Proc. Natl. Acad. Sci. USA Vol. 91, pp. 2041-2045, March 1994 Psychology Delayed emergence of effects of memory-enhancing drugs: Implications for the dynamics of long-term memory CESARE MONDADORI, BASTIAN HENGERER, THOMAS DUCRET, AND JUERGEN BORKOWSKI Pharmaceutical Research Division, CIBA-Geigy Limited, Basle, CH-4002, Switzerland Communicated by L. Weiskrantz, November 22, 1993 (receivedfor review April 27, 1993)

ABSTRACT Many theories of memory polate that pro- facilitation emerges, we hoped not only to help to character- cessing of information outlasts the earning situation and ize the substances themselves but also to shed light on the involves several different physiological substrates. If such dynamics of discrete phases of memory. The strategy of physiologically distinct mecanism or stages of memory do in examining time dependence of memory effects has a well- fact exist, they should be differentially affected by particular established history for substances that interfere with memory experimental manipulations. Accordingly, a selective improve- retention but has hardly been adopted in studies of memory ment ofthe processes underlying short-term memory should be enhancement. Even with interference treatments, reports of detectable only while the information is encoded in the short- delayed effects [as with cerebral electroshock, for example term mode, and a selective influence on long-term memory (14, 15)] and protein synthesis inhibition (e.g., see refs. should be detectable only from the moment when memory is 16-20) are relatively sparse. Moreover, some memory- based on the long-term trace. Our comparative study of the modulating substances have a wide range of unspecific side time course of the effects of the cholinergic agonist arecoline, effects that complicate interpretation-i.e., direct drug- the -aminobutyric acid type B receptor antagonist CGP induced behavioral effects can interfere with the behavioral 36742, the angiotensin-converting enzyme inhibitor captopril, manifestation of the memory effects at the time of retest. In and the nootropic oxiracetam, four substances with completely the present study, we focused largely on the memory- different primary sites of action, show that the memory- enhancing effect of oxiracetam, with which the risks of enhancing effects consistently come into evidence no sooner interference are minimal given that it has no known acute than 16-24 h after the learning trial. On the one hand, this effects on spontaneous behavior at the dose used; but par- finding suggests that all these substances act by way ofthe same ticular parallel experiments were also carried out with arec- type of mechanism; on the other hand, it demonstrates that the oline, CGP 36742, and captopril. substrate modulated by the compounds forms the basis of memory only after 16-24 h. From the observation that animals also show clear signs of retention during the first 16 h-i.e., DESIGN, METHODS, PROCEDURES before the effects of the substances are measurable-it can be General Design. With oxiracetam four questions were inferred that retention during this time is mediated by other addressed. The first was to determine precisely when the mechanisms that are not influenced by any of the substances. enhancement of retention becomes manifest. Retention was therefore assessed in separate groups of animals at various The group ofpreparations already found to display memory- intervals (1, 2, 4, 8, 16, and 24 h) after the learning trial. The enhancing effects in experimental learning situations is sur- second question was whether the timing of the injection in prisingly large. Even more surprising is the fact that the relation to the learning trial and to the retention test itselfhad primary pharmacological sites of action of these substances an influence on the time ofappearance ofenhanced retention. are very heterogeneous. Among others, the list includes For this purpose, oxiracetam was administered 1, 4, or 8 h cholinergic agonists at the muscarinic and nicotinic receptors before or immediately after the learning trial. The third (1, 2); cholinesterase inhibitors such as physostigmine (3); question was whether the time of day had an influence on calcium-channel blockers such as nimodipine (4); angioten- performance during retention. To this end, retention was sin-converting enzyme (ACE) inhibitors such as captopril (5); always measured simultaneously in a group that had already transmitters such as, e.g., norepinephrine (6); t-aminobu- been subjected to training and treatment 1 day beforehand. tyric acid type B (GABAB) receptor blockers such as CGP Since the corresponding controls were also tested at each 36742 (7); peptides such as vasopressin and corticotropin (8); interval, the experiments included an additional seven glucose (9); and, not least, the large group of so-called treated and seven control groups with retention intervals of nootropics including oxiracetam and piracetam (10, 11). 25, 26, 28, 32, 40, 44, and 48 h. The fourth question was how Despite their different primary sites of action, all of the long the enhanced retention could be seen to persist. This was above-mentioned substances have a similar effect in a simple assessed by using retention intervals of4, 8, and 16 days after one-trial passive avoidance task; if administered, e.g., 1 h the learning trial. before the learning trial, they improve the retention perfor- The generality of the delayed emergence of enhanced mance at retest 24 h later (12, 13). At the time of retention retention was then also studied with the muscarinic agonist testing, the changes induced by the substances have already arecoline, the GABAB receptor blocker CGP 36742, and the been expressed at a stage generally reckoned to be in ACE inhibitor captopril. With these three substances, only long-term memory, but it is not known when these changes immediate posttrial administration was used. To test the occurred. Indeed, it is often implicitly assumed that the possible influence of time of day, the same parallel control facilitatory effects occurred concomitantly with the initial measurements were made as in the experiments with oxirac- learning itself, demonstrable at short retention intervals. By etam. making a comparative study of the time at which memory For all substances, the optimal dosage was first determined in pilot experiments in which dose-response profiles for The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviations: ACE, angiotensin-converting enzyme; GABAB, in accordance with 18 U.S.C. §1734 solely to indicate this fact. y-aminobutyric acid type B receptor.

2041 Downloaded by guest on October 1, 2021 2042 Psychology: Mondadori et al. Proc. Natl. Acad Sci. USA 91 (1994) memory enhancement were obtained. Accordingly, the doses retest latencies permitted retests to be carried out within an used were 100 mg of oxiracetam per kg, 0.3 mg of arecoline acceptable time frame (in some cases, five groups had to be per kg, 10 mg of CGP 36742 per kg, and 3 mg of captopril per tested simultaneously); (ii) the conditions were such that only kg. a few animals were included in the analysis, with cutoff Animals. Male mice (MA 01 Tif; 20-22 g) from our own latencies of 150 sec, which enhanced the power of the animal breeding unit were used throughout the experiment. statistical evaluation; and (iii) it provided an optimal baseline They were housed in plastic cages in a room lit from 6 a.m. against which to detect enhancement of retention. to 6 p.m. and had free access to food and water at all times. In all experiments, the animals were randomly assigned to Apparatus and Procedures. The apparatus consisted of an the different groups (n = 25) receiving the drug or saline. All electrifiable grid (50 x 50 cm) of stainless-steel rods (4 mm in administration was i.p. Retests were performed blind. diameter; spaced 13 mm apart), enclosed by gray poly(vinyl Learning was assessed by comparing the drug and saline chloride) (PVC) walls 50 cm high. In the middle of the grid groups with no-shock controls. Drug effects were assessed in was a platform 12 mm high and 67 mm in diameter, which was relation to a group receiving saline injection but otherwise enclosed by a removable gray PVC tube (180 mm high; treated identically to the drug groups. The data deriving from 68-mm inner diameter). The mice were placed one by one on such experiments are comparable to "failure-time data" (21). the platform inside the tube, which was removed after 10 sec. The appropriate statistical procedure to analyze the outcome With a few exceptions, they stepped down from the platform of single sets of experiments is a generalized Wilcoxon test within 10 sec and received a footshock (1 mA; 1 sec). (22). No-shock controls were run in parallel. At a specified interval after training, each animal was again placed on the platform and the step-down latency was recorded ("retest latency") RESULTS up to a maximum cutoff time of 150 sec. The experimental Oxiracetam. Time ofemergence ofretention enhancement. parameters were adjusted to produce minimal indications of If the compound was administered 1 h before the learning learning. There were three reasons for doing so: (i) the short trial, its effect on memory emerged reliably only after a 150- 100-

S 0 0 * *

0 ..' 0 10

.-N 0 1 2 4 8 16 20 24 hours

0) .0-..4-. 150- 0) 0 100- 0

*6I0 *6,; ~~~~~~~~~~m~~~~~~~~10:

0 25 26 28 32 40 44 48 hours

FIG. 1. Step-down passive avoidance in mice. Effects of pretrial oxiracetam on retention performance at various retention intervals after training and treatment. Oxiracetam was always administered i.p. 1 h before the learning trial. Boxplots represent the retest latencies, in seconds, of the various experimental and control groups. Hatched box, saline control; open box, oxiracetam (100 mg/kg, i.p.); stippled box, no-shock controls. Prolongation of the latencies by comparison with the saline-treated controls indicates drug-induced memory facilitation. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by generalized Wilcoxon test. Corresponding retest latencies (Upper, 1- to 24-h interval; Lower, 25- to 48-h interval) were measured in parallel. To depict a distribution, a boxplot (23) consists of a box and tails, with outliers (ifpresent) for each group. Horizontal line through the box depicts the median, and the top and bottom of the box mark the upper and lower quartiles. Thus, the box itself shows the range for exactly that half of the data points in the group that cluster around the median. Tails mark the actual range of 90% of the data in the group. Outliers are represented by small circles. Downloaded by guest on October 1, 2021 Psychology: Mondadori et al. Proc. Natl. Acad. Sci. USA 91 (1994) 2043

retention interval of 20 h. At all shorter retention intervals, 150 * S0 0 the animals showed the usual signs of having learned (i.e., their step-down latencies were longer than those of the 100- no-shock controls), but there were no facilitatory drug ef- fects. The results are shown in Fig. 1 (Upper). * 0 u Time of day for retention testing. The parallel control 00-% * O experiments clearly indicate that the absence ofany enhance- ment effect with retention intervals shorter than 20 h could

- not be due to circadian influences; at all intervals when no % 0 facilitation by oxiracetam could be detected, highly signifi- - 10 - d cant effects of the drug were found in the animals already o trained and treated 1 day earlier (Fig. 1 Lower). These ..1. experiments also show that the effect that emerged after 16 41 h persisted at all intervals up to 48 h. Time ofinjection in relation to learning andretention tests. Regardless of whether the drug was administered 1 or 4 h 0 before or immediately after the learning trial, the emergence 1 2 4 8 16 Days of the memory-facilitation effect was the same; it did not FIG. 3. Step-down passive avoidance in mice. Effects of posttrial become evident until an interval of 16-24 h had elapsed oxiracetam on retention performance at various intervals aftertraining between training and retest. At each of the tests after longer and treatment. Oxiracetam (100 mg/kg) was always administered i.p. intervals, the animals treated with oxiracetam showed a immediately after the learning trial. For details see legend to Fig. 1. significant improvement of retention performance by com- parison with the controls. Again, at all shorter intervals the Arecoline, Captopril, and CGP 36742. Results comparable effect oftrainingper se was present (the retest latencies ofthe to those with oxiracetam were found for the emergence of saline- and oxiracetam-treated animals were significantly memory enhancement; with intervals of <24 h, there was no longer than those ofthe no-shock controls), but there were no facilitation. In addition, the time-of-day data were identical to perceptible differences between the drug and saline groups. those found with oxiracetam; at all intervals from 25 to 48 h, For the sake of clarity, in Fig. 2, the latencies are expressed distinct memory facilitation was found (data not shown). Fig. as percentages of the (saline) controls. The data after 1-h 4 shows the prolongation of the retest latencies as percent- pretrial administration are identical to those shown in Fig. 1. ages of the (saline treated) control values. To allow direct If injected 8 h before the learning test, oxiracetam had no comparisons, the oxiracetam results after posttrial treatment facilitating effects on memory; the retest latencies of the already given in Fig. 2 are presented again. controls and the treated animals were about the same (mean ± SEM: vehicle, 19.1 ± 3.8; oxiracetam, 14.3 ± 2.0; median: vehicle, 11.5; oxiracetam, 11.0). DISCUSSION Stability of the memory-facilitating effect. The memory- The most salient result from these experiments is unques- enhancing effect of oxiracetam (posttrial treatment) was still tionably the observation that the memory-enhancing effects clearly perceptible even after intervals as long as 16 days of all the preparations tested became detectable only after a (Fig. 3), the longest time that was tested. It remains to be seen lapse of 16-24 h. The detailed study with oxiracetam dem- what the maximum interval might be and whether it extends onstrated that this delayed appearance of the effect was beyond the course of natural forgetting of the initial learning evident regardless of whether the compound was adminis- experience. tered 1 or 4 h before the learning trial or immediately after it.

-350 -300 -250 200 1150

10

0

lh pre-trial lh 2h FIG. 2. Emergence of the memory facilitation effect induced by oxiracetam under different treatment conditions. Oxiracetam (100 mg/kg) was given i.p. either 1 or 4 h before the learning trial (pretrial) or immediately after it (posttrial). Histograms represent the mean step-down latencies (percentage of controls) of the various experimental groups. (*, P < 0.05; ***, P < 0.001, by generalized Wilcoxon test.) For details, see also legend to Fig. 1. Downloaded by guest on October 1, 2021 2044 Psychology: Mondadori et al. ~~~~~~~~~~~Proc. Natl.~~350Acad Sci. USA 91 (1994) -1 i t -~~300 rl el. cr .-PI -:7 M. O. r., CA ;.z

-PI 11) as1_|l_ 1r 0 s...-.°;lil-200:.It I C() l." In 0 _,,,i_| *~~~~~~I 7 A) Oxiraceta n = i -=h 24h p Arcol 16h

CGP 36742 lh 2h h FIG. 4. Emergence ofthe memory facilitation effect induced by the ACE inhibitor captopril (3 mg/kg), the muscarinic agonist arecoline (0.3 mg/kg), and the GABAB receptor blocker CGP 36742 (10 mg/kg). For purposes of comparison, oxiracetam is also included (same data as in Fig. 2). All treatments were given i.p. immediately after the learning trial. Histograms represent mean step-down latencies (percentage of controls) of the various experimental groups. (*, P < 0.05; ***, P < 0.001, by generalized Wilcoxon test.) For details see also legend to Fig. 1. Accordingly, the time at which the effect of oxiracetam enhancing effects of all the substances used in the present became manifest had nothing to do with the bioavailability of experiments are steroid dependent (12, 13) and steroids are the drug at the time ofretest: the 4-h difference in the time of potent modulators of gene transcription (30, 31). administration in relation to the learning trial had no effect on It is known that by activating immediate early genes (e.g., the time of the emergence of the memory-enhancement c-fos) (32-34) neuronal signals can, within minutes, stimulate effect. The results also showed that the enhancement effect the production of nuclear transcription factors (35-39). The of oxiracetam was obtained in all retention tests performed lesion-induced elevations of the c-fos RNA level in vivo is between 26 and 48 h and was evident for at least 16 days. already past its peak after 2 h, but even after 12 h the mRNA Circadian variations were shown to be irrelevant. still remains elevated (36). The activation of late-response The fact that the time courses of the emergence of the genes (e.g., the nerve growth factor gene) mediated by a memory-enhancing effect of the various substances were c-fos-responsive element in the promoter region becomes practically identical may suggest that their effects ultimately detectable after =3 h and reaches its maximum in -12 h; but come about by a modulation of the same type of process, even after 24 h it has hot reverted to the initial levels (36). regardless of their disparate pharmacological mechanisms. That is simply to say that biochemical activity triggered by The observation that the animals showed clear signs of neuronal signals can outlast the initiating stimulus by many learning at intervals far shorter than the time at which the hours. In the light of these relationships, the earlier obser- enhancement becomes detectable strongly suggests that the vation that memory-improving substances can still exert their enhancement effect was on a later stage of memory, whereas effect even if they are administered several hours after the the shorter-term memory was unaffected by these com- learning trial (40) seems explicable; it is possible that they pounds. The fact that the enhancement persists for at least 16 modulate the protein synthesis initiated by the learning days clearly places the effect as one on long-term memory in terms of any of the various theories of multistage memory experience. Our idea that this pharmacologically sensitive storage (24, 25, 41, 42) dating back to William James (26) and phase, the so-called consolidation phase, coincides with the even earlier. Accordingly, it can be inferred that in these period during which transcription factors brought into play by animals long-term memory comes into operation after =16 h. the learning situation are present still needs to be tested Speculations about the mechanisms undellying long-term experimentally. The ability of such disparate substances to memory often implicitly assume that it includes protein show similar memory-enhancing effects, all delayed by ap- synthesis. This idea, originally proposed by Katz and Hal- proximately the same interval after learning and all steroid stvadt (27), was strongly supported by experiments showing dependent, might be integrated in a hypothesis whereby the that antibiotics-that inhibit protein synthesis can also impair delayed enhancing effects of the substances might corre- memory (28, 29). In the context of the present findings, it is spond with the time when late-gene products form the interesting to note that the amnesia produced by such anti- substrate of memory. biotics emerged after a certain time lag; however, by com- What already seems quite certain is that the delayed parison with our results the delay was mostly found to be emergence of the drug-induced effects on memory applies much shorter (e.g., see refs. 16-20). The present study has only to memory improvement and to antibiotics: in a series now disclosed clear indications about the dynamics of long- of pilot experiments, we noted that the memory-disturbing term memory in mice, and there would therefore seem to be effects of, e.g., diazepam, MK-801, and scopolamine, are good grounds to consider these results in the light of new immediately perceptible-i.e., after a retention interval of findings concerning the dynamics of protein synthesis in only 1 h-and remain stable thereafter. These amnestic neuronal tissue, especially considering that the memory- effects also were not steroid sensitive (12). Downloaded by guest on October 1, 2021 Psychology: Mondadori et al. Proc. Natl. Acad. Sci. USA 91 (1994) 2045 Special thanks to A. Kirkwood for constructive criticism and 21. Kalbfleisch, J. D. S. & Prentice, R. L. (1980) The Statistical assistance in preparing this manuscript and to R. Maier for drawing Analysis ofFailure Time Data (Wiley, New York). the figures. 22. Gehan, E. (1980) Biometrika 52, 203-223. 23. Tukey, J. W. (1977) Exploratory Data Analysis (Addison- 1. Flood, J. F. & Cherkin, A. (1988) Neurobiol. Aging 9, 5-8. Wesley, London), p. 47. 2. Battig, K. (1970) Psychpharmacologia 18, 68-76. 24. Atkinson, R. C. & Shiffrin, R. M. (1968) in The Psychology of 3. Stratton, L. 0. & Petrinovich, L. F. (1963) Psychopharmaco- Learning and Motivation, eds. Spence, K. W. & Spence, J. T. logia 5, 47-54. (Academic, New York), pp. 89-195. 4. Deyo, R. A., Straube, K. & Disterhoft, J. F. (1989) Science 25. Gibbs, M. E. & Ng, K. T. (1976) Neurosci. Lett. 2, 165-169. 248, 809-811. 26. James, W. (1890) Principles ofPsychology (Holt, New York). 5. Mondadori, C. & Etienne, P. (1990) Psychopharmacology 100, 27. Katz, J. J. & Halstead, W. C. (1950) Comp. Psychol. Monogr. 301-307. 20, 1-38. 6. Gold, P. E. & Van Buskirk, R. B. (1975) Behav. Biol. 13, 28. Flexner, J. B., Flexner, L. B. & Stellar, E. (1963) Science 141, 145-153. 57-59. 7. Mondadori, C., Jaekel, J. & Preiswerk, G. (1993) Behav. 29. Davis, H. P. & Squire, L. R. (1984) Psychol. Bull. %, 518-559. Neural Biol. 60, 62-68. 30. Yamamoto, K. R. (1985) Annu. Rev. Genet. 19, 209-252. 8. De Wied, D. & Bohus, B. (1979) in Brain Mechanisms in 31. Schuetz, G. (1988) J. Biol. Chem. 369, 77-86. Memory and Learning: From the Single Neuron to Man, ed. 32. Cole, A. J., Saffen, D. W., Baraban, J. M. & Worley, P. F. Brazier, M. A. B. (Raven, New York), pp. 139-149. (1989) Nature (London) 340, 474-476. 9. Gold, P. E. (1986) Behav. Neural Biol. 45, 342-349. 33. Dragunow, M. & Robertson, H. A. (1987) Nature (London) 10. Mondadori, C., Petschke, F. & Hausler, A. (1989) Pharma- 329, 441-442. copsychiatry 22 (Suppl. 16), 102-106. 34. Dragunow, M. & Robertson, H. A. (1988) Brain Res. 455, 11. Mondadori, C. (1993) Behav. Brain Res. 59, 1-9. 295-299. 12. Mondadori, C., Gentsch, C., Hengerer, B., Ducret, T., 35. Goelet, P., Castellucci, V. F., Schacher, S. & Kandel, E. R. Borkowski, J., Racine, A., Lederer, R. & Hausler, A. (1992) (1986) Nature (London) 322, 419-422. Psychopharmacology 109, 383-389. 36. Hengerer, B., Lindholm, D., Heumann, R., Ruther, U., Wag- 13. Mondadori, C., Ducret, T. & Hiusler, A. (1992) Psychophar- ner, E. F. & Thoenen, H. (1990) Proc. Natl. Acad. Sci. USA macology 108, 11-15. 87, 3899-3903. 14. Geller, A. & Jarvik, M. E. (1968) Psychon. Sci. 12, 169-170. 37. Bishop, M. J. (1991) Cell 64, 235-248. 15. McGaugh, J. L. & Landfield, P. W. (1970) Physiol. Behav. 5, 38. Greenberg, M. E. & Ziff, E. B. (1984) Nature (London) 311, 1109-1113. 433-438. 16. Barondes, S. H. & Cohen, H. D. (1967) Brain Res. 4, 44-51. 39. Morgan, I. J. & Curran, T. (1986) Nature (London) 322, 552- 17. Barondes, S. H. & Cohen, H. D. (1967) Proc. Natl. Acad. Sci. 555. USA 58, 157-164. 40. Mondadori, C., Ducret, T. & Borkowski, J. (1991) Brain Res. 18. Cohen, H. D. & Barondes, S. H. (1967) Proc. Am. Psychol. 555, 107-111. Assoc. 2, 79-80. 41. Waugh, N. C. & Norman, D. A. (1965) Psychol. Rev. 72, 19. Barondes, S. H. & Cohen, H. D. (1968) Science 160, 556-557. 89-104. 20. Cohen, H. D. & Barondes, S. H. (1968) Nature (London) 218, 42. Squire, L. A. (1987) Memory and Brain (Oxford Univ. Press, 271-273. New York). Downloaded by guest on October 1, 2021