Introduction

During the acute phase of a spinal cord injury, 5-HT may accelerate secondary damage, since it is known to be released at the injury site.1,2,3 We recently showed that both the 5-HT1A and 5-HT2A receptor subtypes are present on spinal dorsal column axons, and that they have opposing effects on axonal excitability. 5-HT2A agonists strongly excite action potentials, whereas 5-HT1A agonists depress them.4,5,6 Further, mianserin (5-HT2A antagonist) has been reported to be neuroprotective in acute spinal cord injuries.7,8 Thus, 5-HT receptors may play a role in the pathological responses of axons to injury. However, in the absence of any previous evidence for 5-HT receptors existing on axons, most hypotheses regarding 5-HT mediated secondary axonal injury have focused on indirect mechanisms, such as posttraumatic ischemia. Our previous results provided a theoretical basis for the direct action of 5-HT mediating excitotoxicity in axons, and suggested a new and interesting therapeutic approach to spinal cord injury. We postulated that 5-HT1A agonists depress axonal excitability and, as a consequence, they may also exert neuroprotective effects. To determine the role of 5-HT1A agonist in acute spinal cord injury, we evaluated methods of 5-HT agonists treatment with in vitro models of dorsal column axon compressive injury.

Materials and methods

A total of 24 male adult Wistar rats weighing 250–350 g were used, after being anesthetized with methoxyflurane and decapitated. Care of the animals was conducted in accordance with the Guide to the Care and Use of Experimental Animals (Shiga University of Medical Science). The dorsal columns were isolated,9,10 placed in a 3 ml recording chamber, and superfused at 120 ml/h with room-temperature (23–27°C) Ringer's solution for 2–3 h to wash out the methoxyflurane and stabilize the responses. The Ringer's solution contained NaCl (124 mM), KCl (3 mM), Na2HPO4 (1 mM), NaHCO3 (26 mM), MgSO4 (1.5 mM), CaCl2 (1.5 mM), and glucose (10 mM). The solution was saturated with O2 (95%) and CO2 (5%) at pH 7.4.

Many 5-HT receptor subtypes have been discovered in central nervous tissues and described.11 5-HT presumably activates all 5-HT receptors, whereas quipazine likely activates primarily 5-HT2A receptors, tandopirone should selectively activate 5-HT1A receptors, and mianserin should block 5-HT2A receptors. Table 1 summarizes the eight drug protocols tested (protocol ABCDEFGH listed in Table 1, n=49). The drugs were mixed into Ringer's solution and superfused at 120 ml/h for the 20 min incubation period. We tested the effects of 5-HT hydrochloride (5-HT HCl, SIGMA, St. Louis, MO, USA, protocol A listed in Table 1, n=8), quipazine dimaleate (quipazine, SIGMA, St. Louis, MO, USA, protocol BC listed in Table 1, n=10), and tandospirone (Sumitomo Pharm. Co. Ltd, Japan, protocol EF listed in Table 1, n=13). The protocols include two superfusate concentrations (10 μM and 100 μM) of quipazine and tandospirone. In the six experiments, we incubated the nerve preparations for 20 min in mianserin HCl (a 5-HT2A receptor antagonist, SIGMA, St. Louis, MO, USA, protocol D listed in Table 1, n=6) before applying the quipazine. In the six experiments, mianserin alone was superfused (protocol G listed in Table 1, n=6). In the six control experiments, Ringer's solution alone was superfused (protocol H listed in Table 1, n=6). Control pre-drug response amplitudes and latencies were 0.785±0.126 mV and 6.71±0.92 ms in experiments using maximal stimuli. The 5-HT agonist solutions included 0.01% ascorbic acid (Sigma, St. Louis, MO, USA) to inhibit oxidation, which had no effect on the action potentials in the preparations.

Table 1 Summary of treament protocols
Full size table

At superfusion rates of 120 ml/h, we predicted that drug concentrations in the chamber would approach 95% of the concentrations in the incoming fluid within 6 min, 99% within 10 min, and 99.9% by 30 min. We applied the agonists for 20 min. Further, the preparations were incubated for 10 min before and 10 min after injury.

To activate the action potentials, we applied 0.2 Hz constant-current 200 μs duration pulses (Pulse Generator, MacLab Software, Australia) with a bipolar platinum electrode placed on the dorsal column (Stimulator, Nihon Kohden SEM4201, Japan). Maximal response by stimulation was determined by progressively elevating the current intensity until the response no longer showed a greater amplitude with additional current. The action potentials were recorded using glass micropipettes filled with 1 M NaCl that were inserted 50 μm deep into the dorsal column. The mean conduction distance between the stimulus and recording electrodes was 2.3±0.54 mm. The voltage signals were amplified (Amplifier (100×), Nihon Kohden AB601G, Japan), analyzed and displayed graphically on a computer (Data Acquisition, MacLab Software, Macintosh, PowerBook550c), as illustrated in Figure 1. Response amplitudes were measured from the initial positive peak (P1) to the first large negative peak (N1) of each compound action potential. Response latencies were measured from stimulus onset to N1.

Figure 1
figure 1

Experimental paradigm: isolation of dorsal column from spinal cord and schematic diagram of stimulation-recording arrangement with isolated dorsal columns

Full size image

The mean action potential amplitudes and latencies are expressed as percentages of the pre-injury control levels, which were obtained before spinal cord compression. All±values indicate standard deviations. To compare the treatment protocols, we applied statistical analysis. Significant differences between the control and treatment groups were determined by analysis of variance (ANOVA). Differences were deemed statistically significant at P<0.05.

Results

The isolated rat dorsal columns were typically 3.0 cm in length. Stimulation with constant-current pulses activated robust compound action potentials that were conducted along the length of the dorsal column preparations. Table 2 summarizes the mean control amplitudes and latencies found in individual groups. Since stimulus and recording positions differed from preparation to preparation, the control response amplitudes and latencies varied. However, the response characteristics remained stable within each experiment, seldom varying more than 5% over periods of 20–30 min.

Table 2 Summary of pre-drug amplitudes and latencies
Full size table

After compression injury, the mean action potential amplitudes were reduced to 33.9±5.4%, and then recovered to 62.8±8.4% after 120 min of perfusion with Ringer's solution. At 10 μM and 100 μM concentrations, perfusion with tandospirone showed recovery values of 68.3±14.5% and 86.2±6.9% of the pre-injury values, respectively, and the 100 μM concentration of tandospirone resulted in a significantly greater recovery of mean action potential amplitudes as compared with the control Ringer's solution 2 h after injury (P<0.05). The latency shifts with 10 μM and 100 μM concentrations of tandospirone were respectively 2.3±4.8% and 2.2±2.0%. In contrast, 100 μM of quipazine (a 5-HT2A agonist) caused significantly less recovery as compared with the control Ringer's solution from 60 to 100 min after injury (P<0.05). Quipazine did not consistently affect response latencies. At a 50 μM concentration, mianserin (a 5-HT2A antagonist) eliminated the inhibitory effects of 100 μM quipazine, whereas 5-HT did not significantly affect the recovery of response amplitudes. Moreover, at 120 min after injury, neither the 10 μM nor 100 μM concentration of quipazine showed a significant reduction of amplitudes, while a 50 μM concentration of mianserin had no effect on the change of action potentials in the in vitro preparations.

Table 3 Summary of amplitude and latency changes (120 minutes post-injury)
Full size table
Figure 2
figure 2

Time course of the protocol F (100 μM tandosporine), which induced excitability changes. The ordinate indicates response amplitude as a percentage of the pre-drug control level

Full size image
Figure 3
figure 3

The effects of (1) 100 μM quipazine (protocol C), (2) 100 μM quipazine with 50 μM mianserin (protocol D), (3) 100 μM tandosporine (protocol F) on compound action potentials in adult rat dorsal columns

Full size image

Discussion

More than three decades ago, Osterholm, et al.12 suggested that the neurotransmitter 5-HT causes tissue damage in the brain. It has since been proposed that 5-HT contributes to the posttraumatic decline of blood flow and edema seen in injured spinal cords.13,14,15,16 Microdialysis studies have shown that a spinal cord injury releases large amounts of 5-HT into extracellular spaces,3,17,18 and we also reported that 5-HT is released from neural elements at the injury site and is transiently taken up by platelets.1,2 Mianserin, a 5-HT antagonist, has been reported to have beneficial effects towards the recovery of neurologic and neurophysiologic functions in acute spinal cord injury models.7,8 However, the mechanisms by which 5-HT causes secondary tissue damage, especially with axons, are not well understood. Most investigators have focused on the possible indirect effects of 5-HT, such as posttraumatic ischemia, and it has been demonstrated that some peripheral and central axon terminals have 5-HT autoreceptors.19,20 We recently found that spinal axons possess 5-HT receptors and suggested that 5-HT may have a direct excitotoxic effect on spinal axons.4,5,6

In previous experiments, 5-HT had robust effects on dorsal columns obtained from rats,4,5,6 and both 5-HT1A and 5-HT2A receptor subtypes were found on spinal dorsal column axons, however, they demonstrated opposing effects on axonal excitability. Further, we found that 5-HT2A agonists strongly excite action potentials, while 5-HT1A agonists depress them.4,5,6 It would be of interest to extend this area of study to investigate actual injured spinal cords, as the ratio of 5-HT2A/5-HT1A receptor expression at various stages after injury may well influence axonal responses to the extracellular neurotransmitter derangements that occur during those processes. In the previous study, it has been demonstrated that the neurotransmitters had the effects on dorsal columns obtained from neonatal rats.4 While the sensitivity diminished with maturation of dorsal column. We thought that the damage of the myelin sheath by spinal cord injury might cause the penetration of neurotransmitter. Our previous results also suggested an interesting and new therapeutic approach to spinal cord injury. Since 5-HT1A agonists probably depress axonal excitability more than the 5-HT2A antagonist, as a consequence, 5-HT1A agonists may also exert neuroprotective effects, perhaps more strongly than the 5-HT2A antagonist. Fortunately, many highly selective 5-HT1A receptor agonists are already available as drugs,11 while tandospirone is commonly used to treat anxiety disorders.21 It has been reported that the 5-HT-induced nociceptive response is mediated by 5-HT2A receptors in the periphery.22 Therefore, our study also suggests a therapeutic approach for the pain and numbness.

The data from this present study indicate that tandospirone and quipazine have opposing effects on an injured spinal cord, as the recovery of the compound action potential amplitude with 100 μM tandosporine significantly exceeded that seen in the control group, and quipazine (100 μM) accelerated the amplitude decrease at 60–100 min after injury. Moreover, mianserin (50 μM) blocked the quipazine induced amplitude change. The most conservative explanation of these results is that 5-HT2A receptors increase the excitotoxicity of dorsal column axons while 5-HT1A receptors decrease it following spinal cord injury. In conclusion, the 5-HT1A receptor agonist was effective for the recovery of spinal action potential amplitudes in a rat spinal cord injury model.