Many FG-labeled neurons were observed in the TG following FG injection into the mental skin in naive rats (Figure 1A). A small number of FG-labeled neurons were observed in the TG at 7 days after IAN transection and many were evident in the TG at 14 days after IAN transection (Figure 1B and 1C). Although the number of FG-labeled TG neurons at 7 and 14 days after IAN transection was significantly smaller than that in naive rats, the number of neurons at 14 days was significantly larger compared to that at 7 days (Figure 1D).

After successful completion of the training, in which rats allowed noxious mechanical stimulation (> 15 g) to be applied to the mental skin region, the IAN was transected. Figure 1E (ipsilateral side to IAN transection) and F (contralateral side to IAN transection) illustrate the mechanical threshold intensity for evoking escape behavior by mechanical stimulation of the mental skin region before and 7-14 days after IAN transection. The threshold value was significantly lower at 11-14 days after IAN transection compared to pre-operative values (11 days: p < 0.05, 12-14 days: p < 0.01) (median values, pre-operative, ipsilateral: 15 g, contralateral: 10 g; 7 days after transection, ipsilateral: 15 g, contralateral: 10 g; 14 days after transection, ipsilateral: 4 g, contralateral: 10 g, n = 5 in each group).

The activity of 76 single fiber activities was recorded from the IAN at 14 days after IAN transection (66 and 10 fibers from IAN-transected rats with and without behavioral changes, respectively) and the activity of 89 IAN fibers was recorded from naive rats. The single unit activities were classified as Aβ-, Aδ- and C-fiber responses according to their conduction velocities determined from the response latency and conduction distance (Figure 2). Fifteen fibers were classified as C-fibers and 34 fibers were as Aβ-fibers and 40 fibers were Aδ-fibers in naive rats (Figure 2C). On the other hand, only 3 fibers were classified as C-fibers and all others were classified as A-fibers (Aβ-: n = 34, Aδ-: n = 29) in IAN-transected rats (Figure 2D). We observed clear differences in background activity and mechanically evoked responses in Aδ-fibers between naive and IAN-transected rats. Five IAN fibers did not have any receptive field (RF) in the face and 24 fibers did have a facial RF. Most of the primary afferents of the IAN fibers did not show background firing in naive rats (Typical data shown in Figure 3Aa). In addition, we also did not observe background activities in IAN-transected rats without behavioral changes (- behav. change in Figure 3Ad). The background activities in the IAN-transected rats were significantly higher as compare with that in naive rats (Figure 3Ad) and those without a RF (RF-) showed the highest background activity (RF- in Figure 3Ac, d). The afterdischarge indicated by the arrow in Figure 3Bb was also significantly higher in the IAN-transected rats compared to naive rats (Figure 3Bc).

We analyzed the effect of mechanical stimulation of the RF on Aδ-fiber units only, because of the small population of C-fibers in IAN-transected rats. Aδ-fibers showed graded firing following increases in mechanical stimulus intensity from the non-noxious to the noxious range in naive and IAN-transected rats, as illustrated in Figure 3C. These fibers showed significantly larger responses to both non-noxious and noxious mechanical stimulation as compared to those of naive rats (Figure 3D).

Mean size of FG-labeled TG neurons for patch-clamp recording was 27.4 ± 0.8 μm in naive rats and 27.8 ± 0.7 μm in IAN-transected rats (n = 17 each). Since there is a positive correlation between neuronal cell size and conduction velocity of A- and C- afferents in DRG neurons 51 , small-diameter TG neurons recorded in the present study were considered to be classified as small to medium Aδ-TG neurons (diameter 21-36 μm) for the patch-clamp recording experiment. Following perforation of the cell membrane with amphotericin B, the series resistance dropped to < 20 MΩ (naive: 17.2 ± 0.8 MΩ; IAN transection: 16.8 ± 0.8 MΩ, n = 17 each) within 5-12 min and remained stable for more than 15 min. In addition, the value for the cell capacitance was 23.1 ± 1.0 pF in naive rats and 23.1 ± 1.3 pF in IAN-transected rats (n = 17 in each group).

Total I _Na in TG neurons was larger in IAN-transected rats compared with naive rats, as illustrated in Figure 4Aa (naive rats) and Figure 4Ab (IAN-transected rats). The neurons were first held at -60 mV and then stepped from -80 mV to +80 mV for 50 ms (conditioning pre-pulse potential) as illustrated in Figure 4A. Since I _Na can be subdivided into TTX-R I _Na and TTX-S I _Na on the basis of their sensitivity to TTX, we also analyzed these two currents in naive and IAN-transected rats. Representative wave forms of the total Na⁺ currents, TTX-R I _Na and TTX-S I _Na, in TG neurons from naive and IAN-transected rats are illustrated in Figure 4A. TTX-R I _Na recorded from TG neurons from IAN-transected rats was significantly larger than these recorded from naive rats (n = 7, p < 0.05) (Figure 4Ad). TTX-S I _Na was also larger than that of naive rats (Figure 4Ae: naive, Figure 4Af: IAN-transected). TTX-R I _Na was much larger than TTX-S I _Na in both naive and IAN-transected rats (Figure 4Ac, d, e, f). The I-V relation curve and total I _Na and TTX-R I _Na are illustrated in Figure 4B. The total I _Na (n = 7), TTX-R I _Na (n = 7) and TTX-S I _Na (n = 7) were all significantly larger in IAN-transected rats than naive rats (p < 0.05). The mean peak current densities of total I _Na (n = 7), TTX-R I _Na (n = 7) and TTX-S I _Na (n = 7) were significantly larger in IAN-transected rats compared to naive rats as illustrated in Figure 4C (p < 0.05). Moreover, the magnitude of the transection-induced potentiation of mean peak current densities was significantly larger for TTX-S I _Na than TTX-R I _Na (78.9 ± 6.2% vs. 37.5 ± 7.5%, n = 7 in each, p < 0.05).

Under current-clamp conditions, the ability of TG neurons to generate action potentials was analyzed. Action potentials were elicited during current injection into TG neurons in both naive and IAN-transected rats (Figure 5A). Before TTX administration, a single spike was generated at threshold current injection in naive and IAN-transected rats (upper traces in Figure 5B). The change in wave form was recorded during current injection of square pulses in 10 pA steps. The first spike amplitude at 1T stimulus and number of spikes at 2-3T stimuli were different between naive and IAN-transected rats (Figure 5B). All tested neurons exhibited action potentials in the presence of 1 μM TTX. In this case (upper 4 traces in Figure 5B), the spike amplitude was slightly decreased in naive and IAN-transected rats after administration of 1 μM TTX compared to those before TTX administration, suggesting that some neurons in both groups included TTX-S as well as TTX-R I _Na components. The administration of 1 μM TTX inhibited 31 ± 6% of the control spike amplitude of action potentials at one threshold in naive rats. Furthermore, 1 μM TTX administration inhibited 23 ± 5% of the control spike amplitude of action potentials at one threshold in IAN-transected rats. We also analyzed the spike amplitude of action potentials, overshoot amplitude of action potentials, threshold intensity for spike generation and also number of spikes during gradual increases in membrane potential under the condition of the presence of 1 μM TTX (Figure 5C, D, E and 5F). Amplitudes of action potentials generated by 1T current application in the presence of 1 μM TTX were significantly larger in TG neurons in IAN-transected rats than in those of naive rats, as illustrated in Figure 5C (n = 10, p < 0.05). The overshoot amplitude of action potentials was also significantly larger in IAN-transected rats than that of naive rats, as shown in Figure 5D (n = 10, p < 0.05). The first spikes were elicited at significantly lower stimulus intensity in IAN-transected rats than in naive rats, as illustrated in Figure 5E (n = 7 in each group, p < 0.05). The resting membrane potential was significantly larger in TG neurons of IAN-transected rats compared to that in naive rats (IAN rats: -47.1 ± 1.1 mV, naive rats: -51.9 ± 1.3 mV, p < 0.05, n = 7 in each). Spike number was increased following an increase in the injected current in naive and IAN-transected rats. The mean number of spikes was significantly higher in IAN-transected rats than that of naive rats, as illustrated in Figure 5F (n = 10 in each threshold group, p < 0.05).

The changes in total K⁺ current, I _k and I _A currents in TG neurons were also studied in IAN-transected rats (Figure 6). The peak values of total K⁺ current, I _k and I _A currents were measured and all currents were significantly smaller in IAN-transected rats compared to naive rats, as illustrated in Figure 6A and 6B (mean ± SEM, total K⁺ current: 4568 ± 188 pA in naive rats, 2522 ± 198 pA in IAN-transected rats, p < 0.05, n = 5; I _k: 2902 ± 114 pA in naive rats, 1650 ± 127 pA in IAN-transected rats, p < 0.05, n = 5; I _A: 2676 ± 155 pA in naive rats, 1012 ± 132 pA in IAN-transected rats, p < 0.05, n = 5). The mean peak current densities of total K⁺ current (n = 5), I _k and I _A currents (n = 5) were significantly smaller in IAN-transected rats compared to naive rats, as illustrated in Figure 6C (p < 0.05).

A total of 94 male Sprague-Dawley rats weighing 100-250 g was used for the present study. Rats were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and placed on a warm mat. A small incision was made on the surface of facial skin over the masseter muscle and the alveolar bone was reached through the masseter muscle. The surface of the alveolar bone was exposed and the bone covering the IAN was removed and the IAN was exposed. The IAN was transected at 7 mm proximal from the angle of alveolar bone and then immediately replaced into the inferior alveolar canal 58 . For patch clamp recording, FG dye (2%, 10 μl; Fluorochrome, Englewood, CO, USA) was subcutaneously injected into the mental region 2 days before the recording experiment. After surgery, Penicillin G potassium (20,000 units, i.m.; Eli Lilly, Indianapolis, IN) was injected to prevent infection.

Fifteen rats were used for the FG tracing study (naive: n = 5, 7 days after transection: n = 5, 14 days after transection: n = 5). Rats were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and 10 μl of 2% FG was subcutaneously injected into the mental skin 2 days before perfusion in IAN-transected rats. Two days after the FG injection, rats were deeply anesthetized with the same anesthetic and perfused with 200 ml 0.9% saline followed by 500 ml of 4% paraformaldehyde. The TG was removed and post-fixed in the same fixative for 2 days and the tissue was then transferred to 20% sucrose (w/v) in phosphate-buffered saline (PBS) for several days for cryoprotection. Thirty-micron-thick sections were cut with a freezing microtome and sections were collected in PBS. FG immunohistochemical staining of TG neurons was carried out as previously described by Saito et al. 58 . TG neurons were defined as FG positive if the cytoplasm was stained with a black deposit. The FG-labeled TG neurons with clear nuclei were counted and their areas were measured. The number of FG-labeled TG neurons at the root of the third branch of the trigeminal nerve was counted in 3 sections (one section with the largest number of FG labeled neurons and next two sections) from each rat.

In daily sessions, rats were trained to stay in a plastic cage and to keep their snout protruding through a hole on the wall during mechanical stimulation of the mental skin with von Frey filaments (Touch-Test, North Coast Medical, Inc., CA, USA). Rats under this condition can escape from von Frey stimulus freely. When rats escaped from the von Frey stimulus, the escape behavior was defined as nocifensive. The maximum intensity used before IAN transection was 15 g. The escape threshold was measured before the IAN was transected and then the IAN of those rats was transected. The escape threshold to mechanical stimulation of the mental skin was measured daily at 8 days before and 7-14 days after IAN transection. Only rats with IAN transection which showed a significant decrement in mechanical escape threshold at 14 days after IAN transection were used for single fiber and patch clamp recording experiments (see below). Quantitative mechanical stimuli were applied to the mental skin region in ascending and descending orders to evaluate the escape threshold. Each von Frey filament was applied 5 times. When rats showed an escape response to a filament, the bending force of that filament was defined as the escape threshold intensity 58 . The median threshold intensity was calculated from the values following one ascending and one descending trial.

We used IAN-transected rats which showed hypersensitivity to mechanical stimulation of the mental skin and IAN-transected rats without behavior changes which did not show any behavioral changes after IAN transection for single fiber recording experiments. Rats were divided into 3 groups: ipsilateral to IAN transection (n = 16), ipsilateral to IAN without behavioral changes after IAN transection (n = 5) and naive groups (n = 22). For IAN fiber recording, each group of rats was anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and the trachea and left femoral veins were cannulated to allow artificial respiration and intravenous administration of drugs, respectively. Anesthesia was maintained with halothane (2-3%) mixed with oxygen during surgery. Rats were mounted in a stereotaxic frame and a craniotomy was performed, 1-4 mm lateral to the midline and 2-5 mm anterior to the interaural line. The skull was rigidly secured to a head holder by stainless-steel screws and dental acrylic resin, and the nose holder was removed. This setup allowed convenient access to fibers which responded to stimulation of the orofacial RFs innervated by the IAN.

After surgery, anesthesia was maintained throughout the experiment by continuous inhalation of halothane (1-2%) mixed with oxygen. During recording sessions, rats were immobilized with pancuronium bromide (1 mg/kg/h, i.v.) and ventilated artificially. Expired CO₂ concentration was monitored (Capstar-100, Cwe, Bioseb, USA) and maintained between 3.0-4.0%. Rectal temperature was maintained at 37-38°C by a thermostatically-controlled heating pad (ATB-1100, Nihon Kohden, Tokyo, Japan) and an electrocardiogram was monitored.

Bipolar electrodes (interpolar distance: 0.5 mm) were inserted into the spinal trigeminal subnucleus caudalis (Vc) 0.8 mm deep from the brainstem surface at the obex level to elicit antidromic spikes from Vc (Figure 7A). An enamel-coated tungsten microelectrode (impedance = 10 MΩ, 1000 Hz) was then advanced carefully through the cortex about 2.5-3.0 mm lateral to the midline and 3.0-4.0 mm anterior to the interaural line until an electrode tip reached the IAN trunk. Then, the electrode was advanced at 1 μm steps and single fiber activity was recorded. IAN unit activities were searched for by applying mechanical stimulation (pressure or brush) to the mental region. When single unit activity was isolated, responses to mechanical stimulation of the facial skin were carefully examined. Because of technical difficulties to approach intraoral structures, only cutaneous facial RFs were mapped. To identify antidromic responses, 1 ms electrical pulses (0.1-0.5 mA) were applied to the Vc. Each neuron was classified as Aβ- (> 7 m/s), Aδ- (7-2 m/s) or C- (< 2 m/s) fibers according to the conduction velocity of the action potentials calculated from the antidromic latency and the distance between recording and stimulating sites 66 .

Graded mechanical stimuli were applied to the most sensitive areas of RFs. Mechanical stimuli consisted of quantitative pressure with von Frey filaments (1, 6, 15, 26 and 60 g), brushing with a camel hair brush and pinch produced by a small arterial clip. After identification of a neuron by brushing the face, graded mechanical stimuli were applied to RFs. To avoid sensitization by noxious stimulation, we did not use repeated noxious stimuli to search for high-threshold mechanosensitive neurons. Neuronal responses were saved on computer disk for subsequent off-line analysis of signals.

The waveforms of each neuron were amplified using a differential amplifier (AB-601G, Nihon Kohden, Tokyo, Japan, high cut: 10 KHz, low cut: 150 Hz) and identified using Spike 2 software (CED, Cambridge, UK). Peristimulus time histograms (bin width = 1 s) were generated in response to each stimulus. Background discharges were first recorded for 120 s before the application of the mechanical stimulation, and they were subtracted from the neuronal responses during the analysis. The mean firing frequency was calculated during mechanical stimulation. Afterdischarges were recorded for 10 s after pinching of the RF. The mechanical stimulation of the RFs was considered to have induced an effect when the mean firing frequency at 5 s after mechanical stimulation differed from mean background discharge rate by ± 2SD.

The IAN-transected rats were tested for mechanical stimulation of the mental region at 14 days after nerve injury. Rats that showed allodynia-like responses to non-noxious mechanical stimulation of the mental skin were used for patch clamp recording experiments. For mental skin stimulation, mechanical stimulation was applied to adjacent regions more than 1 mm distant from the FG injection site.

Rats with mechano-allodynia were sacrificed by decapitation and TG neurons were used for the electrophysiological studies. Acute dissociation of TG neurons was performed as described previously 63 64 . Briefly, rats were anaesthetized with sodium pentobarbital (45 mg/kg, i.p.) and decapitated. The left TG was rapidly removed and incubated for 15-25 min at 37°C in modified Hank's balanced salt solution (130 mM NaCl, 5 mM KCl, 0.3 mM KH₂PO₄, 4 mM NaHCO₃, 0.3 mM Na₂HPO₄, 5.6 mM glucose, 10 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), pH 7.3) containing collagenase type XI and type II (each 2 mg/ml; Sigma-Aldrich, MO, USA). The cells were dissociated by trituration with a fire-polished Pasteur pipette and then plated onto poly-L-lysine-coated coverslips in 35-mm dishes. The plating medium contained Leibovitz's L-15 solution (Invitrogen, Carlsbad, CA, USA) supplemented with 10% newborn calf serum, 26 mM NaHCO₃, and 30 mM glucose. The cells were maintained in 5% CO₂ at 37°C and used for recordings between 2 and 8 h after plating. After incubation, the coverslips were transferred to the recording chamber in a standard external solution containing 155 mM NaCl, 3 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES and 20 mM glucose, pH 7.3.

The composition of the extracellular recording solution used in these experiments is shown in Table 1. The cells were also studied in the presence and absence of TTX to determine which currents were TTX resistant. In these voltage-clamp experiments, 1 μM TTX was added to the extracellular solution, when TTX-R I _Na was recorded in naive and IAN-transected rats. To examine the outward K⁺ current, the solution was replaced with 150 mM choline chloride, 3 mM KCl, 1 mM MgCl₂, 10 mM HEPES, and 20 mM glucose, pH 7.35 24 63 . Some recordings were performed in the current-clamp mode and we used a quasiphysiological recording solution in this study (Table 1). In the current-clamp mode experiments, 1 μM TTX was added to the extracellular solution. All experiments were performed at room temperature (21-26°C).

FG-labeled TG neurons were identified by applying a short pulse of UV light (340-380 nm) and capturing the image of fluorescent cells with a microscope (Nikon, Tokyo, Japan). Locally developed software permitted the superposition of a tracing of the perimeter of the fluorescent cell onto the image of the same cell in the TG visualized with visible light. Whole cell recordings were conducted with the rapid perforated-patch technique 24 50 67 68 69 . Fire-polished patch pipettes (2-5 MΩ) were filled with an internal solution (Table 1) and amphotericin B. In the case of the potassium current recording under voltage clamp, we used the same internal solution as described in current clamp condition (Table 1). Both current- and voltage-clamp recordings were conducted with an Axopatch 200B amplifier (Axon Instr., Foster City, CA, USA). Signals were low-pass filtered at 1 or 5 kHz and digitized at 10 kHz.

Neurons were always bathed in a flowing stream of external solution, except during the application of drugs. After seal formation and membrane perforation, leakage and capacitive transients were reduced by analog circuitry. A series resistance compensation (> 80%) was employed 70 . The recording chamber (volume, 0.5 ml) was mounted on an inverted microscope (Nikon, Tokyo, Japan) equipped with phase-contrast, a video camera and two micromanipulators. The chamber was perfused under gravity with an external solution at approximately 0.5 ml/min. Current density was determined by dividing the peak current evoked by the cell capacitance.

In the voltage-clamp mode, TTX-R I _Na was recorded in naive and IAN-transected rats. TTX-S I _Na was isolated by digitally subtracting TTX-R I _Na from the total I _Na. The current-voltage (I-V) relationship was first monitored by using step pulses (50 ms) from the holding potential of -80 mV to +80 mV in 5 mV increments at 5 s intervals.

In the current-clamp mode, we firstly determined the threshold (1T) for evoking action potential (overshoot of action potential > 30 mV). The threshold was defined as the current values for eliciting depolarizing single pulses (10-400 pA, 300 ms). The firing rate of action potentials was assessed by counting the number of action potentials evoked by the depolarizing pulses (1T, 2T and 3T). The spike amplitude and height of overshoot, and the threshold current, were also assessed in naive and IAN-transected rats as illustrated in Figure 5A.

We also analyzed the K⁺ current in this model. We used voltage protocols modified from a previous study 24 63 64 . Outward K⁺ currents were elicited by stepping to a conditioning voltage of either -40 mV or -120 mV from a holding potential of -60 mV; then the membrane was depolarized from -60 mV to +60 mV in increments of 10 mV; +60 mV produced the largest peak current in each recording. The transient A currents was determined by subtraction of -40 mV protocol from -120 mV protocol. Activation of the currents in standard solution was rapid and decayed only partially during 300 ms depolarization pulses. The amplitudes and rates of rise in the absolute current increased with increasing depolarization.

Results are presented as median for behavioral test, and mean ± SEM for single fibers analysis, patch clamp analysis and FG immunohistochemistry. One-way ANOVA followed by Dunnett's test was used for data from the FG labeling of TG neurons. ANOVA on rank with post-hoc Student-Newman-Keuls test was used for the behavioral data. The Mann-Whitney U test, two-way ANOVA followed by the Dunn's and Holm-Sidak tests were used for data from the single fiber recording experiments, and Duncan's new multiple range test was used for the data from the patch clamp recording experiments. Differences were considered significant at p < 0.05.