NeumbiologyofAging, Vol. 13, pp. 319-323, 1992
0197-4580/92 $5.00 + .00 Copyrightt) 1992PergamonPressLtd.
Printedin the USA.All rightsreserved.
Delayed Acquisition of Eyeblink Conditioning in Aged F1 Hybrid (Fischer-344 × Brown Norway) Rats C R A I G W E I S S I A N D R I C H A R D F. T H O M P S O N
Neurosciences Program, University of Southern California, Los Angeles', CA 90089-2520
Received 28 May 1991; Accepted 18 September 1991 WEISS, C. AND R. F. THOMPSON. Delayedacquisition ~[eyeblink conditioning in aged FI hybrid(Fischer-344 X Brown Norway) rats. NEUROBIOL AGING 13(2) 319-323, 1992.--We previously reported that the freely moving male Fischer344 rat provides a useful model to demonstrate the progressive impairment of eyeblink conditioning associated with aging. However, because the youngest F-344 rats only performed at 60% of maximum, we ran the same experiment with hybrid rats and discovered most (i.e., those age 9-24 months) learned rapidly and exhibited conditioned responses on greater than 80% of trials by the end of two training sessions. In contrast, the aged rats (36 months) exhibited significantly fewer CRs on all four training days. However, unlike all ages of F-344 rats (3-30 months) which were run in our last study, these aged hybrid rats exhibited considerable improvement with extra training. These data indicate clear differences in the rate of learning between the two strains and suggest that even young F-344 rats may have deficits in the neural circuits which mediate eyeblink conditioning. Other anecdotal findings on differences between the two strains are noted. Aging Stress
Classicalconditioning
Nictitating membrane
Hybrid
CLASSICAL conditioning of the eyeblink reflex is an excellent paradigm to analyze learning and memory in various species. Both humans and animals show progressive age related impairments on this task (7,8,9) which become apparent as early as middle age (15,16,17). We reported that the freely moving male Fischer-344 rat showed a progressive age related impairment of eyeblink conditioning ( 14,15). However, the youngest F-344 rats exhibited CRs on no more than 60% of trials. To insure that the young F-344s were in fact poor learners and that our paradigm and apparatus would allow maximal conditioning, we ran the same experiment with F1 hybrid (F-344 X Brown Norway) rats. The F-344 X BN F1 hybrid was selected as an alternative model to the F-344 based on a study of several F1 hybrids (2). The F-344 X BN cross produces progeny with the fewest detrimental pathologies and at a later age of onset. This study was undertaken to determine if the same "hybrid vigor" would appear in an associative learning paradigm (i.e., classical conditioning of the eyeblink reflex). Preliminary data were reported in abstract form (13).
Fischer-344
BrownNorway
EMG
Three sets of cohorts were run: the first consisted of a group of 9-24 month olds housed 3 per cage ( 1 of each age); the second, consisted of 36 month olds housed 2 per cage; the third, consisted of 9-month and 36-month-olds housed 2 per cage ( 1 of each age). All rats were weighed at the time of arrival, kept on a 12 h/ 12h light/dark cycle, and had food and water available ad lib. The rats were run in the light portion of the cycle. Methods were reported previously (15). Briefly, a headstage containing wires to record the EMG activity from obicularis oculi and to deliver a periorbital shock was implanted on the skull with dental acrylic. Rats were then allowed to recover, habituated for one day, and then run on a "delay" conditioning paradigm using a 350 ms white noise conditioning stimulus (CS, 85 dB, 5 ms rise/fall time) and a 100 ms coterminating periorbital shock unconditioned stimulus (US, 2 mA, 60 Hz, AC). Rats were trained for four days. Each day one session was given which consisted of 10 blocks X 10 trials. Each block had one noise alone trial, four paired trials (noise & shock), one shock alone trial and four more paired trials. The intertrial interval (ITI) was randomized between 20 s and 40 s with a 30 s average. A chi-square analysis with Yate's statistical correction was used on each trial to determine the presence of a conditioned response (CR), i.e., if there was significantly more EMG
METHOD The subjects for this experiment were male F1 hybrid (F344 X Brown Norway) rats at 9, 18, 24, and 36 months of age.
t Requests for reprints should be addressed to Craig Weiss, Neurosciences Program, University of Southern California, Los Angeles, CA 900892520.
319
320
WEISS AND THOMPSON
activity in the 100 ms preUS period as compared to the 100 ms preCS period (200 ms periods were used on noise alone trials). Significant differences for all statistical tests were identified by p values ~< 0.05. Standard errors are indicated when means are presented. RESULTS Percent CRs
The 36-month-old rats showed learning deficits as indicated in Fig. I (top panel). A two-way analysis of variance (ANOVA) (Age × Session) for the percent of CRs indicated that there was a significant interaction of age and training session, F(9,57) = 8.0, as well as significant main effects. The interaction is due to
F1
Hybrids
(F-344
x
BN )
~ - - " 9m o - - o 1 8 m v - - V 2 4 m u---U36m N=6 N=4 N=4 N=9 100 90 80 70 ~0 6 0 fl" 50 L) 40 30 20 10 0
I
I
I
I
1
2
3
4
Training
Session
Fischer-344
a--A 3m N=7 100 90 80 70 (b 6 0 rr C) 5 0 40 30 20 10 0
o - - o 1 2 m v - - v 1 8 m u---u30 m N=6 N=5 N=7
I
I
I
I
1
2
3
4
Training
Session
FIG. 1. Percentage of trials with conditioned responses as a function of age and training session. Values are means _+ SE. The top panel presents the data from these FI hybrids (F-344 × Brown Norway), for comparison, the bottom panel shows the data from F-344 rats which were previously studied (14).
the delayed acquisition by the 36-month-old rats. The learning rates for the four age groups is seen in Fig. l; for comparison, the bottom panel of Fig. 1 shows our previously reported data from F-344 rats (15). The effect of age on percent CRs was shown post hoc by one way ANOVAs (i.e., there were significant differences among the age groups on each of the sessions, F(3,19) = 12.0, 43.0, 7.3, and 3.0 for sessions one through four respectively. The Least Significant Difference Test (LSD) indicated that for all sessions the difference was due to the poor performance of the 36 month olds. In addition, during session one, there was a significant difference between the 24 and 18 month olds. The effect of repeated training was also analyzed post hoc by one way ANOVAs. The results indicated that each age group had significant learning across the four sessions. F values for learning across sessions for rats aged 9, 18, 24, and 36 months were: F(3,15) = 16.9, F(3,9) = 12.5, F(3,9) = 5.7, and F(3,24) = 114.8, respectively. Planned comparisons were used to determine between which sessions significant learning had occurred. The results indicated that, in general, significant increases in performance occurred only between sessions one and two. The two exceptions are that 1) the 36 month olds also had significant improvement between sessions two and three, and 2) the 24 month olds had significant improvement between sessions one and three but not between sessions one and two (they were already outstanding by the end of session one). Note that, in these and subsequent analyses, the data of one 9-month-old rat was excluded, Although the rat had orienting responses and a normal blink threshold, it had the lowest percent CRs of any rat in the entire population (4%, 25%, 20%, 25% for sessions 1-4, respectively). Those values are about one SD beyond the mean. Similarly, that rat never met criterion on any session. Its brain is being examined for histopathology. Thus, in contrast to the 36-month-old rats, younger rats (924 months-old) learn the conditioned eyeblink rapidly (they show substantial numbers of CRs within the first session) and show greater than 80% CRs after two days of training (mean = 82.8% _+ 3.5%). Furthermore, although the aged rats approached 70% CRs after four days of training (mean = 68.9% _+ 6.0%), they still had significantly fewer CRs on all four days when compared to younger rats. An example of the gradual increase in conditioned EMG activity for a typical aged rat is shown in Fig, 2. This rat and the younger control rats were selected because their daily percent CRs were closest to the mean of their respective age groups. Trials to criterion. A two-way ANOVA (Age × Session) for trials to criterion (8 CRs out of any block of 9 trials, excluding the shock alone trial) indicated a significant interaction, F(9,57) = 10.7, as well as significant main effects. As shown in Fig. 3, this interaction is due to the poor learning by the 36 month olds on sessions one and two. This is indicated by significant F values for one-way ANOVAs on the effect of age for sessions one and two, F(3,19) = 8.8 and 65.1, respectively. Planned comparisons confirmed that the 36-month-old rats took significantly more trials to reach criterion than the younger rats. In fact, on sessions I and 2, the 36-month-old rats never attained criterion; the maximum score of 100 was assigned if criterion was not met. In contrast, the younger rats rapidly met criterion with a mean of 68.6 _+ 10.1 trials for all 9-24 m old rats combined. A one-way ANOVA on the total trials to criterion (the sum of the four daily trails to criterion) also yielded a significant effect of age F(3,19) = 24.4. These data are presented in Fig. 4.
EYEBLINK CONDITIONING IN HYBRID RATS
321
F1 DAY' 1
~7-- v 9m N=6
Hybrids ( F - 3 4 4 13---1318m 14=4
x BN)
zx.. z~24rn 1"4=4
~36rn N=9
100 E 0
(1)
©Xlo
9O 80 7O 60 50
T "~',~
40
o3 f~
F1
30 20 10 0 1
2
3
4
Training Session
FIG. 3. Trials to criterion (8 CRs out of any 9 consecutive trials) as a function of age and training session. Values are means _+ SE. A score of 100 (the maximum number of trials per session) was assigned for any subject not reaching criterion on a given session. Notice that smaller values indicate better learning and that decreasing values indicate retention of the learned response.
m l 19m ] ]
l ()()ms
,m,
348ms CS
"~ 3f}O-5,000Hz
FIG. 2. Example of the gradual increase in conditoned EMG activity tbr a 36-month-old FI hybrid (F-344 × Brown Norway) rat over four days of training. Each row shows superimposed EMG activity from the 10 noise alone trials on each of four successive days of training. For comparison, the last row shows the increase in conditoned EMG activity for a 9-month-old rat during the first training session. CS onset and olt'set are indicated by the arrow heads (348 ms duration). The EMG was filtered to pass 300-5,000 Hz. The tick marks at the left and boxes at the right are a result of our plotting system.
tended to vocalize (squeal) more. The vocalization was noted in a few subjects while they received the periorbital shock and in many while they were handled to give an overdose of barbiturate prior to perfusion. Vocalizing during the shock suggested that the shock may have been more aversive to the hybrid rats than to the Fischers, but a two-way ANOVA of Age × Session with repeated measures on the daily shock threshold indicated that there were no significant differences among the rats. There was no significant change in threshold across days either. This result, in conjunction with the fact that all rats could hear (i.e., exhibit orienting responses to a test CS) indicated that the learning deficit is not likely due to peripheral mechanisms but is likely to involve central associative sites instead (see Discussion section).
F1
L S D tests indicated that the difference was solely due to the 36 m o n t h olds. As a group, the 9 - 2 4 m o n t h olds had a mean o f 87.4 + 12.4 total trials to criterion: the 36 m o n t h olds had a mean o f 247 _+ 13.1. Also, note that the initial and total trials to criterion are highly correlated (r = .98). Morphoh)gical d~fferences between strains. These hybrid rats differed morphologically from the previously run parental strain o f F - 3 4 4 s (15). Most noticeable was the pigmentation and texture o f the fur. F - 3 4 4 s have coarse white fur whereas hybrids have s m o o t h brown fur (except for a band o f white which runs sagitally along the ventral surface). A n o t h e r obvious difference is that the 9 - 2 4 - m o n t h - o l d hybrid rats weighed from 2 2 % - 2 7 % more than the 3 - 1 8 m o n t h old F - 3 4 4 s . Similarly, the 36 m o n t h - o l d hybrids weighed 9% more than the 30 m o n t h old F 3 4 4 s (see Figure 5). A n o t h e r difference was found during surgery. W e noted that hybrids o f all ages had thinner skin and tended to bleed longer than the F - 3 4 4 rats. Finally, w e noted that, during handling, the hybrids seemed to be as docile as the Fischer rats but
?--?
9m N=6
Hybrids El---l:318m N=4
(F-344 zx.. zX24m N=4
x
I~,N) 0----O36m N=9
3OO E
.9 4J
25O 2OO
0 o
150
4~ O3
100 5O 0
initial
total
FIG. 4. Comparison of the initial and total trials to criterion as a function of age. Values are means _+ SE. Note that smaller values indicate better learning.
322
WEISS AND THOMPSON
~F:44 [~Hybrid ~'E 4oo o v 300 .c
"~
2oo
3
..Z_
ii
12 18 Age ( m o n t h s )
2g
FIG. 5. Comparison of weights for Fischer-344 rats and F 1hybrids (F344 × Brown Norway). The age groups do not necessarily match the age groups reported here because of lapses between weighingand training the rats.
Necropsy resuhs. Within two weeks to one month after the conclusion of the behavioral study, the rats were necropsied for gross abnormalities. The only significant lesions or abnormalities among the 9-24-month-olds were that two rats had a pituitary tumor. In contrast, the 36-month-olds had some noticeable pathology of their internal organs (i.e., spotting (blood clots) of the lungs, liver, and kidneys and tumors of the testes). However, even the worst case was not as bad as typically seen among the old F-344s. Specific pathologies for these hybrid rats are presented in Table 1. Note that two rats had paresis of the back legs, probably due to degenerative radiculoneuropathy, and still learned fairly well. DISCUSSION
Behavioral Data Because all but the very, aged hybrid rats (9-24 vs. 36 months) were found to learn rapidly, our data indicate clear differences in the rate of learning between these hybrid rats and the F-344 rats we previously studied (15). Apparently, putting one shock alone trial within the middle of each block is not enough to explain the poor performance of the F-344 rats. In fact, the Fischers performed so poorly that the oldest group of
TABLE 1 PATHOLOGYAND PERCENTAGEOF CONDITIONEDRESPONSES Session Age
ID
Pathology
18 m 258 Pituitary.Tumor 7 mm All other 18 m hybrid rats (n = 5) 24 m 262 Pituitary Tumor 0.5 mm All other 24 m hybrid rats(n = 3) 36m 218 Pituitary t u m o r l 0 m m X 6 m m Two tumors on body 36 m 220 Pituitary tumor 4 mm 36 m 222 Paresisofhindlimbs, founddead 4 weeks after experiment 36 m 112 Paresisofhindlimbs, found dead 2.5 weeks after experiment All other 36 m hybrid rats (n = 5)
1
2
3
4
28 45 51 70 1
71 58 92 81 6
92 71 89 93 45
47 82 96 94 30
0 1
9 13
48 45
51 64
4
2
61
78
10
25
74
79
Comparison of percentage of CRs for rats (by age) with and without gross pathology at the time of necropsy.
hybrid rats learned better than the youngest group of Fischer344 rats which we previously ran. This is particularly striking when survival rates, rather than absolute ages, are compared (i.e., in comparison to the F-344s we previously used, these FI hybrids are of greater absolute age to compensate for lifespan differences, i.e., 5%-20% of F-344s survived to 30 months and 24% of F l s survived to 36 months (personal communication from NIA). Thus, the learning deficit with our simple delay paradigm is nearly complete by middle age ( 18 months) in F-344 rats and does not appear in the F1 hybrid until extreme old age (36 months). The data from humans (18) is similar to the F-344 rats, i.e., humans have been reported to show deficits in CR amplitude during the fourth decade, and in percent CRs during the fifth decade (18). Data from rabbit studies are more difficult to interpret because the lifespan of rabbits has not been empirically determined and an aging study with standard delay parameters has not been done. The best data on the lifespan of a rabbit comes from Fox (4). He estimated the life expectancy for a rabbit to be about 8 years and noted that reproduction starts to fail (the number of surviving offspring within a litter begins to decrease) between age 18 and 24 months. At 30 months old, rabbits were found to require significantly more trials to criterion than 3-month-old rabbits when trained with the more difficult trace conditioning paradigm (16). Because the hybrids learned to greater than 80% of maximum, the data presented here add credence to the hypothesis that even young F-344 rats may have deficits in the essential neural circuits that mediate eyeblink conditioning (15). The current data suggest that learning deficits on our delay conditioning paradigm only become apparent in very aged FI hybrid rats. This deficit is likely due to deficits in central rather than peripheral mechanisms. Although our methods for determining hearing and shock thresholds were rather subjective, more quantitative methods confirmed this hypothesis (7). Furthermore, associative deficits still occur even when electrical stimulation of mossy fiber input to the cerebellum is used in lieu of an auditory CS (17). Because the oldest hybrid rats learned better than the youngest F-344, it seems as if this crossbreeding results in hybrid vigor as related to our task. In fact, in addition to our F 344 rats, the F 1 hybrids appear to have learned better than two other strains that have been used in eyeblink conditioning (I,10,12). Schmajuk and Christiansen (10) used 2-month-old restrained Sprague-Dawley rats. Those rats were given four days of habituation, then 50 trials per day with a 60 s mean ITI. Those rats took a mean of 253 trials to reach a criterion of 8 CRs out of 10 trials and a mean of 408 trials (approximately 8 days) to reach 80% CRs within a session. Skelton (12) used freely moving 3-month-old Long Evans rats (we converted weight, 350-400 g to age) to show that bilateral lesions of the interpositus nucleus abolish CRs. Those rats were given one day of habituation then 100 trials per day with a mean ITI of 30 s. Those rats took a mean of 140 trials to reach a criterion of 9 CRs out of 10 trials; the percent CRs were not reported. Adams et al. (1) also conditioned Long Evans rats but while the rats were restrained and recorded neural models of CRs from the interpositus nucleus. They used a white noise CS and an airpuff US. After four days of habituation those rats reached criterion (8 CRs out of 9 trials) in about 50 trials (personal communication). As we report in the Method and Results sections, our rats received one day of habituation, then 100 trials per day with a
EYEBLINK CONDITIONING IN HYBRID RATS
323
mean ITI of 30 s. This resulted in remarkably quick and robust learning in all ages of the hybrid rat, especially the 9-monthold rats, i.e., young adults (68 trials to a criterion of 8 CRs in 9 trials, and 83% CRs after two days of training). In contrast, the young (3-month) F-344 rats that we previously studied reached criterion in about 180 trials but never achieved 80% CRs. Thus, we replicated the general finding that there is an age related deficit in eyeblink conditioning using a delay paradigm (9,15). In our previous article (15), we noted Graves and Solomon (7) may not have found a deficit in the rabbit because of the longer CS (500 ms vs. 350 ms) which may facilitate conditioning in aged subjects. Furthermore, assuming that restraint is differentially stressful for rats and rabbits, the relatively slow learning of the restrained rat by Schmajuk and Christiansen (10) suggests that either the stress of restraint was enough to impede acquisition ofeyeblink conditioning, just as it does acquisition of long term potentiation (LTP), another model of learning (5,11 ), or that periorbital shock is a more salient US than airpufl:
36-month-old rats with paresis of the back legs (degenerative radiculoneuropathy) whereas Bronson reported that F-344s have an incidence of 23% with a mean age of occurrence at 30 months. The majority of 36 month old hybrids had minimal pathology. These results contrast with the findings on the four necropsies that were previously done on 30-month-old F-344 rats. They showed more gross lesions including greatly enlarged spleens and testes (Coleman et al. (3) reported 80% incidence of testicular tumors at 18-24 months), and vascular congestion of the lungs, liver, and kidneys. In conclusion, our paradigm allows for rapid acquisition of an associatively learned task whereas not putting the rat under duress (i.e., restraint). This is quite beneficial for behavioral and molecular biological studies. For studies that require single cell recording, examination of the UR or discrete movements (pure eyeblink rather than combined eyeblink and head twitch) the restrained rat paradigm may be a necessary alternative, However, with further technical advances even the aforementioned limitations may be overcome.
Patholog)' Data
ACKNOWLEDGEMENTS
The necropsy results suggest significant differences for the effect of age on hybrid and Fischer rats just as Bronson reported (2). Table 1 shows that 18-36-month-old hybrids yielded four pituitary tumors (21%), whereas Goodman et al. (6) reported an 1 I% incidence for F-344s. We also found two (22%) of the
This research was supported by NSF award BNS 8718300 to R.F.T., NIH awards AG00093 and AG05142, and NIA pilot projects program. We wish to thank M. Gordon, A. Moro, E. Savay, and J. Yoo for technical assistance and our gerontology consultant C. E. Finch.
REFERENCES
1. Adams, R. M.; Zhang, A. A.; Lavond, D. G. Neural unit activity in cerebellar interpositus nucleus models classically conditioned eyelid responses in the rat. Soc. Neurosci. Abstr. 19:890: 1989. 2, Bronson, R. T. Rate of occurrence of lesions in 20 inbred and hybrid genotypes of rats and mice sacrificedat 6 month intervals during the first years of life. In: Harrison, D. E., ed., Genetic effects on aging II. Caldwell, NJ: Telford Press: 1990. 3. Coleman, G. L.; Barthold, S. W.: Obaldistan, S. J.; Foster, S. J.: Jonas, A. M. Pathological changes during aging in barrier-reared Fischer-344 male rats. J. Gerontol. 32:358-278; 1977. 4. Fox, R. R. The rabbit (Oryctolagus cuniculus) and research on aging. Exp. Aging Res. 6:235-248: 1980. 5. Foy, M. F.; Stanton, M. E.; Levine, S.; Thompson, R. F. Behavioral stress impairs long-term potentiation in rodent hippocampus. Behav. Neural Biol. 48:138-149:1987. 6. Goodman, D. G.; Ward, J. M.; Squire, R. A.: Chu, K. C.; Linhart, M. S. Neoplastic and non-neoplastic lesions in aging F344 rats. Toxicol. Appl. Pharmacol. 48:237-248; 1977. 7. Graves, C. A.: Solomon, P. R. Age-relateddisruption of trace but not delay classical conditioning of the rabbit's nictitating membrane response. Behav, Neurosci. 99:88-96; 1985. 8. Kimble, G. A.: Pennypacker, H. S. Eyelid conditioningin young and aged subjects. J. Gen. Psychol. 103:283-289; 1963. 9. Powell, D. A.; Buchanan, S. L.; Hernandez, L. L. Age related changes in classical (Pavlovian) conditioningin the New Zealand albino rabbit. Exp. Aging Res. 7:453-465, 1981.
10. Schmajuk, N. A.; Christiansen, B. A. Eyeblink conditioning in rats. Physiol. Behav. 48:755-758, 1990. 11. Shors, T. J.: Foy, M. R.: Levine, S.; Thompson, R. F. Unpredictable and uncontrollable stress impairs neuronal plasticity in the rat hippocampus. Brain Res. Bull. 24:663-667: 1990. 12. Skelton, R. W. Bilateralcerebellar lesionsdisrupt conditioned eyelid response in unrestrained rats. Behav. Neurosci. 102:586-590; 1988. 13. Weiss, C.; Savay, E.: Thompson, R. F. Efliectsof age on eyeblink conditioningin freely moving Fischer-344 and FI hybrid rats. Soc. Neurosci. Abstr. 16:269; 1990. 14. Weiss, C.: Shors, T. J.; Tram L. B.: Thompson, R. F. Eyeblink conditioningin the freely moving middle aged rat. Soc. Neurosci. Abstr. 15:81; 1989. 15. Weiss, C.; Thompson, R. F. The effects of age on eyeblink conditioning in the freely moving Fischer-344 rat. Neurobiol. Aging 12(3):249-254; 1991. 16. Woodrufl:Pak, D. S.; Lavond, D. G.; Logan, C. G.; Thompson, R. F. Classical conditioningin 3-, 3%, and 45-month-old rabbits: behavioral learning and hippocampal unit activity. Neurobiol. Aging 8:101-108: 1987. 17. Woodruff-Pak, D. S.; Steinmetz, J. E.; Thompson, R. F. Classical conditioningof rabbits 2~.~to 4 years old using mossy fiber stimulation as a CS. Neurobiol. Aging, 9:187-193:1987. 18. Woodruff-Pak, D. S.; Thompson, R. F. Classical conditioningof the eyeblink response in the delay paradigm in adults aged 18-83 years. Psychol. Aging 3:219-229:1988.
0197-4580/92 $5.00 + .00 Copyrightt) 1992PergamonPressLtd.
Printedin the USA.All rightsreserved.
Delayed Acquisition of Eyeblink Conditioning in Aged F1 Hybrid (Fischer-344 × Brown Norway) Rats C R A I G W E I S S I A N D R I C H A R D F. T H O M P S O N
Neurosciences Program, University of Southern California, Los Angeles', CA 90089-2520
Received 28 May 1991; Accepted 18 September 1991 WEISS, C. AND R. F. THOMPSON. Delayedacquisition ~[eyeblink conditioning in aged FI hybrid(Fischer-344 X Brown Norway) rats. NEUROBIOL AGING 13(2) 319-323, 1992.--We previously reported that the freely moving male Fischer344 rat provides a useful model to demonstrate the progressive impairment of eyeblink conditioning associated with aging. However, because the youngest F-344 rats only performed at 60% of maximum, we ran the same experiment with hybrid rats and discovered most (i.e., those age 9-24 months) learned rapidly and exhibited conditioned responses on greater than 80% of trials by the end of two training sessions. In contrast, the aged rats (36 months) exhibited significantly fewer CRs on all four training days. However, unlike all ages of F-344 rats (3-30 months) which were run in our last study, these aged hybrid rats exhibited considerable improvement with extra training. These data indicate clear differences in the rate of learning between the two strains and suggest that even young F-344 rats may have deficits in the neural circuits which mediate eyeblink conditioning. Other anecdotal findings on differences between the two strains are noted. Aging Stress
Classicalconditioning
Nictitating membrane
Hybrid
CLASSICAL conditioning of the eyeblink reflex is an excellent paradigm to analyze learning and memory in various species. Both humans and animals show progressive age related impairments on this task (7,8,9) which become apparent as early as middle age (15,16,17). We reported that the freely moving male Fischer-344 rat showed a progressive age related impairment of eyeblink conditioning ( 14,15). However, the youngest F-344 rats exhibited CRs on no more than 60% of trials. To insure that the young F-344s were in fact poor learners and that our paradigm and apparatus would allow maximal conditioning, we ran the same experiment with F1 hybrid (F-344 X Brown Norway) rats. The F-344 X BN F1 hybrid was selected as an alternative model to the F-344 based on a study of several F1 hybrids (2). The F-344 X BN cross produces progeny with the fewest detrimental pathologies and at a later age of onset. This study was undertaken to determine if the same "hybrid vigor" would appear in an associative learning paradigm (i.e., classical conditioning of the eyeblink reflex). Preliminary data were reported in abstract form (13).
Fischer-344
BrownNorway
EMG
Three sets of cohorts were run: the first consisted of a group of 9-24 month olds housed 3 per cage ( 1 of each age); the second, consisted of 36 month olds housed 2 per cage; the third, consisted of 9-month and 36-month-olds housed 2 per cage ( 1 of each age). All rats were weighed at the time of arrival, kept on a 12 h/ 12h light/dark cycle, and had food and water available ad lib. The rats were run in the light portion of the cycle. Methods were reported previously (15). Briefly, a headstage containing wires to record the EMG activity from obicularis oculi and to deliver a periorbital shock was implanted on the skull with dental acrylic. Rats were then allowed to recover, habituated for one day, and then run on a "delay" conditioning paradigm using a 350 ms white noise conditioning stimulus (CS, 85 dB, 5 ms rise/fall time) and a 100 ms coterminating periorbital shock unconditioned stimulus (US, 2 mA, 60 Hz, AC). Rats were trained for four days. Each day one session was given which consisted of 10 blocks X 10 trials. Each block had one noise alone trial, four paired trials (noise & shock), one shock alone trial and four more paired trials. The intertrial interval (ITI) was randomized between 20 s and 40 s with a 30 s average. A chi-square analysis with Yate's statistical correction was used on each trial to determine the presence of a conditioned response (CR), i.e., if there was significantly more EMG
METHOD The subjects for this experiment were male F1 hybrid (F344 X Brown Norway) rats at 9, 18, 24, and 36 months of age.
t Requests for reprints should be addressed to Craig Weiss, Neurosciences Program, University of Southern California, Los Angeles, CA 900892520.
319
320
WEISS AND THOMPSON
activity in the 100 ms preUS period as compared to the 100 ms preCS period (200 ms periods were used on noise alone trials). Significant differences for all statistical tests were identified by p values ~< 0.05. Standard errors are indicated when means are presented. RESULTS Percent CRs
The 36-month-old rats showed learning deficits as indicated in Fig. I (top panel). A two-way analysis of variance (ANOVA) (Age × Session) for the percent of CRs indicated that there was a significant interaction of age and training session, F(9,57) = 8.0, as well as significant main effects. The interaction is due to
F1
Hybrids
(F-344
x
BN )
~ - - " 9m o - - o 1 8 m v - - V 2 4 m u---U36m N=6 N=4 N=4 N=9 100 90 80 70 ~0 6 0 fl" 50 L) 40 30 20 10 0
I
I
I
I
1
2
3
4
Training
Session
Fischer-344
a--A 3m N=7 100 90 80 70 (b 6 0 rr C) 5 0 40 30 20 10 0
o - - o 1 2 m v - - v 1 8 m u---u30 m N=6 N=5 N=7
I
I
I
I
1
2
3
4
Training
Session
FIG. 1. Percentage of trials with conditioned responses as a function of age and training session. Values are means _+ SE. The top panel presents the data from these FI hybrids (F-344 × Brown Norway), for comparison, the bottom panel shows the data from F-344 rats which were previously studied (14).
the delayed acquisition by the 36-month-old rats. The learning rates for the four age groups is seen in Fig. l; for comparison, the bottom panel of Fig. 1 shows our previously reported data from F-344 rats (15). The effect of age on percent CRs was shown post hoc by one way ANOVAs (i.e., there were significant differences among the age groups on each of the sessions, F(3,19) = 12.0, 43.0, 7.3, and 3.0 for sessions one through four respectively. The Least Significant Difference Test (LSD) indicated that for all sessions the difference was due to the poor performance of the 36 month olds. In addition, during session one, there was a significant difference between the 24 and 18 month olds. The effect of repeated training was also analyzed post hoc by one way ANOVAs. The results indicated that each age group had significant learning across the four sessions. F values for learning across sessions for rats aged 9, 18, 24, and 36 months were: F(3,15) = 16.9, F(3,9) = 12.5, F(3,9) = 5.7, and F(3,24) = 114.8, respectively. Planned comparisons were used to determine between which sessions significant learning had occurred. The results indicated that, in general, significant increases in performance occurred only between sessions one and two. The two exceptions are that 1) the 36 month olds also had significant improvement between sessions two and three, and 2) the 24 month olds had significant improvement between sessions one and three but not between sessions one and two (they were already outstanding by the end of session one). Note that, in these and subsequent analyses, the data of one 9-month-old rat was excluded, Although the rat had orienting responses and a normal blink threshold, it had the lowest percent CRs of any rat in the entire population (4%, 25%, 20%, 25% for sessions 1-4, respectively). Those values are about one SD beyond the mean. Similarly, that rat never met criterion on any session. Its brain is being examined for histopathology. Thus, in contrast to the 36-month-old rats, younger rats (924 months-old) learn the conditioned eyeblink rapidly (they show substantial numbers of CRs within the first session) and show greater than 80% CRs after two days of training (mean = 82.8% _+ 3.5%). Furthermore, although the aged rats approached 70% CRs after four days of training (mean = 68.9% _+ 6.0%), they still had significantly fewer CRs on all four days when compared to younger rats. An example of the gradual increase in conditioned EMG activity for a typical aged rat is shown in Fig, 2. This rat and the younger control rats were selected because their daily percent CRs were closest to the mean of their respective age groups. Trials to criterion. A two-way ANOVA (Age × Session) for trials to criterion (8 CRs out of any block of 9 trials, excluding the shock alone trial) indicated a significant interaction, F(9,57) = 10.7, as well as significant main effects. As shown in Fig. 3, this interaction is due to the poor learning by the 36 month olds on sessions one and two. This is indicated by significant F values for one-way ANOVAs on the effect of age for sessions one and two, F(3,19) = 8.8 and 65.1, respectively. Planned comparisons confirmed that the 36-month-old rats took significantly more trials to reach criterion than the younger rats. In fact, on sessions I and 2, the 36-month-old rats never attained criterion; the maximum score of 100 was assigned if criterion was not met. In contrast, the younger rats rapidly met criterion with a mean of 68.6 _+ 10.1 trials for all 9-24 m old rats combined. A one-way ANOVA on the total trials to criterion (the sum of the four daily trails to criterion) also yielded a significant effect of age F(3,19) = 24.4. These data are presented in Fig. 4.
EYEBLINK CONDITIONING IN HYBRID RATS
321
F1 DAY' 1
~7-- v 9m N=6
Hybrids ( F - 3 4 4 13---1318m 14=4
x BN)
zx.. z~24rn 1"4=4
~36rn N=9
100 E 0
(1)
©Xlo
9O 80 7O 60 50
T "~',~
40
o3 f~
F1
30 20 10 0 1
2
3
4
Training Session
FIG. 3. Trials to criterion (8 CRs out of any 9 consecutive trials) as a function of age and training session. Values are means _+ SE. A score of 100 (the maximum number of trials per session) was assigned for any subject not reaching criterion on a given session. Notice that smaller values indicate better learning and that decreasing values indicate retention of the learned response.
m l 19m ] ]
l ()()ms
,m,
348ms CS
"~ 3f}O-5,000Hz
FIG. 2. Example of the gradual increase in conditoned EMG activity tbr a 36-month-old FI hybrid (F-344 × Brown Norway) rat over four days of training. Each row shows superimposed EMG activity from the 10 noise alone trials on each of four successive days of training. For comparison, the last row shows the increase in conditoned EMG activity for a 9-month-old rat during the first training session. CS onset and olt'set are indicated by the arrow heads (348 ms duration). The EMG was filtered to pass 300-5,000 Hz. The tick marks at the left and boxes at the right are a result of our plotting system.
tended to vocalize (squeal) more. The vocalization was noted in a few subjects while they received the periorbital shock and in many while they were handled to give an overdose of barbiturate prior to perfusion. Vocalizing during the shock suggested that the shock may have been more aversive to the hybrid rats than to the Fischers, but a two-way ANOVA of Age × Session with repeated measures on the daily shock threshold indicated that there were no significant differences among the rats. There was no significant change in threshold across days either. This result, in conjunction with the fact that all rats could hear (i.e., exhibit orienting responses to a test CS) indicated that the learning deficit is not likely due to peripheral mechanisms but is likely to involve central associative sites instead (see Discussion section).
F1
L S D tests indicated that the difference was solely due to the 36 m o n t h olds. As a group, the 9 - 2 4 m o n t h olds had a mean o f 87.4 + 12.4 total trials to criterion: the 36 m o n t h olds had a mean o f 247 _+ 13.1. Also, note that the initial and total trials to criterion are highly correlated (r = .98). Morphoh)gical d~fferences between strains. These hybrid rats differed morphologically from the previously run parental strain o f F - 3 4 4 s (15). Most noticeable was the pigmentation and texture o f the fur. F - 3 4 4 s have coarse white fur whereas hybrids have s m o o t h brown fur (except for a band o f white which runs sagitally along the ventral surface). A n o t h e r obvious difference is that the 9 - 2 4 - m o n t h - o l d hybrid rats weighed from 2 2 % - 2 7 % more than the 3 - 1 8 m o n t h old F - 3 4 4 s . Similarly, the 36 m o n t h - o l d hybrids weighed 9% more than the 30 m o n t h old F 3 4 4 s (see Figure 5). A n o t h e r difference was found during surgery. W e noted that hybrids o f all ages had thinner skin and tended to bleed longer than the F - 3 4 4 rats. Finally, w e noted that, during handling, the hybrids seemed to be as docile as the Fischer rats but
?--?
9m N=6
Hybrids El---l:318m N=4
(F-344 zx.. zX24m N=4
x
I~,N) 0----O36m N=9
3OO E
.9 4J
25O 2OO
0 o
150
4~ O3
100 5O 0
initial
total
FIG. 4. Comparison of the initial and total trials to criterion as a function of age. Values are means _+ SE. Note that smaller values indicate better learning.
322
WEISS AND THOMPSON
~F:44 [~Hybrid ~'E 4oo o v 300 .c
"~
2oo
3
..Z_
ii
12 18 Age ( m o n t h s )
2g
FIG. 5. Comparison of weights for Fischer-344 rats and F 1hybrids (F344 × Brown Norway). The age groups do not necessarily match the age groups reported here because of lapses between weighingand training the rats.
Necropsy resuhs. Within two weeks to one month after the conclusion of the behavioral study, the rats were necropsied for gross abnormalities. The only significant lesions or abnormalities among the 9-24-month-olds were that two rats had a pituitary tumor. In contrast, the 36-month-olds had some noticeable pathology of their internal organs (i.e., spotting (blood clots) of the lungs, liver, and kidneys and tumors of the testes). However, even the worst case was not as bad as typically seen among the old F-344s. Specific pathologies for these hybrid rats are presented in Table 1. Note that two rats had paresis of the back legs, probably due to degenerative radiculoneuropathy, and still learned fairly well. DISCUSSION
Behavioral Data Because all but the very, aged hybrid rats (9-24 vs. 36 months) were found to learn rapidly, our data indicate clear differences in the rate of learning between these hybrid rats and the F-344 rats we previously studied (15). Apparently, putting one shock alone trial within the middle of each block is not enough to explain the poor performance of the F-344 rats. In fact, the Fischers performed so poorly that the oldest group of
TABLE 1 PATHOLOGYAND PERCENTAGEOF CONDITIONEDRESPONSES Session Age
ID
Pathology
18 m 258 Pituitary.Tumor 7 mm All other 18 m hybrid rats (n = 5) 24 m 262 Pituitary Tumor 0.5 mm All other 24 m hybrid rats(n = 3) 36m 218 Pituitary t u m o r l 0 m m X 6 m m Two tumors on body 36 m 220 Pituitary tumor 4 mm 36 m 222 Paresisofhindlimbs, founddead 4 weeks after experiment 36 m 112 Paresisofhindlimbs, found dead 2.5 weeks after experiment All other 36 m hybrid rats (n = 5)
1
2
3
4
28 45 51 70 1
71 58 92 81 6
92 71 89 93 45
47 82 96 94 30
0 1
9 13
48 45
51 64
4
2
61
78
10
25
74
79
Comparison of percentage of CRs for rats (by age) with and without gross pathology at the time of necropsy.
hybrid rats learned better than the youngest group of Fischer344 rats which we previously ran. This is particularly striking when survival rates, rather than absolute ages, are compared (i.e., in comparison to the F-344s we previously used, these FI hybrids are of greater absolute age to compensate for lifespan differences, i.e., 5%-20% of F-344s survived to 30 months and 24% of F l s survived to 36 months (personal communication from NIA). Thus, the learning deficit with our simple delay paradigm is nearly complete by middle age ( 18 months) in F-344 rats and does not appear in the F1 hybrid until extreme old age (36 months). The data from humans (18) is similar to the F-344 rats, i.e., humans have been reported to show deficits in CR amplitude during the fourth decade, and in percent CRs during the fifth decade (18). Data from rabbit studies are more difficult to interpret because the lifespan of rabbits has not been empirically determined and an aging study with standard delay parameters has not been done. The best data on the lifespan of a rabbit comes from Fox (4). He estimated the life expectancy for a rabbit to be about 8 years and noted that reproduction starts to fail (the number of surviving offspring within a litter begins to decrease) between age 18 and 24 months. At 30 months old, rabbits were found to require significantly more trials to criterion than 3-month-old rabbits when trained with the more difficult trace conditioning paradigm (16). Because the hybrids learned to greater than 80% of maximum, the data presented here add credence to the hypothesis that even young F-344 rats may have deficits in the essential neural circuits that mediate eyeblink conditioning (15). The current data suggest that learning deficits on our delay conditioning paradigm only become apparent in very aged FI hybrid rats. This deficit is likely due to deficits in central rather than peripheral mechanisms. Although our methods for determining hearing and shock thresholds were rather subjective, more quantitative methods confirmed this hypothesis (7). Furthermore, associative deficits still occur even when electrical stimulation of mossy fiber input to the cerebellum is used in lieu of an auditory CS (17). Because the oldest hybrid rats learned better than the youngest F-344, it seems as if this crossbreeding results in hybrid vigor as related to our task. In fact, in addition to our F 344 rats, the F 1 hybrids appear to have learned better than two other strains that have been used in eyeblink conditioning (I,10,12). Schmajuk and Christiansen (10) used 2-month-old restrained Sprague-Dawley rats. Those rats were given four days of habituation, then 50 trials per day with a 60 s mean ITI. Those rats took a mean of 253 trials to reach a criterion of 8 CRs out of 10 trials and a mean of 408 trials (approximately 8 days) to reach 80% CRs within a session. Skelton (12) used freely moving 3-month-old Long Evans rats (we converted weight, 350-400 g to age) to show that bilateral lesions of the interpositus nucleus abolish CRs. Those rats were given one day of habituation then 100 trials per day with a mean ITI of 30 s. Those rats took a mean of 140 trials to reach a criterion of 9 CRs out of 10 trials; the percent CRs were not reported. Adams et al. (1) also conditioned Long Evans rats but while the rats were restrained and recorded neural models of CRs from the interpositus nucleus. They used a white noise CS and an airpuff US. After four days of habituation those rats reached criterion (8 CRs out of 9 trials) in about 50 trials (personal communication). As we report in the Method and Results sections, our rats received one day of habituation, then 100 trials per day with a
EYEBLINK CONDITIONING IN HYBRID RATS
323
mean ITI of 30 s. This resulted in remarkably quick and robust learning in all ages of the hybrid rat, especially the 9-monthold rats, i.e., young adults (68 trials to a criterion of 8 CRs in 9 trials, and 83% CRs after two days of training). In contrast, the young (3-month) F-344 rats that we previously studied reached criterion in about 180 trials but never achieved 80% CRs. Thus, we replicated the general finding that there is an age related deficit in eyeblink conditioning using a delay paradigm (9,15). In our previous article (15), we noted Graves and Solomon (7) may not have found a deficit in the rabbit because of the longer CS (500 ms vs. 350 ms) which may facilitate conditioning in aged subjects. Furthermore, assuming that restraint is differentially stressful for rats and rabbits, the relatively slow learning of the restrained rat by Schmajuk and Christiansen (10) suggests that either the stress of restraint was enough to impede acquisition ofeyeblink conditioning, just as it does acquisition of long term potentiation (LTP), another model of learning (5,11 ), or that periorbital shock is a more salient US than airpufl:
36-month-old rats with paresis of the back legs (degenerative radiculoneuropathy) whereas Bronson reported that F-344s have an incidence of 23% with a mean age of occurrence at 30 months. The majority of 36 month old hybrids had minimal pathology. These results contrast with the findings on the four necropsies that were previously done on 30-month-old F-344 rats. They showed more gross lesions including greatly enlarged spleens and testes (Coleman et al. (3) reported 80% incidence of testicular tumors at 18-24 months), and vascular congestion of the lungs, liver, and kidneys. In conclusion, our paradigm allows for rapid acquisition of an associatively learned task whereas not putting the rat under duress (i.e., restraint). This is quite beneficial for behavioral and molecular biological studies. For studies that require single cell recording, examination of the UR or discrete movements (pure eyeblink rather than combined eyeblink and head twitch) the restrained rat paradigm may be a necessary alternative, However, with further technical advances even the aforementioned limitations may be overcome.
Patholog)' Data
ACKNOWLEDGEMENTS
The necropsy results suggest significant differences for the effect of age on hybrid and Fischer rats just as Bronson reported (2). Table 1 shows that 18-36-month-old hybrids yielded four pituitary tumors (21%), whereas Goodman et al. (6) reported an 1 I% incidence for F-344s. We also found two (22%) of the
This research was supported by NSF award BNS 8718300 to R.F.T., NIH awards AG00093 and AG05142, and NIA pilot projects program. We wish to thank M. Gordon, A. Moro, E. Savay, and J. Yoo for technical assistance and our gerontology consultant C. E. Finch.
REFERENCES
1. Adams, R. M.; Zhang, A. A.; Lavond, D. G. Neural unit activity in cerebellar interpositus nucleus models classically conditioned eyelid responses in the rat. Soc. Neurosci. Abstr. 19:890: 1989. 2, Bronson, R. T. Rate of occurrence of lesions in 20 inbred and hybrid genotypes of rats and mice sacrificedat 6 month intervals during the first years of life. In: Harrison, D. E., ed., Genetic effects on aging II. Caldwell, NJ: Telford Press: 1990. 3. Coleman, G. L.; Barthold, S. W.: Obaldistan, S. J.; Foster, S. J.: Jonas, A. M. Pathological changes during aging in barrier-reared Fischer-344 male rats. J. Gerontol. 32:358-278; 1977. 4. Fox, R. R. The rabbit (Oryctolagus cuniculus) and research on aging. Exp. Aging Res. 6:235-248: 1980. 5. Foy, M. F.; Stanton, M. E.; Levine, S.; Thompson, R. F. Behavioral stress impairs long-term potentiation in rodent hippocampus. Behav. Neural Biol. 48:138-149:1987. 6. Goodman, D. G.; Ward, J. M.; Squire, R. A.: Chu, K. C.; Linhart, M. S. Neoplastic and non-neoplastic lesions in aging F344 rats. Toxicol. Appl. Pharmacol. 48:237-248; 1977. 7. Graves, C. A.: Solomon, P. R. Age-relateddisruption of trace but not delay classical conditioning of the rabbit's nictitating membrane response. Behav, Neurosci. 99:88-96; 1985. 8. Kimble, G. A.: Pennypacker, H. S. Eyelid conditioningin young and aged subjects. J. Gen. Psychol. 103:283-289; 1963. 9. Powell, D. A.; Buchanan, S. L.; Hernandez, L. L. Age related changes in classical (Pavlovian) conditioningin the New Zealand albino rabbit. Exp. Aging Res. 7:453-465, 1981.
10. Schmajuk, N. A.; Christiansen, B. A. Eyeblink conditioning in rats. Physiol. Behav. 48:755-758, 1990. 11. Shors, T. J.: Foy, M. R.: Levine, S.; Thompson, R. F. Unpredictable and uncontrollable stress impairs neuronal plasticity in the rat hippocampus. Brain Res. Bull. 24:663-667: 1990. 12. Skelton, R. W. Bilateralcerebellar lesionsdisrupt conditioned eyelid response in unrestrained rats. Behav. Neurosci. 102:586-590; 1988. 13. Weiss, C.; Savay, E.: Thompson, R. F. Efliectsof age on eyeblink conditioningin freely moving Fischer-344 and FI hybrid rats. Soc. Neurosci. Abstr. 16:269; 1990. 14. Weiss, C.: Shors, T. J.; Tram L. B.: Thompson, R. F. Eyeblink conditioningin the freely moving middle aged rat. Soc. Neurosci. Abstr. 15:81; 1989. 15. Weiss, C.; Thompson, R. F. The effects of age on eyeblink conditioning in the freely moving Fischer-344 rat. Neurobiol. Aging 12(3):249-254; 1991. 16. Woodrufl:Pak, D. S.; Lavond, D. G.; Logan, C. G.; Thompson, R. F. Classical conditioningin 3-, 3%, and 45-month-old rabbits: behavioral learning and hippocampal unit activity. Neurobiol. Aging 8:101-108: 1987. 17. Woodruff-Pak, D. S.; Steinmetz, J. E.; Thompson, R. F. Classical conditioningof rabbits 2~.~to 4 years old using mossy fiber stimulation as a CS. Neurobiol. Aging, 9:187-193:1987. 18. Woodruff-Pak, D. S.; Thompson, R. F. Classical conditioningof the eyeblink response in the delay paradigm in adults aged 18-83 years. Psychol. Aging 3:219-229:1988.
