz stimulation with a maximum frequency of ~3 Hz, Aδ fibers were activated by 250 and 5 Hz (maximum frequency of ~22 Hz with 250 Hz stimulation) and Aβ fibers were activated by all three stimuli (maximum frequency of ~140 Hz with 2000 Hz stimulation, Figures 1,2,3). One of the most important factors for affecting the maximum frequency of sensory fibers is the inactivation of voltage-dependent sodium channels. In other words, discharge frequency of fibers is controlled by the time taken to recover the sodium channels from their inactivation state. Recent studies have shown that TTX-sensitive sodium channels expressed in large diameter DRG neurons exhibit faster recovery from the inactivation than the TTX-resistant ones expressed in small DRG neurons 26272829. Therefore, it can be speculated that high frequency sine-wave stimuli are unable to initiate APs in small C fibers due to slow repriming of TTX-resistant sodium channels.

Passive membrane property is also an important factor for the ability of neurons to discharge. Previous report has shown that the membrane properties of the three subgroups of rat DRG neurons were varied 17. In particular, the membrane time constant of C neurons (6.5 ms) is much longer than those of Aδ (2.6 ms) and Aβ (1.8 ms) neurons. On the other hand, the durations of single sine-wave stimuli at frequencies of 2000 Hz and 250 Hz are 0.5 ms and 4 ms, respectively. The membrane time constant is the time required for the voltage change across the membrane to reach ~63% of its final value 30. Therefore, the longer membrane time constant might contribute to the failure of AP in C neurons with 2000 and 250 Hz stimuli, since the membrane potential is slowly depolarized, resulting that the membrane potential can not reach AP threshold within the short depolarizing cycles with the high frequency stimuli as compared to 5 Hz one. In C fiber of rat DRG neurons, AP failure was increased when repetitive stimulation was applied at high frequencies 1. Consistent with the lines of evidence, in the present study, there was no firing in most C fibers with 250 Hz and 2000 Hz stimuli even when the high intensity stimuli were applied as described in the results.

Male Sprague-Dawley rats (6–9-week old), weighing 250–300 g were used for behavioral tests, intracellular recordings and in vivo patch-clamp recordings. A skin patch dispersion electrode was fixed to attach the skin of left hind limb of rats. Transcutaneous nerve stimuli were applied through this electrode using the nociceptive mode of the Neurometer^® (NUROTRON INCORPORATED, Baltimore, USA). In this mode, the intensity was automatically increased in step from 0 to 9.99 mA (29 steps for 5 Hz stimulation, 20 steps for 250 Hz and 2000 Hz stimuli). The durations were 2.5 sec for 5 Hz and 2.1 sec for 250 Hz and 0.72 sec for 2000 Hz stimuli. When rats escaped and/or vocalized, the stimulation was immediately stopped and the stimulus intensity was defined as behavioral threshold.

The methods used for the current experiment were similar to those described previously 1. Under anesthesia with diethyl ether, a rat was decapitated and the lumbar laminectomy was performed. L4-6 DRGs with an attached proximal dorsal root were isolated from the animal. The isolated DRG was submerged in Krebs solution (in mM: NaCl 117, KCl 3.6, CaCl₂ 2.5, MgCl₂ 1.2, NaH₂PO₄ 1.2, NaHCO₃ 2.5 and glucose 11) equilibrated with 95% O₂-5% CO₂ and maintained at 36 ± 1°C and set in a recording chamber. Intracellular recordings of APs were made from DRG neurons with glass-microelectrodes having a DC tip resistance of 50–100 MΩ, filled with 4 M potassium-acetate. Signals were amplified with a high input-impedance bridge amplifier (Axoclamp 2B; Axon Instruments, Foster City CA, USA). Artifacts were minimized with low-pass (1000 Hz) or notch (250 Hz) filters using pCLAMP program (Axon Instruments). Data from neurons with resting membrane potentials less than – 50 mV and AP amplitudes smaller than 60 mV were excluded in the present study. AP duration was determined at half of the peak amplitude of the AP. Antidromic stimulation (duration 100 μsec) was given to the central end of the dorsal root with a suction electrode. The stimulus intensity was monitored with a digitized output isolator (ss-202J; Nihon Kohden, Tokyo, Japan). The value of conduction velocity was calculated from the latency of AP and the length of the dorsal root retained.

The methods for in vivo patch-clamp recording from SG neurons have been described in detailed elsewhere 2224. Briefly, under artificial ventilation, a lumbar laminectomy was performed at the level of L4–L5 and the animal was then placed in a stereotaxic apparatus (Model ST-7, Narishige, Japan). Under a binocular microscope with ×8 ~ × 40 magnification, the dura mater was cut and reflected, and then either L4 or L5 dorsal root was shifted laterally using a glass hook. The pia-arachnoid membrane was cut to make a window to allow the patch electrode into the spinal cord. The patch pipette had a tip resistance of 8–10 MΩ when filled with a solution of the following composition (in mM): Cs₂SO₄ 110, CaCl₂ 0.5, MgCl₂ 2, EGTA 5, HEPES 5, Mg-ATP 5, tetraethylammonium 5. Signals were acquired with an Axopatch 200B amplifier (Axon Instruments), low-pass-filtered at 5 kHz. Data were analyzed using Axograph program (Axon Instruments). The surface of the exposed spinal cord was perfused with Krebs solution equilibrated with 95% O₂-5% CO₂ at 38 ± 0.5°C. Drugs were dissolved in Krebs solution and applied to the perfusing line. Whole-cell patch-clamp recordings could be obtained from in vivo preparations for more than 8 hours and stable recordings were made from single SG neurons for up to 2 hours. After the end of the experiments the animals were killed by exanguination.

Data were presented as mean ± SEM. Statistical significance was determined as P < 0.05 using Dunnett's multiple comparison.