- Email: [email protected]
Reference and working memory of rats following hippocampal damage induced by transient forebrain ischemia
Physiology& Behavior.Vol. 37, pp. 387-392. Copyrighte PergamonPress Ltd., 1986. Printed in the U.S.A.
0031-9384/86$3.00 + .00
Reference and Working Memory of Rats Following Hippocampal Damage Induced by Transient Forebrain Ischemia H A S K E R P. D A V I S
Department o f Psychology, University o f Colorado, Colorado Springs, CO 80933 JOSEPH TRIBUNA
School o f Pharmacy, St. John's University, Jamaica, N Y 11439 W I L L I A M A. P U L S I N E L L I
A N D B R U C E T. V O L P E
Department of Neurology, Cornell University Medical College 1300 York Avenue, New York, N Y 10021 R e c e i v e d 1 N o v e m b e r 1985 DAVIS, H. P., J. TR1BUNA, W. A. PULS1NELLI AND B. T. VOLPE. Reference and workingmemory of ratsfollowing hippocampal damage induced by transientforebrain ischemia. PHYSIOL BEHAV 37(3) 387-392, 1986.--Acquisition of reference and working memory was evaluated in an animal model of cerebral ischemia. Rats were subjected to 30 minutes of transient forebrain ischemia, allowed to recover, and then tested for 95 trials on an 8-arm maze with 5 arms baited. During the 95 trials post ischemic (PI) rats made significantly more working and reference errors than controls (.o<0.05). However, an analysis of the last 20 trials (75-95) showed that while PI rats and control rats had comparable reference memory (p>0.8). P1 rats tended to have a persistent working memory deficit compared to controls (p<0.06). Subsequent morphologic analysis showed that PI rats had almost complete loss of pyramidal neurons in the anterior CA 1 region of the hippocampus, moderate to severe loss in mid-dorsal posterior hippocampus, and less damage to the dorsolateral striatum. These results suggest that the PI animal is a reasonable model for the permanent behavioral impairment and pathologic damage found in some human survivors of cardiac arrest. Amnesia Cardiac arrest Hippocampus Reference memory Working memory
Ischemia
CARDIAC arrest is a common medical problem, and although mortality has improved, sensorimotor deficits and cognitive disturbances may cause persistent morbidity in many of these survivors [1, 3, 5, 12]. Some survivors have demonstrated a fairly circumscribed amnesic syndrome characterized by impaired learning and memory of events after their injury, and a gradient of memory loss for premorbid events [3, 20, 21, 25]. Post mortem neuropathologic examination of patients with cardiac arrest has revealed marked loss of specific populations of neurons highly vulnerable to ischemia [2]. Post mortem studies of patients who had cardiac arrest and who also had ante mortem behavioral studies have shown the most severe damage in hippocampus, particularly the CA1 region [4,22]. We have been investigating an animal model that demonstrates some of the morphologic and the behavioral consequences of cerebral ischemia that follow cardiac arrest in humans. Investigators have shown that rats subjected to
Learning and memory
Radial maze
controlled transient occlusion of the major vessels to the brain develop reproducible and quantifiable morphologic brain injury that appears most severe in the CA1 region of the hippocampus [11, 16, 17, 19]. However, few studies have objectively addressed long-term functional changes in the behavior of animals exposed to global cerebral ischemia [23,24]. One of the advantages of this model is that the majority of animals subjected to 30 minutes of ischemia survive indefinitely, and it is therefore possible to characterize and quantify their behavior long after recovery from the acute insult. The present study replicates and extends a previous investigation to determine whether ischemic-induced neuronal damage in the rat, and in particular damage to CAI hippocampal neurons, results in permanent learning and memory impairments similar to humans [24]. A radial 8-arm maze task was used to assess behavior because it has been demonstrated that rats with hippocampal damage perform differently on two aspects of this task [13].
387
I)AVIS E7 AI~.
388 In the radial maze it is possible to measure both reference performance (entering baited arms only) and working performance (not re-entering a baited arm after the food is taken) by always baiting the same 5 arms of the 8-arm maze on all trials. That is, reference performance requires the animal to learn that 5 of 8 arms are baited, and that these 5 baited arms remain contant relative to room cues for all trials. Working performance requires the rat to remember from which arms food has been taken so as not to re-enter that arm during a particular trial. The reference memory component includes many repetitions of invarant material that is useful over any and all trials, while the working memory component requires the retention of trial specific information that varies across trials. If post ischemic (PI) rats showed dissociable reference and working memory, then these results would suggest similarities to post cardiac arrest amnesia in which patients have deficits in remembering new and/or variable information [7].
METHOD
Subjects Male Wistar rats (Hilltop, Scottsdale, PA) weighing between 250-300 grams served as subjects. Animals were individually housed in polyurethane cages (33×21x19 cm) throughout the experiment and provided ad lib access to water.
Surgical Procedure Rats were subjected to forebrain ischemia by a method described in detail previously [16]. Briefly, on Day 1 rats were anesthetized by the inhalation of 2.5% halothane mixed with 30% oxygen and 70% nitrogen. Atraumatic clasps were placed loosely around each common carotid artery without interrupting blood flow. The vertebral arteries were permanently occluded by electrocautery at the first cervical vertebra. Animals were allowed 24 hours of recovery. After recovery, operated animals are indistinguishable from control animals by electrocephalographic, physiologic, and subjective behavioral criteria [16]. On Day 2, forebrain ischemia was produced in awake rats by tightening the carotid artery clasps. Animals that did not become unresponsive within 1 minute were classified as noncriterion shams. The carotid artery clasps were released after 30 minutes of occlusion. Body temperature was maintained at a minimum of 37°C by a heating lamp connected to a rectal thermistor until thermal homeostasis was restored. Control rats were subjected to halothane anesthesia and skin incisions, Previous behavioral assessments detected no differences between controls and vertebral cautery control rats exposed to halothane and occlusion of the vertebral arteries [24]. Animals were allowed 30 days of postoperative recovery. After 2-3 weeks there was no weight difference between control rats and operated rats. It was not possible to distinguish operated from unoperated rats by observation of feeding, grooming, or exploring habits.
ter) was at the end of each arm. The test room was rich in stationary extramaze stimuli. Thirty days postoperatively the PI and control rats were started on a partial food deprivation schedule and by trial 15 all animals bad reached 80e/ of their ad lib weight. Maze adaptation was started after 1 day of food deprivation. Individual rats were allowed to explore the maze for 15 rain on 3 consecutive days. On Day 1 food pellets (94 rag, Bio-Serv, Frenchtown, N J) were scattered over the entire surface of all eight arms. On Day 2, pellets were scattered at the end of five of the eight arms. The five baited arms for all rats were the same relative to room cues. However, the maze was rotated 90° between the testing of each rat to prevent the use of odor cues on a given day and the use of any intramaze cues over days. On Day 3, pellets were placed in the recessed food holes of the five baited arms. For training, rats were given 1 trial each day during which a single pellet was placed in the food hole of the 5 baited arms. A rat was placed on the center platform facing a randomly selected arm and allowed to make arm choices until either all 5 pellets were taken, a total of 16 choices were made, or 10 minutes elapsed. Immediately after a trial the rat was returned to its home cage and given its additional food for the day. During a trial, an initial entry of a baited arm was considered a correct choice, an initial entry of an unbaited arm was considered a reference memory error, and re-entry of a previously chosen baited arm was considered a working memory error. Thus, a rat could make a reference error or a working error. Rats received a total of 95 trials. Trials 1-20 were not included in the behavioral analysis because all rats were not making 16 choices or obtaining the 5 pellets within 10 minutes. By trial 20, all rats were completing their trial within 10 minutes. Rats were assigned a code number at the time of transient forebrain ischemic injury so that behavioral training was blind. All trials were between 7 a.m. and 11 a.m.
Neuropatholoj4ic Procedure After a 1 month period of recovery from transient ischemia and 98 days of behavioral training, all PI and sham operated rats were decapitated. Their brains were rapidly removed from the skull and frozen in Freon-12 chilled to -70°C in dry ice as described previously [18]. Coronal sections (20/zm thick) were taken at the approximate level of the anterior commissure (Bregma - 0 . 3 ram), anterior hippocampus (Bregma - 3 . 3 ram) and posterior hippocampus (Bregma - 5 . 3 ram), and were stained with hematoxylin and eosin. Regional tissue damage was graded with a light microscope by two of the authors without knowledge of the experimental conditions. Ischemic neuronal damage was graded on a scale of 0-3 with 0=normal brain, l = a few neurons damaged (as few as one neuron damaged), 2 = m a n y neurons damaged, 3=majority of neurons damaged. A neuropathologic score for each region was calculated by summing the grade of damage for both hemispheres of rats within a group and dividing by the number of rats to obtain a mean grade of damage.
Behavioral Apparatus and Procedure Rats were tested in a radial 8-arm maze described previously [15,24]. Briefly, each a n n (60x 10 cm) projected from a center platform (65 cm wide) and had clear Plexiglas rails (20 cm high) on all sides. A recessed food hole (3.4 cm in diame-
RESULTS
Neuropathology The distribution and severity of ischemic neuronal injury are presented in Table 1.
MEMORY A F T E R T R A N S I E N T F O R E B R A I N I S C H E M I A TABLE 1 SEVERITY AND DISTRIBUTION OF ISCHEMIC BRAIN DAMAGE Mean (± SD) Neuropathologic Score Post Ischemic n = 10
Non-Criterion Shem n=6
Sham Operated n = 13
0 0
0 0
0 0
Anterior Hippocampus CA-I CA-2 CA-3
2.95 + 0.22 0 0
0 0 0
0 0 0
Posterior Hippocampus Superior Anterior Subiculum CA-I CA-2 CA-3
1.38 --- 1.06 1.78 ± 0.84 0 0
0 0 0 0
0 0 0 0
Striatum Thalamus
1.77 ± 0.93 0.1 ± 0.3
0.75 ± 1.2 0
0 0
Neocortex Anterior Posterior
The five month survival period after forebrain ischemia allowed sufficient time for astrocytic and microglial reactions to obscure lesser grades of neuronal damage. F o r example, microvacuolization, shrunken neurons with incrustation, or homogenizing cell change, typical transition stages that characterize ischemic injury were not present [2]. Therefore, quantification of neuronal damage in this study was accurate only in brain regions showing moderate to severe injury. Brain sections from the anterior hippocampal section (Bregma - 3 . 3 mm) of a control rat and PI rat are shown in Fig. 1A and 2A, respectively. These sections show the selective loss in the anterior CA1 hippocampal neurons in PI rats. Loss of CA1 neurons appeared complete in anterior hippocampal secions. Moderate to severe damage to the CA1 region extended to the mid-dorsal posterior hippocampus (Bregma - 5 . 3 mm), but, no PI animal had lateral dorsal posterior CA 1 injury. Mild to moderate damage to the subiculum was confined to the mid-dorsal posterior hippocampus. Behavior Individual trials were averaged into blocks of five trials, and then a multiple analysis o f variance (MANOVA) with repeated measures using group and blocks of trials as independent variables was performed on each type of error. The scores o f control and noncriterion sham rats were pooled because there was no detectable behavioral difference, F(I, 17)< 1.0, p > 0 . 5 , and noncriterion shams showed no hippocampal pathology. The mean number of reference errors and working errors in blocks of 5 trials are shown in Fig. 3 and Fig. 4, respectively. Control rats (combined control and noncriterion shams, N = 19) and PI rats ( N = 10) demonstrated improved reference performance over trials, F(14,378)=17.70, p<0.001. The PI
389 rats, however, made significantly more reference errors than control rats, F(1,27)=6.85, p<0.025. There was a significant group × trial interaction, F(14,378)=3.40, p<0.001. Similarly, PI rats and control rats showed significantly improved working memory performance over trials, F(14,378)=9.51, p<0.001, and PI rats made more working errors than control rats, F(1,27)=4.13, p<0.05. There was an interaction between group and trial, F(14,378)=2.17, p<0.01. The significant group x trial interaction for both reference and working performance is likely due to the initial poor performance of the PI rats. In later trials, the difference between PI rats and control rats appears small for working performance, and there appears to be no difference for reference performance (see Figs. 3 and 4). To examine the possibility that reference and working memory performance recovered, a M A N O V A with repeated measures was performed on the last 20 trials. No significant difference was detected between PI rats and control rats for reference performance, F(1,27)<1.0, p>0.60, and no significant trial or group × trial effect was detected, F(3,81)<1.0, p>0.50; F(3,81)<1.0, p>0.60, respectively. Thus, both PI and control rats were performing maximally on the reference aspect of this task. For working performance, there was a tendency for PI rats to make more working errors than control rats, F(I,27)=3.94, p<0.06. However, there was no significant effect over trials, F(3,81)<1.0, p>0.50, and there was no group × trial interaction, F(3,81)<1.0, p>0.90, suggesting that working performance would not further improve.
DISCUSSION
This study confirms that reference memory recovers in PI rats with CA 1 hippocampal damage [24]. That is, there was no difference in the performance of PI rats and control rats during their last 20 training trials, a point in training when reference performance of all rats was maximal. These results are also consistent with a previous finding of a small but persistent deficit in working memory [24]. Working performance during the last 20 trials was maximal for both PI and control rats as indicated by the lack of an interaction or an effect over trials. However, there was a tendency for PI rats to make more working errors than control rats fp<0.06). The behavior of PI rats on a radial maze is similar in several respects to that observed in rats with focal hippocampal damage caused by aspiration or neurotoxin lesions. First, like rats with focal hippocampal damage, PI rats demonstrate impaired working memory when they are not pretrained. For example, Jarrard [10] has demonstrated in rats impaired acquisition of working memory after aspiration of CA1. Handelmann and Olton [6] reported similar results for rats subjected to kainic acid lesion of CA3. Secondly, rats pretrained and then subjected to focal hippocampal damage have normal retention of reference memory. Rats pretrained on a radial maze prior to either CAl-alveus lesions [9] or CA3 lesions [8] demonstrated normal retention of reference memory when returned to the maze. Similarly, we reported preliminary results showing that PI rats trained preoperatively in an 8-arm maze with 5 baited arms, and then subjected to ischemia, performed normally on the reference aspect of the task when returned to the maze after 1 month of recovery [23]. Thus, like rats with kainic acid or aspiration lesions of hippocampus, PI rats have impaired working memory during acquisition, but when pretrained show normal retention of the reference component of the radial arm maze.
390
D A V I S E1 A I
FIG. 1 IA) Photomicrograph of normal rat brain demonstrating coronal section through the anterior hippocampus. I B) Inset indicates normal pyramidal neurons. H and E, original magnification × 10 for 1A, ;,<400 for lB.
FIG. 2. (A) Photomicrograph of post ischemic rat brain demonstrating coronal section throughthe anterior hippocampus. (B) Note the absence of pyramidal neurons. Inset indicates pyramidal cell loss, original magnification x 10 for 2A, x400 for 2B.
MEMORY AFTER T R A N S I E N T FOREBRAIN ISCHEMIA
391
WORKING ERRORS
REFERENCE ERRORS c o n t ro I
iI, Mean
\
iI
Reference
con tro I
i schemi c
!
Errors
2.0
Mean W o r k i n g
• t rake
Errors
i
" z"*k
"°
! i
°°~ Blocks
of
Five
Trials
A A 2o A ;o ;5 & & A ¢5 ~o ;5 io & Blocks
of
Five
Trials
FIG. 3. Mean reference errors for PI (N=I0) and control rats (N= 19) in blocks of 5 trials. The SEM for PI rats ranged between 0.12 and 0.37, and for control rats between 0.09 and 0.18.
FIG. 4. Mean working errors for PI (N= 10) and control rats ( N : 19) in blocks of 5 trials. The SEM for P1 rats ranged between 0.19 and 0.53, and for control rats between 0.04 and 0.16.
In contrast to these similarities, the initial acquisition of reference memory is markedly delayed in PI rats first trained on the radial maze postoperatively. Olton et al. [13] suggest that reference memory is spared in rats with damage to the hippocampal system. Indeed, if rats are pretrained prior to fimbria fornix transection, reference memory is impaired in the early postoperative trials compared to controls, but rapidly recovers [14]. The improvement in our PI animals is slower and occurs over 65 trials. However, this impairment of reference memory does not appear to be a nonspecific performance effect that follows ischemic injury. We have shown that animals pretrained on a reference task, and then subjected to ischemic injury had reference memory comparable to control animals when returned to the maze after recovery from ischemic injury [23]. It may be that ischemic injury has a more severe effect on acquisition of reference memory than fimbria fornix section. That is, ischemic injury in our hands appears to have a more disruptive effect on acquisition than on retention of already learned information, particularly if the to-be-remembered material does not vary over trials. In contrast, the retention of variable working memory material is disrupted by either fimbria fornix section or ischemia-induced CAI hippocampal damage.
Like post arrest human amnesics, PI rats have impaired learning of new information (both reference and working) and permanent impairment of the task specific information that varies from trial-to-trial (working memory). Thus, rats with CAI hippocampal damage following transient ischemia may provide a model for the cognitive disturbances that occur after global ischemic injury in humans. This animal model has functional impairments and morphologic injury that is similar to those that occur in humans after global ischemic insult [4,22]. It is possible that the mechanism of neuronal damage might also be similar. Thus, in addition to providing a model of the functional and morphologic consequences of cardiac arrest in humans, PI rats may provide a useful model for testing the effects of pharmacologic agents on these sequelae of ischemic insult.
ACKNOWLEDGEMENTS
This research was supported by funds from the Burke Foundation, USPHS Grant N503346, and a Grant-in-Aid (851098) from the American Heart Association to B.T.V. We thank Ellen Wright for data analysis.
REFERENCES 1. Bedell, S. E., T. L. Delbance, E. F. Cook and F. H. Epstein. Survival after cariopulmonary resuscitation in the hospital. N Engl J Med 309: 569-576, 1983. 2. Brierley, J. B. and D. I. Graham. Hypoxia and vascular disorders of the central nervous system. In: Greenfield's Neuropathology, edited by J. H. Adams, J. A. N. Corsellis and L. W. Duchen. London: Arnold, Ltd., 1984, pp. 125-208. 3. Caronna, J. J. Diagnosis, prognosis and treatment of hypoxic coma. In: Advances in Neurology, vol 26, edited by S. Fahn, S. Davis and L. P. Rowland, Jr. New York: Raven Press, 1979, pp. 1-19. 4. Cummings, J. L , U. Tomiyasu, S. Read and D. Benson. Amnesia with hippocampal lesions after cardiopulmonary arrest. Neurology 334: 679-681, 1984.
5. Earnest, M. P., P. R. Yarnell, S. L. Merrill and G. L. Knapp. Long term survival and neurologic status after resuscitation from out of hospital cardiac arrest. Neurology 30: 1298--1302, 1980. 6, Handelmann, G. E. and D. S. Olton. Spatial memory following damage to hippocampal CA3 pyramidal cells with kainic acid: Impairment and recovery with preoperative training. Brain Res 217: 41-58, 1981. 7. Hirst, W. and B. T. Volpe. Cognitive processes in the neurologic patient--Automatic and effortful encoding with amnesia. In: Handbook of Cognitive Neuroscience, edited by M. S. Gazzaniza. New York: Plenum Press, 1984, pp. 369-386. 8. Jarrard, L. E. Selective hippocampal lesions and behavior: Effects of kainic acid lesions on performance of place and cue tasks. Behav Neurosci 97: 873-889, 1983.
392
9. Jarrard, L. E. Selective hippocampal lesions and spatial discrimination in the rat. Soc Neurosci Abstr 4: 222, 1978. 10. Jarrard, L. E. Selective hippocampal lesions: Differential effects on performance by rats of a spatial task with preoperative versus postoperative training. J Comp Physiol Psychol 92; 111%1127, 1978. 11. Johansen, F. F., M. B. Jorgensen, D. K. G. Ekstrom von Lubitz and N. H. Diemer. Selective dendrite damage in hippocampal CA1 stratum radiatum with unchanged axon ultrastructure and glutamate uptake after transient cerebral ischemia in the rat. Brain Res 291; 373-377, 1984. 12. Longstreth, W. T., T. S. lnui, L. A. Cobb and M. Copass. Neurologic recovery after out-of-hospital cardiac arrest. Attn Intern Med 95: 580-592, 1983. 13. Olton, D. S., J. T. Becker and G. E. Handelmann. Hippocampus, space, and memory. Behav Brain Sci 2: 313-365, 1979. 14. Olton, D. S. and B. C. Papas. Spatial memory and hippocampal function. Neuropsychologia 17: 66%682, 1979. 15. Olton, D. S. and R. J. Samuelson. Remembrance of places passed: Spatial memory in rats. J Exp Psychol [Anita Behav] 2: 97-116, 1976. 16. Pulsinelli, W. A. and J. B. Brierley. A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke 10: 267-272, 1979. 17. Pulsinelli. W. A., J. B. Brierley and F. Plum. Temporal profile of neuronal damage in a model of transient forebrain ischemia. Ann Neurol 11: 491-498, 1982.
DAVIS ET AL
18. Pulsinelli, W. A., D. E. Levy and T. E. Duffy. Regional cerebral blood flow and glucose metabolism lollowing transient forebrain ischemia. Ann Neurol 11: 499-5(19, 1982. 19. Pulsinelli, W. A. and T. E. Duffy. Regional energy balance in rat brain after transient forebrain ischemia. ,/Neurochem I 1: 499509, 1982. 20. Volpe, B. T. and W. Hirst. rl~he characterization of an amnesic syndrome following hypoxic ischemic injury. Arch Neurol 40: 436-440, 1983. 21. Volpe, B. T., J. Holtzman and W. Hirst. Further characterization of patients with amnesia after cardiac arrest: Perserved recognition memory. Neurology 35: 1793-1797, 1985. 22. Volpe, B. T. and C. K, Petito. Dementia with bilateral medial temporal lobe ischemia. Neurology. in press. 23. Volpe, B. T., W. A. Pulsinelli and H. P. Davis. Amnesia in humans and animals after ischemic cerebral injury. Ann N Y Acad Sci 444: 492-493, 1985. 24. Volpe. B. T., W. A. Pulsinelli, J. Tribuna and H. P. Davis. Behavioral performance of rats following transient forebrain ischemia. Strol, e 15: 558-562, 1984. 25. Zola-Morgan, S., L. R. Squire and D. G. Amaral. Human amnesia and the medial temporal region: Memory impairment following a bilateral lesion limited to the CAI field of the hippocampus. Soc Neuro~sci Ahstr I1: 459, 1985.
0031-9384/86$3.00 + .00
Reference and Working Memory of Rats Following Hippocampal Damage Induced by Transient Forebrain Ischemia H A S K E R P. D A V I S
Department o f Psychology, University o f Colorado, Colorado Springs, CO 80933 JOSEPH TRIBUNA
School o f Pharmacy, St. John's University, Jamaica, N Y 11439 W I L L I A M A. P U L S I N E L L I
A N D B R U C E T. V O L P E
Department of Neurology, Cornell University Medical College 1300 York Avenue, New York, N Y 10021 R e c e i v e d 1 N o v e m b e r 1985 DAVIS, H. P., J. TR1BUNA, W. A. PULS1NELLI AND B. T. VOLPE. Reference and workingmemory of ratsfollowing hippocampal damage induced by transientforebrain ischemia. PHYSIOL BEHAV 37(3) 387-392, 1986.--Acquisition of reference and working memory was evaluated in an animal model of cerebral ischemia. Rats were subjected to 30 minutes of transient forebrain ischemia, allowed to recover, and then tested for 95 trials on an 8-arm maze with 5 arms baited. During the 95 trials post ischemic (PI) rats made significantly more working and reference errors than controls (.o<0.05). However, an analysis of the last 20 trials (75-95) showed that while PI rats and control rats had comparable reference memory (p>0.8). P1 rats tended to have a persistent working memory deficit compared to controls (p<0.06). Subsequent morphologic analysis showed that PI rats had almost complete loss of pyramidal neurons in the anterior CA 1 region of the hippocampus, moderate to severe loss in mid-dorsal posterior hippocampus, and less damage to the dorsolateral striatum. These results suggest that the PI animal is a reasonable model for the permanent behavioral impairment and pathologic damage found in some human survivors of cardiac arrest. Amnesia Cardiac arrest Hippocampus Reference memory Working memory
Ischemia
CARDIAC arrest is a common medical problem, and although mortality has improved, sensorimotor deficits and cognitive disturbances may cause persistent morbidity in many of these survivors [1, 3, 5, 12]. Some survivors have demonstrated a fairly circumscribed amnesic syndrome characterized by impaired learning and memory of events after their injury, and a gradient of memory loss for premorbid events [3, 20, 21, 25]. Post mortem neuropathologic examination of patients with cardiac arrest has revealed marked loss of specific populations of neurons highly vulnerable to ischemia [2]. Post mortem studies of patients who had cardiac arrest and who also had ante mortem behavioral studies have shown the most severe damage in hippocampus, particularly the CA1 region [4,22]. We have been investigating an animal model that demonstrates some of the morphologic and the behavioral consequences of cerebral ischemia that follow cardiac arrest in humans. Investigators have shown that rats subjected to
Learning and memory
Radial maze
controlled transient occlusion of the major vessels to the brain develop reproducible and quantifiable morphologic brain injury that appears most severe in the CA1 region of the hippocampus [11, 16, 17, 19]. However, few studies have objectively addressed long-term functional changes in the behavior of animals exposed to global cerebral ischemia [23,24]. One of the advantages of this model is that the majority of animals subjected to 30 minutes of ischemia survive indefinitely, and it is therefore possible to characterize and quantify their behavior long after recovery from the acute insult. The present study replicates and extends a previous investigation to determine whether ischemic-induced neuronal damage in the rat, and in particular damage to CAI hippocampal neurons, results in permanent learning and memory impairments similar to humans [24]. A radial 8-arm maze task was used to assess behavior because it has been demonstrated that rats with hippocampal damage perform differently on two aspects of this task [13].
387
I)AVIS E7 AI~.
388 In the radial maze it is possible to measure both reference performance (entering baited arms only) and working performance (not re-entering a baited arm after the food is taken) by always baiting the same 5 arms of the 8-arm maze on all trials. That is, reference performance requires the animal to learn that 5 of 8 arms are baited, and that these 5 baited arms remain contant relative to room cues for all trials. Working performance requires the rat to remember from which arms food has been taken so as not to re-enter that arm during a particular trial. The reference memory component includes many repetitions of invarant material that is useful over any and all trials, while the working memory component requires the retention of trial specific information that varies across trials. If post ischemic (PI) rats showed dissociable reference and working memory, then these results would suggest similarities to post cardiac arrest amnesia in which patients have deficits in remembering new and/or variable information [7].
METHOD
Subjects Male Wistar rats (Hilltop, Scottsdale, PA) weighing between 250-300 grams served as subjects. Animals were individually housed in polyurethane cages (33×21x19 cm) throughout the experiment and provided ad lib access to water.
Surgical Procedure Rats were subjected to forebrain ischemia by a method described in detail previously [16]. Briefly, on Day 1 rats were anesthetized by the inhalation of 2.5% halothane mixed with 30% oxygen and 70% nitrogen. Atraumatic clasps were placed loosely around each common carotid artery without interrupting blood flow. The vertebral arteries were permanently occluded by electrocautery at the first cervical vertebra. Animals were allowed 24 hours of recovery. After recovery, operated animals are indistinguishable from control animals by electrocephalographic, physiologic, and subjective behavioral criteria [16]. On Day 2, forebrain ischemia was produced in awake rats by tightening the carotid artery clasps. Animals that did not become unresponsive within 1 minute were classified as noncriterion shams. The carotid artery clasps were released after 30 minutes of occlusion. Body temperature was maintained at a minimum of 37°C by a heating lamp connected to a rectal thermistor until thermal homeostasis was restored. Control rats were subjected to halothane anesthesia and skin incisions, Previous behavioral assessments detected no differences between controls and vertebral cautery control rats exposed to halothane and occlusion of the vertebral arteries [24]. Animals were allowed 30 days of postoperative recovery. After 2-3 weeks there was no weight difference between control rats and operated rats. It was not possible to distinguish operated from unoperated rats by observation of feeding, grooming, or exploring habits.
ter) was at the end of each arm. The test room was rich in stationary extramaze stimuli. Thirty days postoperatively the PI and control rats were started on a partial food deprivation schedule and by trial 15 all animals bad reached 80e/ of their ad lib weight. Maze adaptation was started after 1 day of food deprivation. Individual rats were allowed to explore the maze for 15 rain on 3 consecutive days. On Day 1 food pellets (94 rag, Bio-Serv, Frenchtown, N J) were scattered over the entire surface of all eight arms. On Day 2, pellets were scattered at the end of five of the eight arms. The five baited arms for all rats were the same relative to room cues. However, the maze was rotated 90° between the testing of each rat to prevent the use of odor cues on a given day and the use of any intramaze cues over days. On Day 3, pellets were placed in the recessed food holes of the five baited arms. For training, rats were given 1 trial each day during which a single pellet was placed in the food hole of the 5 baited arms. A rat was placed on the center platform facing a randomly selected arm and allowed to make arm choices until either all 5 pellets were taken, a total of 16 choices were made, or 10 minutes elapsed. Immediately after a trial the rat was returned to its home cage and given its additional food for the day. During a trial, an initial entry of a baited arm was considered a correct choice, an initial entry of an unbaited arm was considered a reference memory error, and re-entry of a previously chosen baited arm was considered a working memory error. Thus, a rat could make a reference error or a working error. Rats received a total of 95 trials. Trials 1-20 were not included in the behavioral analysis because all rats were not making 16 choices or obtaining the 5 pellets within 10 minutes. By trial 20, all rats were completing their trial within 10 minutes. Rats were assigned a code number at the time of transient forebrain ischemic injury so that behavioral training was blind. All trials were between 7 a.m. and 11 a.m.
Neuropatholoj4ic Procedure After a 1 month period of recovery from transient ischemia and 98 days of behavioral training, all PI and sham operated rats were decapitated. Their brains were rapidly removed from the skull and frozen in Freon-12 chilled to -70°C in dry ice as described previously [18]. Coronal sections (20/zm thick) were taken at the approximate level of the anterior commissure (Bregma - 0 . 3 ram), anterior hippocampus (Bregma - 3 . 3 ram) and posterior hippocampus (Bregma - 5 . 3 ram), and were stained with hematoxylin and eosin. Regional tissue damage was graded with a light microscope by two of the authors without knowledge of the experimental conditions. Ischemic neuronal damage was graded on a scale of 0-3 with 0=normal brain, l = a few neurons damaged (as few as one neuron damaged), 2 = m a n y neurons damaged, 3=majority of neurons damaged. A neuropathologic score for each region was calculated by summing the grade of damage for both hemispheres of rats within a group and dividing by the number of rats to obtain a mean grade of damage.
Behavioral Apparatus and Procedure Rats were tested in a radial 8-arm maze described previously [15,24]. Briefly, each a n n (60x 10 cm) projected from a center platform (65 cm wide) and had clear Plexiglas rails (20 cm high) on all sides. A recessed food hole (3.4 cm in diame-
RESULTS
Neuropathology The distribution and severity of ischemic neuronal injury are presented in Table 1.
MEMORY A F T E R T R A N S I E N T F O R E B R A I N I S C H E M I A TABLE 1 SEVERITY AND DISTRIBUTION OF ISCHEMIC BRAIN DAMAGE Mean (± SD) Neuropathologic Score Post Ischemic n = 10
Non-Criterion Shem n=6
Sham Operated n = 13
0 0
0 0
0 0
Anterior Hippocampus CA-I CA-2 CA-3
2.95 + 0.22 0 0
0 0 0
0 0 0
Posterior Hippocampus Superior Anterior Subiculum CA-I CA-2 CA-3
1.38 --- 1.06 1.78 ± 0.84 0 0
0 0 0 0
0 0 0 0
Striatum Thalamus
1.77 ± 0.93 0.1 ± 0.3
0.75 ± 1.2 0
0 0
Neocortex Anterior Posterior
The five month survival period after forebrain ischemia allowed sufficient time for astrocytic and microglial reactions to obscure lesser grades of neuronal damage. F o r example, microvacuolization, shrunken neurons with incrustation, or homogenizing cell change, typical transition stages that characterize ischemic injury were not present [2]. Therefore, quantification of neuronal damage in this study was accurate only in brain regions showing moderate to severe injury. Brain sections from the anterior hippocampal section (Bregma - 3 . 3 mm) of a control rat and PI rat are shown in Fig. 1A and 2A, respectively. These sections show the selective loss in the anterior CA1 hippocampal neurons in PI rats. Loss of CA1 neurons appeared complete in anterior hippocampal secions. Moderate to severe damage to the CA1 region extended to the mid-dorsal posterior hippocampus (Bregma - 5 . 3 mm), but, no PI animal had lateral dorsal posterior CA 1 injury. Mild to moderate damage to the subiculum was confined to the mid-dorsal posterior hippocampus. Behavior Individual trials were averaged into blocks of five trials, and then a multiple analysis o f variance (MANOVA) with repeated measures using group and blocks of trials as independent variables was performed on each type of error. The scores o f control and noncriterion sham rats were pooled because there was no detectable behavioral difference, F(I, 17)< 1.0, p > 0 . 5 , and noncriterion shams showed no hippocampal pathology. The mean number of reference errors and working errors in blocks of 5 trials are shown in Fig. 3 and Fig. 4, respectively. Control rats (combined control and noncriterion shams, N = 19) and PI rats ( N = 10) demonstrated improved reference performance over trials, F(14,378)=17.70, p<0.001. The PI
389 rats, however, made significantly more reference errors than control rats, F(1,27)=6.85, p<0.025. There was a significant group × trial interaction, F(14,378)=3.40, p<0.001. Similarly, PI rats and control rats showed significantly improved working memory performance over trials, F(14,378)=9.51, p<0.001, and PI rats made more working errors than control rats, F(1,27)=4.13, p<0.05. There was an interaction between group and trial, F(14,378)=2.17, p<0.01. The significant group x trial interaction for both reference and working performance is likely due to the initial poor performance of the PI rats. In later trials, the difference between PI rats and control rats appears small for working performance, and there appears to be no difference for reference performance (see Figs. 3 and 4). To examine the possibility that reference and working memory performance recovered, a M A N O V A with repeated measures was performed on the last 20 trials. No significant difference was detected between PI rats and control rats for reference performance, F(1,27)<1.0, p>0.60, and no significant trial or group × trial effect was detected, F(3,81)<1.0, p>0.50; F(3,81)<1.0, p>0.60, respectively. Thus, both PI and control rats were performing maximally on the reference aspect of this task. For working performance, there was a tendency for PI rats to make more working errors than control rats, F(I,27)=3.94, p<0.06. However, there was no significant effect over trials, F(3,81)<1.0, p>0.50, and there was no group × trial interaction, F(3,81)<1.0, p>0.90, suggesting that working performance would not further improve.
DISCUSSION
This study confirms that reference memory recovers in PI rats with CA 1 hippocampal damage [24]. That is, there was no difference in the performance of PI rats and control rats during their last 20 training trials, a point in training when reference performance of all rats was maximal. These results are also consistent with a previous finding of a small but persistent deficit in working memory [24]. Working performance during the last 20 trials was maximal for both PI and control rats as indicated by the lack of an interaction or an effect over trials. However, there was a tendency for PI rats to make more working errors than control rats fp<0.06). The behavior of PI rats on a radial maze is similar in several respects to that observed in rats with focal hippocampal damage caused by aspiration or neurotoxin lesions. First, like rats with focal hippocampal damage, PI rats demonstrate impaired working memory when they are not pretrained. For example, Jarrard [10] has demonstrated in rats impaired acquisition of working memory after aspiration of CA1. Handelmann and Olton [6] reported similar results for rats subjected to kainic acid lesion of CA3. Secondly, rats pretrained and then subjected to focal hippocampal damage have normal retention of reference memory. Rats pretrained on a radial maze prior to either CAl-alveus lesions [9] or CA3 lesions [8] demonstrated normal retention of reference memory when returned to the maze. Similarly, we reported preliminary results showing that PI rats trained preoperatively in an 8-arm maze with 5 baited arms, and then subjected to ischemia, performed normally on the reference aspect of the task when returned to the maze after 1 month of recovery [23]. Thus, like rats with kainic acid or aspiration lesions of hippocampus, PI rats have impaired working memory during acquisition, but when pretrained show normal retention of the reference component of the radial arm maze.
390
D A V I S E1 A I
FIG. 1 IA) Photomicrograph of normal rat brain demonstrating coronal section through the anterior hippocampus. I B) Inset indicates normal pyramidal neurons. H and E, original magnification × 10 for 1A, ;,<400 for lB.
FIG. 2. (A) Photomicrograph of post ischemic rat brain demonstrating coronal section throughthe anterior hippocampus. (B) Note the absence of pyramidal neurons. Inset indicates pyramidal cell loss, original magnification x 10 for 2A, x400 for 2B.
MEMORY AFTER T R A N S I E N T FOREBRAIN ISCHEMIA
391
WORKING ERRORS
REFERENCE ERRORS c o n t ro I
iI, Mean
\
iI
Reference
con tro I
i schemi c
!
Errors
2.0
Mean W o r k i n g
• t rake
Errors
i
" z"*k
"°
! i
°°~ Blocks
of
Five
Trials
A A 2o A ;o ;5 & & A ¢5 ~o ;5 io & Blocks
of
Five
Trials
FIG. 3. Mean reference errors for PI (N=I0) and control rats (N= 19) in blocks of 5 trials. The SEM for PI rats ranged between 0.12 and 0.37, and for control rats between 0.09 and 0.18.
FIG. 4. Mean working errors for PI (N= 10) and control rats ( N : 19) in blocks of 5 trials. The SEM for P1 rats ranged between 0.19 and 0.53, and for control rats between 0.04 and 0.16.
In contrast to these similarities, the initial acquisition of reference memory is markedly delayed in PI rats first trained on the radial maze postoperatively. Olton et al. [13] suggest that reference memory is spared in rats with damage to the hippocampal system. Indeed, if rats are pretrained prior to fimbria fornix transection, reference memory is impaired in the early postoperative trials compared to controls, but rapidly recovers [14]. The improvement in our PI animals is slower and occurs over 65 trials. However, this impairment of reference memory does not appear to be a nonspecific performance effect that follows ischemic injury. We have shown that animals pretrained on a reference task, and then subjected to ischemic injury had reference memory comparable to control animals when returned to the maze after recovery from ischemic injury [23]. It may be that ischemic injury has a more severe effect on acquisition of reference memory than fimbria fornix section. That is, ischemic injury in our hands appears to have a more disruptive effect on acquisition than on retention of already learned information, particularly if the to-be-remembered material does not vary over trials. In contrast, the retention of variable working memory material is disrupted by either fimbria fornix section or ischemia-induced CAI hippocampal damage.
Like post arrest human amnesics, PI rats have impaired learning of new information (both reference and working) and permanent impairment of the task specific information that varies from trial-to-trial (working memory). Thus, rats with CAI hippocampal damage following transient ischemia may provide a model for the cognitive disturbances that occur after global ischemic injury in humans. This animal model has functional impairments and morphologic injury that is similar to those that occur in humans after global ischemic insult [4,22]. It is possible that the mechanism of neuronal damage might also be similar. Thus, in addition to providing a model of the functional and morphologic consequences of cardiac arrest in humans, PI rats may provide a useful model for testing the effects of pharmacologic agents on these sequelae of ischemic insult.
ACKNOWLEDGEMENTS
This research was supported by funds from the Burke Foundation, USPHS Grant N503346, and a Grant-in-Aid (851098) from the American Heart Association to B.T.V. We thank Ellen Wright for data analysis.
REFERENCES 1. Bedell, S. E., T. L. Delbance, E. F. Cook and F. H. Epstein. Survival after cariopulmonary resuscitation in the hospital. N Engl J Med 309: 569-576, 1983. 2. Brierley, J. B. and D. I. Graham. Hypoxia and vascular disorders of the central nervous system. In: Greenfield's Neuropathology, edited by J. H. Adams, J. A. N. Corsellis and L. W. Duchen. London: Arnold, Ltd., 1984, pp. 125-208. 3. Caronna, J. J. Diagnosis, prognosis and treatment of hypoxic coma. In: Advances in Neurology, vol 26, edited by S. Fahn, S. Davis and L. P. Rowland, Jr. New York: Raven Press, 1979, pp. 1-19. 4. Cummings, J. L , U. Tomiyasu, S. Read and D. Benson. Amnesia with hippocampal lesions after cardiopulmonary arrest. Neurology 334: 679-681, 1984.
5. Earnest, M. P., P. R. Yarnell, S. L. Merrill and G. L. Knapp. Long term survival and neurologic status after resuscitation from out of hospital cardiac arrest. Neurology 30: 1298--1302, 1980. 6, Handelmann, G. E. and D. S. Olton. Spatial memory following damage to hippocampal CA3 pyramidal cells with kainic acid: Impairment and recovery with preoperative training. Brain Res 217: 41-58, 1981. 7. Hirst, W. and B. T. Volpe. Cognitive processes in the neurologic patient--Automatic and effortful encoding with amnesia. In: Handbook of Cognitive Neuroscience, edited by M. S. Gazzaniza. New York: Plenum Press, 1984, pp. 369-386. 8. Jarrard, L. E. Selective hippocampal lesions and behavior: Effects of kainic acid lesions on performance of place and cue tasks. Behav Neurosci 97: 873-889, 1983.
392
9. Jarrard, L. E. Selective hippocampal lesions and spatial discrimination in the rat. Soc Neurosci Abstr 4: 222, 1978. 10. Jarrard, L. E. Selective hippocampal lesions: Differential effects on performance by rats of a spatial task with preoperative versus postoperative training. J Comp Physiol Psychol 92; 111%1127, 1978. 11. Johansen, F. F., M. B. Jorgensen, D. K. G. Ekstrom von Lubitz and N. H. Diemer. Selective dendrite damage in hippocampal CA1 stratum radiatum with unchanged axon ultrastructure and glutamate uptake after transient cerebral ischemia in the rat. Brain Res 291; 373-377, 1984. 12. Longstreth, W. T., T. S. lnui, L. A. Cobb and M. Copass. Neurologic recovery after out-of-hospital cardiac arrest. Attn Intern Med 95: 580-592, 1983. 13. Olton, D. S., J. T. Becker and G. E. Handelmann. Hippocampus, space, and memory. Behav Brain Sci 2: 313-365, 1979. 14. Olton, D. S. and B. C. Papas. Spatial memory and hippocampal function. Neuropsychologia 17: 66%682, 1979. 15. Olton, D. S. and R. J. Samuelson. Remembrance of places passed: Spatial memory in rats. J Exp Psychol [Anita Behav] 2: 97-116, 1976. 16. Pulsinelli, W. A. and J. B. Brierley. A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke 10: 267-272, 1979. 17. Pulsinelli. W. A., J. B. Brierley and F. Plum. Temporal profile of neuronal damage in a model of transient forebrain ischemia. Ann Neurol 11: 491-498, 1982.
DAVIS ET AL
18. Pulsinelli, W. A., D. E. Levy and T. E. Duffy. Regional cerebral blood flow and glucose metabolism lollowing transient forebrain ischemia. Ann Neurol 11: 499-5(19, 1982. 19. Pulsinelli, W. A. and T. E. Duffy. Regional energy balance in rat brain after transient forebrain ischemia. ,/Neurochem I 1: 499509, 1982. 20. Volpe, B. T. and W. Hirst. rl~he characterization of an amnesic syndrome following hypoxic ischemic injury. Arch Neurol 40: 436-440, 1983. 21. Volpe, B. T., J. Holtzman and W. Hirst. Further characterization of patients with amnesia after cardiac arrest: Perserved recognition memory. Neurology 35: 1793-1797, 1985. 22. Volpe, B. T. and C. K, Petito. Dementia with bilateral medial temporal lobe ischemia. Neurology. in press. 23. Volpe, B. T., W. A. Pulsinelli and H. P. Davis. Amnesia in humans and animals after ischemic cerebral injury. Ann N Y Acad Sci 444: 492-493, 1985. 24. Volpe. B. T., W. A. Pulsinelli, J. Tribuna and H. P. Davis. Behavioral performance of rats following transient forebrain ischemia. Strol, e 15: 558-562, 1984. 25. Zola-Morgan, S., L. R. Squire and D. G. Amaral. Human amnesia and the medial temporal region: Memory impairment following a bilateral lesion limited to the CAI field of the hippocampus. Soc Neuro~sci Ahstr I1: 459, 1985.