Perineural α^2A-Adrenoceptor Activation Inhibits Spinal Cord Neuroplasticity and Tactile Allodynia after Nerve Injury

After receiving Animal Care and Use Committee approval at each institution, male Wistar rats were prepared for behavioral (Universite Catholique de Louvain) or immunohistochemical (Wake Forest University) studies. Briefly, under halothane anesthesia, the right sciatic nerve was exposed at the proximal thigh level and liberated from surrounding tissues. One-third to one-half of the nerve was then tightly ligated with silicon-treated silk suture (Prolene 7–0) as previously described, 8then muscle and skin layers were closed. The animals were allowed to recover for 4–6 weeks, at which time hypersensitivity to light touch was clearly evident. To investigate the effect of intrathecal clonidine, lumbar spinal catheters were inserted in some animals through an incision in the cisternal membrane as previously described. 10 

Sensitivity to mechanical stimulation was assessed by application of calibrated von Frey filaments (Stoelting, Wood Dale, IL). Animals were placed on a plastic mesh floor in individual clear boxes. After accommodation to the environment, von Frey filaments were applied vertically to the plantar surface of the ligated hind paw and pressed to the point of bending 3 to 4 times over approximately 2 s. If no response was elicited, a larger filament was applied in the same manner. The filaments were applied in increasing order until a brisk withdrawal or paw flinching occurred, which was considered a positive response. This withdrawal threshold was determined three times, with testing separated by 5–10 min, and the mean withdrawal threshold was used for data analysis.

For perineural injection, rats were briefly anesthetized with 4% halothane in 100% oxygen by inhalation. Percutaneous sciatic nerve block was performed on the injured side by introducing a 25-gauge needle between the greater trochanter and the ischial tuberosity as previously described. 11A total volume of 0.3 ml was injected in a fanning motion along the path of the sciatic nerve. Animals received either perineural injection of saline (n = 6) or clonidine, 3 μg (n = 9), 15 μg (n = 8), or 30 μg (n = 11). Withdrawal threshold to von Frey filament testing was determined 5 h and 1, 3, 7, 10, 14, and 21 days after perineural injection.

Additional studies examined the site and mechanism of action of clonidine. To determine the site of action, both systemic and intrathecal controls were performed. In these studies, animals received perineural saline in addition to either intramuscular clonidine, 30 μg (n = 11) or intrathecal clonidine (n = 12). Intrathecal clonidine was administered in a cumulative manner, with total doses of 7.5, 15, and 30 μg and injections separated by 30 min. Because the duration of inhibition of mechanical hypersensitivity from intrathecal clonidine after nerve injury has been demonstrated to be approximately 3 h, 12a 21 day time course was not performed in these animals. To determine the receptor type activated by clonidine, animals received perineural clonidine, 30 μg alone or with α^2A-preferring adrenergic antagonist, BRL44408, 135 μg, 15 μg, and 45 μg (n = 6 in each group). To determine the role of transient neural blockade, animals received perineural ropivacaine, 1,500 μg (n = 11).

Four weeks after nerve injury and 1 week after perineural injection of saline or clonidine, 30 μg, rats were anesthetized with sodium pentobarbital and perfused intracardially with cold phosphate buffered saline (PBS) containing 1% sodium nitrite followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB, pH 7.4). The L4–L5 spinal cord segments and the ipsilateral and contralateral sciatic nerve segments at both proximal and mid-thigh levels were removed and postfixed in the same fixative for 3–6 h. Then all tissues were transferred to 30% sucrose in 0.1 M PB at 40°C for cryoprotection.

The spinal cord segments were cut on a cryostat at a 40 μm thickness. The free-floating sections were collected in PBS. After being pretreated with 0.3% hydrogen peroxide (H²O²) and 10% normal goat serum (NGS, Vector Laboratories, Burlingame, CA), sections were incubated for 36 h at 40°C in a rabbit polyclonal antiphospho-CREB antibody (1:1,000, New England Biolabs, Beverly, MA, USA) diluted in phosphate buffered saline containing 0.3% Triton-X 100 (PBS + T) and 3% NGS. Subsequently, the sections were incubated in biotinylated goat antirabbit IgG (Vector) and further processed using Elite Vectastain ABC kit (Vector) according to the instructions of the manufacturer. Between each incubation, sections were washed in PBS + T (15 min, twice). Finally, the immunoprecipitates were developed by 3, −3’ diaminobenzedine and the chromogen was enhanced by the glucose oxidase-nickel-3, −3’ diaminobenzedine method. 14 

Four sciatic nerve segments, ipsilateral at proximal thigh level (containing ligation), contralateral at proximal thigh level, ipsilateral at mid-thigh level, and contralateral at mid-thigh level, were mounted on the same block using O.C.T. mounting media. Sections were cut at 10 μm thickness, thaw mounted on precleaned Superfrost-plus slides (Fisher Scientific, Pittsburgh, PA) and kept at −80°C until used. After pretreatment in 0.1% H²O²and 10% NGS in 0.1 M Tris buffered saline (TBS), sections were incubated in a polyclonal rabbit antibody directed against α^2A-adrenergic receptor (generously provided by Robert Elde, Ph.D., Dean of Neuroscience, University of Minnesota, Minneapolis, MN). Subsequently, the sections were incubated in biotinylated goat antirabbit IgG (Vector) and further processed using Elite Vectastain ABC kit (Vector) according to the instructions of the manufacturer. Finally, the immunoprecipitates were developed by the glucose oxidase-nickel-diaminobenzedine method described previously. Between each incubation, sections were washed in TBS containing 0.05%+ T (15 min, twice).

Four normal rats,four nerve injured rats, and four nerve injured rats 1 week after perineural injection of clonidine, 30 μg were used for quantification. Five L4–L5 spinal cord sections were randomly selected from each rat. Images of both ipsilateral and contralateral dorsal horns of nerve injured rats or from the dorsal horn of normal rats were captured at ×250 magnification using a CCD camera. The number of digitized pixels overlaid pCREB-IR cells and the number of pCREB-IR cells in the fixed area on each side of the dorsal horn was measured automatically using image analysis software (SigmaScan, Jandel Scientific Inc., San Rafael, CA). In a fixed area (235 μm × 235 μm) that covered the entire LI to LIII in depth and was located in the middle one-third of the mediolateral extent of the dorsal horn, a threshold at which all pCREB-IR cells were overlaid by digitized pixels was applied to all sections measured. To examine possible changes in expression of pCREB within cells, the mean optical density of all positive (supra-threshold) objects was compared.

To determine if α^2A-adrenoceptor IR cells in injured sciatic nerve also coexpress ED1, a marker for macrophages, CD2, a marker for mature T lymphocytes, or S100, a marker for Schwann cells, 10 μm thick sciatic nerve sections were used. Sections were incubated in 10% NGS in 0.1 M TBS for 2 h and then in a rabbit polyclonal anti-α^2A-adrenoceptor antibody 1:1,000 diluted in TBS + T containing 10% NGS. The incubation lasted for 18 h at 40°C. After thorough rinsing, sections were incubated in a goat antirabbit IgG conjugated with Alexa Fluor 488 (1:400, Molecular Probes, Eugene, OR) diluted in TBS containing 10% NGS for 1 h. After rinsing with TBS + T, sections were further incubated in either a mouse monoclonal antibody raised against ED1 (1:500, Serotec, Raleigh, NC), against CD2 (1:50, Serotec), or a mouse monoclonal antibody raised against S100A1 (1:100, DAKO Corp., Carpinteria, CA) for 18 h at 40°C and then in biotinylated goat antimouse IgG (Vector) for 1 h. Sections were then incubated in StrepAvidin conjugated with Alexa Fluor 568 (1:400, Molecular Probes). Between each incubation, sections were rinsed thoroughly with TBS + T. Finally, sections were coverslipped with antifading mounting material (Vector) and examined under a laser confocal microscope (LSM 510, Carl Zeiss Microscopy, Jena, Germany).

Values are expressed as mean + SEM. Groups were compared using two-way analysis of variance (ANOVA) followed by the Dunnett or Fisher LSD test. To compare the mean number of the pixels occupied by pCREB-IR cells between two-side dorsal horn, a paired t  test or one-way ANOVA with Student Newman-Keuls multiple comparison method was used. P < 0.05 was considered significant.

Four to six weeks after partial sciatic nerve ligation, animals developed hypersensitivity to von Frey filament testing, with paw withdrawal thresholds reduced from 29 + 9.6 g before surgery to 2.9 + 1.4 g after surgery (P < 0.01). Perineural saline injection did not affect paw withdrawal threshold (fig. 1). In contrast, perineural clonidine injection produced a dose-dependent increase in paw withdrawal threshold (fig. 1). In terms of percent return to prenerve ligation withdrawal threshold, the peak effect of perineural clonidine was 34 ± 5.1% after 3 μg, 41 ± 6.4% after 15 μg, and 77 ± 9.2% after 30 μg. Peak effect occurred between 1 and 7 days after injection of clonidine. Perineural clonidine injection had no other effect on behavior, and animals ambulated normally immediately after recovery from the brief anesthetic for perineural clonidine injection.

Intramuscular injection of clonidine, 30 μg, did not affect withdrawal threshold at any time for up to 21 days after injection (withdrawal threshold was 2.6 ± 1.8 g before clonidine and a maximum of 7.9 ± 4.3 g at 24 h after injection;P > 0.4). Intrathecal clonidine dose-dependently inhibited mechanical hypersensitivity (withdrawal threshold 2.8 ± 0.4 g before clonidine and 6.4 ± 0.9, 12 ± 2.2, and 17 ± 2.1 g after 7.5, 15, and 30 μg, respectively;P < 0.05 for 15 and 30 μg). However, the effect of 30 μg intrathecal clonidine was not as great as that produced by perineural clonidine, and was transient, with the withdrawal threshold no different than the control 3 days after injection.

Perineural injection of the α^2A-preferring antagonist, BRL44408, alone did not modify the paw withdrawal threshold in rats after nerve injury (data not shown). In contrast, BRL44408 produced a dose-dependent blockade of the antihypersensitivity effect of perineural clonidine, 30 μg (fig. 2). The lowest dose of BRL44408 studied, 5 μg, delayed the onset of clonidine's effect until 7 days after injection, whereas larger doses prevented the effect of clonidine (fig. 2).

Finally, we tested whether transient peripheral nerve blockade would reproduce clonidine's effect. Perineural injection of the local anesthetic, ropivacaine, caused unilateral motor dysfunction in all animals, lasting approximately 1 h, consistent with peripheral nerve blockade. However, the withdrawal threshold was not affected beyond this period (withdrawal threshold was 3.2 ± 0.6 g before ropivacaine injection and a maximum of 6.6 ± 0.9 g at 4 days after injection;P > 0.6).

In the normal sciatic nerve (fig. 3A), there were no α^2A-adrenoceptor IR profiles in either endoneural or epineural tissues. Two and four weeks after partial sciatic nerve ligation, intensely stained small size α^2A-adrenoceptor IR cell profiles were present at the injury site, particularly around the ligation sutures (fig. 3B). Some strongly stained α^2A-adrenoceptor IR cell profiles were also observed in a region of 1 cm proximal and distal to the ligation injury site (fig. 3C). The nuclei of some these cells could be discerned clearly. However, at the mid-thigh level of the ipsilateral sciatic nerve (2 cm distal to the ligation site), no such strongly stained α^2A-adrenoceptor IR cells were observed (not shown). In the epineural connective tissue, strongly stained α^2A-adrenoceptor IR cells were also observed (fig. 3D). Numerous labeled cells were located around small blood vessels. Interestingly, some intensely stained α^2A-adrenoceptor IR cells also appeared at the proximal thigh level of the contralateral sciatic nerve (fig. 3E). Similarly, some labeled cells were present around the small blood vessels in the epineural tissue (fig. 3F).

To characterize these α^2A-IR cells, we carried out the double immunofluorescent staining of α^2Awith the macrophage marker ED1 (fig. 4A–C), with the T lymphocyte marker CD2 (fig. 4D) or with Schwann cell marker S100 (data not shown). In the proximal (fig. 4A) and distal (fig. 4B) areas adjacent to the lesion site, all ED1-IR cells displayed characteristic morphology of macrophages. A majority of these macrophages coexpressed α^2A-IR. In the ligation site (fig. 4C), where both ED1-IR and α^2A-IR cells were most abundant, some cells also coexpress both markers. It is noticeable that some α^2A-IR cells did not coexpress ED1. Interestingly, in the ligation site (fig. 4D) and the proximal and distal areas adjacent to the lesion site, some α^2A-IR cells coexpress T lymphocyte marker CD2-IR, but that majority of α^2A-IR and ED1-IR cells were not colocalized. In contrast, double staining of α^2A-IR and Schwann cell marker S100 revealed that none of S100-IR cells coexpressed α^2A-IR (not shown). Thus, the majority of α^2A-IR cell profiles were likely macrophage (approximately 2/3 of costained cells), some of them were T lymphocytes (approximately 1/3 of costained cells), and none of them were Schwann cells.

Numerous pCREB-IR cells were observed in the dorsal horn of L4–L6 spinal cord of normal rats (fig. 5A and B). The immunoprecipitates were all localized in the nuclei of these cells. pCREB-IR cells were concentrated in the superficial laminae. Similar to a previous report, 9compared with the contralateral side (fig. 5C), the number of pCREB-IR cells was significantly increased in the ipsilateral dorsal horn (fig. 5D) 4 weeks after partial sciatic nerve ligation. One week after a single perineural injection of clonidine at the injury site the number of pCREB-IR cells in the contralateral (fig. 5E) and ipsilateral dorsal horn (fig. 5F) was significantly decreased. After a single injection of saline into the injury site of nerve-ligated rats, the number of pCREB-IR cells in the ipsilateral dorsal horn remained significantly higher than the contralateral side (data not shown), similar to nerve-ligated rats without injection.

The mean number of pixels in a fixed area occupied by pCREB-IR cells in the ipsilateral dorsal horn was significantly greater than the contralateral side in perineural saline injected controls or in comparison with normal animals (fig. 6A, P < 0.01). However, the mean optical density of pCREB-IR cells in this fixed area was not significantly different between the ipsilateral and contralateral dorsal horn or in comparison with normal animals (fig. 6B). After perineural injection of clonidine in nerve-ligated rats, both the mean number of pixels and the mean optical density in a fixed area occupied by pCREB-IR cells in the dorsal horn were significantly reduced ipsilateral and contralateral to the ligation-injection site compared with controls (P < 0.001).

Spinally and epidurally administered α²-adrenergic agonists reduce nerve injury-induced hypersensitivity in animals 12relieve chronic pain in humans, 15and are commonly used to treat neuropathic pain. 16Unlike perispinal injection, systemic administration of clonidine provides minimal analgesia in patients with chronic pain, 17and it is assumed that activation of peripheral α²-adrenergic receptors should exacerbate, rather than alleviate pain. 4The current studies after nerve injury in rats argue against this assumption, and point to a novel method to treat nerve injury-induced pain.

Many consider the site of nerve injury to be an important “pain generator.” Abnormal spontaneous (ectopic) activity originating from this site 18and in uninjured afferents 19may drive changes in DRG cells as well as the spinal cord, resulting in ongoing pain and hypersensitivity to sensory stimuli. Blockade of neural activity with perineural injection of local anesthetic reduces mechanical hypersensitivity after spinal nerve ligation, 20although the effect is brief. It is unlikely that clonidine diminished tactile allodynia in the present study by inhibition of ectopic activity at the site of nerve injury. For one reason, although clonidine is commonly added to local anesthetics for perineural injection, it is inactive alone in the normal animal or human, and prolongs the duration of local anesthetics by a few hours only. 21For another reason, silencing ectopic activity by local anesthetic injection at the nerve injury site in the current study had only a very transient effect on tactile allodynia.

Perl postulates that injury to a mixed peripheral nerve results in increased expression of excitatory α²adrenoceptors that maintain hypersensitivity and pain. 4Approximately 30–50% of C fiber nociceptors are excited by application of norepinephrine after nerve injury in rabbit 2or rat, 22whereas few or none are excited in the normal animal. This excitation is blocked by α²-, but not by α¹-adrenergic receptor antagonists. 2α²-Adrenoceptor agonists stimulate DRG cells after nerve injury, 23but not in normals. Noradrenergic fibers at the site of nerve injury and surrounding DRG cells provide the presumed anatomic substrate for these changes. 3 

However, other data fail to support α^2A-adrenergic receptor mediated excitation. Some α²-adrenoceptor agonists stimulate C fiber polymodal nociceptors from rabbit injured nerve, but others, including clonidine, do not. 24In nerve injured monkeys 25and Lewis rats, 26hypersensitivity from norepinephrine is blocked by α¹- rather than α²-adrenoceptor antagonists. Finally, there are no clinical case reports of pain or increased hyperalgesia or allodynia from peripheral or central administration of clonidine in patients with chronic pain.

Only one previous study examined the functional role of α-adrenergic receptors at the site of nerve injury. 27In that study, the whole nerve, from injury site to DRG to central projection, was removed and studied in vitro . Norepinephrine applied at the nerve injury site excited C and Aδ fibers, and, to a much lesser extent, Aβ fibers, and excitation was blocked by α²-, but not α¹-adrenergic receptor antagonists. One would expect that excitation of C and Aδ fibers would be perceived as noxious in the conscious state. In contrast, we observed no evidence of nociception after perineural injection of clonidine after partial sciatic nerve ligation in the current study, and we observed inhibition, rather than excitation of A fiber-mediated mechanical hypersensitivity. These discrepant results could reflect different methods of nerve injury, methods of drug delivery, in vitro  versus  in vivo  application, or divergent end-points (electrophysiology vs.  behavior).

Others have demonstrated increased α²-adrenoceptor ligand binding at the site of neuroma formation 28and a buildup of α^2A-adrenoceptor immunoreactivity at the site of nerve ligation 48 h after injury, 7but fine anatomic localization, subtype characterization, and identity of adrenoceptor expressing structures have not previously been performed. We demonstrate for the first time that nerve injury results in increased α^2A-adrenergic receptor immunoreactivity at the site of injury several weeks after the insult. Schwann cell origin of the staining can be excluded by lack of colocalization with a specific marker for this cell type. At least some of the immunoreactivity was in nerve fibers, as suggested by staining sandwiched between and parallel to Schwann cell profiles. This is consistent with the increased α^2A-adrenoceptor immunoreactivity observed in DRG cell bodies following this lesion. 29 

In other cases, immunostaining was present on cell profiles with obvious nuclear staining, most likely representing immune cells. Some of these had foamy vacuole structures resembling macrophages, and this was confirmed by specific labeling with the macrophage marker, ED1. As indicated in figure 5, some of the α^2A-adrenoceptor immunostaining in these macrophages was consistent with ingested cellular debris containing these receptors, and in other cases the receptor staining was on the plasma membrane and likely represents functional receptors. Other cells expressing α^2Aadrenoceptors had morphology consistent with lymphocytes, and this was confirmed by specific labeling with the T helper and Natural Killer lymphocyte marker, CD2. Interestingly, these lymphocytes were present not only at the injury site, but both proximal and distal to the site and in the proximal sciatic nerve contralateral to the injury. This is consistent with bilateral changes often observed after unilateral injury (like pCREB expression), and with the known expression of α²adrenoceptors by lymphocytes. 30Sham surgery did not result in new expression of α²-adrenoceptor immunostained cells in either contralateral or ipsilateral lesions (Personal Communication, Weiya Ma, Ph.D., Department of Anesthesiology, Wake Forest University School of Medicine, Winston-Salem, NC, March, 2002).

Clonidine could reduce hypersensitivity by actions on peripheral nerve terminals, at the injury site, or in the central nervous system. Topical clonidine reduces pain in the surrounding skin in patients with complex regional pain syndrome I, 5and intrathecal clonidine is analgesic in patients with neuropathic pain, most likely by an action in the spinal cord. 31In the current study, neither systemic nor intrathecal clonidine mimicked the effect of perineural clonidine on mechanical hypersensitivity. Similarly, perineural injection of a local anesthetic only briefly diminished hypersensitivity. Thus, clonidine potently reduces hypersensitivity after perineural injection by a local action not related to conduction blockade. This is consistent with the observation that the increased potency of α²-adrenergic agonists after nerve injury reflects actions outside the central nervous system. 32 

The mechanisms by which clonidine acts at the site of nerve injury are currently uncertain. Clonidine is rapidly absorbed into the systemic circulation following peripheral injection, 33making a local depot effect unlikely. In addition, the effect of clonidine was reduced in a dose-dependent manner by perineural injection of the α^2A-preferring antagonist, BRL44408, which is most consistent with a local, specific action with delayed consequences.

It is quite possible that perineural clonidine produces a long-lasting and delayed action by affecting immune cell function near the site of nerve injury. α²-Adrenergic agonists can reduce cytokine release from monocytes, including TNF-α34and such factors sensitize nerve endings and drive central sensitization to innocuous stimuli. T lymphocytes contain norepinephrine, and incubation of lymphocytes with norepinephrine suppresses mitogen-induced proliferation and induces apoptosis. 35Peripheral and central sensitization and plasticity can also be driven by uptake at the site of nerve injury of nerve growth factors, 36,37and clonidine could affect either generation, production, or central trafficking of these factors. We propose that clonidine most likely acts by reducing cytokine and growth factor production by immune cells, perhaps in addition inducing apoptosis in some immune cells, thereby reducing centrally transported cytokines and growth factors and reducing central sensitization.

Phosphorylation of CREB, stimulated primarily by increases in intracellular calcium, alters gene transcription. 38Nerve injury induces early immediate gene expression cfos , 39as well as somatostatin, 40dynorphin, 41and enkephalin, 42all products important to spinal sensory neurotransmission. However, cfos  expression wanes in established hypersensitivity, yet pCREB expression remains elevated. 9For this reason, continued pCREB expression, driven by ongoing abnormal afferent input, may play an important role in maintaining central sensitization after nerve injury. One week after perineural clonidine injection, at the time of maximum reduction in hypersensitivity, pCREB expression was diminished in the spinal cord. Although causality is not proven by these observations, we propose that this central effect on this transcription factor reflects reduced central trafficking of factor(s), which enhance spontaneous activity of afferents and convergent neurons in the spinal cord.

In summary, behavioral and immunohistochemical studies in rats with peripheral nerve injury fail to support the hypothesis that peripheral α²-adrenoceptor stimulation causes and maintains pain and hypersensitivity. Rather, they are consistent with a site- and subtype- specific action on novelly expressed α^2Aadrenoceptors at the site of injury that reduce hypersensitivity. The delayed onset and prolonged duration observed in the patient and in this animal model, accompanied by reduction in a clonidine-induced reduction in a spinal transcription factor, suggest that perineural clonidine acts by altering generation of factors released by immune cells and their central trafficking that regulates and maintains hypersensitivity. These findings support trials of a new approach to treatment of nerve injury-associated pain in humans and investigation of peripherally derived factors altering spinal cord gene expression in models of chronic pain.