INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Neuropathic pain continues to pose a major clinical challenge. To explore neural mechanisms of chronic pain and new approaches to therapy, several animal models of experimental neuropathic pain have been developed and studied. Animals that receive injury to primary sensory neurons exhibit behavioral symptoms of neuropathic pain manifested as mechanical and thermal hyperalgesia or as allodynia. Injury to primary afferents distal to the DRG, i.e., the spinal nerve or sciatic (Bennett and Xie 1988; Devor 1994; Kim and Chung 1992; Seltzer et al. 1990; Wall and Gutnick 1974) or to somata within the DRG (Hu and Xing 1998; Song et al. 1999), produces hyperalgesia accompanied with allodynia. Cellularly, increased excitability and other intrinsic alterations in membrane properties of DRG cells have been demonstrated following peripheral nerve injury or DRG compression. For example, nerve injury or DRG compression reduces the amount of depolarizing current required to evoke an action potential as well as lowering action potential threshold (Abdulla and Smith 2001a,b; Devor 1994; Gurtu and Smith 1998; Stebbing et al. 1999; Zhang et al. 1999). These findings support the hypothesis that increased excitability of DRG cells is associated with the generation and maintenance of hyperalgesia and plays important roles in neuropathic pain.

Recently, it has been demonstrated that injuries to the dorsal root can lead to neuropathic pain behavior (Colburn et al. 1999; Eschenfelder et al. 2000; Sheth et al. 2002; Tabo et al. 1999). Tabo et al. (1999) found that dorsal root ligation produces a pronounced mechanical allodynia and enlarges the cutaneous receptive fields of dorsal horn neurons but does not produce thermal hyperalgesia. On the other hand, Sheen and Chung (1993) did not find neuropathic pain behavior after a dorsal rhizotomy. The reasons for the discrepancies among these studies on the dorsal root injury are unclear. However, these studies suggest that the behavioral and cellular effects of nerve injury to the central branches (dorsal root)compared with injury to peripheral branches-of primary afferent neurons or to their somata, i.e., DRG neurons, may be not comparable. While peripheral nerve injury is known to increase excitability of DRG cells and contribute to chronic pain, cutaneous hyperalgesia and allodynia via central sensitization of nociceptive neurons in the dorsal horn (Hökfelt 1997; Hökfelt et al. 1997; Ji and Woolf 2001), the effect of dorsal root injury that produces behavioral hyperalgesia on the excitability of DRG neurons is unknown. Therefore this study was designed to investigate and compare the possible differences in hyperalgesia produced by injuries to central and peripheral branches of axons and somata of DRG. The central and peripheral branches of axons of DRG neurons were injured by partial dorsal rhizotomy (PDR) (Song et al. 2000) and chronic constriction injury of sciatic nerve (CCI) (Bennett and Xie 1988), respectively. The DRG somata were injured by chronic compression of DRG (CCD) (Hu and Xing 1998; Song et al. 1999). Second, we examined if the increased excitability of DRG cells is associated with hyperalgesia induced by PDR as well as that by CCD and CCI.

Furthermore, it is assumed in chronic pain states that increased excitability of DRG cells triggers chronic changes in excitability and/or synaptic plasticity of dorsal horn neurons, which results in hyperalgesia. Central sensitization is thought to be mediated by inflammation- or injury-evoked activation of N-methyl-D-aspartate (NMDA) receptors on postsynaptic dorsal horn neurons (Dickenson et al. 1997). In addition, it has been shown that NMDA receptor antagonists can reduce neuropathic pain (Chaplan et al. 1997; Kim et al. 1997). However, there is no data showing the roles of NMDA receptors in CCD- and PDR-induced hyperalgesia. Therefore the third goal of the present study was to examine and compare the spinal antinociceptive effects of NMDA receptor antagonists in rats that received PDR, CCD, or CCI.

Preliminary results of the present study have been published in abstract form (Song et al. 2000).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Surgical procedure

ANIMALS. Experiments were performed on adult, male Sprague-Dawley rats (n = 212, 200-220 g). The rats were housed in groups of four in plastic cages (40 × 60 × 30 cm) with soft bedding and free access to food and water under a 12-h day/12-h night cycle. They were kept 5-7 days, under these conditions, before and 100 days, for some rats, after surgery. The animals were divided into different groups as described below (PDR, CCD, CCI, Sham surgery, and unoperated control). All surgeries were done under anesthesia induced by sodium pentobarbital (40 mg/kg, ip). After surgery, the muscle and skin layers were sutured. An oral antibiotic, Augmentin, was administered after surgery in the drinking water for each rat (7.52 g in 500 ml) for 7 days. These procedures were reviewed and approved by Parker Research Institute Animal Care Committee.

PDR. Rats (n = 60) were anesthetized, and a midline incision was made from T₁₂-L₃. The paraspinal muscles were separated from the spinal processes on the left side. The transverse processes of L₁ and L₂ were exposed by scraping off attached ligaments and a small "window" laminectomy was performed unilaterally at L₁-L₂ to expose L₄ and L₅ dorsal roots central to the DRGs and close to the spinal cord (Fig. 1). The dura was opened approximately 5 mm in length, and the L₄ and L₅ dorsal roots were identified under a dissecting microscope. The rootlets of L₄ and L₅ were carefully isolated from each other. Normally there are four rootlets within each dorsal root. The rostral half of the rootlets (2 of 4) were transected (approximately 1-2 mm), whereas the caudal half were kept intact. Procaine (2%) was dropped onto the rootlets before transection.

View larger version (55K): [in this window] [in a new window]   Fig. 1. A schematic of the animal models of neuropathic pain produced by partial dorsal rhizotomy (PDR), chronic constriction injury of sciatic nerve (CCI), and chronic compression of dorsal root ganglion (DRG) (CCD).

CCD. Compression was produced by surgically implanting stainless steel rods unilaterally into the intervertebral foramen at L₄ and L₅ (Fig. 1, n = 60). The procedure for CCD has been described previously (Hu and Xing 1988; Song et al. 1999). In brief, the rats were anesthetized, paraspinal muscles were separated from the mammillary and transverse processes, and the intervertebral foramina of L₄ and L₅ was exposed. A stainless steel L-shaped rod, 4 mm in length and 0.6 mm in diameter, was implanted into each foramen, one at L₄ and the other at L₅.

CCI. The procedure was the same as that described in the CCI model by Bennett and Xie (1988). The rats (n = 60) were anesthetized, and the left common sciatic nerve was exposed at the level of the middle of the thigh. Proximal to the sciatica's trifurcation, about 7 mm of nerve was freed of adhering tissue and four ligatures (4-0 chronic gut) were tied loosely around it with about 1 mm spacing. The length of nerve affected was about 4-5 mm long (Fig. 1).

SHAM SURGERY. Rats (n = 12) were evenly divided into three groups and received sham surgery. The surgical procedure was identical to that described in PDR, CCD, or CCI, but without injury to the dorsal root, DRG, or sciatic nerve (4 rats in each group).

CONTROL. Another group of rats (n = 20) served as control and did not receive surgery or injury.

IMPLANTATION OF INTRATHECAL CATHETERS AND DRUG APPLICATION. For the purpose of intrathecal injection (it) and examination of the effects of NMDA receptor antagonists on chronic pain, intrathecal catheters were implanted in rats (n = 152) using a modification of the method described by Yaksh and Rudy (1976). The catheters were extended to the rostral edge of the lumbar enlargement (approximately 7.0-7.5 cm from the cisterna magna). The catheter was inserted 30 min before PDR, CCD, or CCI on the day of surgery. Rats showing postoperative neurological deficits and/or inflammation were not studied. D-2-Amino-5-phosphonovaleric acid (APV) and dizocilpine maleate (MK-801) were used for examining the role of NMDA receptors in modulating chronic pain in the spinal cord. APV (Sigma, 20 µg) and MK-801 (RBI, 20 µg) were dissolved in 0.9% saline and intrathecally administered 20 min prior to surgery (pretreatment) or after the third postoperative test day (posttreatment) in a 10 µl volume. Saline was intrathecally injected as a control.

Behavioral testing

The Hargreaves test (Hargreaves et al. 1988) was used to determine the presence of thermal hyperalgesia by measuring foot withdrawal latency to heat stimulation. Each rat was placed in a box (17 × 22 × 14 cm) containing a smooth glass floor. The temperature of the glass was measured and maintained at 25 ± 1°C. A heat source (IITC Model 336 Analgesia Meter) was focused on a portion of the hindpaw, which was flush against the glass, and a radiant thermal stimulus was delivered to that site. The stimulus shut off automatically when the hindpaw moved (or after 20 s to prevent tissue damage). The intensity of the heat stimulus was constant throughout all experiments. The elicited paw movement occurred at a latency of approximately 9-12 s in control rats. Thermal stimuli were delivered six times to each hind paw at 5- to 6-min intervals. For assessment of thermal hyperalgesia, the withdrawal latencies on the contralateral side were subtracted from those on the experimental side, and the result was expressed as a difference score. The rats were tested on each of 2 successive days prior to surgery. Postoperative tests were conducted 1, 3, 5, 7, 10, and 14 days after surgery and once weekly for 14 wk for examining the time course of hyperalgesia. For the rats receiving pretreatment of NMDA receptor antagonists, postoperative tests were conducted 1, 2, 3, 5, 7, 10, and 14 days after injection. For the rats receiving postoperative treatment of NMDA receptors antagonists on the third day after surgery, additional tests were conducted 6, 12, 24, and 36 h after injection.

Electrophysiological studies

An in vitro preparation of DRG was made from L₄ and/or L₅ ganglia for further electrophysiological studies. The procedure was similar to that we previously described (Zhang et al. 1999). In brief, the rat was anesthetized with sodium pentobarbital. The sciatic nerve was isolated from surrounding tissue, transected at the mid-thigh level, and its proximal portion traced to the ganglia. A laminectomy was then performed and the L₄ and L₅ DRGs, and their dorsal rootlets were identified. The locations of the rods and the previously transected dorsal roots were checked immediately on exposing the ganglia and the dorsal roots in CCD and PDR rats. Oxygenated artificial cerebrospinal fluid (ACSF), consisting of (in mM) 130 NaCl, 3.5 KCl, 1.25 NaH₂PO₄, 24 NaHCO₃, 10 dextrose, 1.2 MgCl₂, and 1.2 CaCl₂ (pH =7.3), was dripped periodically onto the surface of the ganglion during the surgical procedure to prevent drying and hypoxia. The ganglion was removed from the rat and placed in a 35-mm Petri dish filled with oxygenated ACSF. Under a dissecting microscope, the perineurium and epineurium were peeled away from the ganglion with fine forceps, and the attached sciatic nerve and dorsal roots were transected adjacent to the ganglion. The ganglion was then placed in the recording chamber and mounted on the stage of an upright microscope (BX50-WI, Olympus). A U-shaped stainless steel rod with four pieces of silver wire crossed from one side to the other was used to gently hold the ganglion in place within the recording chamber. The DRG was perfused continuously with oxygenated ACSF at a rate of 2 ml/min. The temperature was maintained at 35 ± 1°C (SD) by a temperature controller (TC-344B, Warner Instruments, Hamden, CT).

Intracellular recordings were made in vitro from the DRG somata using conventional bridge-balance techniques (Axoclamp-2B, Axon Instruments, Foster City, CA) and analyzed with PCLAMP-8 under Windows 98 (Axon Instruments). DRG cells were visualized under a microscope using differential interference contrast. Somata of the DRG cells were classified visually by their diameter as small (30 µm), medium (31-49 µm), or large (50 µm). Glass microelectrodes were fabricated with a Flaming/Brown micropipette puller (Model P-97/PC, Sutter Instruments.) and filled with 2 M potassium acetate (pH = 7.2). Satisfactory recordings were obtained with electrodes having DC resistance ranging from 20 to 60 M.

To compare the excitability of DRG neurons from PDR, CCD, CCI, and unoperated control rats, we examined the threshold current, action potential (AP) threshold, resting membrane potential (V_m), input resistance (R_in), afterhyperpolarization (AHP), patterns of discharges evoked by depolarizing current, and other electrophysiological properties of DRG neurons. The V_m was taken 2-3 min after a stable recording was first obtained. Depolarizing currents of 0.05-2 nA (100 ms duration) were delivered in increments of 0.05-0.1 nA (for small cells) or 0.1-0.2 nA (for medium and large cells) until an AP was evoked. Threshold current was defined as the minimum current required to evoke an AP. AP voltage threshold was defined as the first point on the upstroke of an action potential where the rising rate exceeded 50 mV/ms (Anderson et al. 1987). AP amplitude was measured from the AP threshold to the peak. AP duration was measured at threshold voltage. AHP amplitude was measured from the base line to the valley peak, and AHP duration was measured as an interval from the onset of AHP to the point of 50% decay of AHP. The R_in for each cell was obtained from the slope of a steady-state I-V plot in response to a series of hyperpolarizing currents, 100 ms duration, delivered in steps of 0.1-0.2 nA from -2 to 0.5 nA. Repetitive discharge of each cell was measured by counting the spikes evoked by intracellular injection of depolarizing currents at 2.5 × threshold strength (1,000 ms). We classified the discharge patterns of the DRG neurons into two types: 1) cells firing either one or two APs and 2) cells firing >2 APs.

Statistical tests

Difference scores of the latency over time were tested with one-way ANOVA followed by Newman-Keuls posttests. Student's t-test was used to identify the significant differences in electrophysiological measurements between controls and any of the operated groups. All data are presented as mean ± SE. Unless otherwise stated, statistical results described as significant are based on a criterion of P < 0.01.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Hyperalgesia produced by PDR, CCD, and CCI

All rats that received PDR, CCD, or CCI in the different groups described below produced clear behavioral signs of thermal hyperalgesia. The hyperalgesia was indicated by the significantly decreased latency of foot withdrawal to heat stimulation of plantar surface of hindpaw ipsilateral to the injury and expressed as the negative difference scores (Figs. 2-4).

View larger version (20K): [in this window] [in a new window]   Fig. 2. Time course of thermal hyperalgesia produced by PDR, CCD, and CCI. Thermal hyperalgesia is indicated by a negative difference score (±SE) obtained on each day of testing for rats tested ipsilateral and contralteral to PDR, CCD, or CCI. Eight rats were tested in each group. CCD and CCI rats exhibited significant thermal hyperalgesia compared with the average of the preoperative tests and the tests from control rats (*P < 0.01). Hyperalgesia lasted 9 and 12 wk, respectively. PDR rats exhibited significant thermal hyperalgesia compared to the control (*P < 0.01). However, the severity and duration of hyperalgesia was significantly less and shorter than that produced by CCD or CCI (#P < 0.01).

View larger version (18K): [in this window] [in a new window]   Fig. 3. Pretreatment of D-2-amino-5-phosphonovaleric acid (APV) and dizocilpine maleate (MK-801) occluded and delayed production of thermal hyperalgesia produced by (A) PDR, (B) CCD, and (C) CCI. Thermal hyperalgesia is indicated by a negative difference score (±SE). Eight rats were tested in each group. APV and MK-801 were intrathecally applied 30 min prior to surgery (arrow). Saline (0.9% NaCl) was applied as control.

View larger version (18K): [in this window] [in a new window]   Fig. 4. Posttreatment of APV and MK-801 (at arrow) inhibited thermal hyperalgesia produced by (A) PDR, (B) CCD, and (C) CCI. Eight rats were tested in each group. APV and MK-801 were injected, respectively, 30 min after the 3rd day test. Saline (0.9% NaCl) was applied as control.

Figure 2 shows the time course of the thermal hyperalgesia in PDR, CCD, and CCI rats (n = 8 in each group). Before surgery, there was no significant difference between the values of latency of withdrawal from both feet in each rat in groups of PDR, CCD, CCI, and unoperated control, and therefore the difference scores from two preoperative test sessions in each group clustered around zero. In the first postoperative day test, PDR, CCD, and CCI rats showed profound thermal hyperalgesia. The difference scores were significantly reduced from the preoperative values of 0.14 ± 0.18, 0.09 ± 0.15, and 0.11 ± 0.17 (average values of the preoperative tests) to -1.7 ± 0.38, 4.5 ± 0.42, and -3.0 ± 0.35 for PDR, CCD, and CCI rats, respectively. Peak hyperalgesia was seen in the first 4 wk, and then it waned gradually. It is worthy to note that the severity of thermal hyperalgesia produced by PDR was significantly less than that produced by CCD or CCI. This was shown by comparing the total population of difference scores for each group across all postoperative days using a one-way ANOVA followed by Newman-Keuls posttests. The negative difference scores of PDR rats were approximately 40-50% of those from CCD or CCI rats during the first 5 wk, while the duration of PDR-induced hyperalgesia was 5 wk, which is shorter than that of CCD (9 wk) and CCI (12 wk).

In addition, slightly shortened withdrawal latency in the contralateral feet during the first postoperative week was also found in 5/8 CCI, 4/8 CCD, and 1/8 PDR rats. These changes were much less than that in the ipsilateral feet and therefore did not alter the direction of the negative difference scores as shown in Fig. 2. There was no significant difference between withdrawal latencies of the feet in sham and unoperated control rats. The other rats that received PDR, CCD, or CCI and were used for examining antinociceptive effects of NMDA receptor antagonists (Figs. 3 and 4) and for electrophysiological studies also showed thermal hyperalgesia as discussed in the related paragraphs below.

The chronic pain states of the affected hindpaw were confirmed not only by the shortened latency to heating, but also by abnormal topography of the responses. The rats that received PDR, CCD, or CCI developed varying degrees of abnormality in gait and posture, which may serve to minimize aversive sensory stimulation, as previously suggested by Bennett and Xie (1988) and Song et al. (1999) in models of CCI and CCD. For example, the rats were often seen to raise the affected hind paw from the floor and hold it in a protected position next to the flank while standing or sitting. The affected hind paw was placed clumsily while walking, and the toes, which are normally spread apart while walking or standing, were together and slightly ventroflexed. When a mechanical or heat stimulus was applied to the hindpaw during preoperative testing, the reflex withdrawal was of small amplitude and brief, typically lasting 1-2 s. Postoperative withdrawal to the same stimuli delivered ipsilateral to the injured side was typically of greater amplitude and longer duration, with the paw held in the air 2-5 s and sometimes 20-60 s. Such prolonged paw lifts were accompanied by exaggerated aversive behavior such as licking the stimulated paw and/or pulling on the nail with the mouth. PDR rats appeared to exhibit much less spontaneous pain-like and aversive behavior, which is consistent with less thermal hyperalgesia. Unfortunately, we did not collect data for quantitatively analyzing this apparent difference.

Effects of NMDA receptor antagonists on thermal hyperalgesia

Antinociceptive effects of NMDA receptor antagonists APV and MK-801 on thermal hyperalgesia were examined in PDR, CCD, and CCI rats (n = 48 in each group). The expression of hyperalgesia was transiently blocked by pre- and postoperative i.t. of APV and MK-801, respectively. Figure 3 shows that pretreatment of APV or MK-801 (n = 8 in each group) delayed the emergence of PDR-, CCD-, and CCI-induced hyperalgesia for approximately 2-3 days. Similarly, as shown in Fig. 4, postoperative i.t. of APV and MK-801 (n = 8 in each group) transiently blocked the hyperalgesia produced by PDR, CCD, and CCI. Such inhibition lasted for approximately 24-36 h. In contrast, saline i.t. (n = 8 in each group) did not affect thermal hyperalgesia in either group. An additional test showed that neither APV (n = 4) nor MK-801 (n = 4) affected the withdrawal response to thermal stimulation on unoperated control rats.

Effects of PDR, CCD, and CCI on excitability of DRG cells

Table 1 summarizes the electrophysiological properties of DRG cells (n = 335) recorded from the PDR, CCD, CCI, and the unoperated control rats. These cells were silent but not spontaneously active. They were classified as small, medium, and large according to the diameter of its soma visualized under a microscope as described in the methods. Evaluation of the differences in excitability of DRG cells from each group of rats was made for cells of comparable sizes.

Excitability of DRG cells was evaluated by any changes in the threshold current, V_m, AP threshold, R_in, AHP, etc. Significant changes were found in threshold current and AP threshold for all of the three categories of cells from CCD and CCI DRGs. The mean threshold currents decreased approximately 50-70% (Fig. 5A), and the mean AP voltage thresholds lowered approximately 20% (Fig. 5B). In addition, AP duration in small-sized cells was significantly longer in CCD and CCI cells (Table 1). There were no significant changes found in the other parameters measured such as V_m, R_in, AHP, etc. These results are generally consistent with previous findings observed in CCD and CCI rats (Abdulla and Smith 2001a; Zhang et al. 1999). However, we did not find significant changes in the threshold currents and AP voltage thresholds and any of the parameters measured for each class of cells from PDR-rats (Fig. 5, A and B; Table 1).

View larger version (34K): [in this window] [in a new window]   Fig. 5. Alterations in threshold current (A) and AP threshold (B) produced by PDR, CCD, and CCI. Excitability of the DRG cells was significantly increased indicated by the decreased threshold current and the lowered AP voltage threshold. *P < 0.01 (Student's t-test).

Excitability of DRG cells was also evaluated by examining the discharge patterns during the response to intracellular injection of depolarizing current at 2.5 × threshold strength (1,000 ms) as described in METHODS. Examples of typical responses of DRG cells to the depolarizing current are shown in Fig. 6, A-D. About 50-60% of CCD (n = 78) and CCI (n = 83) DRG cells exhibited multiple APs to the depolarizing current, and the rest of the cells exhibited one or two APs. In contrast, only 21.5% of the cells (n = 79) from unoperated control ganglia discharged multiple APs, and most of the cells exhibited one or two APs. The incidence of multiple APs was significantly higher in CCD and CCI cells than in control cells (P < 0.01 in each case, Fisher's exact test). However, the incidence of multiple APs in PDR cells (24.2%, n = 95) was about the same as that in control cells. The data are summarized in Fig. 6, E-G.

View larger version (22K): [in this window] [in a new window]   Fig. 6. Alterations in the repetitive discharge characteristics of DRG cells produced by PDR, CCD, and CCI. As stated in the method that the threshold current for each cell was obtained 1st by a series intracellular injection of depolarizing current at 50 ms duration. The increased depolarizing current (2.5 × threshold) with longer duration (1,000 ms) evoked different patterns of discharges from DRG cells. Examples of the different discharge patterns recorded from control cells are shown in A-D. A: a small-sized cell (22 µm diam) with single spike. B: a small-sized cell (26 µm diam) with 2 spikes. C: a small-sized cell (25 µm diam) with multiple discharges. D: a small-sized cell (21 µm diam) with multiple spikes. Effects of PDR, CCD, and CCI on the discharges of the 3 categories of cells are summarized in E-G. Both CCD and CCI significantly increased the number of discharges and changed the discharge patterns (*P < 0.01, 2 test), while PDR did not change the discharge patterns (#P > 0.05) compared with those in control.

In addition, spontaneous activity, one of the signs of hyperexcitability of the DRG cells, was also recorded in the present experiment. A cell was defined as spontaneously active if the spontaneous firing lasted 2 min after a stable recording was first obtained. Only one large cell exhibiting spontaneous activity was observed from control cells (1.3%, n = 80). In contrast, 8 of 86 (9.3%) CCD and 7 of 90 (7.7%) CCI cells were spontaneously active. These results are consistent to the previous findings in the nerve injured (e.g., Devor 1994; Wall and Devor 1983) and DRG compressed (Song et al. 1999; Zhang et al. 1999) rats. However, of the 98 PDR cells, only 3 (3.1%) were spontaneously active. This incidence is closer to that in the control group but less than that in CCD and CCI groups. The discharge patterns of the spontaneous active cells were similar to those described previously (Song et al. 1999; Xie et al. 1995; Zhang et al. 1999).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study investigated and compared changes in hyperalgesia, excitability of DRG neurons, and spinal antinociception of NMDA receptor antagonists in rats that received injuries to the central (PDR) and peripheral (CCI) branches of axons and somata (CCD) of DRG neurons. There are four principle findings: 1) as described previously for CCD and CCI, PDR can induce hyperalgesia; 2) PDR-induced hyperalgesia is briefer and less severe than CCD- and CCI-induced hyperalgesia; 3) the excitability of DRG cells increases following CCD and CCI, but not PDR; and 4) NMDA receptor antagonists APV and MK-801 inhibit the expression of hyperalgesia in PDR, CCD, and CCI rats. These studies provide new evidence to support a role for injury-induced plasticity of primary sensory neurons to chronic pain and also indicate that injury to different regions of the sensory neuron can have different effects on hyperalgesia.

The result of PDR induced-hyperalgesia in our present study is somewhat different from those recently reported by Tabo et al. (1999) and Sheth et al. (2002) in their models of dorsal root constriction (DRC) and dorsal rhizotomy, respectively. Tabo et al. reported that DRC produced mechanical allodynia with a lack of thermal hyperalgesia. Sheth et al. found that completely dorsal rhizotomy led to a transient decrease in mechanical threshold. A possible reason for these discrepancies may be that the injuries applied to the dorsal root were different among the studies. In the present study, the two rostral rootlets were transected, whereas the two caudal rootlets were kept intact so that impulses from the connected peripheral nerve or receptive field could still be conducted to the CNS. In Sheth et al.'s model, the dorsal root was completely cut. In Tabo et al.'s study, the dorsal root was constricted with silk ligatures and might have been completely cut if ligated too tightly. Therefore injuries in their models caused less hyperalgesia (Sheth et al.), hypoalgesia that lasted for 1 wk (Tabo et al.), and even an absence of hyperalgesia after dorsal rhizotomy (Sheen and Chung 1993). In light of these differences, we speculate that the remaining intact dorsal rootlets in the present model are important for producing enhanced hyperalgesia. In addition, the present study further confirms and extends our previous findings that CCD produces rapid onset hyperalgesia in which mechanical and thermal hyperalgesia lasts up to 5 wk (Song et al. 1999). We now present relatively complete behavioral tests showing that the thermal hyperalgesia following CCD can last up to 9 wk. The CCI-induced hyperalgesia is in general consistent with previous findings first described by Bennett and Xie (1988).

Why PDR produced significantly less severe and briefer hyperalgesia than CCD and CCI is unknown. However, it has been hypothesized that increased excitability of DRG cells contributes to the generation and maintenance of hyperalgesia in chronic pain states. Several lines of study on the primary sensory neuron injury support this hypothesis for peripheral nerve or DRG somata injury (Burchiel 1984; Hu and Xing 1998; Hu et al. 2000; Ji and Woolf 2001; Kajander and Bennett 1992; Song et al. 1999; Wall and Devor 1983; Xie et al. 1995; Zhang et al. 1997, 1999). Does this mean that the excitability of the DRG cells may be different in the three models we tested? Interestingly, our present studies show that PDR, unlike CCD and CCI, does not increase the excitability of DRG neurons. These findings suggest that the lesser severity and shorter duration of the PDR-induced hyperalgesia may, in part, result from the lack of hyperexcitability of the DRG cells.

Nerve injury can trigger and sensitize specific nociceptive or wide-dynamic-range (WDR) spinal dorsal horn neurons, a mechanism widely considered to induce hyperalgesia (Hökfelt 1997; Hökfelt et al. 1997). On the other hand, nerve injury can also trigger chronic changes in DRG cells, resulting in spontaneous activity arising from ectopic foci in the injured and/or neighboring intact C-fibers. The sensory bombardment is thought to produce central sensitization that accounts, at least partially, for the hyperalgesia (Hökfelt 1997; Hökfelt et al. 1997). Therefore we speculate that PDR initially triggers and sensitizes the specific nociceptive or WDR-type dorsal horn neurons, but does not alter the excitability of DRG cells. Consequently, there is a lack of constant injury signal from the somata of DRG cells to facilitate the enhanced excitability of dorsal horn neurons, thus increasing and maintaining the hyperalgesia. Previously, a study showed that compression of DRG excited WDR-type dorsal horn neurons in cat spinal cord throughout the duration of the compression stimulus, while similar compression of the dorsal roots only transiently excited dorsal horn neuron (Hanai et al. 1996). This indicates that the dorsal root is less sensitive than DRG to a mechanical stimulus, providing further evidence for a difference between injuries to the DRG and dorsal root.

Why do injuries to the peripheral branches of axon or somata increase excitability of DRG cells, but injury to the central branches of axon do not? Although there are no clear answers to this question, some studies from mammals and other species have provided evidence that may offer an explanation. The peripheral axons of primary sensory neurons retrogradely transport cytokines, such as nerve growth factor (NGF) and tumor necrosis factor (TNF), from their peripheral targets back to their cell bodies (somata) in the DRG. Immune and Schwann cells, activated by axonal injury and inflammation, also release cytokines that may activate receptors or be taken up by injured and intact axons. This is likely to lead to the transport of signals to the somata, which may trigger additional cellular reactions such as glial cell activation and immune cell infiltration into the DRG (George et al. 1999; Hu and McLachlan 2002; Schafers et al. 2002; Tonra et al. 1998; Wagner and Myers 1996). Walters and Ambron (1995) hypothesized that axon injury unmasks nuclear localization signals in certain axonplasmic proteins at the injured site. This causes the proteins ("positive injury signals") to be transported retrogradely to the somata of sensory neurons and activate transcription factors. These factors then induce the expression of early and late genes, which finally induce functional alterations (Ambron and Walters 1996; Ambron et al. 1996; Gunstream et al. 1995). These positive signals coupled with negative signals produced by interrupted transport of trophic signals from peripheral targets trigger alterations in expression of neuropeptides, receptors, and ion channels in neuronal somata (Hökfelt et al. 1997; Waxman 1999; Waxman et al. 2000). For example, 2 wk after transection of the sciatic nerve, DRG somata exhibit novel expression of a TTX-S type III Na⁺ current, a decrease in TTX-R Na⁺ current (Rizzo et al. 1995; Waxman 1999), a reduction in K⁺ currents (Everill and Kocsis 1999), and a reduction in an N-type component of high voltage-activated Ca²⁺ currents (Abdulla and Smith 2001; Baccei and Kocsis 2000). In addition, inhibitors of axonal transport can block axonal injury-induced hyperexcitability of sensory neurons (Gunstream et al. 1995). Based on these studies, we suggest that injury produced by CCD or CCI activated positive injury signals that were transported retrogradely to the DRG soma to induce hyperexcitability. In contrast, PDR may have activated injury signals in the dorsal root that were mainly transported to the central terminals in the spinal cord rather than to DRG cells and their nuclei. Consequently, injury to the dorsal rootlets might not induce long-term alterations in DRG cells. Future studies are necessary to test this hypothesis.

NMDA receptors have been established to mediate sensitization of the dorsal horn neurons, which are critical for hyperalgesia (Burchiel 1984). Our present study provides new data to support this hypothesis and confirms and extends previous findings that NMDA receptor antagonists can block neuropathic pain produced by peripheral nerve injury (Baccei and Kocsis 2000; Hökfelt et al. 1997; Kim et al. 1997). The present study also demonstrates, for the first time, that thermal hyperalgesia produced by CCD and PDR are inhibited by intrathecal administration of NMDA receptor antagonists APV or MK-801. These results indicate that NMDA receptors may play key roles in mediating and/or modulating hyperalgesia resulting from injury to the central as well as peripheral branches of axon, and to somata of the primary sensory neurons.

In summary, our studies show that hyperalgesia can be induced by injury to the central as well as the peripheral branches of axons and somata of primary sensory neurons, while the severity and duration of the hyperalgesia is less in injury to the central branches compared with that to the peripheral branches and DRG somata. Increased excitability of DRG cells is associated with hyperalgesia produced by injury to the peripheral branches of axon and somata, but not the central branches of axon of the primary sensory neurons. NMDA receptors may play important roles in mediating and/or modulating generation and persistence of hyperalgesia resulting from injuries to any regions of the primary sensory neurons in the spinal cord. These findings indicate that injury to the dorsal root, compared with injury to the peripheral nerve or DRG somata, has different effects on the development of hyperalgesia. These contributions involve different excitability mechanisms in the primary sensory neurons, but each involves pathways (presumably in the spinal cord) that depend on NMDA receptors.