The Effects of Reversible Inactivation of the Red Nucleus on Learning

Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

The Effects of Reversible Inactivation of the Red Nucleus on Learning-Related and Auditory-Evoked Unit Activity in the Pontine Nuclei of Classically Conditioned Rabbits M. Claire Cartford, 1 Elizabeth B. Gohl, Maria Singson, and David G. Lavond Departments of Psychology and Biological Sciences University of Southern California Los Angeles, California 90089-2520

Abstract unit activity were attenuated but not abolished by red nucleus cooling. The pontine nuclei carry auditory conditioned stimulus information to the cerebellum during classical conditioning of Introduction the nictitating membrane response in rabbits. In well-trained animals In simple delay classical conditioning of the learning-related as well as stimulus-evoked rabbit nictitating membrane response, neural unit unit activity can be recorded throughout the activity that models the learned behavioral eye- pontine nuclei but particularly in the lateral blink motor response occurs in brainstem and cer- and dorsolateral pons. Recent work in our ebellum (Clark and Lavond 1993, 1996; Krupa et laboratory has provided evidence that the al. 1993; Krupa et al. 1996; Clark et al. 1992, this learning-related unit activity in the pons is issue). Permanent and reversible lesion studies as dependent on the interpositus nucleus and well as anatomical pathway tracing studies have that the pons is not a site of essential produced a growing body of evidence describing plasticity for the learned response. In the the circuitry of the eye-blink response and the flow present study we considered the question of of learning-related unit activity within that cir- whether learning-related unit activity might cuitry. Much of the data presented to date supports be projected from the interpositus nucleus the hypothesis that the interpositus (IP) nucleus of to the pons through the red nucleus, a the cerebellum is the locus of the essential associa- primary output target of the interpositus tion and the source of learning-related unit activity and a structure known to be essential for recorded in other structures within the circuit expression of the learned response. Multiple (Chapman et al. 1990; Clark et al. 1984, 1992, unit recordings were taken from lateral and 1996; Clark and Lavond 1993; Clark et al., this is- dorsolateral pontine locations in sue). Less is known about the functional roles that well-trained rabbits before and during may be played by each of the other structures that cooling of the red nucleus. Analysis of participate in the acquisition and expression of the pooled data for all recording locations learned behavior. within the lateral and dorsolateral pons Within the conditioned nictitating membrane indicated that reversible inactivation of red response circuitry the lateral and dorsolateral pon- nucleus abolished both stimulus-evoked and tine nuclei carry conditioned stimulus information learning-related unit activity. However, we to the cerebeUum and receive projections from also found discrete recording locations cerebellar deep nuclei (Eller and Chan-Pelay 1976; where stimulus-evoked and learning-related Watt and Mihailoff 1983; Gerrits and Voogd 1987; Steinmetz and Sengelaub 1992; Gould et al. 1993; J.K. Thompson, W.J. Spangler, and R.F. Thompson, unpubl.). There are known primary auditory pro- 1Corresponding author. jections to the lateral pontine area (Aitkin and Boyd

LEARNING & MEMORY 3:519-531 9 1997by Cold Spring Harbor Laboratory Press ISSN1072-0502/97 $5.00

L E A R N / N G & M E M O R Y 519 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

Cartford et al.

1978; Steinmetz et al. 1987). Electrophysiological cycle. The rabbits were cared for by the experi- data indicate that stimulus-evoked responses occur menters and by the staff and veterinarians of the in the pons to both light- and tone-conditioned University of Southern California. All care and treat- stimuli, as well as learning-related "models" that ment of the animals was approved and governed are increases in unit activity correlated to the rab- by the guidelines of the American Association bit eye-blink behavior or conditioned response for Accreditation of Laboratory Animal Care (Steinmetz et al. 1987; McCormick et al. 1983). (AAALAC), the American Psychological Association There is also some evidence that the pontine nu- (APA), and the National Institutes of Health. clei receive unconditioned stimulus somatosensory information (Lavond et al. 1981; McCormick et al. 1983). These anatomical and electrophysiological SURGERY findings suggest that the pons could be a site of convergence of conditioned stimulus and uncondi- Animals were operated before initial training tioned stimulus information, and therefore, could to control for any changes in learning that might be possibly be a locus of essential association for the caused by the presence of the implanted cold learned response. probe or recording electrode. The rabbits were Recently, Clark et al. (this issue) undertook in- preanesthetized with xylazine (5 mg/kg) and ket- vestigation of the question of whether the pontine amine (50 mg/kg) and maintained on 1.5%-2.5% nuclei might be a site of essential plasticity for the halothane for the duration of aseptic surgery. learned response. Their results indicate that learn- A cooling probe consisting of 19-gauge stain- ing-related unit activity in the pons is dependent less steel tubing (based on a design by Zhang et al. on the activity of the IP nucleus. This result lends 1986) was implanted in one of two possible loca- further support to the hypothesis that the IP tions. We inserted the cold probe into the brain nucleus is the site of essential association and that from a location anterior to the red nucleus so that learning-related unit activity that develops in the IP the body of the cold probe that remained above is projected subsequently to the pontine nuclei. the surface of the rabbit brain would not interfere On the basis of this result and the known anatomi- with placement of the base stand for the pontine cal and physiological connections between the IP recording electrode micromanipulator. To achieve and the lateral pons, we explored the question of proper placement of the probe it was inserted at a whether inactivation of red nucleus, the major out- 15 ~ angle from perpendicular to the brain surface put source of the conditioned response and the with the distal cooling tip of the probe located major efferent of the IP, would have an effect on either at coordinates AP +10, ML -2.5, DV -16, or the learning-related unit activity in the pons. We AP +12, ML -1.5, DV -16 (coordinates were de- used the technique of reversibly inactivating red rived from previous studies in this laboratory and nucleus by cooling probe while simultaneously re- were based originally on the stereotaxic atlas of cording from multiple units in varying locations McBride and Klemm 1968). From these locations within the lateral and dorsolateral pontine nuclei. cooling spread to the red nucleus either from a location lateral to the nucleus or from a location anterior to the nucleus. All coordinates are in mil- Materials and Methods limeters from lambda with lambda being placed 1.5 mm lower than bregma. The minus sign for ML All methods used in this study have been out- coordinates indicates placement on the right side lined in detail in previous submissions (see Clark et of the rabbit's head. al. 1992). Briefly, our procedures included the fol- In addition to the cold probe, a recording elec- lowing. trode constructed from a 00 stainless steel insect pin, insulated with a minimum of six baked coats of Epoxylite, with 20-40 lam exposed tip was chronically implanted in the contralateral IP using SUBJECTS coordinates AP +0.5, ML +5, DV -14.5. The subjects were 14 adult male New Zealand To obtain recordings from pontine cells, a white rabbits initially weighing 2.0 (+0.2) kg. Ani- 5-mm diameter hole was made in the right cranium mals were housed individually with free access to using coordinates AP +8.5, ML -2.5 (at the center food and water and were kept on a 12-hr light/dark of the hole). The hole was filled with sterile bone

L E A R N / N G & M E M O R Y 520 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

PONTINE NUCLEI AND EYE-BLINK CONDITIONING

wax and a stainless steel base stand was cemented Recording electrodes identical to those described in place over the hole using dental acrylic. The above were lowered using the Narishige microma- base stand would later be used for attaching a Na- nipulator. Stereotaxic atlas coordinates were used rishige microelectrode manipulator (MO-95) so to guide in sampling locations along the rostral/ that recording electrodes could be lowered into caudal axis of the pontine nuclei (from lambda AP varying regions of the pons during testing days. +8.5 + 2.0; ML -2.0 + 0.5). Finally, a socket for attachment of a mini- Each test session consisted of the following torque potentiometer and air puff hose was fixed procedures. An electrode was lowered into the to the anterior portion of the cranium. This socket, pons, trait activity was isolated, and then a block of along with the cold probe, base stand, and Fixed trials was begun. Unit activity was monitored for recording electrode were secured to the rabbit recordings that reflected stimulus-evoked or learn- skull using dental acrylic anchored by three stain- ing-related elevations in unit activity for at least one less steel screws. full block of trials. The cooling probe was then After all surgical procedures were completed, activated and a second block of trials was run at but before the animal was revived, a small loop of that recording location. The electrode was then 6-0 Ethilon monofilament surgical suture was se- moved to another location and the process was cured in the distal margin of the left nictitating begun again. Marking lesions were made at the end membrane of the rabbit. This loop was used later of testing along each electrode pathway by passing to secure the minitorque potentiometer for mea- 100 ~ of current through the recording electrode surement of the eye-blink response. for 5 sec. Lesions were made every 5 mm starting Two of the 14 subjects underwent surgery as at the lowest recording location, then moving dor- described above with the exception that they did sally through the brain tissue. In some instances not have a fixed recording electrode placed in the additional marking lesions were made at locations IP nucleus. These subjects were naive controls that where both recording and cooling blocks had been were tested for tone-evoked unit activity in the completed. The testing procedure was repeated, lateral pons before and during cooling of red one session each day, for as long as we were able nucleus. We were interested in the effects of cool- to cool the red nucleus or until three marking le- ing on tone-evoked activity for comparison with sion tracks had been made. In our experience, the trained animals. clear deciphering of recording locations became too difficult when more than three tracks were marked. The two control animals were not given eye- TRAINING blink conditioning. After surgery and 7 days of re- Rabbits were allowed to recover from surgery covery these animals were given 1 hr of habitua- for 7 days. tion to the test chamber followed the next day by On the first day after the recovery period each testing. Testing procedures were as described animal was habituated for 1 hr to a plexiglass rabbit above for recording from the lateral pontine nu- restrainer placed inside an Industrial Acoustics clei. Rabbits were placed in the test chamber and sound-attenuating chamber. Training sessions be- given tone alone training trials that followed all gan on the day after habituation. Each daily training normal parameters with the exception that they session consisted of 12 blocks of 9 trials for a total had no unconditioned stimulus delivery (no air of 108 trials. Each block of trials consisted of one puff). Recording locations in the pons were tone alone trial, one air puff alone trial, and seven sampled until tone-evoked responses were found. paired trials. The tone alone trial occurred at the When locations with evoked responses were beginning of each block and was followed by three found, data were collected for several blocks of paired trials, then the air alone trial, and finally, trials. Cooling was then initiated for one block of four more paired trials. The conditioned stimulus trials. Subsequent to cooling the electrode was (CS) was a 1-kHz, 85-dB (SPL), 352-msec tone. The moved to a new location as described above. unconditioned stimulus (US) was a 2.1-N/cm 2 (source pressure), 98-msec corneal air puff. DATA COLLECTION AND ANALYSIS After each animal reached criterion in acquisition training it was overtrained 1 day. The animal Each trial had a 252-msec baseline period of no was then ready for recording from pontine cells. stimuli, a 252-msec CS period, and a 252-msec US

L E A R N / N G & M E M O R Y 521 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

Cartford et aL

period. During the triM, data were sampled and to calculate z scores for the three 84-msec CS pe- extracted at 4-msec intervals. The behavioral-de- riod epochs and the three 84-msec US period ep- pendent measure was an extension of the nictitat- ochs for each individual trial. Any increase in unit ing membrane as measured by a minitorque poten- activity that resulted in a z score of 1.65 or greater tiometer. The neural unit activity-dependent mea- (a one-tailed significance level of 0.05) in CS ep- sure was amplified with an A-M Systems four- ochs 2 and 3 was scored as a learned unit response channel amplifier (model 1700, 10k gain, or model. Any individual animal that showed these 300-5000 Hz bandpass filtering). These neural learned unit responses on at least 80% of the trials spikes were passed to an amplitude discriminator for a given day was considered to show activity that digitized the activity. This information was that modeled the learned behavior. then collected by the computer to construct a peri- For each testing session, baseline means and stimulus time histogram. All stimulus presenta- standard errors were calculated separately for be- tions, sessions, data collection, and data analyses havioral and unit responses for each block of trials. were controlled by IBM-PC/XT clones that were For statistical analysis, z scores were calculated by equipped with an interface (Lavond and Steinmetz blocks to take into account differences in normal 1989) and programs written in 8088 machine code versus cooling blocks and differences in recording and Forth. locations. Behavioral-conditioned responses (CR) were counted if nictitating membrane closure exceeded 0.5 mm during the CS period on paired trials, and HISTOLOGY during the CS and US periods on tone alone trials. The learning criterion was the first time that eight Each rabbit was sacrificed with an overdose of CRs occurred on nine consecutive trials. We de- intravenous sodium pentobarbital and perfused in- fined bad trials as 0.7 mm of eyelid movement in tra-aortically with saline, followed by 10% formalin. the 160-msec before CR onset or 0.5 mm of eyelid Marking lesions were made of the chronically im- closure in the first 25 msec of the CS period. Con- planted IP recording electrode locations by passing ditioned responses, trials to criterion, amplitude, 100 pA through the electrode for 10 sec. The brain latency, and area under the curve for both CS and was removed and postfixed in 10% sucrose forma- US periods were calculated and printed at the end lin solution. Each brain was blocked and embed- of each training and testing session. ded in a gelatin-albumin matrix, frozen and cut on We used a minimum criterion of 2:1 signal-to- a microtome at a thickness of 80 jxm, and mounted noise ratio based on visualization of unit data on onto chrome-alum subbed slides. The tissues were the oscilloscope. Most recordings were 3"1 or bet- stained with Prussian Blue for marking lesions and ter. We estimate the number of units we typically cresyl violet for cells, and were analyzed using a discriminated (based on height) to be 3 to 5 units. light microscope to reconstruct electrode and cold Unit activity from IP nucleus and from the probe placements. pons were analyzed with peristimulus time histograms and statistical analysis with z scores to quan- tify activity associated with learning (Thompson et Results al. 1976). Each trial had three 252-msec periods: (1) a preconditioned stimulus (baseline) period, BEHAVIORAL RESULTS (2) a CS period, and (3) a US period. These three stimulus periods were each subdivided into three Eleven subjects trained to a 1-kHz tone 84-msec epochs (epochs 1, 2, and 3). Significant reached criterion within 5 days of training in activity in CS epoch 1 was defined as tone-evoked 256 _+ 102 (mean _+ S.D.) trials to criterion. Clark et unit activity. Significant activity in CS epochs 2 and al. (1992) found comparable trims to criterion data 3 was defined as learning-related unit activity. for 12 animals with implanted cold probes with the Significant unit activity was determined by z mean trials to criterion equal to 271 +_ 43.0. In the score analysis. For each daily training session a present study, one additional rabbit required 7 grand mean and standard error for behavioral re- days of training to reach criterion using a 5-kHz sponses were calculated across all trials using the tone as the CS. The 5-kHz tone was used to achieve 84-msec epoch immediately before the onset of the greater filtering of recorded neural units from pos- CS. The grand mean and standard error were used sible tone artifacts in the recording. Those data are

L E A R N I N G & M E M O R Y 522 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

PONTINE NUCLEI AND EYE-BLINK CONDITIONING

not included in our report on acquisition, but are RECORDING RESULTS FROM NORMAL ANIMALS included in all other aspects of the study.

Figure 1 illustrates behavioral results of our ex- LATERAL PONS periment. Inactivating red nucleus abolished the learned nictitating membrane behavior. This is For the 12 normal subjects, 16 recording loca- consistent with prior inactivation of red nucleus tions within the pons had complete recording and studies (Clark and Lavond 1993, 1996; Krupa et al. cooling data. z Scores based on the block of trials 1993). Conditioned response amplitude data for all preceding cooling and during cooling were ana- subjects was pooled for paired trials during each lyzed for significant levels of unit activity recorded block preceding cooling and for each block during in the pons during each of the three epochs within cooling. Pooled conditioned response amplitude the CS period. For pooled data from all recording during normal training was 5.1 _+ 1.2 mm and dur- locations a two-factor analysis of variance ing cooling was 0.2 _+ 0.1 mm, t(15) = 4.063, (ANOVA) (normal versus cooling) with six re- P < 0.001. Similar to previous studies (Clark and peated measures (three CS epochs and three US Lavond 1993, 1996), cooling red nucleus also at- epochs) showed a main effect of cooling tenuated but did not abolish unconditioned re- (F1,32 = 7.33, P < 0.01). sponse (UR) amplitude. Unconditioned response z Scores for baseline activity in the pons over data were pooled for all subjects using air puff blocks preceding and those during cooling were alone trials from each block preceding cooling and analyzed using a related measures t-test and from each block during cooling. Pooled UR ampli- showed that cooling did not significantly change tude during normal training was 5.9 -+ 1.5 mm and baseline neural activity, t(15) = -0.8885, P > 0.05. during cooling was 2.8 _+ 1.0 ram, t(15) = 2.4717, Of the 16 recording locations 13 showed an in- P < 0.05. crease in unit activity during the baseline period with cooling and 3 showed a decrease. Based on comparison of the group mean z score precooling and during cooling there was an overall increase of baseline unit activity by 27% during cooling. Figure 2 is a map of the 16 pontine recording locations, z Scores for each location are listed on Table 1 for precooling and cooling blocks of CS epoch 1 and CS epoch 3 unit activity. In addition, we have represented changes in unit activity that occurred attributable to cooling by listing in the boxed columns of Table 1 the arithmetic difference of the precooling and cooling z scores. We fotmd three types of unit activity response patterns within our pontine recording sites: (1) stimulus-only unit activity, (2) learning-related unit activity, and (3) both stimulus and learning-related unit activity. Within the 16 recording locations we found nine sites that reflected stimulus-evoked unit responses, that is, responses with z scores above the critical value in CS epoch 1. We found 10 sites with learning-related unit activity or activity with z scores above the critical value in CS epoch 3, and we found six sites with both types of activity. Table 2 provides a list of those locations where stimulus-evoked responses were recorded and indicates those locations where these responses Figure 1: Results of pooled behavioral measure of nic- were abolished (z scores fell below the critical titating membrane extension for conditioned responses level for unit activity during cooling). Table 3 pro- (CRs) and unconditioned responses (URs) during normal vides a list of those locations where learning-re- and cooling sessions. lated unit activity was recorded. Locations are in-

L E A R N I N G & M E M O R Y 523 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

Cartford et aL

Figure 2: Summary of recording locations. Each location is noted (0) and identified by a corresponding number represented to the side of the location. Location numbers correspond to those listed in Tables 1,2, and 3 and in Figs. 3 and 4. dicated where these models were abolished (z sures t-test. Cooling did not cause a statistically sig- scores fell below the critical level for unit activity nificant reduction in tone-evoked unit activity z during cooling). scores in these locations, t(4) = 1.15, P > 0.05. Re- The data for individual recording sites indicate cording location 12 showed an increase in stimu- that cooling did not cause stimulus-evoked or lus-evoked unit activity during cooling. We believe learning-related unit activity to fall below the sig- the cooling may have interrupted an inhibitory ef- nificant level in all locations. Figure 3 shows that fect in this location. there are four recording locations where cooling Attenuation of unit activity was observed for red nucleus attenuated but did not abolish learn- both stimulus-evoked as well as learning-related ing-related unit activity (locations 1, 2, 8, and 16). unit activity in the pontine nuclei in locations z Scores for unit data from locations where models where cooling did not abolish unit activity. In the were attenuated but not abolished by red nucleus recording locations where stimulus-evoked unit ac- cooling were analyzed using a related measures t- tivity was not abolished by cooling (Table 2 z test. Cooling did not cause a statistically significant scores not marked with asterisk), there was an reduction in learning-related unit activity z scores, overall attenuation of activity of 24% in comparison t(4) - 2.91, P > 0.05. z Score changes for recording to group mean z scores for CS epoch 1 before and location 2 before and during cooling suggest that during cooling. However, no z score for an indi- cooling may have unmasked a normally inhibitory vidual recording location falls below the 1.65 cri- effect at this location. teflon for significant activity as compared to base- Figure 4 shows that there are four sites where line. Likewise, when mean z scores for CS epoch 3 stimulus-evoked unit activity was not abolished by are compared before and during cooling for loca- red nucleus cooling (locations 1, 3, 10, and 16). z tions where cooling did not abolish unit activity Scores for unit data from locations where stimulus- (Table 3 z scores not marked with an asterisk), a evoked activity was not abolished by red nucleus 44% reduction in overall activity is noted. How- cooling were also analyzed using a related mea- ever, once again, no individual z scores fell below

L E A R N I N G & M E M O R Y 524 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

PONTINE NUCLEI AND EYE-BLINK CONDITIONING

Table 1 : Pontine recording locations stimulus-evoked (CS1) and learning-related (CS3) unit activity and the effects of red nucleus cooling on that unit activity (as reflected in the arithmetic difference between normal and cooling z-scores from CS epoch 1 and CS epoch 3 blocks of data) Recording Normal Cooling Norm-Cool Normal Cooling Norm-Cool location CS epoch 1 CS epoch 1 CS epoch 1 CS epoch 3 CS epoch 3 CS epoch 3

1. 3.610 2.820 .790 4.990 3.270 1.720 2. 2.120 4.600 -2.480 1.460 4.180 -2.720 3. 1.770 .520 1.250 2.590 1.700 .890 4. 4.290 1.080 3.210 5.140 1.41 0 3.730 5. 3.490 .800 2.690 5.870 .530 5.340 6. .740 .150 .590 3.340 1.61 0 1.730 7. .1 70 .320 -.015 1.360 .640 .720 8. 3.340 2.260 1.080 1.450 1.930 -0.480 9. 1.020 .010 1.01 0 4.430 .290 4.140 10. .800 -.160 .960 6.380 2.570 3.810 11. .480 .360 .120 1.550 .400 1.1 50 12. .120 -1.240 1.360 2.230 -1.150 3.380 13. 2.830 .570 2.260 8.690 .850 7.840 14. 1.1 60 -1.020 2.180 0.000 -.830 -0.830 15. 5.450 -.430 5.880 .370 -.050 0.420 16. 6.440 2.290 4.150 3.510 2.140 1.370

the 1.65 criterion for significant levels of activity when compared to baseline.

INTERPOSITUS

Table 4 includes z scores for CS epoch 3 unit activity recorded in IP nucleus (there were no

Table 2- Effects of red nucleus cooling on stimulus-evoked unit activity in pontine recording locations (reflected in z scores from CS epoch 1) Stimulus-evoked unit activity location a normal cool

1. 3.610 2.820 2. 2.120 4.600 3.* 1.770 .520 Figure 3: z Scores calculated for tone-evoked unit ac- 4.* 4.290 1.080 tivity for individual recording locations represent unit 5.* 3.490 .800 activity for one block (nine trials) before cooling and one 8. 3.340 2.260 block (nine trials) of cooling. Four of the nine recording 13.* 2.830 .570 locations continue to show significant levels of unit ac- 15.* 5.450 -.430 tivity during cooling of the red nucleus. These data are 16. 6.440 2.290 rank-ordered from highest z score to lowest (before cooling). The numbers below the bars identify the locations aft) Locations where cooling had a significant effect on from which these recordings were made as plotted on unit activity. Fig. 2.

L E A R N / N G & M E M O R Y 525 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

Cartford et al.

Table 3: Effects of red nucleus cooling on learning-related unit activity in pontine recording locations (reflected in z scores from CS epoch 3)

Learning-related unit activity

location a normal cool

1. 4.990 3.27O 3. 2.590 1.700 4.* 5.140 1.410 5.* 5.870 .530 6.* 3.340 1.610 9.* 4.430 .290 O. 6.380 2.570 2.* 2.230 -1.150 3.* 8.690 .850 6. 3.510 2.140

Figure 4: z Scores calculated for learning-related unit a(,) Locations where cooling had a significant effect on activity for individual recording locations represent unit unit activity. activity for one block (nine trials) before cooling and one block (nine trials) of cooling. Four of the 10 recording locations continue to show significant levels of unit ac- the unit activity increased during cooling of red tivity during cooling of the red nucleus. These data are nucleus. rank ordered from highest z score to the lowest (before Baseline unit activity for the IP nucleus record- cooling). The numbers below the bars identify the loca- ing locations showed an overall increase in activity tions from which these recordings were made as plotted by 14% during cooling. However, three of the re- on Fig. 2. cording locations showed a modest decrease in activity and one showed a large increase. The difference in z scores before and during cooling was not statistically significant, t(4) = 0.40618, P > 0.05. cases where significant increases in unit activity Two different cold probe placements near the were observed in CS epoch 2). Four animals had red nucleus were used in this study to control for accurate implanted IP electrode placements. The interference that might be caused by cooling when data from these animals were analyzed for effects recordings were taken in the nearby rostral lateral of red nucleus cooling on learning-related unit ac- pons. Using t-tests for independent means we veri- tivity. A related measures t-test showed no signifi- fied that no differences in unit data activity, either cant reduction in learning-related unit activity z tone-evoked or learning-related activity, could be scores during cooling of red nucleus, accounted for by the different location of the cold t(4) = 0.9261, P > 0.05. Interpositus data from these subjects closely matches that reported by Clark and Lavond (1993) where cooling the red Table 4: Effects of red nucleus cooling on nucleus had no statistically significant effect on learning-related unit activity in interpositus learning-related unit activity in IP during acquisi- nucleus (reflected in z scores from CS epoch 3) tion and retention of the eye-blink response. Like recordings from pontine areas where I nterpositus learning-related cooling did not abolish unit activity, the z scores unit activity for IP indicate an attenuation of activity attributable to red nucleus cooling. This attenuation rep- normal cool resents a 10% drop in unit activity based on group 7.49 5.75 mean z scores for CS epoch 3. With one exception, 2.0 3.15 this attenuation of activity did not cause unit activ- 2.71 4.65 ity to decrease below the critical level of 1.65 as 4.03 1.01 compared to baseline and, in fact, in two locations

L E A R N / N G & M E M O R Y 526 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

PONTINE NUCLEI AND EYE-BLINK CONDITIONING probes. For tone-evoked unit activity z scores, with successful cooling in each subject. Subject t(14) = 0.2773, P > 0.05, and for learning-related 96-279 showed a clear short latency stimulus- unit activity, t(14) = 0.1817, P> 0.05. evoked response that was significantly reduced by Figure 5 is a peristimulus time histogram of red nucleus cooling. Subject 96-280 showed a long multiple unit recordings from IP nucleus and pon- onset latency and long duration response to the tine nuclei during a block of trials preceding cool- tone stimulus before cooling. During red nucleus ing and a block of trials during cooling. Learning- cooling this response changed to a short onset la- related model activity can be seen in both record- tency. ings before and during cooling. These data The initial recording from rabbit 96-280 looks replicate earlier findings of Chapman et al. (1990) very much like learning-related unit activity re- and Clark and Lavond (1993) for IP recording with corded in well-trained animals (see Fig. 5), yet ex- red nucleus inactivation. The data illustrate atten- amination of our data from normal subjects shows tuation but not loss of learning-related unit models no such shift in unit activity onset latency during in a dorsolateral pontine recording location before cooling. It is possible that changes in unit activity and during red nucleus cooling. associated with learning had already masked or The variable results that we report both in caused changes in unit activity associated with the pontine as well as IP recording locations may be an conditioning stimulus. The fact that we have seen indication of a widespread inhibitory influence ex- evidence in normal subjects for inhibitory feed- erted on the eye-blink system perhaps controlled back processes being unmasked by red nucleus by or routed through the red nucleus. This could cooling (recording location 15 in particular) leads explain, for instance, locations where unit activity us to believe that red nucleus cooling is somehow increased during cooling of red nucleus. However, releasing or interfering with an inhibitory process with our small number of observations it is difficult that normally controls the onset of the stimulus- to find significant effects. What the data indicate evoked response. consistently is a variability of response patterns during cooling, including increases in unit activity, decreases in unit activity, as well as a few locations Discussion where no changes occur whatsoever. Our results show that temporary inactivation of the red nucleus has a significant effect on learning-related and stimulus-evoked unit activity found RECORDING RESULTS FROM CONTROL ANIMALS in the pontine nuclei. The results also show that Figure 6 includes histology (A), z scores 03), sites within central regions of the rostral/caudal and peristimulus time histograms (C) for each na- extent of the dorsolateral and lateral pons maintain ive subject tested for stimulus-evoked responses in significant, although attenuated learning-related the lateral pons before and during cooling of red unit activity and stimulus-evoked activity during nucleus. We obtained one excellent recording red nucleus inactivation.

Figure 5: Peristimulus time histograms taken from one animal showing behavioral nictitating membrane responses (nmr) and unit activity recorded from lateral pons (LPN) and interpositus nucleus (IP). These data represent the average activity over one block (nine trials) before cooling of red nucleus (Normal) and one block (nine trials) during cooling of red nucleus (Cooling). The behavioral response is abolished with cooling. Tone- evoked unit activity is abolished and learning-related unit activity is reduced in both interpositus and lateral pons recordings.

L E A R N / N G & M E M 0 R Y 527 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

Cartford et aL

Figure 6: Data and histology for control subjects 96-279 and 96-280. (A) Peristimulus time histograms for neural recording during one block of normal and one block of cooling trials. (B) z Scores for CS epoch 1 and CS epoch 3 for the block of normal and the block of cooling trials represented in the peristimulus time histograms. (C) Recording locations where multiple unit data was obtained.

These data support earlier findings that the The data from our study, considered with that pontine nuclei are an integral part of the condi- of Clark et al. (this issue), confirm the integral re- tioned stimulus pathway in rabbit eye-blink classi- lationship of the pontine nuclei within the eye- cal conditioning. Extensive mapping of recording blink circuitry and support the hypothesis that the information taken from rabbits trained using a learning-related unit activity we and others have tone-conditioned stimulus has verified that models seen in the pons is dependent on the function of of learning-related unit activity as well as tone- the IP. We now also see an indication that feed- evoked unit activity occur extensively throughout back to the pontine nuclei from the red nucleus the pontine midbrain (McCormick et al. 1983). may play a functional role in the eye-blink circuit, Steinmetz et al. (1985 and 1986) have shown that although we have found no evidence in the litera- direct stimulation of either the middle cerebellar ture for a direct anatomical pathway from red peduncle or the dorsolateral, lateral, and medial nucleus to the pons. pontine nuclei may serve as the conditioned stimu- Involvement of the red nucleus in the eye- lus in this paradigm. Lesions of the dorsolateral blink circuitry has been established from both pons, the caudal lateral pons, and the lateral lem- anatomical and physiological data. This nucleus niscus abolish conditioned responses to a tone- has been shown to be an essential part of the conditioned stimulus (Steinmetz et al. 1987). Ana- output of the cerebellum for the conditioned tomical and physiological data substantiate a direct eye-blink response (Tsukahara et al. 1981; Des- and reciprocal connection between the IP nucleus mond and Moore 1982; Tarnecki 1988; Chap- and the dorsolateral and lateral pontine nuclei man et al. 1990; Clark and Lavond 1993; D.A. (Eller and Chan-Palay 1976; Watt and Mihailoff Haley, D.G. Lavond, and R.F. Thompson, unpubl.) 1983; Gerrits and Voogd 1987; Steinmetz et al. and is now known to have reciprocal connec- 1987; Steinmetz and Segelaub 1992; Gould et al. tions to the IP (Courville and Brodal 1966; J. Lock- 1993; J.K. Thompson, w.J. Spangler, and R.F. ard and D.G. Lavond, unpubl.). However, there Thompson, unpubl.). is no evidence of direct anatomical connections

L E A R N / N G & M E M O R Y 528 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press PONTINE NUCLEI AND EYE-BLINK CONDITIONING between red nucleus and dorsolateral or lateral Ocular formation cells indicated that the majority pons. of projections come from the lateral lemniscal au- Our recording data from the IP do not provide ditory system. If our lateral placement of the cold conclusive evidence for a theory that changes in probe effectively inactivated this system as well as unit activity in the pons could be caused by anti- the red nucleus, we might expect to see a loss or dromic effects from red nucleus cooling passing reduction of tone-evoked unit activity within the back to IP and from there to the pons. Attenuation projection area of these cells. of unit activity in the IP during red nucleus cooling We do not believe cooling had a generalized is not significant either in our findings or those of effect on the lemniscal system in our subjects. If Clark and Lavond (1993) and is far less than attenu- there were such a generalized effect from cooling ation that occurs in the pons. It is possible that the it would be difficult to explain the retention of effect in the pons is somehow amplified over that tone-evoked information in the middle regions of seen in the IP. Given the evidence for both ana- the dorsolateral and lateral pons. Descending as tomical and physiological connections between well as ascending auditory fibers travel through the dorsolateral and lateral pons and the IP, further lemniscal projection and both have direct connec- studies are warranted to explore possible interac- tions within the lateral pons (Matano et al. 1966; tions occurring among these structures. Aitkin and Boyd 1978; Kandler and Herbert 1991). In 1993 Clark and Lavond found that cooling Also, if the effects were caused by the cold probe red nucleus during acquisition prevented expres- placement we would expect that there be no ef- sion of the learned response that was formed in the fect from cooling in those subjects with more an- IP during training. Very recently our laboratory has terior and medial cold probe placements. We saw shown that the trigeminal complex, a major locus no significant differences in unit activity in the for unconditioned stimulus information traveling pons that corresponded to different cold probe to the cerebellum, receives learning-related infor- placements. Finally, there were no significant ef- mation from IP through red nucleus excitatory pro- fects on baseline neural unit activity between nor- jections (Clark and Lavond 1996). This study also mal and cooling blocks of recordings for any cold showed evidence of inhibitory connections be- probe placement. Thus, it does not seem that the tween red nucleus and pars oralis of the trigeminal location of the cold probe was having any gener- complex. Our current data show attenuation or alized effect on the regions from which we were abolition of neuronal unit activity related to the taking recordings. conditioned stimulus and to learned behavior in all The role of the pontine nuclei within the eye- parts of the lateral and dorsolateral pontine nuclei blink circuit remains to be elucidated. The impor- as a consequence of red nucleus inactivation. tance of the pontine projection for carrying CS in- Taken together, these data suggest that the red formation to the cerebellum has been well estab- nucleus is serving as more than output for the con- lished (Steinmetz and Sengelaub 1992; Steinmetz ditioned response. It also seems to function as an et al. 1986, 1987, 1989). More recently, Tracy important feedback source of information to mid- (1995) found that stimulation thresholds in the dor- brain structures that are active in sending stimulus solateral pons were lowered during training with a information to the cerebellum. tone-conditioned stimulus, whereas IP thresholds The effects of red nucleus cooling on stimulus- were not. In addition, J. Steinmetz (pers. comm.) evoked unit responses in the pons surprised us and has found evidence that tuning curves of pontine led us to question whether at least some of our cells become sharper as a consequence of training. results may be explained by the location of the This suggests that modulation in cells sensitive to cold probe. For instance, we wondered whether auditory stimuli may be occurring at the level of there could be a generalized effect from cooling in the pons, although it is also possible that tuning of the lateral reticular formation adjacent to the red cells responsive to auditory stimuli occurs afferent nucleus. Irvine and Jackson (1983) found stimulus- to the pons and shows in the response properties sensitive cells in the central tegmental field with of pontine neurons (Woody et al. 1992, 1994; complex and broad response properties that Weinberger 1993). The tuning of pontine neurons would suggest both a convergence of sensory mo- may also be a result of feedback from efferent con- ralities as well as multiple origins (i.e., inputs nections such as IP, or red nucleus by IP. We plan through the auditory pathways as well as cerebel- to explore the pontine area further for evidence of lar components). Anatomical analysis of these re- adaptations to the conditioned stimulus that might

L E A R N I N G I M E M 0 R Y Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press Cartford et al. suggest a functional role for the pons in a mecha- for his generous and invaluable assistance throughout this nism such as selective attention. project 9 The results of this study as well as those of Clark et al. (this issue) provide support for our References hypothesis regarding memory formation within Aitkin, L.M. and J. Boyd. 1978. Acoustic input to the lateral pontine nuclei. Hearing Res. 1" 67-77 9 the putative eye-blink circuitry. Our cooperative memory hypothesis states that the fundamental as- Chapman, P.F., J.E. Steinmetz, L.L. Sears, and R.F. Thompson. sociation between tone and air puff and the 1990. Effects of lidocaine injection in the interpositus nucleus learned behavior occurs in the IP nucleus, that dif- and red nucleus on conditioned behavioral and neuronal ferent memories about different aspects of the con- responses. Brain Res. 537" 140-156. ditioning are represented in a distributed manner Chen, L., S. Bao, J.M. Lockard, J.J. Kim, and R.F. Thompson. in other brain structures, and when combined, re- 1996. Impaired classical eyeblink conditioning in sult in a coordinated memorial representation. cerebellar-lesioned and purkinje cell degeneration (pcd) The exact nature of the role of each of the mutant mice. J. Neurosci. 16: 2829-2838. structures involved in the eye-blink circuit remains to be discovered. For instance, the roles of the Clark, R.E. and D.G. Lavond. 1993. Reversible lesions of the red nucleus during acquisition and retention of a classically cerebellar cortex and deep nuclei have been of conditioned behavior in rabbits. Behav. Neurosci. particular interest in this learning paradigm. In con- 107: 264-270. trast to our results (Clark et al. 1997a) that have shown that cooling IP nucleus abolished learning- 91996. Neural unit activity in the trigeminal complex related unit activity in cerebellar cortex, and re- with interpositus or red nucleus inactivation during classical sults from the Thompson laboratory (Chen 1996) eyeblink conditioning 9Behav. Neurosci. 110" 13-21. showing eye-blink learning in mutant mice lacking Clark, G.A., D.A. McCormick, D.G. Lavond, and R.F. Purkinje cells, Katz and Steinmetz (1997) report Thompson 91984. Effects of lesions of cerebellar nuclei on that learning-related unit activity remains in cer- conditioned behavioral and hippocampal neuronal responses. ebellar cortex after chemical lesions of the IP. Brain Res. 291 : 125-136. The results of our present study suggest that Clark, R.E., A.A. Zhang, and D.G. Lavond. 1992. Reversible the pontine nuclei are subject to modifying influ- lesions of the cerebellar interpositus nucleus during ences from the output pathway of the learned be- acquisition and retention of a classically conditioned havior in addition to the influence of the IP nucleus behavior. Behav. Neurosci. 106: 879-888 9 shown by Clark et al. (this issue). In combination with studies suggesting that there may be plasticity 91997a. The importance of cerebellar cortex and associated with auditory stimuli in the pontine nu- facial nucleus in acquisition and retention of eyeblink/NM conditioning: Evidence for critical unilateral regulation of the clei, our results support the notion of a unique, conditioned response. Neurobiol. Learn 9Mem. 67:96-111. integral role for the pontine nuclei within the circuitry of the learned eye-blink. Clark, R.E., E.B. Gohl, and D.G. Lavond. 1997b. The In conclusion, this study has shown that, like learning-related activity that develops in the pontine nuclei the IP, the red nucleus exerts an influence on during classically eye-blink conditioning is dependent on the interpositus nucleus 9Learn. & Mem. (this issue) 9 learning-related unit activity within the lateral and dorsolateral pontine nuclei. Unlike the IP, the red Courville, J. and A. Brodal. 1966. Rubrocerebellar nucleus also exerts an influence on stimulus- connections in the cat: An experimental study with silver evoked unit activity within the lateral and dorso- impregnation methods 9J. Comp. Neurol. 126" 471-485. lateral pontine nuclei. Given these results it seems likely that the pontine nuclei participate in the neu- Desmond, J.E. and J.W. Moore. 1982. A brain stem region essential for classically conditioned but not unconditioned ral circuitry of the learned nictitating membrane nictitating membrane response 9Physiol. Behav. response in a role that is more complex than the 28:1029-1033. simple relay of conditioned stimulus information to the cerebellum. What exactly this role might be is Eller, T. and V. Chan-Palay. 1976. Afferents to the cerebellar a subject for future investigation. lateral nucleus. Evidence from retrograde transport of horseradish peroxidase after pressure injections through micropipettes. J. Comp. Neurol. 166: 285-302. Acknowledgments This research was supported by National Institute of Gerrits, N.M. and J. Voogd. 1987. The projection of the Mental Health 1 R01 MH51197. Special thanks to Bob Clark nucleus reticularis tegmenti pontis and adjacent regions of

L E A R N / N G & M E M O R Y 530 Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

PONTINE NUCLEI AND EYE-BLINK CONDITIONING

the pontine nuclei to the central cerebellar nuclei in the cat. projections from the pontine nuclei to the cerebellar J. Comp. Neurol. 258: 52-69. interpositus nucleus. Behav. Neural Biol. 57:103-115.

Gould, T.J., L.L. Sears, and J.E. Steinmetz. 1993. Possible CS Steinmetz, J.E., D.G. Lavond, and R.F. Thompson. 1985. and US pathways for rabbit classical eyelid conditioning: Classical conditioning of the rabbit eyelid response with Electrophysiological evidence for projections from the mossy fiber stimulation as the conditioned stimulus. Bull. pontine nuclei and inferior olive to cerebellar cortex and Psychonomic Soc. 23: 245-248. nuclei. Behav. Neural Biol. 60" 172-185. Steinmetz, J.E., D.J. Rosen, P.F. Chapman, D.G. Lavond, and R.F. Thompson. 1986. Classical conditioning of the rabbit lrvine, D.R.F. and G.D. Jackson. 1983. Auditory input to eyelid response with a mossy fiber stimulation CS. I. Pontine neurons in mesencephalic and rostral pontine reticular nuclei and middle cerebellar peduncle stimulation. Behav. formation: An electrophysiological and horseradish Neurosci. 100: 871-880. peroxidase study in the cat. J. Neurophysiol. 49:1319-1333. Steinmetz, J.E., C.G. Logan, D.J. Rosen, J.K. Thompson, D.G. Kandler, K. and H. Herbert. 1991. Auditory projections from Lavond, and R.F. Thompson. 1987. Initial localization of the the cochlear nucleus to pontine and mesencephalic reticular acoustic conditioned stimulus projection system to the nuclei in the rat. Brain Res. 562: 230-242. cerebellum essential for classical eyelid conditioning. Proc. Natl. Acad. Sci. 84" 3531-3535. Katz, D.B. and J.E. Steinmetz. 1997. Single-unit evidence for eye-blink conditioning in cerebellar cortex is altered, but not Steinmetz, J.E., D.G. Lavond, and R.F. Thompson. 1989. eliminated, by interpositions nucleus lesions. Learn. & Mem. Classical conditioning in rabbits using pontine nucleus (in press). stimulation as a conditioned stimulus and inferior olive stimulation as an unconditioned stimulus. Synapse Krupa, D.J., J.K. Thompson, and R.F. Thompson. 1993. 3" 225-233. Localization of a memory trace in the mammalian brain. Science 260: 989-991. Tarnecki, R. 1988. Functional connections between neurons of interpositus nucleus of the cerebellum and the red Krupa, D.J., J. Weng, and R.F. Thompson. 1996. Inactivation nucleus. Behav. Brain Res. 28" 117-125. of brainstem motor nuclei blocks expression but not acquisition of the rabbit's classically conditioned eyeblink Thompson, R.F., T.W. Berger, D.F. Cegavski, M.M. response. Behav. Neurosci. 110:219-227. Pattersonn, R.A. Roemer, T.J. Teyler, and R.A. Young. 1976. The search for the engram. Am. Psychol. 31" 209-227. Lavond, D.G. and J.E. Steinmetz. 1989. An inexpensive Tracy, J. 1995. "Brain and behavior correlates in classical interface for IBM-PC/XT and compatibles. Behav. Res. conditioning of the rabbit eyeblink response." Ph.D. Methods Instrum. Computers 21: 435-440. dissertation. University of Southern California, Los Angeles, CA. Lavond, D.G., D.A. McCormick, G.A. Clark, D.T. Holmes, and R.F. Thompson. 1981. Effects of ipsilateral rostral pontine Tsukahara, N., Y. Oda, and T. Notsu. 1981. Classical reticular lesions on retention of classically conditioned conditioning mediated by the red nucleus in the cat. J. nictitating membrane and eyelid responses. Physiol. Psychol. Neurosci. 1" 72-77. 9" 335-339. Watt, C.B. and G.A. Mihailoff. 1983. The cerebellopontine Lavond, D.G., J.J. Kim, and R.F. Thompson. 1993. system in the rat. I. Autoradiographic studies. J. Comp. Mammalian brain substrates of aversive classical Neurol. 215:312-330. conditioning. Ann. Rev. Psychol. 44:317-342. Weinberger, N.M. 1993. Learning-induced changes of Matano, S., T. Ariviki, T. Kozeki, and T. Ban. 1966. Some auditory receptive fields. Curr. Opin. Neurobiol. 3: 570-577. observations on the efferent fibers of the inferior colliculus in the rabbit. Med. J. Osaka Univ. 16." 259-269. Woody, C.D., X.F. Wang, E. Gruen, and J. Landeira-Fernandez. 1992. Unit activity to click CS changes McBride, R.L. and W.R. Klemm. 1968. Stereotaxis atlas of in dorsal cochlear nucleus after conditioning. NeuroReport rabbit brain, based on the rapid method of photography of 3" 385-388. frozen unstained sections. Comm. Behav. Biol. Part A Woody, C.D., X.F. Wang, and E. Gruen. 1994. Response to 2: 179-215. acoustic stimuli increases in the ventral cochlear nucleus after stimulus pairing. NeuroReport 5:513-515. McCormick, D.A., D.G. Lavond, and R.F. Thompson. 1983. Neuronal responses of the rabbit brainstem during Zhang, A.A., H. Ni, and R.M. Harper. 1986. A miniaturized performance of the classically conditioned nictitating cryoprobe for functional neuronal blockade in freely moving membrane (NM)/eyelid response. Brain Res. 271: 73-88. animals. J. Neurosci. 16" 79-87.

Steinmetz, J.E. and D.R. Sengelaub. 1992. Possible conditioned stimulus pathway for classical eyelid Received January 28, 1997; accepted in revised form March conditioning in rabbits. I. Anatomical evidence for direct 25, 1997.

L E A R N I N G I M E M 0 R Y Downloaded from learnmem.cshlp.org on September 27, 2021 - Published by Cold Spring Harbor Laboratory Press

The effects of reversible inactivation of the red nucleus on learning-related and auditory-evoked unit activity in the pontine nuclei of classically conditioned rabbits.

M C Cartford, E B Gohl, M Singson, et al.

Learn. Mem. 1997, 3: Access the most recent version at doi:10.1101/lm.3.6.519

References This article cites 36 articles, 4 of which can be accessed free at: http://learnmem.cshlp.org/content/3/6/519.full.html#ref-list-1

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the Service top right corner of the article or click here.