- Email: [email protected]
Instrumental learning within the spinal cord
Physiology & Behavior 77 (2002) 259 – 267
Instrumental learning within the spinal cord II. Evidence for central mediation Eric D. Crowna,*, Adam R. Fergusona, Robin L. Joynesb, James W. Graua a
Department of Psychology, Texas A&M University, College Station, TX 77843, USA b Kent State University, Kent, OH, USA
Received 19 September 2001; received in revised form 18 December 2001; accepted 27 June 2002
Abstract Rats spinally transected at the second thoracic vertebra can learn to maintain their leg in a flexed position if they receive legshock for extending the limb. These rats display an increase in the duration of a flexion response that minimizes net shock exposure. The current set of experiments was designed to determine whether the acquisition of this behavioral response is mediated by the neurons of the spinal cord (i.e., is centrally mediated) or reflects a peripheral modification (e.g., a change in muscle tension). Experiment 1 found that preventing information from reaching the spinal cord by severing the sciatic nerve blocked the acquisition of this behavioral response. Spinalized rats also failed to learn if the spinal cord was anesthetized with lidocaine during exposure to response-contingent shock (Experiment 2). Experiment 3 demonstrated that prior exposure to response-contingent shock on one hindleg facilitated acquisition of the response when subjects were later tested on the opposite leg. These findings suggest that acquisition of the instrumental response depends on neurons within the spinal cord. D 2002 Elsevier Science Inc. All rights reserved. Keywords: Plasticity; Spinal cord; Lidocaine; Sciatic nerve; Instrumental learning
1. Introduction Recent studies of pain modulation, locomotion, and learning have shown that the spinal cord is far more plastic than originally imagined (for recent review, see Ref. [1]). For example, researchers in the pain literature have demonstrated that nociceptive stimulation can sensitize neurons within the spinal cord, a process known as central sensitization [2 –4]. Researchers in the locomotion literature have found that the spinal cord contains the neural circuitry necessary to support organized stepping and can adapt to changing environmental conditions (e.g., obstacles placed in their path during stepping [5,6]). Our laboratory has used traditional learning paradigms to examine the functional properties of spinal cord neurons. We have demonstrated that the neurons of the spinal cord are capable of Pavlovian and instrumental conditioning [7 – 13]. To explore the capacity for instrumental learning, we have used a modification of the master – yoke paradigm of
*
Corresponding author. Tel.: +1-979-862-4852; fax: +1-979-845-4727. E-mail address: [email protected] (E.D. Crown).
Horridge [14]. In this paradigm, one group of spinally transected (spinalized) rats (master) is given shock to a hindleg whenever that limb is extended (response-contingent shock), causing a flexion of the foot at the ankle joint. Instituting a contingency between the response (leg flexion) and the reinforcer (shock onset) produces a change in the duration of the flexion response that minimizes net shock exposure. Subjects in another group (yoked) receive legshock at the same time and for the same duration as the master subjects; however, these subjects receive shock independent of leg position (noncontingent shock). We have shown that rats given response-contingent shock display a progressive increase in the duration of the flexion response, whereas rats given noncontingent shock do not exhibit such an increase in response duration [9,10]. This suggests that response-contingent shock has a differential effect, a defining feature of instrumental learning (additional evidence that the increase in response duration depends on the instrumental relation is provided elsewhere [9]). We have assumed that the change in flexion duration produced by response-contingent shock reflects a centrally mediated phenomenon—an example of learning within the spinal cord. However, there is no empirical evidence for this
0031-9384/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 2 ) 0 0 8 5 9 - 4
260
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
assumption. Indeed, as it stands, one could argue that a peripheral mechanism underlies the change in flexion duration. For example, perhaps response-contingent stimulation of the muscles produces a cumulative increase in muscle tension that maintains the flexion response. What is especially worrisome is that this could occur without any input from the spinal cord. The present experiments examine whether the change in flexion duration (our measure of learning) observed in spinalized rats depends on neurons within the spinal cord. We address this issue by testing whether severing the connection with the spinal cord (Experiment 1), or inhibiting neural activity within the spinal cord through intrathecal application of the Na + channel blocker lidocaine (Experiment 2), affects learning. We also examine whether response-contingent shock applied to one hindlimb affects learning on the opposite (contralateral) leg (Experiment 3). Our results demonstrate that spinal cord neurons play an essential role in the acquisition of the instrumental response.
2. General methods 2.1. Subjects The subjects were male Sprague – Dawley rats obtained from Harlan (Houston, TX). The rats were approximately 100 days old and were individually housed with food and water continuously available. Rats were maintained on a 12h light/dark cycle and were tested during the last 6 h of the light cycle.
vocalize to the legshock, and (c) examining the spinal cord postmortem in a randomly selected subset of the subjects. 2.2.2. Severing the sciatic nerve In Experiment 1, the sciatic nerve innervating one hindleg was severed in a subset of the subjects. Rats were spinalized at T2 as described above. While still anesthetized, a small longitudinal incision was made approximately 1 cm from the midline at the level of the hip. The sciatic nerve was exposed using a glass probe to gently separate the nerve from the surrounding muscle tissue. For half of the subjects, the nerve was severed approximately 1 cm lateral to the dorsal root, with a cautery device. The remaining subjects served as the sham-operated controls. Whether this surgery was conducted on the left or right hindleg was counterbalanced across subjects. 2.2.3. Intrathecal catheter implantation In Experiment 2, an intrathecal catheter was inserted down the cord after cauterization following the procedure of Yaksh and Rudy [15]. This was accomplished by fitting a 25-cm segment of PE-10 tubing with a 0.23-cm (diameter) stainless steel guide wire (SWGX-090; Small Parts) and sliding it caudally 9 cm into the subarachnoid space at T2, so that the tubing lay along the dorsal surface of the cord. The wire was then removed and the tubing was secured with adhesive (Superglue). The exposed cord was then covered with Oxycel (Parke Davis) and the wound closed with Michel clips. In this way, approximately 10 cm of tubing protruded from the wound. 2.3. Apparatus
2.2. Surgery 2.2.1. Spinalization Rats were anesthetized with pentobarbital (50 mg/kg ip). To stabilize and position the rat’s body for surgery, its head was held in a stereotaxic instrument and a small gauze ‘‘pillow’’ was placed under its chest. After the second thoracic vertebra (T2) was localized by touch, an anterior – posterior incision was made on the back over T2. Next, the tissue in front of T2 was cleared away and the cord was transected by cauterization. The exposed cord was then covered with Oxycel (Parke Davis) and the wound closed with Michel clips. Rats were maintained in a temperaturecontrolled environment (approximately 25.5 °C) during recovery and testing. During recovery, the rat’s rear legs were maintained in a normal flexed position by a piece of porous tape (Orthaletic, 1.3 cm width) gently wrapped once around the rat’s body. For all experiments, the recovery period between surgery and behavioral testing was approximately 24 h. Transections were confirmed by (a) inspecting the cord during the operation, (b) observing the behavior of the subjects after they have recovered to insure that they exhibit paralysis below the level of the forepaws and do not
Instrumental training was conducted while spinal rats were loosely restrained in specially designed tubes. A detailed description of the apparatus and experimental setup is provided in Grau et al. [9]. Briefly, the tubes have two slots that allow both hindlegs to hang freely. To minimize the effects of upper body movements on leg position, the rat’s midsection was gently secured with a wire ‘‘belt.’’ Legshock was applied by attaching one lead from a BRS/ LVE shock generator (Model SG-903) to an intracutaneous wire electrode that was implanted through the skin over the tibia. The other lead was attached to a 2.5-cm stainless steel pin that was inserted 0.4 cm into the tibialis anterior muscle. Leg position was monitored using an insulated contact electrode taped to the plantar surface of the rat’s foot, just distal to the plantar protuberance. A fine wire attached to the proximal end of the contact electrode extended from the rear of the foot and was connected to a digital input monitored by a Macintosh computer. A plastic dish containing a NaCl solution was placed below the restraining tube, so that the tip of the rod could contact the solution. A drop of detergent was added to the solution to reduce surface tension. A ground wire was connected to a 1-mm stainless steel rod,
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
which was placed in the solution. When the contact electrode attached to the rat’s paw touched the solution, it completed the circuit monitored by the computer. The state of this circuit was sampled at a rate of 30 times per second. Flexion force was measured by attaching a monofilament plastic line (‘‘4 lb test,’’ Stren, Dupont) to the rat’s foot immediately proximal to the plantar protuberance. The 40 cm length of line was passed through an eyelet attached to the apparatus directly under the paw. The end of the line was attached to a strain gauge that was fastened to a ringstand. After the line was connected to the rat’s paw, the ringstand was positioned so that the line was taught, just barely engaging the gauge. Shock intensity was adjusted to produce a flexion force of a predetermined value (0.4 N). The strain gauge was then removed from the rat’s foot. 2.4. Procedure 2.4.1. Testing with response-contingent shock Prior to testing, the rear legs were shaved and marked for placement of the shock leads. The wire electrode was then inserted over the tibia at the distal mark and the rats were placed in the restraining tubes. Next, the contact electrode used to monitor leg position was taped to the paw. To minimize lateral leg movements, a 20-cm piece of porous tape (Orthaletic, 1.3 cm) was wrapped around the leg and taped to a bar extending across the apparatus directly under the front panel of the restraining tube. The tape was adjusted so that it was taut enough to extend the joint between the tibia and femur. One lead from the shock generator was attached to the stainless steel wire inserted over the tibia. The shock generator was set to deliver a 0.1-mA shock and the region over the second mark was probed to find a site that elicited a vigorous flexion response. The pin was then inserted perpendicular to the body into the tibialis anterior muscle. The electrode placements were checked by verifying that a single intense (1.6 mA) test shock (0.3 s) elicited a flexion response of at least 0.8 N. The shock intensity necessary to induce a 0.4-N flexion response was obtained for the hindleg as described above. Finally, three short (0.15s) shock pulses were applied and the level of the salt solution was adjusted so that the tip of the rod was submerged 4 mm below the surface. During testing with response-contingent legshock, each rat received shock immediately after its leg contacted the salt solution for the duration of the 30-min observation period. We have shown previously that shock onset is a critical variable in this paradigm, as delaying shock onset by as little as 100 ms disrupts instrumental performance [9]. 2.4.2. Experiment 1 Rats (N = 12) were spinalized at T2 and either had the sciatic nerve cut or underwent a sham operation as described above. Twenty-four hours after surgery, subjects were tested with response-contingent shock applied to the previously operated leg, as described above.
261
2.4.3. Experiment 2 Twelve spinalized rats with intrathecal catheters served as subjects. A 0.05-ml intrathecal injection of lidocaine (Xylocaine) was administered to half of the subjects (Lidocaine) at a concentration of 20 mg/ml immediately before the rats were placed in the training apparatus. This was accomplished by inserting a 27-gauge needle into the exposed end of the catheter tubing and injecting the drug slowly over a period of 1 min. The lidocaine injection was followed by 0.01 ml of 0.9% saline. To verify the efficacy of the drug, tail-flick responses to a slight pinch were monitored to make sure reflexive activity had ceased. The other half of the subjects (Sham) underwent the same surgical procedures but received no injection. They were then placed in the apparatus and tested with responsecontingent shock as described above. Whether testing occurred on the left or right hindleg was counterbalanced across subjects. 2.4.4. Experiment 3 Twenty-four rats were divided into three groups for Experiment 3. This experiment involved two phases. In Phase 1, subjects were prepared for legshock as described above. Instead of placing shock leads into one leg, however, both legs were prepared for training at the same time so that the left leg was in one solution and the right leg was in another solution. After 30 min of training with either response-contingent legshock or nothing and a 10-min period that allowed for the previously trained leg to return to resting position, Phase 2 testing began. First, the water level of both salt solutions for each subject was reset by adding 100 ml of water. Following this, subjects received 30 min of response-contingent shock applied to either the pretrained leg (ipsilateral; n = 8) or the opposite leg (contralateral; n = 8) (Phase 2). The unshocked subjects from Phase 1 (n = 8) received response-contingent shock to the left or right leg in a counterbalanced fashion. Because Phase 2 required that the subject’s leg not be disturbed before testing, a few modifications had to be made to our testing procedure. Prior to Phase 1, the shock intensity needed to induce a 0.4-N change in flexion force was determined for both legs and the water level needed to submerge the contact electrode 4 mm for each leg was set. In Phase 2, addition of 100 ml of water elevated the water level by 4 mm, effectively raising the response criterion for the test leg from 4 to 8 mm. 2.5. Behavioral measures Three behavioral measures were used to monitor performance: time in solution, response number, and response duration (see Ref. [9]; Fig. 2). The computer recorded when the contact electrode touched the underlying solution (time in solution). Whenever the electrode left the solution, the number of responses was increased by one (response number). To obtain a measure of performance
262
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
over time, we divided the session into thirty 1-min time bins. We have previously shown how instrumental learning can be distinguished from a reactive system that is insensitive to response –reinforcer relations [9,16]. A key difference concerns response duration; only the instrumental account anticipates that contingent shock will produce a progressive increase in response duration. Response duration was derived from time in solution and response number using the following equation: Response durationi=(60 s time in solutioni)/(Response numberi + 1) where i was the current time bin. To address the possibility that differences in response duration during testing reflect a loss of responding in the previously shocked rats, we also present response number. 2.6. Statistics The results were analyzed using an analysis of variance (ANOVA). Post hoc comparisons were made using Duncan’s New Multiple Range test. In all cases, a criterion of P < .05 was used to judge statistical significance.
3. Experiment 1 The shock stimulus used to study instrumental learning in spinalized rats should have two consequences: (1) the direct elicitation of a motor response, and (2) the induction of an afferent barrage that engages neurons within the spinal cord. We have assumed that learning (the increase in flexion duration) depends on the second component. If so, then eliminating communication with the spinal cord should prevent learning. The present experiment tests this by severing the sciatic nerve, the primary source of innervation for the region where our shock stimulus is applied [17]. 3.1. Results The values for the shock intensity necessary to induce a 0.4-N change in flexion force ( ± S.E.) ranged from 0.21 mA ( ± 0.04) to 0.21 mA ( ± 0.06) and the values for initial response duration ranged from 0.42 s ( ± 0.02) to 0.52 s ( ± 0.06). Independent ANOVAs confirmed that these differences did not approach statistical significance (both Fs < 2.5, P>.05).
Fig. 1. Performance over time during testing in Experiment 1. The change in response duration (top panels) and response number (bottom panels) observed across the testing phase for rats that received either a sciatic nerve lesion (filled circles) or sham surgery (open circles). The group means ( ± S.E.M.) are depicted in the right panels.
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
Sham-operated rats exhibited a progressive increase in flexion duration over the 30 min of testing with responsecontingent shock (Fig. 1, top panels). Rats that had the sciatic nerve severed failed to learn. An ANOVA revealed a significant effect of Surgery Condition (sciatic transection versus sham), Time, and Surgery Time interaction (all Fs>2.76, P < .0001). Spinalized rats that had received a sciatic nerve transection failed to maintain the instrumental flexion response, as evidenced by the fact that this group made more contacts with the salt solution (Fig. 1, bottom panels). An ANOVA found a significant main effect of Surgery Condition [ F(1,10) = 7.33, P < .05]. The main effect of Time and its interaction with Surgery Condition were not significant (both Fs < 1.46, P>.05). Because our reinforcer involves direct electrical stimulation of the tibialis anterior muscle, we expected that cutting the sciatic nerve would have little effect on the elicitation of the motor response. Of greater concern was whether repeatedly performing the response would produce an increase in the flexion response when input from the sciatic nerve was eliminated. We found that rats given
263
response-contingent shock after the sciatic nerve was cut did not exhibit an increase in flexion duration, suggesting that the acquisition of the instrumental response depends on communication with spinal cord neurons.
4. Experiment 2 Experiment 1 demonstrated that input from the sciatic nerve is necessary for the acquisition of the instrumental flexion response in spinalized rats. Experiment 2 used a pharmacological manipulation, administration of intrathecal lidocaine, to further evaluate whether spinal cord neurons play an essential role in the acquisition of the instrumental response. Lidocaine generally depresses neural activity within the spinal cord and has been shown to prevent the increase in neural excitability observed in response to intermittent shocks (wind-up) and peripheral inflammation (central sensitization) [18,19]. If instrumental learning depends on spinal cord neurons, lidocaine-treated rats should not exhibit an increase in flexion duration when given response-contingent shock.
Fig. 2. Performance over time during testing in Experiment 2. The change in response duration (top panels) and response number (bottom panels) observed across the testing phase for subjects given intrathecal lidocaine (filled circles) or nothing (open circles) subjects. The group means ( ± S.E.M.) are depicted in the right panels.
264
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
4.1. Results The shock intensity needed to elicit a 0.4-N flexion varied from 0.58 mA ( ± 0.04) to 0.59 mA ( ± 0.06) and initial flexion durations ranged from 0.16 s ( ± 0.01) to 0.19 s ( ± 0.01). Independent ANOVAs verified that these differences did not approach statistical significance (both Fs < 1.0, P>.05). The untreated controls exhibited an increase in flexion duration over the 30 min of testing (Fig. 2, top panels). Pretreatment with lidocaine blocked learning. An ANOVA confirmed that Drug Treatment had a significant impact [ F(1, 10) = 14.73, P < .01]. There was also a significant effect of Time, and a Drug Treatment Time interaction (both Fs>1.65, P < .05). Subjects that received lidocaine also made more contacts with the salt solution (Fig. 2, bottom panels). An ANOVA yielded a main effect for Drug Treatment that approached statistical significance [ F(1, 10) = 4.37, P < .063]. Neither the main effect of Time nor its interaction with Drug Treatment was significant (both Fs < 1.25, P>.05). In summary, we found that intrathecal lidocaine prevented performance of the instrumental flexion response. The drug did not, however, eliminate the shock-elicited response. As observed in Experiment 1, repeated execution of the response did not, in the absence of a functional spinal cord, bring about an increase in flexion duration, implying that the spinal cord is essential.
5. Experiment 3 We have previously shown that exposure to noncontingent shock produces a behavioral deficit that undermines the capacity for instrumental learning [9,10]. To test whether this effect depends on spinal cord neurons, we showed that blocking the afferent input and pretreatment with intrathecal lidocaine block the development of the behavioral deficit [20]. We also showed that rats given noncontingent shock to one hindleg fail to learn when contingent shock is later applied to the opposite leg. This transfer suggests that a common system, within the spinal cord, must mediate the behavioral deficit. Thus, it appears that noncontingent shock has a disabling effect that generally undermines behavioral potential. Perhaps exposure to contingent shock has the opposite effect—one that generally enables behavioral potential. If it does, training with contingent shock applied to one hindlimb could potentially facilitate acquisition when responsecontingent shock is applied to the contralateral leg. A problem with testing this hypothesis is that the rapid learning exhibited by our untreated controls could potentially mask the beneficial effects of prior training with contingent shock. We address this issue by testing subjects with the response criterion set to a higher level (a contact electrode depth of 8 mm). A pilot study indicated that, under
these conditions, untreated controls often fail to learn. We also modified the procedure used in earlier studies in another regard. Normally, we equate flexion force and contact electrode depth prior to testing. Although such a procedure helps to guard against the negative consequences of motor fatigue and other nonassociative habituation-like effects, it undermines our capacity to resolve a beneficial effect of training with response-contingent shock on the pretreated (ipsilateral) leg. The problem is that subjects trained with contingent shock will, as a consequence of the training, maintain their leg at a higher position. If contact electrode depth is then equated based on this new position, we effectively eliminate the behavioral benefit of the training and force subjects to acquire the instrumental response to meet the new (higher) criterion. To avoid this problem, we equated contact electrode depth prior to training. At the end of training, the criterion was raised by an equal amount for all subjects. 5.1. Results 5.1.1. Training The shock intensity needed to elicit a flexion response of 0.4 N prior to training ranged from 0.40 mA ( ± 0.01) to 0.44 mA ( ± 0.04). These group differences did not approach significance [ F(2,22) = 1.02, P>.05]. During training, rats given response contingent shock showed a progressive increase in the duration of the flexion response relative to unshocked rats (Fig. 3, top left panel). An ANOVA yielded a main effect of Shock Condition, Time, as well as a significant Shock Condition Time interaction (all Fs>3.66, P < .0001). Post hoc tests found that both response-contingent shocked groups displayed significantly longer flexion durations than did unshocked rats (both Ps < .05) and that the response-contingent shocked groups did not differ ( P>.05). The groups that received response-contingent shock also made more contacts with the salt solution than did the unshocked rats during training (Fig. 3, bottom left panel). An ANOVA revealed a significant main effect of Shock Condition, Time, and a significant Shock Condition Time interaction (all Fs>5.05, P < .001). Post hoc analyses showed that both response-contingent shocked groups made more responses (solution contacts) than the unshocked group and that the two shocked groups did not differ. 5.1.2. Testing Subjects that had previously received response-contingent shock learned to maintain their leg at a higher level when the response criterion was raised prior to testing (Fig. 3, top right panel). The magnitude of this effect was the same independent of whether subjects were tested on the ipsilateral or contralateral leg. Subjects that had not received training with response-contingent shock failed to learn under these conditions. An ANOVA yielded a significant
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
265
Fig. 3. Performance over time during training and testing in Experiment 3. The change in response duration (top panels) and response number (bottom panels) observed across both the training and testing phases for subjects trained and tested on the ipsilateral (open circles) or contralateral leg (closed triangles). A control group remained unshocked during the training phase (closed squares). The group means for testing ( ± S.E.M.) are depicted in the rightmost panels.
main effect of Shock Condition (ipsilateral, contralateral, or, unshocked) and Time (both Fs>2.78, P < .01). The Shock Condition Time interaction, however, was not significant [ F(58,609)>1.0, P>.05]. Post hoc analyses determined that the ipsilateral and contralateral groups displayed significantly longer flexion durations than the unshocked group (both Ps < .05). The ipsilateral and contralateral groups were not significantly different from each other ( P>.05). As usual, rats that failed to learn exhibited a greater number of responses (Fig. 3, bottom right panel). An ANOVA revealed a significant main effect of Shock Condition and Time, as well as a significant Shock Condition Time interaction (all Fs>2.04, P < .01). Post hoc tests found that the ipsilateral and contralateral groups made significantly fewer contacts with the salt solution than the unshocked group (both Ps < .05). Again, there were no significant differences between the ipsilateral and contralateral groups ( P>.05). We found that raising the response criterion disrupted learning in the untreated controls. Subjects that had previously experienced response-contingent shock were able to learn under these more difficult test conditions. Most importantly, this benefit of training was equally robust on the contralateral leg. This suggests that training with contingent shock induces a modification within the spinal cord that fosters (enables) instrumental behavior.
In some regard, it may seem surprising that training with contingent shock did not produce a greater effect on the ipsilateral leg. This would appear to indicate that instrumental training in this paradigm has only a general enabling effect—that it does not selectively augment the trained reflex. However, it is important to remember that shock treatment also induces a response-selective decrease in shock reactivity, a habituation-like effect that develops independent of the response –reinforcer relation [9]. This habituation effect would selectively undermine instrumental behavior on the pretrained limb, reducing the likelihood of observing both a response-specific and a more general benefit of training with response-contingent shock. To observe the former, we would need to equate flexion force after training. But with our current equipment, we cannot readjust flexion force without readjusting contact electrode depth. As noted above, this would introduce a confound in our test of contralateral transfer. Thus, while the present results clearly show that there is an enabling effect that generally fosters instrumental behavior, the data do not specifically address the relative contribution of the limb-specific component. Of course, such a component is evident from the behavior observed during, and immediately after, training with contingent shock, for this training only affects leg position in the pretrained limb.
266
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
6. Discussion
6.2. Mechanisms of instrumental learning
We had previously shown that spinalized rats exposed to response-contingent shock learn to maintain their leg in a flexed position, effectively minimizing net shock exposure. Although we assumed that this form of instrumental learning was due to an intraspinal modification, the possibility existed that this phenomenon reflected a peripheral modification. The current set of experiments discount this possibility by showing that the behavioral modification relies on spinal cord neurons. Experiment 1 found that removing the afferent barrage to the spinal cord, by severing the sciatic nerve, prevented spinalized rats from displaying an increase in flexion duration in response to contingent legshock. Experiment 2 determined that anesthetizing the spinal cord with lidocaine prevented instrumental learning. Importantly for both Experiments 1 and 2, the failure to learn occurred even though shock directly elicited a leg flexion. Experiment 3 revealed that prior exposure to response-contingent shock has a general enabling effect on spinal cord plasticity, facilitating the acquisition of the instrumental response when subjects are tested on the contralateral leg.
Instrumental learning can be conceived of as involving three mechanisms. First, a process that is sensitive to the response –reinforcer relation is needed. In the absence of this process, contingent and noncontingent shock would have an identical effect (and our behavioral modification would fail the definition of instrumental learning). The second represents the modification of a particular stimulus – response pathway, a selective effect that preserves our index of learning (increased flexion duration in the pretreated limb) over time. In physiological terms, this might reflect a homosynaptic modification within a specific neural pathway. The third mechanism is modulatory in nature, influencing the propensity to learn. In intact animals, this component is often characterized in terms of motivation or drive [21]. In our spinal preparation, the modulatory component is evident from the enabling/disabling of behavioral potential on the contralateral limb. Physiologically, such a diffuse effect would seem heterosynaptic in nature. Both the second and third mechanisms can be thought of as reflecting a kind of memory, in one case preserving a selective modification in a particular neural pathway and in the other maintaining a diffuse change in the capacity for learning. The latter represents a type of metaplasticity, a factor that modulates the capacity for change [22]. Does each of these mechanisms depend on neurons within the spinal cord? From the results of Experiment 3, it seems clear that response-contingent shock engages a process within the spinal cord that enables behavioral potential. Conversely, other studies suggest that noncontingent shock engages an intraspinal mechanism that disables behavioral potential [20]. These forms of metaplasticity appear to be spinally mediated.
6.1. Behavioral consequences These results suggest that the application of shock has three distinct consequences. The first is the direct elicitation of the response due to stimulation of the tibialis anterior muscle. While the magnitude of this effect may habituate [9], it does not appear to be influenced by the instrumental relation; both contingent and noncontingent shock generate a flexion response and, in both cases, the force of this response habituates as a function of shock exposure. The second component is tied to the increase in flexion duration, our primary measure of instrumental learning. We know from prior work that the development of this effect depends on the response –reinforcer relation. Rats given contingent shock exhibit a progressive increase in flexion duration whereas subjects given an equivalent amount of noncontingent shock do not. The results from Experiments 1 and 2 show that the acquisition of this behavioral response depends on neurons within the spinal cord. The third component was revealed in Experiment 3, where we found that response-contingent shock fosters learning when subjects are tested on the contralateral leg. This effect also depends on the response –reinforcer relation, for a different process is engaged when this relation is degraded. Specifically, rats given shock independent of leg position fail to learn when response-contingent shock is applied to the contralateral leg [20]. Neither contingent nor noncontingent shock affects contralateral limb position or initial flexion force. Rather, these shock treatments appear to engage a modulatory process that has a general effect on behavioral plasticity; contingent shock appears to enable behavioral potential while noncontingent shock has a disabling effect.
6.3. Physiological mechanisms Experiments 1 and 2 showed that the acquisition of an instrumental response depends on spinal cord neurons. From this, we can conclude that spinal cord neurons are essential because this system abstracts the response –reinforcer relation and/or stores the consequences of learning. Further physiological work will be needed to determine whether both components depend on spinal cord neurons. Moreover, it is possible that response-contingent shock has a differential effect because it activates a system that directly alters a nociceptive pathway, blurring the distinction between the process that detects the essential behavioral – environmental relation and the structural modification that preserves the consequence of learning over time (memory). Using pharmacological techniques, we have begun to gain some insight into the underlying mechanisms. Administration of the NMDA antagonist APV blocks the acquisition of the instrumental response, suggesting that this behavioral modification depends on the NMDA receptor [23]. Others have shown that nociceptive stimulation can
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
sensitize neurons within the spinal cord (central sensitization) and that this effect depends on a form of NMDAmediated long-term potentiation (LTP) [2]. If the induction of this process is influenced by instrumental contingencies, it could underlie our learning effect. Other studies have focused on the mechanisms that disable behavioral plasticity. In an extensive series of experiments, we showed that the expression of the behavioral deficit depends on a ligand that acts at the kappa opioid receptor [24]. The disabling effect also appears to involve GABAergic systems because administration of the GABA-A antagonist bicuculline can block both the induction and expression of the behavioral deficit [25]. Interestingly, both systems have been shown to modulate the development of LTP [4]. 6.4. Clinical implications Our discovery that instrumental relations can engage mechanisms that modulate behavioral potential within the spinal cord could have important implications for the recovery of function after a spinal cord injury. Wernig et al. have shown that the rules of spinal locomotion can be used to shape locomotive behavior in humans and foster the recovery of stepping after a spinal cord injury [26,27]. Behavioral, physiological, and pharmacological treatments that enable instrumental learning may promote reacquisition of stepping. Conversely, coincident nociceptive stimulation associated with a spinal cord injury could engage processes that disable behavioral potential and undermine recovery. Identifying the mechanisms that underlie this destructive process could help preserve/restore behavioral function.
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14] [15] [16] [17] [18]
[19]
Acknowledgements [20]
This work was funded by the National Institute of Mental Health grant MH60157 to J.W.G. The authors would like to thank Amy Sieve, Stephanie Washburn, Guadalupe Garcia, Shivali Dhruv, Jill Olson, and Brianne Patton for their comments on a previous version of this manuscript.
[21] [22] [23]
References [1] Patterson MM, Grau JW. Spinal cord plasticity: alterations in reflex function. Boston: Kluwer Academic Publishing; 2001. [2] Coderre TJ, Katz J, Vaccarino AL, Melzack R. Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 1993;52:259 – 85. [3] Dickenson AH. Balances between excitatory and inhibitory events in the spinal cord and chronic pain. Prog Brain Res 1996;110:225 – 31. [4] Herrero JF, Laird JMA, Lopez-Garcia JA. Wind-up of spinal cord neurons and pain sensation: much ado about something? Prog Neurobiol 2000;61:169 – 203. [5] Hodgson JA, Roy RR, de Leon R, Dobkin B, Edgerton VR. Can the
[24]
[25]
[26]
[27]
267
mammalian lumbar spinal cord learn a motor task? Med Sci Sports Exerc 1994;26:1491 – 7. Nakada K, Hodgson JA, de Leon R, Roy RR, Edgerton VR. Prolonged modification of the mechanics of the step cycle by single and repetitive mechanical stimuli in chronic spinal cats. Soc Neurosci Abstr 1994;20:1755. Crown ED, Grau JW. Preserving and restoring behavioral potential within the spinal cord using an instrumental training paradigm. J Neurophys 2001;86:845 – 55. Grau JW. Instrumental conditioning. The Corsini encyclopedia of psychology and behavioral science. 3rd ed. New York: Wiley; 2000. p. 767 – 9. Grau JW, Barstow DG, Joynes RL. Instrumental learning within the spinal cord: I. Behavioral properties. Behav Neurosci 1998;112: 1366 – 88. Grau JW, Penland HR, Joynes RL. Inescapable shock disrupts spinal learning: evidence for spinal mediation and naltrexone reversibility. Soc Neurosci Abstr 1996;21:1840. Grau JW, Salinas JA, Illich PA, Meagher MW. Associative learning and memory for an antinociceptive response in the spinalized rat. Behav Neurosci 1990;104:489 – 94. Illich PA, Salinas JA, Grau JW. Latent inhibition, overshadowing, and blocking of a conditioned antinociception response in spinalized rats. Behav Neural Biol 1994;62:140 – 50. Joynes RL, Grau JW. Mechanisms of Pavlovian conditioning: role of protection from habituation in spinal conditioning. Behav Neurosci 1996;110:1375 – 87. Horridge GA. Learning of the leg position by the ventral nerve cord in headless insects. Proc R Soc Lond 1962;157:33 – 52. Yaksh TH, Rudy TA. Chronic catheterization of the spinal cord subarachnoid space. Physiol Behav 1976;17:1031 – 6. Church RM, Lerner ND. Does the headless roach learn to avoid? Physiol Psychol 1976;4:439 – 42. Hebel R, Stromberg MW. Anatomy and embryology of the laboratory rat. Worthsee: BioMed Verlag; 1986. Coderre TJ. Noxious stimulus-induced plasticity in spinal cord dorsal horn: evidence and insights on mechanisms obtained using the formalin test. In: Patterson MM, Grau JW, editors. Spinal cord plasticity: alterations in reflex function. Norwell, MA: Kluwer Academic Publishing; 2001. p. 163 – 85. Coderre TJ, Yashpal K, Herny JL. Specific contribution of lumbar spinal mechanisms to persistent nociceptive responses in the formalin test. NeuroReport 1994;5:1337 – 40. Joynes RL, Ferguson AR, Crown ED, Patton BC, Grau JW. Instrumental learning within the spinal cord: V. Evidence the behavioral deficit observed after noncontingent nociceptive stimulation reflects an intraspinal modification [submitted for publication]. Hull CL. Essentials of behavior. New Haven: Yale Univ. Press; 1951. Abraham WC, Tate WP. Metaplasticity: a new vista across the field of synaptic plasticity. Prog Neurobiol 1997;52:303 – 23. Joynes RL, Janjua KR, Grau JW. Instrumental learning within the spinal cord: VI. Disruption of learning by the NMDA antagonist APV. Behav Neurosci [in press]. Joynes RL, Grau JW. Instrumental learning within the spinal cord: III. Prior exposure to noncontingent shock induces a behavioral deficit that is blocked by an opioid antagonist. Behav Brain Res [in press]. Ferguson AR, Crown ED, Dhruv S, Washburn SN, Grau JW. Instrumental learning in spinalized rats: the GABA antagonist bicuculline prevents the behavioral deficit observed after noncontingent shock. Soc Neurosci Abstr 2000;26:2207. Wernig A, Muller S, Nanassy A, Cagol E. Laufband therapy based on ‘‘rules of spinal locomotion’’ is effective in spinal cord injured persons. Eur J Neurosci 1995;7:823 – 9. Wernig A, Nanassy A, Muller S. Maintenance of locomotor abilities following Laufband (treadmill) therapy in para- and tetraplegic persons: follow-up studies. Spinal Cord 1998;36:744 – 9.
Instrumental learning within the spinal cord II. Evidence for central mediation Eric D. Crowna,*, Adam R. Fergusona, Robin L. Joynesb, James W. Graua a
Department of Psychology, Texas A&M University, College Station, TX 77843, USA b Kent State University, Kent, OH, USA
Received 19 September 2001; received in revised form 18 December 2001; accepted 27 June 2002
Abstract Rats spinally transected at the second thoracic vertebra can learn to maintain their leg in a flexed position if they receive legshock for extending the limb. These rats display an increase in the duration of a flexion response that minimizes net shock exposure. The current set of experiments was designed to determine whether the acquisition of this behavioral response is mediated by the neurons of the spinal cord (i.e., is centrally mediated) or reflects a peripheral modification (e.g., a change in muscle tension). Experiment 1 found that preventing information from reaching the spinal cord by severing the sciatic nerve blocked the acquisition of this behavioral response. Spinalized rats also failed to learn if the spinal cord was anesthetized with lidocaine during exposure to response-contingent shock (Experiment 2). Experiment 3 demonstrated that prior exposure to response-contingent shock on one hindleg facilitated acquisition of the response when subjects were later tested on the opposite leg. These findings suggest that acquisition of the instrumental response depends on neurons within the spinal cord. D 2002 Elsevier Science Inc. All rights reserved. Keywords: Plasticity; Spinal cord; Lidocaine; Sciatic nerve; Instrumental learning
1. Introduction Recent studies of pain modulation, locomotion, and learning have shown that the spinal cord is far more plastic than originally imagined (for recent review, see Ref. [1]). For example, researchers in the pain literature have demonstrated that nociceptive stimulation can sensitize neurons within the spinal cord, a process known as central sensitization [2 –4]. Researchers in the locomotion literature have found that the spinal cord contains the neural circuitry necessary to support organized stepping and can adapt to changing environmental conditions (e.g., obstacles placed in their path during stepping [5,6]). Our laboratory has used traditional learning paradigms to examine the functional properties of spinal cord neurons. We have demonstrated that the neurons of the spinal cord are capable of Pavlovian and instrumental conditioning [7 – 13]. To explore the capacity for instrumental learning, we have used a modification of the master – yoke paradigm of
*
Corresponding author. Tel.: +1-979-862-4852; fax: +1-979-845-4727. E-mail address: [email protected] (E.D. Crown).
Horridge [14]. In this paradigm, one group of spinally transected (spinalized) rats (master) is given shock to a hindleg whenever that limb is extended (response-contingent shock), causing a flexion of the foot at the ankle joint. Instituting a contingency between the response (leg flexion) and the reinforcer (shock onset) produces a change in the duration of the flexion response that minimizes net shock exposure. Subjects in another group (yoked) receive legshock at the same time and for the same duration as the master subjects; however, these subjects receive shock independent of leg position (noncontingent shock). We have shown that rats given response-contingent shock display a progressive increase in the duration of the flexion response, whereas rats given noncontingent shock do not exhibit such an increase in response duration [9,10]. This suggests that response-contingent shock has a differential effect, a defining feature of instrumental learning (additional evidence that the increase in response duration depends on the instrumental relation is provided elsewhere [9]). We have assumed that the change in flexion duration produced by response-contingent shock reflects a centrally mediated phenomenon—an example of learning within the spinal cord. However, there is no empirical evidence for this
0031-9384/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 0 3 1 - 9 3 8 4 ( 0 2 ) 0 0 8 5 9 - 4
260
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
assumption. Indeed, as it stands, one could argue that a peripheral mechanism underlies the change in flexion duration. For example, perhaps response-contingent stimulation of the muscles produces a cumulative increase in muscle tension that maintains the flexion response. What is especially worrisome is that this could occur without any input from the spinal cord. The present experiments examine whether the change in flexion duration (our measure of learning) observed in spinalized rats depends on neurons within the spinal cord. We address this issue by testing whether severing the connection with the spinal cord (Experiment 1), or inhibiting neural activity within the spinal cord through intrathecal application of the Na + channel blocker lidocaine (Experiment 2), affects learning. We also examine whether response-contingent shock applied to one hindlimb affects learning on the opposite (contralateral) leg (Experiment 3). Our results demonstrate that spinal cord neurons play an essential role in the acquisition of the instrumental response.
2. General methods 2.1. Subjects The subjects were male Sprague – Dawley rats obtained from Harlan (Houston, TX). The rats were approximately 100 days old and were individually housed with food and water continuously available. Rats were maintained on a 12h light/dark cycle and were tested during the last 6 h of the light cycle.
vocalize to the legshock, and (c) examining the spinal cord postmortem in a randomly selected subset of the subjects. 2.2.2. Severing the sciatic nerve In Experiment 1, the sciatic nerve innervating one hindleg was severed in a subset of the subjects. Rats were spinalized at T2 as described above. While still anesthetized, a small longitudinal incision was made approximately 1 cm from the midline at the level of the hip. The sciatic nerve was exposed using a glass probe to gently separate the nerve from the surrounding muscle tissue. For half of the subjects, the nerve was severed approximately 1 cm lateral to the dorsal root, with a cautery device. The remaining subjects served as the sham-operated controls. Whether this surgery was conducted on the left or right hindleg was counterbalanced across subjects. 2.2.3. Intrathecal catheter implantation In Experiment 2, an intrathecal catheter was inserted down the cord after cauterization following the procedure of Yaksh and Rudy [15]. This was accomplished by fitting a 25-cm segment of PE-10 tubing with a 0.23-cm (diameter) stainless steel guide wire (SWGX-090; Small Parts) and sliding it caudally 9 cm into the subarachnoid space at T2, so that the tubing lay along the dorsal surface of the cord. The wire was then removed and the tubing was secured with adhesive (Superglue). The exposed cord was then covered with Oxycel (Parke Davis) and the wound closed with Michel clips. In this way, approximately 10 cm of tubing protruded from the wound. 2.3. Apparatus
2.2. Surgery 2.2.1. Spinalization Rats were anesthetized with pentobarbital (50 mg/kg ip). To stabilize and position the rat’s body for surgery, its head was held in a stereotaxic instrument and a small gauze ‘‘pillow’’ was placed under its chest. After the second thoracic vertebra (T2) was localized by touch, an anterior – posterior incision was made on the back over T2. Next, the tissue in front of T2 was cleared away and the cord was transected by cauterization. The exposed cord was then covered with Oxycel (Parke Davis) and the wound closed with Michel clips. Rats were maintained in a temperaturecontrolled environment (approximately 25.5 °C) during recovery and testing. During recovery, the rat’s rear legs were maintained in a normal flexed position by a piece of porous tape (Orthaletic, 1.3 cm width) gently wrapped once around the rat’s body. For all experiments, the recovery period between surgery and behavioral testing was approximately 24 h. Transections were confirmed by (a) inspecting the cord during the operation, (b) observing the behavior of the subjects after they have recovered to insure that they exhibit paralysis below the level of the forepaws and do not
Instrumental training was conducted while spinal rats were loosely restrained in specially designed tubes. A detailed description of the apparatus and experimental setup is provided in Grau et al. [9]. Briefly, the tubes have two slots that allow both hindlegs to hang freely. To minimize the effects of upper body movements on leg position, the rat’s midsection was gently secured with a wire ‘‘belt.’’ Legshock was applied by attaching one lead from a BRS/ LVE shock generator (Model SG-903) to an intracutaneous wire electrode that was implanted through the skin over the tibia. The other lead was attached to a 2.5-cm stainless steel pin that was inserted 0.4 cm into the tibialis anterior muscle. Leg position was monitored using an insulated contact electrode taped to the plantar surface of the rat’s foot, just distal to the plantar protuberance. A fine wire attached to the proximal end of the contact electrode extended from the rear of the foot and was connected to a digital input monitored by a Macintosh computer. A plastic dish containing a NaCl solution was placed below the restraining tube, so that the tip of the rod could contact the solution. A drop of detergent was added to the solution to reduce surface tension. A ground wire was connected to a 1-mm stainless steel rod,
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
which was placed in the solution. When the contact electrode attached to the rat’s paw touched the solution, it completed the circuit monitored by the computer. The state of this circuit was sampled at a rate of 30 times per second. Flexion force was measured by attaching a monofilament plastic line (‘‘4 lb test,’’ Stren, Dupont) to the rat’s foot immediately proximal to the plantar protuberance. The 40 cm length of line was passed through an eyelet attached to the apparatus directly under the paw. The end of the line was attached to a strain gauge that was fastened to a ringstand. After the line was connected to the rat’s paw, the ringstand was positioned so that the line was taught, just barely engaging the gauge. Shock intensity was adjusted to produce a flexion force of a predetermined value (0.4 N). The strain gauge was then removed from the rat’s foot. 2.4. Procedure 2.4.1. Testing with response-contingent shock Prior to testing, the rear legs were shaved and marked for placement of the shock leads. The wire electrode was then inserted over the tibia at the distal mark and the rats were placed in the restraining tubes. Next, the contact electrode used to monitor leg position was taped to the paw. To minimize lateral leg movements, a 20-cm piece of porous tape (Orthaletic, 1.3 cm) was wrapped around the leg and taped to a bar extending across the apparatus directly under the front panel of the restraining tube. The tape was adjusted so that it was taut enough to extend the joint between the tibia and femur. One lead from the shock generator was attached to the stainless steel wire inserted over the tibia. The shock generator was set to deliver a 0.1-mA shock and the region over the second mark was probed to find a site that elicited a vigorous flexion response. The pin was then inserted perpendicular to the body into the tibialis anterior muscle. The electrode placements were checked by verifying that a single intense (1.6 mA) test shock (0.3 s) elicited a flexion response of at least 0.8 N. The shock intensity necessary to induce a 0.4-N flexion response was obtained for the hindleg as described above. Finally, three short (0.15s) shock pulses were applied and the level of the salt solution was adjusted so that the tip of the rod was submerged 4 mm below the surface. During testing with response-contingent legshock, each rat received shock immediately after its leg contacted the salt solution for the duration of the 30-min observation period. We have shown previously that shock onset is a critical variable in this paradigm, as delaying shock onset by as little as 100 ms disrupts instrumental performance [9]. 2.4.2. Experiment 1 Rats (N = 12) were spinalized at T2 and either had the sciatic nerve cut or underwent a sham operation as described above. Twenty-four hours after surgery, subjects were tested with response-contingent shock applied to the previously operated leg, as described above.
261
2.4.3. Experiment 2 Twelve spinalized rats with intrathecal catheters served as subjects. A 0.05-ml intrathecal injection of lidocaine (Xylocaine) was administered to half of the subjects (Lidocaine) at a concentration of 20 mg/ml immediately before the rats were placed in the training apparatus. This was accomplished by inserting a 27-gauge needle into the exposed end of the catheter tubing and injecting the drug slowly over a period of 1 min. The lidocaine injection was followed by 0.01 ml of 0.9% saline. To verify the efficacy of the drug, tail-flick responses to a slight pinch were monitored to make sure reflexive activity had ceased. The other half of the subjects (Sham) underwent the same surgical procedures but received no injection. They were then placed in the apparatus and tested with responsecontingent shock as described above. Whether testing occurred on the left or right hindleg was counterbalanced across subjects. 2.4.4. Experiment 3 Twenty-four rats were divided into three groups for Experiment 3. This experiment involved two phases. In Phase 1, subjects were prepared for legshock as described above. Instead of placing shock leads into one leg, however, both legs were prepared for training at the same time so that the left leg was in one solution and the right leg was in another solution. After 30 min of training with either response-contingent legshock or nothing and a 10-min period that allowed for the previously trained leg to return to resting position, Phase 2 testing began. First, the water level of both salt solutions for each subject was reset by adding 100 ml of water. Following this, subjects received 30 min of response-contingent shock applied to either the pretrained leg (ipsilateral; n = 8) or the opposite leg (contralateral; n = 8) (Phase 2). The unshocked subjects from Phase 1 (n = 8) received response-contingent shock to the left or right leg in a counterbalanced fashion. Because Phase 2 required that the subject’s leg not be disturbed before testing, a few modifications had to be made to our testing procedure. Prior to Phase 1, the shock intensity needed to induce a 0.4-N change in flexion force was determined for both legs and the water level needed to submerge the contact electrode 4 mm for each leg was set. In Phase 2, addition of 100 ml of water elevated the water level by 4 mm, effectively raising the response criterion for the test leg from 4 to 8 mm. 2.5. Behavioral measures Three behavioral measures were used to monitor performance: time in solution, response number, and response duration (see Ref. [9]; Fig. 2). The computer recorded when the contact electrode touched the underlying solution (time in solution). Whenever the electrode left the solution, the number of responses was increased by one (response number). To obtain a measure of performance
262
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
over time, we divided the session into thirty 1-min time bins. We have previously shown how instrumental learning can be distinguished from a reactive system that is insensitive to response –reinforcer relations [9,16]. A key difference concerns response duration; only the instrumental account anticipates that contingent shock will produce a progressive increase in response duration. Response duration was derived from time in solution and response number using the following equation: Response durationi=(60 s time in solutioni)/(Response numberi + 1) where i was the current time bin. To address the possibility that differences in response duration during testing reflect a loss of responding in the previously shocked rats, we also present response number. 2.6. Statistics The results were analyzed using an analysis of variance (ANOVA). Post hoc comparisons were made using Duncan’s New Multiple Range test. In all cases, a criterion of P < .05 was used to judge statistical significance.
3. Experiment 1 The shock stimulus used to study instrumental learning in spinalized rats should have two consequences: (1) the direct elicitation of a motor response, and (2) the induction of an afferent barrage that engages neurons within the spinal cord. We have assumed that learning (the increase in flexion duration) depends on the second component. If so, then eliminating communication with the spinal cord should prevent learning. The present experiment tests this by severing the sciatic nerve, the primary source of innervation for the region where our shock stimulus is applied [17]. 3.1. Results The values for the shock intensity necessary to induce a 0.4-N change in flexion force ( ± S.E.) ranged from 0.21 mA ( ± 0.04) to 0.21 mA ( ± 0.06) and the values for initial response duration ranged from 0.42 s ( ± 0.02) to 0.52 s ( ± 0.06). Independent ANOVAs confirmed that these differences did not approach statistical significance (both Fs < 2.5, P>.05).
Fig. 1. Performance over time during testing in Experiment 1. The change in response duration (top panels) and response number (bottom panels) observed across the testing phase for rats that received either a sciatic nerve lesion (filled circles) or sham surgery (open circles). The group means ( ± S.E.M.) are depicted in the right panels.
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
Sham-operated rats exhibited a progressive increase in flexion duration over the 30 min of testing with responsecontingent shock (Fig. 1, top panels). Rats that had the sciatic nerve severed failed to learn. An ANOVA revealed a significant effect of Surgery Condition (sciatic transection versus sham), Time, and Surgery Time interaction (all Fs>2.76, P < .0001). Spinalized rats that had received a sciatic nerve transection failed to maintain the instrumental flexion response, as evidenced by the fact that this group made more contacts with the salt solution (Fig. 1, bottom panels). An ANOVA found a significant main effect of Surgery Condition [ F(1,10) = 7.33, P < .05]. The main effect of Time and its interaction with Surgery Condition were not significant (both Fs < 1.46, P>.05). Because our reinforcer involves direct electrical stimulation of the tibialis anterior muscle, we expected that cutting the sciatic nerve would have little effect on the elicitation of the motor response. Of greater concern was whether repeatedly performing the response would produce an increase in the flexion response when input from the sciatic nerve was eliminated. We found that rats given
263
response-contingent shock after the sciatic nerve was cut did not exhibit an increase in flexion duration, suggesting that the acquisition of the instrumental response depends on communication with spinal cord neurons.
4. Experiment 2 Experiment 1 demonstrated that input from the sciatic nerve is necessary for the acquisition of the instrumental flexion response in spinalized rats. Experiment 2 used a pharmacological manipulation, administration of intrathecal lidocaine, to further evaluate whether spinal cord neurons play an essential role in the acquisition of the instrumental response. Lidocaine generally depresses neural activity within the spinal cord and has been shown to prevent the increase in neural excitability observed in response to intermittent shocks (wind-up) and peripheral inflammation (central sensitization) [18,19]. If instrumental learning depends on spinal cord neurons, lidocaine-treated rats should not exhibit an increase in flexion duration when given response-contingent shock.
Fig. 2. Performance over time during testing in Experiment 2. The change in response duration (top panels) and response number (bottom panels) observed across the testing phase for subjects given intrathecal lidocaine (filled circles) or nothing (open circles) subjects. The group means ( ± S.E.M.) are depicted in the right panels.
264
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
4.1. Results The shock intensity needed to elicit a 0.4-N flexion varied from 0.58 mA ( ± 0.04) to 0.59 mA ( ± 0.06) and initial flexion durations ranged from 0.16 s ( ± 0.01) to 0.19 s ( ± 0.01). Independent ANOVAs verified that these differences did not approach statistical significance (both Fs < 1.0, P>.05). The untreated controls exhibited an increase in flexion duration over the 30 min of testing (Fig. 2, top panels). Pretreatment with lidocaine blocked learning. An ANOVA confirmed that Drug Treatment had a significant impact [ F(1, 10) = 14.73, P < .01]. There was also a significant effect of Time, and a Drug Treatment Time interaction (both Fs>1.65, P < .05). Subjects that received lidocaine also made more contacts with the salt solution (Fig. 2, bottom panels). An ANOVA yielded a main effect for Drug Treatment that approached statistical significance [ F(1, 10) = 4.37, P < .063]. Neither the main effect of Time nor its interaction with Drug Treatment was significant (both Fs < 1.25, P>.05). In summary, we found that intrathecal lidocaine prevented performance of the instrumental flexion response. The drug did not, however, eliminate the shock-elicited response. As observed in Experiment 1, repeated execution of the response did not, in the absence of a functional spinal cord, bring about an increase in flexion duration, implying that the spinal cord is essential.
5. Experiment 3 We have previously shown that exposure to noncontingent shock produces a behavioral deficit that undermines the capacity for instrumental learning [9,10]. To test whether this effect depends on spinal cord neurons, we showed that blocking the afferent input and pretreatment with intrathecal lidocaine block the development of the behavioral deficit [20]. We also showed that rats given noncontingent shock to one hindleg fail to learn when contingent shock is later applied to the opposite leg. This transfer suggests that a common system, within the spinal cord, must mediate the behavioral deficit. Thus, it appears that noncontingent shock has a disabling effect that generally undermines behavioral potential. Perhaps exposure to contingent shock has the opposite effect—one that generally enables behavioral potential. If it does, training with contingent shock applied to one hindlimb could potentially facilitate acquisition when responsecontingent shock is applied to the contralateral leg. A problem with testing this hypothesis is that the rapid learning exhibited by our untreated controls could potentially mask the beneficial effects of prior training with contingent shock. We address this issue by testing subjects with the response criterion set to a higher level (a contact electrode depth of 8 mm). A pilot study indicated that, under
these conditions, untreated controls often fail to learn. We also modified the procedure used in earlier studies in another regard. Normally, we equate flexion force and contact electrode depth prior to testing. Although such a procedure helps to guard against the negative consequences of motor fatigue and other nonassociative habituation-like effects, it undermines our capacity to resolve a beneficial effect of training with response-contingent shock on the pretreated (ipsilateral) leg. The problem is that subjects trained with contingent shock will, as a consequence of the training, maintain their leg at a higher position. If contact electrode depth is then equated based on this new position, we effectively eliminate the behavioral benefit of the training and force subjects to acquire the instrumental response to meet the new (higher) criterion. To avoid this problem, we equated contact electrode depth prior to training. At the end of training, the criterion was raised by an equal amount for all subjects. 5.1. Results 5.1.1. Training The shock intensity needed to elicit a flexion response of 0.4 N prior to training ranged from 0.40 mA ( ± 0.01) to 0.44 mA ( ± 0.04). These group differences did not approach significance [ F(2,22) = 1.02, P>.05]. During training, rats given response contingent shock showed a progressive increase in the duration of the flexion response relative to unshocked rats (Fig. 3, top left panel). An ANOVA yielded a main effect of Shock Condition, Time, as well as a significant Shock Condition Time interaction (all Fs>3.66, P < .0001). Post hoc tests found that both response-contingent shocked groups displayed significantly longer flexion durations than did unshocked rats (both Ps < .05) and that the response-contingent shocked groups did not differ ( P>.05). The groups that received response-contingent shock also made more contacts with the salt solution than did the unshocked rats during training (Fig. 3, bottom left panel). An ANOVA revealed a significant main effect of Shock Condition, Time, and a significant Shock Condition Time interaction (all Fs>5.05, P < .001). Post hoc analyses showed that both response-contingent shocked groups made more responses (solution contacts) than the unshocked group and that the two shocked groups did not differ. 5.1.2. Testing Subjects that had previously received response-contingent shock learned to maintain their leg at a higher level when the response criterion was raised prior to testing (Fig. 3, top right panel). The magnitude of this effect was the same independent of whether subjects were tested on the ipsilateral or contralateral leg. Subjects that had not received training with response-contingent shock failed to learn under these conditions. An ANOVA yielded a significant
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
265
Fig. 3. Performance over time during training and testing in Experiment 3. The change in response duration (top panels) and response number (bottom panels) observed across both the training and testing phases for subjects trained and tested on the ipsilateral (open circles) or contralateral leg (closed triangles). A control group remained unshocked during the training phase (closed squares). The group means for testing ( ± S.E.M.) are depicted in the rightmost panels.
main effect of Shock Condition (ipsilateral, contralateral, or, unshocked) and Time (both Fs>2.78, P < .01). The Shock Condition Time interaction, however, was not significant [ F(58,609)>1.0, P>.05]. Post hoc analyses determined that the ipsilateral and contralateral groups displayed significantly longer flexion durations than the unshocked group (both Ps < .05). The ipsilateral and contralateral groups were not significantly different from each other ( P>.05). As usual, rats that failed to learn exhibited a greater number of responses (Fig. 3, bottom right panel). An ANOVA revealed a significant main effect of Shock Condition and Time, as well as a significant Shock Condition Time interaction (all Fs>2.04, P < .01). Post hoc tests found that the ipsilateral and contralateral groups made significantly fewer contacts with the salt solution than the unshocked group (both Ps < .05). Again, there were no significant differences between the ipsilateral and contralateral groups ( P>.05). We found that raising the response criterion disrupted learning in the untreated controls. Subjects that had previously experienced response-contingent shock were able to learn under these more difficult test conditions. Most importantly, this benefit of training was equally robust on the contralateral leg. This suggests that training with contingent shock induces a modification within the spinal cord that fosters (enables) instrumental behavior.
In some regard, it may seem surprising that training with contingent shock did not produce a greater effect on the ipsilateral leg. This would appear to indicate that instrumental training in this paradigm has only a general enabling effect—that it does not selectively augment the trained reflex. However, it is important to remember that shock treatment also induces a response-selective decrease in shock reactivity, a habituation-like effect that develops independent of the response –reinforcer relation [9]. This habituation effect would selectively undermine instrumental behavior on the pretrained limb, reducing the likelihood of observing both a response-specific and a more general benefit of training with response-contingent shock. To observe the former, we would need to equate flexion force after training. But with our current equipment, we cannot readjust flexion force without readjusting contact electrode depth. As noted above, this would introduce a confound in our test of contralateral transfer. Thus, while the present results clearly show that there is an enabling effect that generally fosters instrumental behavior, the data do not specifically address the relative contribution of the limb-specific component. Of course, such a component is evident from the behavior observed during, and immediately after, training with contingent shock, for this training only affects leg position in the pretrained limb.
266
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
6. Discussion
6.2. Mechanisms of instrumental learning
We had previously shown that spinalized rats exposed to response-contingent shock learn to maintain their leg in a flexed position, effectively minimizing net shock exposure. Although we assumed that this form of instrumental learning was due to an intraspinal modification, the possibility existed that this phenomenon reflected a peripheral modification. The current set of experiments discount this possibility by showing that the behavioral modification relies on spinal cord neurons. Experiment 1 found that removing the afferent barrage to the spinal cord, by severing the sciatic nerve, prevented spinalized rats from displaying an increase in flexion duration in response to contingent legshock. Experiment 2 determined that anesthetizing the spinal cord with lidocaine prevented instrumental learning. Importantly for both Experiments 1 and 2, the failure to learn occurred even though shock directly elicited a leg flexion. Experiment 3 revealed that prior exposure to response-contingent shock has a general enabling effect on spinal cord plasticity, facilitating the acquisition of the instrumental response when subjects are tested on the contralateral leg.
Instrumental learning can be conceived of as involving three mechanisms. First, a process that is sensitive to the response –reinforcer relation is needed. In the absence of this process, contingent and noncontingent shock would have an identical effect (and our behavioral modification would fail the definition of instrumental learning). The second represents the modification of a particular stimulus – response pathway, a selective effect that preserves our index of learning (increased flexion duration in the pretreated limb) over time. In physiological terms, this might reflect a homosynaptic modification within a specific neural pathway. The third mechanism is modulatory in nature, influencing the propensity to learn. In intact animals, this component is often characterized in terms of motivation or drive [21]. In our spinal preparation, the modulatory component is evident from the enabling/disabling of behavioral potential on the contralateral limb. Physiologically, such a diffuse effect would seem heterosynaptic in nature. Both the second and third mechanisms can be thought of as reflecting a kind of memory, in one case preserving a selective modification in a particular neural pathway and in the other maintaining a diffuse change in the capacity for learning. The latter represents a type of metaplasticity, a factor that modulates the capacity for change [22]. Does each of these mechanisms depend on neurons within the spinal cord? From the results of Experiment 3, it seems clear that response-contingent shock engages a process within the spinal cord that enables behavioral potential. Conversely, other studies suggest that noncontingent shock engages an intraspinal mechanism that disables behavioral potential [20]. These forms of metaplasticity appear to be spinally mediated.
6.1. Behavioral consequences These results suggest that the application of shock has three distinct consequences. The first is the direct elicitation of the response due to stimulation of the tibialis anterior muscle. While the magnitude of this effect may habituate [9], it does not appear to be influenced by the instrumental relation; both contingent and noncontingent shock generate a flexion response and, in both cases, the force of this response habituates as a function of shock exposure. The second component is tied to the increase in flexion duration, our primary measure of instrumental learning. We know from prior work that the development of this effect depends on the response –reinforcer relation. Rats given contingent shock exhibit a progressive increase in flexion duration whereas subjects given an equivalent amount of noncontingent shock do not. The results from Experiments 1 and 2 show that the acquisition of this behavioral response depends on neurons within the spinal cord. The third component was revealed in Experiment 3, where we found that response-contingent shock fosters learning when subjects are tested on the contralateral leg. This effect also depends on the response –reinforcer relation, for a different process is engaged when this relation is degraded. Specifically, rats given shock independent of leg position fail to learn when response-contingent shock is applied to the contralateral leg [20]. Neither contingent nor noncontingent shock affects contralateral limb position or initial flexion force. Rather, these shock treatments appear to engage a modulatory process that has a general effect on behavioral plasticity; contingent shock appears to enable behavioral potential while noncontingent shock has a disabling effect.
6.3. Physiological mechanisms Experiments 1 and 2 showed that the acquisition of an instrumental response depends on spinal cord neurons. From this, we can conclude that spinal cord neurons are essential because this system abstracts the response –reinforcer relation and/or stores the consequences of learning. Further physiological work will be needed to determine whether both components depend on spinal cord neurons. Moreover, it is possible that response-contingent shock has a differential effect because it activates a system that directly alters a nociceptive pathway, blurring the distinction between the process that detects the essential behavioral – environmental relation and the structural modification that preserves the consequence of learning over time (memory). Using pharmacological techniques, we have begun to gain some insight into the underlying mechanisms. Administration of the NMDA antagonist APV blocks the acquisition of the instrumental response, suggesting that this behavioral modification depends on the NMDA receptor [23]. Others have shown that nociceptive stimulation can
E.D. Crown et al. / Physiology & Behavior 77 (2002) 259–267
sensitize neurons within the spinal cord (central sensitization) and that this effect depends on a form of NMDAmediated long-term potentiation (LTP) [2]. If the induction of this process is influenced by instrumental contingencies, it could underlie our learning effect. Other studies have focused on the mechanisms that disable behavioral plasticity. In an extensive series of experiments, we showed that the expression of the behavioral deficit depends on a ligand that acts at the kappa opioid receptor [24]. The disabling effect also appears to involve GABAergic systems because administration of the GABA-A antagonist bicuculline can block both the induction and expression of the behavioral deficit [25]. Interestingly, both systems have been shown to modulate the development of LTP [4]. 6.4. Clinical implications Our discovery that instrumental relations can engage mechanisms that modulate behavioral potential within the spinal cord could have important implications for the recovery of function after a spinal cord injury. Wernig et al. have shown that the rules of spinal locomotion can be used to shape locomotive behavior in humans and foster the recovery of stepping after a spinal cord injury [26,27]. Behavioral, physiological, and pharmacological treatments that enable instrumental learning may promote reacquisition of stepping. Conversely, coincident nociceptive stimulation associated with a spinal cord injury could engage processes that disable behavioral potential and undermine recovery. Identifying the mechanisms that underlie this destructive process could help preserve/restore behavioral function.
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14] [15] [16] [17] [18]
[19]
Acknowledgements [20]
This work was funded by the National Institute of Mental Health grant MH60157 to J.W.G. The authors would like to thank Amy Sieve, Stephanie Washburn, Guadalupe Garcia, Shivali Dhruv, Jill Olson, and Brianne Patton for their comments on a previous version of this manuscript.
[21] [22] [23]
References [1] Patterson MM, Grau JW. Spinal cord plasticity: alterations in reflex function. Boston: Kluwer Academic Publishing; 2001. [2] Coderre TJ, Katz J, Vaccarino AL, Melzack R. Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 1993;52:259 – 85. [3] Dickenson AH. Balances between excitatory and inhibitory events in the spinal cord and chronic pain. Prog Brain Res 1996;110:225 – 31. [4] Herrero JF, Laird JMA, Lopez-Garcia JA. Wind-up of spinal cord neurons and pain sensation: much ado about something? Prog Neurobiol 2000;61:169 – 203. [5] Hodgson JA, Roy RR, de Leon R, Dobkin B, Edgerton VR. Can the
[24]
[25]
[26]
[27]
267
mammalian lumbar spinal cord learn a motor task? Med Sci Sports Exerc 1994;26:1491 – 7. Nakada K, Hodgson JA, de Leon R, Roy RR, Edgerton VR. Prolonged modification of the mechanics of the step cycle by single and repetitive mechanical stimuli in chronic spinal cats. Soc Neurosci Abstr 1994;20:1755. Crown ED, Grau JW. Preserving and restoring behavioral potential within the spinal cord using an instrumental training paradigm. J Neurophys 2001;86:845 – 55. Grau JW. Instrumental conditioning. The Corsini encyclopedia of psychology and behavioral science. 3rd ed. New York: Wiley; 2000. p. 767 – 9. Grau JW, Barstow DG, Joynes RL. Instrumental learning within the spinal cord: I. Behavioral properties. Behav Neurosci 1998;112: 1366 – 88. Grau JW, Penland HR, Joynes RL. Inescapable shock disrupts spinal learning: evidence for spinal mediation and naltrexone reversibility. Soc Neurosci Abstr 1996;21:1840. Grau JW, Salinas JA, Illich PA, Meagher MW. Associative learning and memory for an antinociceptive response in the spinalized rat. Behav Neurosci 1990;104:489 – 94. Illich PA, Salinas JA, Grau JW. Latent inhibition, overshadowing, and blocking of a conditioned antinociception response in spinalized rats. Behav Neural Biol 1994;62:140 – 50. Joynes RL, Grau JW. Mechanisms of Pavlovian conditioning: role of protection from habituation in spinal conditioning. Behav Neurosci 1996;110:1375 – 87. Horridge GA. Learning of the leg position by the ventral nerve cord in headless insects. Proc R Soc Lond 1962;157:33 – 52. Yaksh TH, Rudy TA. Chronic catheterization of the spinal cord subarachnoid space. Physiol Behav 1976;17:1031 – 6. Church RM, Lerner ND. Does the headless roach learn to avoid? Physiol Psychol 1976;4:439 – 42. Hebel R, Stromberg MW. Anatomy and embryology of the laboratory rat. Worthsee: BioMed Verlag; 1986. Coderre TJ. Noxious stimulus-induced plasticity in spinal cord dorsal horn: evidence and insights on mechanisms obtained using the formalin test. In: Patterson MM, Grau JW, editors. Spinal cord plasticity: alterations in reflex function. Norwell, MA: Kluwer Academic Publishing; 2001. p. 163 – 85. Coderre TJ, Yashpal K, Herny JL. Specific contribution of lumbar spinal mechanisms to persistent nociceptive responses in the formalin test. NeuroReport 1994;5:1337 – 40. Joynes RL, Ferguson AR, Crown ED, Patton BC, Grau JW. Instrumental learning within the spinal cord: V. Evidence the behavioral deficit observed after noncontingent nociceptive stimulation reflects an intraspinal modification [submitted for publication]. Hull CL. Essentials of behavior. New Haven: Yale Univ. Press; 1951. Abraham WC, Tate WP. Metaplasticity: a new vista across the field of synaptic plasticity. Prog Neurobiol 1997;52:303 – 23. Joynes RL, Janjua KR, Grau JW. Instrumental learning within the spinal cord: VI. Disruption of learning by the NMDA antagonist APV. Behav Neurosci [in press]. Joynes RL, Grau JW. Instrumental learning within the spinal cord: III. Prior exposure to noncontingent shock induces a behavioral deficit that is blocked by an opioid antagonist. Behav Brain Res [in press]. Ferguson AR, Crown ED, Dhruv S, Washburn SN, Grau JW. Instrumental learning in spinalized rats: the GABA antagonist bicuculline prevents the behavioral deficit observed after noncontingent shock. Soc Neurosci Abstr 2000;26:2207. Wernig A, Muller S, Nanassy A, Cagol E. Laufband therapy based on ‘‘rules of spinal locomotion’’ is effective in spinal cord injured persons. Eur J Neurosci 1995;7:823 – 9. Wernig A, Nanassy A, Muller S. Maintenance of locomotor abilities following Laufband (treadmill) therapy in para- and tetraplegic persons: follow-up studies. Spinal Cord 1998;36:744 – 9.