Psychopharmacology

Psychopharmacology (1987) 93: 36~3

© Springer-Verlag 1987

Fluoxetine enhances memory processing in mice J.F. Flood t and A. Cherkin 2

x Psychobiology Research Laboratory and Geriatric Research, Education and Clinical Center (GRECC), 151A2, Veterans Administration Hospital, Sepulveda, CA 91343, USA 2 Department of Psychiatry and Biobehavioral Sciences, School of Medicine, University of California, Los Angeles, CA 90024, USA

Abstract. Fluoxetine (FLU) increases brain concentrations

of serotonin by blocking its uptake, without appreciably affecting the dopamine or norepinephrine systems. The present experiments provide evidence that a subcutaneous injection of F L U enhanced post-memory processing (" consolidation") and retrieval, but not acquisition in young adult mice. F L U (15 mg/kg) enhanced 1-week memory retention when injected 2 min post-training. Similar enhancement was obtained with intracerebroventricular injection (20 Ixg per mouse). F L U enhanced retention when administered prior to training (l 5 mg/kg). F L U (2.5 mg/kg) enhanced recall scores when injected 1 h before the 1-week retention test, indicating an enhancing effect on memory retrieval. Neither the pre-training nor pre-testing effects depended on improved acquisition, since F L U did not improve acquisition of T-maze foot-shock avoidance over the dose range 0.5-35 mg/kg. The sensitive period for posttraining enhancement by F L U (15 mg/kg) was less than 90 rain, as shown by the temporal gradient typical of memory-enhancing drugs. The amnesia induced by a protein synthesis inhibitor anisomycin, or by an anticholinergic drug scopolamine, was blocked by F L U (15 mg/kg) injected post-training. Finally, F L U (15 mg/kg) injected after onetrial passive avoidance training enhanced l-week retention, demonstrating effectiveness in this task as well as in the active avoidance task. Key words: Fluoxetine - Memory - Mice - Recall - Reten-

tion

Scopolamine

Serotonin

The role of serotonin in memory processing is not well established, due to the lack of compounds that specifically affect serotonin rather than dopamine and norepinephrine transmitter systems. Fluoxentine (FLU) is one of a few compounds that with a relatively high degree of specificity blocks serotonin uptake without significantly affecting dopamine or norepinephrine (Fuller et al. 1974, 1975; Wong et al. 1975). F L U does not bind to dopamine, GABA hydroxybutyric acid, acetylcholine, histamine, opiate, or benzodiazepine receptors (Wong 1983). The primary evidence suggesting that serotonin affects memory processing is that reserpine, which reduces brain Offprint requests to: J.F. Flood

serotonin for several days, impairs retention (Dismukes and Rake 1972; Rake 1973). However, an interpretation of reserpine-induced amnesia as demonstrating serotonin involvement in memory processing is clouded by its effects on other systems (i.e., dopamine and norepinephrine) and by reserpine-induced illness (unpublished observations). Administration of 5-hydroxytryptophan, a precursor of serotonin, increases brain levels of serotonin and acts as an anti-amnestic agent to reserpine (Dismukes and Rake 1972; Rake 1973). However, in these studies dopa and amphetamine also acted as anti-amnesties, indicating non-specificity of the 5-hydroxytryptophan effect. Drugs such as nicotine, caffeine, amphetamine, strychnine and picrotoxin have a reputation for improving memory because they improve memory retention under a variety of experimental conditions (McGaugh 1973; Ii'yuchenok 1976; Myers 1977). Other drugs, like corticosterone and ACTH, improve retention under enough conditions so that there is little doubt that they result in better memory retention test performance (Gold and Delanoy 1981; SoumireuMourat et al. 1981). We and others have found that the best of the memory improving drugs: (a) show a relationship between drug dose and degree of memory improvement; (b) have a time-dependent action such that the longer the drug is administered after training the less effectively it improves memory; (c) act as anti-amnesties; (d) improve memory in more than one task; (e) improve retention when administered before training, at a lower dose than when administered after training; (f) improve retention when administered just prior to a retention test; and (g) improve retention when administered centrally. The purpose of this study was to determine across a series of experiments how well F L U compares with drugs having established reputations for enhancing memory retention in animal models. Experiment 1. Effect of post-training subcutaneous administration of fluoxetine on memory retention

Materials ancl methods Subjects. After 1 week in the laboratory, CD-1 male mice obtained from Charles River Breeding Laboratories, Wilmington, MA, were individually caged 24-48 h prior to training and remained singly housed until retention was tested 1 week later. The median body weight was 35 g, with a

37 range o f 33 38 g. The mice were trained on a T-maze active avoidance task between 0700 and 1500 hours. The solutions o f F L U were p r e p a r e d with 70 gl 95% alcohol and sufficient distilled water to prepare the desired dilution; the highest concentration o f alcohol was 0.7%. The vehicle control was a 0.7% solution o f alcohol in saline. Fluoxetine hydrochloride ( F L U , F.W. 345.8) was generously d o n a t e d by Eli Lilly and C o m p a n y .

Apparatus. The T-maze was constructed o f a black plastic start alley with a start box at one end and two goal boxes at the other; a stainless steel r o d floor ran t h r o u g h o u t the maze ( F l o o d et al. 1974). Each goal box was fitted with a slotted plastic liner (the b o t t o m o f which went below the shock grid) which was used to remove the mice from the goal box without h a n d contact. The start box was separated from the start alley by a plastic guillotine d o o r which prevented the mouse from moving down the alley until the training started. The intertrial interval was 30 s with a muffled doorbell-type buzzer as the conditioned stimulus and footshock ( C o u l b o u r n Instruments Inc. M o d e l E10-08) set nominally at 0.30 Ma.

Training procedures. Mice were not permitted to explore the maze before training. A training trial started when a mouse was placed into the start box. The guillotine d o o r was raised and the buzzer sounded simultaneously, then 5 s later footshock was applied. The goal box that the mouse first entered on this trial was designated as " i n c o r r e c t " and the footshock was continued until the mouse entered the other goal box, which on all subsequent trials was designated " c o r r e c t " for the particular mouse. A t the end o f each trial, the mouse was removed from the goal box by lifting the plastic liner and the mouse was carefully returned to its home cage. A new trial began by placing the mouse in the start box, sounding the buzzer and raising the guillotine door, with footshock beginning 5 s later if the mouse did not move into its correct goal box. As training proceeded, a mouse m a d e one o f two types of responses. A response latency longer than 5 s was classed as an escape from the footshock. A response latency less than or equal to 5 s was considered an avoidance, since the mouse avoided receiving a footshock. Two exclusion criteria were applied to reduce learning v a r i a b i l i t y a m o n g mice, as follows. On the first training trials, mice with escape latencies greater than 20 s were discarded. Mice n o t having at least one errorless escape latency between 1.5 and 3.5 s on training trials 2 or 3 were excluded. The total exclusions were fewer than 20%. Mice received a 0.35 ml subcutaneous injection o f the vehicle or freshly p r e p a r e d drug solution within 2 rain after training. All solutions were blind-coded to eliminate experimenter bias. One week after training and drug administration, the T-maze training was resumed until the mice m a d e five avoidance responses in six consecutive training trials. Two parametric measures of m e m o r y retention were analyzed. The first measure is the m e a n n u m b e r o f trials to the first avoidance response for all mice within a group. The second measure was the mean number o f trials to reach the five out o f six criterion. The overall significance of the drug treatment effect was determined by a one-way analysis o f variance (Winer 1971; K e p p e l 1973). D u n n e t t ' s t-test was used to m a k e multiple comparisons between each drug groups and the control group (Winer 1971).

Table 1. Dose-response enhancement of t-maze footshock avoidance retention with post-training subcutaneous administration of fluoxetine Measure of retention

Dose of fluoxetine (mg/kg, SC) 0

2.5

5.0

Mean trials tst avoidance (SEM) Pvalue a

4.65 0.27

3.95 3.40 2.90 2.20 0.38 0.38 0.29 0.27 ns P<0.05 P<0.01 P<0.01

Mean trials to criterion (SEM) Pvalue"

8.95 0.27

8.15 7.60 7.05 6.20 0.37 0.35 0.29 0.26 ns P<0.05 P<0.01 P<0.0I

Recall score (%)

20

35

50

10.0

70

15.0

85

a p values are the statistical difference from control group (0 mg/ kg) as determined by Dunnett's t-test

A non-parametric measure of retention was derived from trials to first avoidance response to visualize better the effects of drug treatments on retention test performance and to correspond with usual reporting practice. F o r this, the n u m b e r of trials to the first avoidance response was dichotomized to yield a per cent recall score. Those mice m a k i n g their first avoidance in three trials or less were classed as remembering the original training. This criterion was a d o p t e d because it has provided optimal separation between the retention test performance o f naive mice (with no T-maze training) and well-trained mice ( F l o o d et al. 1975). M o s t drugs that are recognized as having the ability to improve m e m o r y retention do so when administered shortly after training. The purpose o f this experiment was to determine the range o f doses over which F L U i m p r o v e d retention (70% or greater recall score) c o m p a r e d to controls (25% recall score). The mice were trained in a T-maze as described above and received one o f four doses o f F L U (2.5, 5, 10 or 15 mg/kg, SC), or vehicle, within 2 min after training. This range o f doses was selected because it exerts serotinergic and behavioral effects (Fuller et al. 1978; G a r rigou et al. 1981; L o r d e n and N u n n 1982; M c E l r o y e t a l . 1982). The N per group was 20. The mice were tested 1 week after training and drug administration as described above.

Results As intended, the retention test performance o f the vehicle control group was poor, with only 20% recall. Higher recall scores by the fluoxetine-treated groups are reported in Table 1. A one-way analysis by A N O V A o f mean trials to first avoidance indicated a significant effect o f F L U on m e m o r y retention [F(4,95)=9.11, P < 0 . 0 0 1 ] as well as a significant effect based on mean trials to criterion [F(4,95) = 11.81, P < 0 . 0 0 1 ] . A further analysis of either mean trials to first avoidance response or to criterion using D u n n e t t ' s t-test showed that 2.5 mg/kg h a d no significant effect on retention, b u t 5, 10, and 15 mg/kg improved retention by either measure ( P < 0 . 0 5 , P < 0 . 0 1 , P < 0 . 0 1 , respectively). The descriptive statistics are given in Table 1.

38 [F(5,114)= 3.08, P<0.025]. A further analysis of the data by D u n n e t t ' s t-test on mean trials either to first avoidance response or to criterion showed that groups receiving F L U had significantly improved recall score at 8 gg and 16 gg ( P < 0.05) but 20 Ixg and 24 gg yielded the lowest mean trials to criterion (P<0.01). The descriptive statistics are given in Table 2.

Table 2. Dose-response enhancement of t-maze footshock avoidance retention with post-training intracerebroventricular administration of fluoxetine Measure of retention

Dose of fluoxetine (gg/brain, ICV) 0

4

8

16

20

24

Mean trials 1st avoidance (SEM) Pvalue a

4.40 0,22

3.55 0.33

3.35 0.30

3.40 0.34

2.90 0.33

2.95 0.34

Mean trials to criterion (SEN) Pvalue"

8.60 0.24

Recall score

15

(%)

ns 7.85 0.33 ns 40

Experiment 3. Effect of pre-training subcutaneous administration of flnoxetine on memory retention

P<0.05 P<0.05 P<0.01 P<0.01 7.45 0.33

7.45 0.34

7.10 0.33

We find that drugs which improve retention when administered after training do so at a lower dose when administered prior to training. To determine if this is also true for F L U , we administered 0.25, 0.5, 1.0, 2.5, 5.0, I0.0 mg/ kg, or the vehicle, 1 h prior to training. The mice were trained and tested as in experiment 1. The N per group was 20. The retention test was given 1 week after training.

7.20 0.35

P<0.05 P<0.05 P<0.01 P<0.01 55

55

75

70

a p values are the statistical difference from control group (0 mg/ kg) as determined by Dunnett's t-test

Results The A N O V A for mean trials to first avoidance response yielded a significant drug effect [F(6,133) = 6.59, P < 0.001 ] as well as the trials to criterion [F(6,133)= 8.24, P < 0.001]. Analysis of either mean trials to first avoidance response or to criterion indicated that F L U at 0.25 mg/kg did not significantly affect retention compared to the control group. The group receiving F L U at 0.5 mg/kg differed significantly from the control by either measure of retention (P<0.05). F L U at 1.0, 2.5 and 5.0 mg/kg differed significantly from the control group at P < 0 . 0 1 (Table 3). F L U between 1.0 and 5.0 mg/kg had the highest recall scores (75-80%), while 10 mg/kg was less facilitating (50% recall score). The difference in mean trials to criterion was not significantly different between F L U administered at 5.0 and 10.0mg/kg group.

Experiment 2. Effect of post-training intracerebroventricular administration of fluoxentine on memory retention The purpose of experiment 2 was to test if F L U can affect memory processing by acting directly on the central nervous system. Mice were trained and tested as in experiment 1. Twenty-four to forty-eight hours prior to training a single hole was drilled over the third ventricle ( - 0 . 5 m m relative to bregma, 0.5 right of the central suture) while mice were in a stereotaxic instrument under methoxyflurane (Metofane) anesthesia. Within 3 m i n after training, mice were anesthetized with enflurane, and received one of five doses of F L U in 2 gl vehicle or of the vehicle alone. The doses of F L U per mouse were: 4, 8, 16, 20 a n d 24 gg. The N per group was 20. The mice were tested 1 week after training.

Experiment 4. Dose-response improvement of recall score by pre-test administration of fluoxetine

Results

The purpose of experiment 4 was to test if F L U improved recall of information poorly stored in memory because of weak training. In previous studies, we found that the dose which is optimal for improving retention when administered prior to training is also effective at improving retention when administered prior to a retention test (Flood et al.

The vehicle control group showed poor retention (15% recall score) with fluoxetine-treated mice receiving higher recall scores, shown in Table 2. The A N O V A yielded a significant drug effect for mean trials to first avoidance response [F(5,114) = 3.14, P < 0.01 ] as well as mean trials to criterion

Table 3. Dose-response improvement of t-maze footshock avoidance retention with pretest subcutaneous administration of fluoxetine Measure of retention

Dose of fluoxetine (mg/kg, SC) 0

0.25

0.50

Mean trials 1st avoidance (SEM) P value a

5.00 0.32

3.65 0.41 ns

3.65 0.30 P < 0.05

2.85 0.25 P < 0.01

2.75 0.27 P < 0.0l

2.50 0.28 P < 0.01

3.70 0.41 P < 0.05

Mean trials to criterion (SEM) P value"

9.15 0.31

8.40 0.38 ns

7.85 0.31 P<0.05

7.20 0.23 P<0.05

7.00 0.29 P<0.01

6.60 0.27 P<0.01

8.00 0.38 P<0.05

Recall score (%)

20

45

50

1.0

75

2.50

75

P values are the statistical difference from control group (0 mg/kg) as determined by Dunnett's t-test

5.0

80

10.0

50

39

Table 4. Effect of pretest administration of fluoxetine on recall of mice trained to make an avoidance to the same or opposite side as in original training

Errors (SEM) P value" Mean trials tst avoidance (SEM) P value a Mean trials to criterion (SEM) P value" Recall score (%) P value b

Fluoxetine pretreated

Vehicle pretreated

Same side

Same side

0 2.73 0.36 6.73 0.36 87

Opposite side 2.73 0.31 P<0.001 4.93 0.46 P<0.001 9.13 0.50 P < 0.001 20 P < 0.001

Opposite side

0.20 1.27 0.15 0.19 P<0.001 4.87 5.48 0.50 0.28 P<0.01 9.07 9.67 0.53 0.31 ns 20 7 ns

Statistical difference determined by Student's t-test except for errors between the fluoxetine treated groups. Since the fluoxetinetreated mice tested to the same side as in the original training made no error and therefore there was no variance, the statistical difference was based on the number of mice in each group making 1 or more errors vs the number of mice making no errors. The dichotomized scores were statistically evaluated by Eisher's exact probability test b Statistical difference based on Fisher's exact probability test

1985). The mice were trained and tested as in experiment 1. One hour before the 1-week retention test, mice were injected subcutaneously with 2.5 mg/kg F L U which facilitated retention when F L U was injected prior to training (experiment 3) or with the vehicle. The N per group was 15.

Results The controls performed poorly on the retention test as intended, with only 13% recall or 9.13_+ 1.51 trials to criterion. The FLU-injected group showed improved retention test performance with a recall score of 80% or 6.87_+ 0.99 trials to criterion. A n A N O V A showed that the difference between groups was significant by mean trials either to the first avoidance [F(1,28)=24.93, P < 0 . 0 0 1 ] or to criterion [F(1,28) = 23.73, P < 0.001].

Experiment 5. Effect of fluoxetine on acquisition Improved retention test performance resulting from administration of F L U prior to training in experiment 3 or prior to testing in experiment 4 could have resulted from improved acquisition rather than from enhanced memory processing or recall. To determine if improved retention test performance was due to facilitation of acquisition of Tmaze footshock active avoidance, F L U at 0.5, l, 2.5, 5, 10, 20, 25, 30 or 35 mg/kg, or the vehicle, was administered subcutaneously 1 h prior to training. The mice were trained as in experiment 1 except that training continued until five avoidances where made in six consecutive trials. Mice not reaching criterion in 15 training trials were given a score of 15. The N per group was 10. I n experiments 3 and 4, 2.5 mg/kg F L U enhanced retention. I n a second study, mice

were administered vehicle or 2.5 mg/kg F L U 1 h prior to training with an N of 20 per group.

Results Fluoxetine, injected 1 h prior to training did n o t affect acquistion of footshock avoidance habit in the T-maze over the dose range 0.5-20 mg/kg b u t significantly impaired acquisition at 25-35 mg/kg. A one-way A N O V A , r u n on log trials to criterion to correct for lack of homogeneity of variance, indicated a significant effect of drug dose on acquisition [F(9,90= 15.98, P<0.001]. A further analysis of the mean trials to criterion between the control and each experimental group using D u n n e t t ' s t-test indicated that F L U significantly impaired acquisition at doses of 25 35 mg/kg. A separate A N O V A done on m e a n trials to first avoidance response also indicated that acquisition was significantly impaired by F L U administered at 2 ~ 3 5 mg/ kg. In the separate study, employing vehicle and 2.5 mg/kg F L U with Ns equal 20 per group, the mean trials to first avoidance, criterion a n d standard errors were 5.45_+0.19, 9.85+0.19 for the control and 5.35_0.26, 10.0_0.21 for F L U ; acquisition was not significantly affected by either measure. A n analysis of discrimination error showed that controls made 1.4+0.11 errors before reaching criterion and mice injected with F L U made 1.7___0.22 errors. This difference in discrimination errors was not significant.

Experiment 6. Effect of fluoxetine on recall of T-maze footshock

Avoidance reversal training As a further test of whether acquisition or recall was enhanced by pre-test administraton of F L U , mice were trained as in experiment 1. One hour before 1 week retention test, mice were injected subcutaneously with 2.5 mg/kg F L U or the vehicle. Half the F L U - and vehicle-injected mice were trained to make an avoidance to the same side as during training; the remaining mice had to make an avoidance to the opposite side. The rationale of this reversal test procedure is that if F L U improved acquisition then reversed and non-reversed groups should have similar retention test scores. However, if F L U enhanced recall, then FLU-treated mice trained to make an avoidance to the opposite side should show an increase in errors and trials to criterion and have poorer recall scores, since better recall of the conditions of original training would interfere with learning the new habit.

Results Table 5 shows the mean left-right discrimination errors made during testing. Since the FLU-treated group trained on the retention test to make avoidances to the same side as during the original training made no errors and F L U tested to the opposite side made an average of 2.73 errors during testing, it is clear that enhanced recall of original learning interfered with retention test performance when mice were required to respond to the side opposite that of original training. Based on the per cent mice making one or more errors during testing, the FLU-treated mice tested to the same (0% trials) or opposite (100% trials) side as during original training differed from each other

4O

Table 5. Time-dependent facilitation of retention for t-maze footshock active avoidance training Treatment

Vehicle

Injection delay (min) a 0

Fluoxetine 90

0

30

60

90

Mean trials 4.73 4.93 2.67 3.53 3.60 4.40 lst avoidance _+0.37 _+0.34 _+0.26 +0.43 _+0.36 -+0.44 (SEM) Pvalue b P<0.01 P<0.05 P<0.05 ns Mean trials 8.87 8.93 criterion ___0.37 +0.34 (SEM) Pvalue b Recall score (%)

20

13

6.80 7.73 _+0.32 +0.42

7.67 8.60 _+0.35 ±0.43

P<0.01 P<0.05 P<0.05 ns 80

60

47

33

a Injection delay is the time in min from the end of training until vehicle or fluoxetine (15 mg/kg, SC) was administered b p values are the statistical difference between vehicle given 0 min after training and each fluoxetine-treated group as determined by Dunnett's t-test

nal training but did not affect recall when forced to respond to the opposite side (Table 5). The F L U - t r e a t e d groups differed at P = 0.00006 (Fisher's exact probability test). The recall scores o f the vehicle control group did not differ significantly from each other, though the vehicle controls, tested to the same side as they were during training, achieved a higher recall score.

Experiment 7. Time-dependent improvement of memory retention by post-training subcutaneous administration of fluoxetine Drugs that are recognized as improving m e m o r y retention show a decline in effectiveness as the time from the end o f training until the drug is administered increases ( M c G a u g h 1973). The p u r p o s e o f experiment 7 was to determine if F L U showed a time-dependent facilitation o f retention. Separate groups o f mice received a single injection o f F L U (15 mg/kg, SC) 0, 30, 60 or 90 rain after training. Two control groups received injections o f the vehicle 0 or 90 rain after training. The N p e r group was 10. Training and testing were done as in experiment 1. Results

at P = 0.00008 (Fisher's exact probability test). The vehicle controls trained on the retention test to m a k e on avoidance to the same side as during original training differed significantly from the vehicle control trained to respond to the opposite side ( P < 0 . 0 0 2 , t-test). It should be noted that during original training all mice made mean errors o f 1.20-+0.04 (-+SEM). Thus the vehicle- and F L U - t r e a t e d group tested to the same side as during training remembered where to escape. The vehicle group trained to the opposite m a d e a b o u t the same n u m b e r of errors as all mice did during original training. The nearly three times greater error rate o f the F L U - t r e a t e d mice tested to the opposite side indicates a considerable degree o f impairment considering how easily mice learn to which side to direct their responses. C o m p a r i n g the per cent mice making multiple errors on the first test trial between F L U - and vehicle-treated mice tested to the side opposite training indicated that 93% o f F L U - t r e a t e d mice m a d e two or more errors while 13% o f the vehicle-treated mice made multiple errors on the first retention test trial ( P = 0.00001, Fisher exact probability test). Thus, the vehicle-treated mice more readily learned to reverse sides than the F L U - t r e a t e d mice. A separate two-way A N O V A run on trials to criterion indicated that the interaction o f drug state and test side was significant [F(1,56)=4.57, P < 0 . 0 5 ] as were the main effects of drug state [F(1,56)= 11.59, P < 0.001] and test side [F(1,56) = 12.69, P < 0.001]. A further analysis o f mean trials to criterion indicated that vehicle-treated mice tested to the same side as during training only differed significantly from F L U - t r e a t e d mice tested to the same side ( P < 0 . 0 1 , Dunnett's t-test). The F L U - t r e a t e d mice tested to the opposite side required significantly more trials to reach criterion than F L U - t r e a t e d mice trained to the same side as during training (P < 0.01, Tukey's t-test) but did not differ from vehicletreated mice tested to the opposite side (Table 5). A twoway A N O V A run on trials to first avoidance resulted in the same statistical decision to reject the null hypothesis. The recall scores clearly indicated that F L U improved recall when mice were trained to the same side as the origi-

The vehicle controls injected 0 or 90 min after training had p o o r retention with recall scores o f 20% and 10%, respectively. The recall scores for the fluoxetine-treated mice indicated that fluoxetine resulted in time-dependent facilitation of retention typical o f most drugs which improve retention, with injections immediately after training resulting in 80% recall, injections 30 and 60 min after training resulting in 50% recall and fiuoxetine injection 90 rain after training in 40% recall. The A N O V A for the drug effect was significant for mean trials to first avoidance [F(5,54)=3.75, P < 0.01] and for mean trials to criterion [F(5,54)=3.93, P < 0.01]. A comparison o f each fluoxetine-injected group with the control group (0 or 90 rain) using Dunnett's t-test yielded significantly lower mean trials to first avoidance response, and lower m e a n trials to criterion for F L U administered at 0 or 30 min ( P < 0 . 0 1 ) and at 60 rain ( P < 0 . 0 5 ) . F L U injected 90 rain after training did not signifiantly affect retention by either measure. The mean trials to first avoidance ( ± S E M ) were as follows: vehicle 0 m i n 4 . 8 0 + 0 . 4 4 ; F L U 0 rain 2.6-+0.36; F L U 30 rain 3 . 9 0 _ 0 . 6 0 ; F L U 60 min 3.70 -+ 0.47; F L U 90 min 4.20 4- 0.62 and vehicle 9 0 m i n 5.30___0.44 and to criterion ( ± S E M ) vehicle 0 m i n 8.90_0.46; F L U 0 _ m i n 6.6___0.42; F L U 3 0 r a i n 8.10-+0.56; F L U 6 0 m i n 7.70-+0.47; F L U 9 0 r a i n 8.40-+0.61 and vehicle 90 min 9.30-+0.45.

Experiment 8. Anti-amnestic effect of fluoxetine W e and others have reported that amnesia caused by inhibitors o f brain protein synthesis can be blocked by a variety o f neurotransmitter agonists and hormones ( N a k a j i m a 1975; F l o o d et al. 1977, 1978a, b, 1985; Davis and Squire 1984; F o o d and Cherkin 1986). Recently, we found that A C T H 4 _ lo or drugs affecting six neurotransmitter systems blocked scopolamine-induced amnesia ( F l o o d and Cherkin 1986). The purpose of experient 8 was to determine if F L U had anti-amnestic properties against both anisomycin (ANI), a protein synthesis inhibitor, and scopolamine (SCO), an acetylcholine receptor blocker.

41 The training conditions in this experiment were altered from those o f experiment 1 to assure that control mice would have high recall scores (70-80%) in o r d e r to permit detecting amnesia caused by A N I or SCO. To accomplish this, the nominal footshock level was increased from 0.30 to 0.35 m A , the buzzer was loud rather than muffled, and the intertrial interval was 45 s instead of 30 s. The experiment was done in two parts because o f the different experimental conditions under which A N I and SCO must be used. Retention was tested 1 week after training. Because the amnestic effects o f A N I and SCO are robust, the N per group was 10. Anisomycin (ANI, F.W. 265.5) was prepared at 2 mg/ml in saline and scopolamine h y d r o b r o m i d e trihydrate (SCO, F.W. 438.31) at 0.1 mg/ml in saline. In this study, A N I (20 mg/kg) or vehicle was injected subcutaneously 15 min prior to training, F L U (15 mg/kg) or the vehicle was injected immediately after training, then a second injection o f A N I or vehicle was given 1.75 h after training. The two injections of A N I inhibit protein synthesis in the brain for 4-6 h ( F l o o d et al. 1978a). The three groups were A(V)A, A ( F ) A , and V(V)V, where A = A N I , F = F L U , and V = v e h i c l e . In the second study, SCO (1 mg/kg) or vehicle was injected immediately after training and F L U (15 mg/kg) or the vehicle was administered 45 min later. Three groups were used: SCO(V), SCO(F) and V(V). Results In the study using anisomycin, the results yielded a control group [V(V)V] with g o o d retention (73 % recall score). A N I (V)ANI treatment induced amnesia (13% recall score). The A N I ( F ) A N I group h a d a recall score of 80%, indicating that F L U counteracted the amnesia. A n A N O V A indicated a significant drug effect for mean trials either to first avoidance [F(2,27)=11.38, P < 0 . 0 0 1 ] or to criterion [F(2,27)= 13.83, P < 0.01]. A c o m p a r i s o n by either measure o f retention indicated that A ( F ) A differed significantly from A ( V ) A ( P < 0 . 0 1 ) but not from V(V)V. In the study using scopolamine, the results indicated that the V(V) group remembered well, with 80% recall score. SCO(V) induced amnesia (10% recall score) and S C O ( F ) h a d a recall score o f 70%, indicating the F L U counteracted scopolamine-induced amnesia. A n A N O V A indicated a significant drug effect for mean trials either to first avoidance [F(2,27)= 16.00, P < 0 . 0 1 ] or to criterion [F(2,27)=14.13, P < 0 . 0 0 1 ] . A n analysis o f mean trials to first avoidance and o f mean trials to criterion yielded a significant difference ( P < 0 . 0 1 ) between SCO(V) and SCO(F).

Experiment 9. Effect of fluoxetine on retention for passive avoidance conditioning Drugs that enhance m e m o r y retention usually improve retention in passive avoidance as well as active avoidance paradigms. The purpose o f experiment 9 was to determine if F L U would improve retention for one-trial passive avoid-' ance. The procedure for training and testing mice for passive avoidance has been described previously ( F l o o d et al. 1972, 1974). In brief, the a p p a r a t u s consisted o f a black start c o m p a r t m e n t j o i n e d to a white shock c o m p a r t m e n t by a partition containing a hole through which the mice could enter the white compartment. In the white compartment,

100

8 bd U

eo

C3 Q

60

,-4 bo

40

<> to

o_ 20

0

5

10

15

20

25

SUBCUTANEOUSDOSE (mg/k9)

Fig. 1. The effect of fluoxetine on retention for passive avoidance. Fluoxetine results in a U-shaped dose response curve as with active avoidance. Mean and standard error are not given because the distribution of latencies are bimodal. Higher means are interpreted as showing greater retention. * The per cent mice classed as recalling original training differed significantly from the vehicle control group at P = 0.004 using Fisher's exact probability test

footshock was given until the mouse returned to the black compartment. Acquisition o f this task is determined by the latency-to-enter and latency-to-escape from the shock comp a r t m e n t and by the footshock intensity ( F l o o d et al. 1972). To reduce individual differences in acquisition, only mice with latencies of 1.5-3.4 s to enter and to escape from the shock c o m p a r t m e n t were used; other mice were discarded. Less than 15% o f the mice were discarded. The footshock intensity, used to control overall training strength, was set at 0.20 m A so that control mice would show p o o r recall on the retention test I week after training. In order to test retention, the mice were again placed into the black comp a r t m e n t and t h e latency to enter the white c o m p a r t m e n t was taken as a measure o f retention. Mice not entering the white c o m p a r t m e n t within 180 s were removed and the test was terminated. A latency to enter the white shock c o m p a r t m e n t on the test day o f 180 s was defined as successful passive avoidance. Because test latencies for this task show a clear b i m o d a l distribution ( F l o o d et al. 1972), the results for each group were dichotomized at 180 s and expressed as per cent passive avoidance. The n o n - p a r a m e t r i c Fisher's exact probability test (Winer 1971) was used to determine statistical significance between the control group and the group showing the highest per cent passive avoidance. Training and testing were done between 0900 and 1400 hours. Mice were injected within 2 min after training with 5, 10, 15, 20, or 25 mg/kg F L U or the vehicle. The N per group was 15. The retention test was given i week after training.

Results The per cent mice showing passive avoidance in the control group was low (20%). F L U improved retention test scores optimally at 15 mg/kg, yielding 73% o f the mice showing passive avoidance (Fig. 1). The difference between the control and F L U at 15 mg/kg was significant (P = 0.004, Fisher exact probability test).

42 Discussion

Summary of findings F L U (10 and 15 mg/kg, SC) enhanced 1-week memory retention of young adult mice when injected within 2 min post-training (experiment 1). Similar results were obtained with intracerebroventricular injection of F L U (16 and 24 lag per mouse), providing evidence of a central mechanism of action (experiment 2), although not disproving peripheral effects as well with subcutaneous administration. Fluoxetine (2.5 mg/kg, SC) enhanced retention at a lower dose when injected prior to training (experiment 3) than when injected immediately after training (15 mg/kg, SC; experiment 1). Fluoxetine at 2.5 mg/kg SC also enhanced recall scores when injected 1 h before the 1-week retention test (experiment 4), indicating an enhancing effect upon memory retrieval. Neither the pre-training nor pre-testing effects depended on improved acquisition (experiment 5), since F L U did not improve acquisition of T-maze footshock avoidance over the dose range 0.5-25 mg/kg, SC. F L U enhanced recall and not acquisition when administered prior to testing because retention was impaired in FLU-treated mice trained on the retention test to make an avoidance to the side opposite that of original training compared to the enhanced retention test performance of FLU-treated mice trained to respond to the same side of the T-maze as during original training (experiment 6). The sensitive period for post-training enhancement by F L U was less than 90 min as shown by the temporal gradient (experiment 7). This gradient is typical of memoryenhancing drugs. The post-training gradient rules out a proactive effect of F L U when administered shortly after training on retention test performance, because injections given 90 rain after training, and therefore slightly closer in time to the retention test, did not enhance retention. The amnesia induced by either a protein synthesis inhibitor (ANI), or by an anticholinergic drug (SCO), was blocked by F L U (15 mg/kg, SC) injected post-training (experiment 8). Finally, F L U (15 mg/kg, SC), injected after one-trial passive avoidance training, enhanced 1-week retention, demonstrating effectiveness in this task (experiment 9) as well as in active avoidance.

Related findings Passive avoidance impairment due to olfactory bulbectomy was reversed in rats by 10 mg/kg F L U administered subcutaneously prior to training (Broekkamp et al. 1980) or when administered centrally at 10 lag (Garrigou et al. 1981). Lotden and Nunn (1982) reported that F L U 10 mg/kg impaired acquisition of conditioned taste aversion but 5 mg/kg did not. Shuttlebox acquisition was impaired by pre-training administration of F L U at 2.5, 5, and 10 mg/kg. An analysis of escape latencies showed no effect but there was an increase in freezing and a decrease in inter-trial responding. Even 2.5 mg/kg F L U may have been too high a dose in this stressful task.

Interpretation Fluoxetine is a potent and specific inhibitor of serotonin uptake by competing with serotonin for active transport into serotoninergic nerve terminals (Fuller etal. 1975;

Wong et al. 1975). As the major mechanism for inactivation of neuronally released serotonin is uptake into serotoninergic nerve endings, a single injection of F L U increases the concentration of serotonin reaching the post-synaptic receptor sites by reducing transport back to the pre-synaptic nerve terminals, as demonstrated by reduced synaptosomal accumulation of labeled 5-hydroxytryptophan (Wong et al. 1975; Fuller and Wong 1977; Fuller 1980). One interpretation of the results based on these biochemical findings is that increased serotoninergic activity resulting from administration of F L U enhanced memory processing, as suggested by both improved recall scores of groups receiving appropriate doses of F L U and by the antiamnestic action of F L U against A N I and SCO. However, other effects of increased serotoninergic activity such as analgesia, reduced REM sleep, and increased plasma corticosterone may account for FLU-induced improvement of retention. Any effect of F L U on activity and footshock sensitivity does not affect the interpretation of experiments 1, 2 and 7-9, since F L U was injected after training and 1 week elapsed before retention was tested. In experiment 3, F L U was administered prior to training; the doses of 1.0, 2.5 and 5.0 mg/kg F L U enhanced memory retention but did not enhance acquisition. In fact, FLU impaired acquisition at 30 and 35 mg/kg (experiment 5). F L U has been reported to cause analgesia for footshock (Messing etal. 1975; Buckett 1984) but the results are conflicting (Fuller 1982; McElroy et al. 1982) and the analgesic doses are high (5 20 mg/kg) compared to the effective pretraining doses (1 5 mg/kg) that we found to improve memory retention (experiment 3). Bloch and co-workers (1976) hypothesized that reducing R E M sleep impaired retention and presented data in support of the hypothesis. Fluoxetine reduces REM sleep (Vogel et al. 1975; Fuller etal. 1976; Slater et al. 1978) but it enhances memory retention. An alternative interpretation of our results may be related to reports of increased plasma corticosterone following increased serotoninergic activity. The administration of F L U causes a dose-dependent increase in corticosterone from 2.5 to 20 mg/kg (Fuller et al. 1975; Fuller et al. 1976; Fuller and Snoddy 1979). Since administration of corticosteroids improves memory retention (Flood et al. 1978b) and is anti-amnestic (Barondes and Cohen 1968; Nakajima 1975; Flood et al. 1978 b), it is possible that F L U improved memory retention by increasing plasma corticosterone. However, increased plasma corticosteroid levels are not necessarily associated with improved retention. Cycloheximide and anisomycin, inhibitors of protein synthesis which cause amnesia, also result in a large increase in plasma corticosterone levels (Flood et al. 1978b). Kovacs etal. (1976) found that doses of corticosterone which facilitated recall also increased 5-HT content in brain tissue, while amnestic doses caused a decrease in 5-HT content. Thus it is not clear whether the increased concentration of corticosterone associated with F L U administration is causal in improving memory retention. It remains for additional research to determine if FLU-enhanced memory retention is necessarily related to increased release of corticosterone. We conclude that F L U meets all of the major criteria for a memory-enhancing drug and that its mechanism of action is to be sought in enhancement of post-training memory processing (" consolidation") and of retrieval, but not of acquisition. It is unclear whether memory improvement is directly related to increased activity at serotoninergic syn-

43 apses that store the m e m o r y " e n g r a m " , or due to F L U induced increased corticosteroid release which in turn aids m e m o r y processing.

Acknowledgement. The research was supported by Medical Research Service of the Veterans Administration and by the Sepulveda Geriatric Research, Education and Clinical Center (GRECC). Fluoxetine was generously supplied by Eli Lilly Laboratories, Eli Lilly & Company through Dr. David Wong. We also express our appreciation for Dr. Wong's helpful advice. References

Barondes SH, Cohen ttD (1968) Arousal and the conversion of "short term" to "long-term" memory. Proc Natl Acad Sci USA 61 : 923-929 Bloch V (1976) Brain activation and memory consolidation. In: Rosenzweig MR, Bennett EL (eds) Neuronal mechanisms of learning and memory, MIT Press, Cambridge, MA, pp 583-590 Broekkamp CL, Garrigou D, Lloyd KG (1980) Serotonin-mimetic and antidepressant drugs on passive avoidance learning by olfactory bulbectomised rats. Pharmacol Biochem Behav 13 : 643 646 Buckett WR (1984) Pharmacological studies o n stimulation-produced analgesia in mice. Eur J Pharmacol 69:2181-2190 Davis HP, Squire LR (1984) Protein synthesis and memory: A review. Psychol Bull 96: 518-559 Dismukes RK, Rake AV (1972) Involvement of biogenic amines in memory formation. Psychopharmacologia 23:17-25 Flood JF, Bennett EL, Rosenzweig MR, Orme AE (1972) Influence of training strength on amnesia induced by pretraining injections of cycloheximide. Physiol Behav 9:589-600 Flood JF, Rosenzweig MR, Bennett EL, Orme AE (1974) Comparison of the effects of anisomycin on memory across six strains of mice. Behav Biol 10:147-160 Flood JF, Bennett EL, Rosenzweig MR, Orme AE (1975) The influence of duration of protein synthesis inhibition on memory. Physiol Behav 10:555-562 Flood JF, Jarvik ME, Bennett EL, Orme AE, Rosenzweig MR (1977) The effect of stimulants, depressants, and protein synthesis inhibition on retention. Behav Biol 20:168-183 Flood JF, Bennett EL, Orme AE, Rosenzweig MR, Jarvik ME (1978a) Memory: Modifiation of anisomycin-induced amnesia by stimulants and depressants. Science 199:324-326 Flood JF, Vidal D, Bennett EL, Orme AE, Vasquez S, Jarvik ME (1978b) Memory facilitating and anti-amnesic effects of corticosteroids. Pharmacol Biochem Behav 8:81 87 Flood JF, Smith GE, Cherkin A (1985) Hydergine enhances memory in mice. J Pharmacol [Suppl III] 16:39 49 Flood JF, Cherkin A (1986) Scopolamine effects on memory retention in mice: A model of dementia? Behav Neural Biol 45 : 16%184 Fuller RW (1980) Pharmacology of central serotonin neurons. Annu Rev Pharmacol Toxicol 20:111-127 Fuller RW (1982) Functional consequences of inhibition serotonin uptake with fluoxetine in rats. In: Ho BT et al. (eds) Serotonin in biological psychiatry. Raven, New York, pp 219-228 Fuller RW, Clemens JA, Slater IH, Rathbun RC (1978) Neuroendocrine and behavioral studies with fluoxetine, an inhibitor of serotonin uptake in brain. In: Deniker P, Radouco-Thomas C, Villeneuve A (eds) Neuro-psychopharmacology: Proceedings of the Tenth Congress of the C.I.N.P., Pergamon, Oxford New- York, pp 641-646 Fuller RW, Perry KW, Snoddy HD, Molloy BB (1974) Comparison of the specificity of 3-(p-trifluoromethylphenoxy)-n-methyl3-phenylpropylamine and clorimipramine as amine uptake inhibitors in mice. Eur J Pharmacol 28 : 233-236 Fuller RW, Rathbun RC, Parli CJ (1976) Inhibition of drug metab-

olism by fluoxetine. Res Commun Chem Pathol Pharmacol 19 : 353-356 Fuller RW, Snoddy HD (1979) The effects of metergoline and other serotonin receptor antagonists on serum corticosterone in rats. Endocrinology 105:923-928 Fuller RW, Snoddy HD, Molloy BB (1975) Potentiation of the 1-5-hydroxytryptophan-induced elevation of plasma corticosterone levels in rats by a specific inhibitor of serotonin uptake. Res Commun Chem Pathol Pharmacol 10:193-196 Fuller RW, Wong DT (1977) Inhibition of serotonin reuptake. Fed Proc 36 : 2154-2158 Garrigou D, Broekkamp CL, Lloyd KG (1981) Involvement of the amygdala in the effect of antidepressants on the passive avoidance deficit in bulbectomised rats. Psychopharmacology 74: 66-70 Gold PE, Delanoy RL (1981) ACTH modulation of memory storage processing. In: Martinez JL, Jr, Jensen RA, Messing RB, Rigter H, McGaugh JL (eds) Endogenous peptides and learning and memory processes. Academic, New York, pp 7%98 II'yuchenok R Yu (1976) Pharmacology of behavior and memory. Hemisphere Printing, Washington Keppel G (1973) Design and analysis: a researcher's handbook. Prentice-Hall, Englewood Cliffs, pp 556-559 Kovacs GL, Tlegdy G, Lissak K (1976) 5-Hydroxytryptamine and the mediation of pituitary-adrenocortical hormones in the extinction of active avoidance behavior. Neuroendocrinology 1:219-230 Lorden JF, Nnnn WB (1982) Effects of central and peripheral pretreatmeut with fluoxetine in gustatory conditioning. Pharmacol Biochem Behav 17:435443 McElroy JF, DuPont AF, Feldman RS (1982) The effects of fenfluramine and fluoxetine on the acquisition of a conditioned avoidance response in rats. Psychopharmacology 77:356-359 McGaugh JL (1973) Drug facilitation of learning and memory. Annu Rev Pharmacol Toxicol 13:229-241 Messing RB, Phebus L, Fisher LA, Lytle LD (1975) Analgesic effect of fluoxetine hydrochloride. Psychopharmacol Commun 1:511 521 Myers RD (1974) Handbook of drug and chemical stimulation of the brain. Van Nostrand, New York, pp 596-647 Nakajima S (1975) Amnesic effect of cycloheximide in the mouse mediated by adrenoeortical hormones. J Comp Physiol Psychol 88:378-385 Rake AV (1973) Involvement of biogenic amines in memory formation: The central nervous system indole amine involvement. Psychopharmacologia 29 : 91 100 Slater HI, Jones GT, Moore RA (1978) Inhibition of REM sleep by fluoxetine, a specific inhibitor of serotonin uptake. Neuropharmacology 17: 383-389 Soumireu-Mourat B, Micheau J, Franc C (1981) ACTH4 9 analog (ORG2766) and memory processes in mice. In: Martinez JL, Jr, Jensen RA, Messing RB, Rigter H, McGaugh JL (eds) Endogenous peptides and learning and memory processes.Academic, New York, pp 143-158 Vogel GW, Thurmond A, Gibbons P, Sloan P, Boyd M, Walker M (1975) REM sleep reduction effects on depression syndromes. Arch Gen Psychiatry 32:765 777 Winer BJ (1971) Statistical principles in experimental design. McGraw-Hill, New York, pp 196-210, 39%402 Wong DT, Bymaster EP, Horng JS, Molloy BB (1975) A new selective inhibitor for uptake of serotonin into synaptosomes of rat brain: 3-(p-tri-fluoromethylphenoxy)-n-methyl-3-phenylpropylamine. J Pharmacol Exp Ther 193 : 804-811 Wong DT, Bymaster FP, Reid LR, Threlkeld PG (1983) Fluoxetine and two other serotonin uptake inhibitors without affinity for neuronal receptors. Biochem Pharmacol 32:1287-1293 Received March 20, 1987 / Final version April 21, 1987