The Inverted “U-Shaped” Dose-Effect Relationships in Learning and Memory: Modulation of Arousal and Consolidation

INTRODUCTION

The so called “inverted U-shaped dose-effect curve” (IUSDEC) is a nonlinear relationship which has been frequently reported when studying the negative or positive actions of pharmacological and non-pharmacological treatments on cognitive functions and memory. The effects of increasing dosages of a given compound appear to increase up to a maximum, and then the effects decrease. At the present time it is not easy to elucidate the mechanisms on which the IUSDEC is based and in many instances researchers have described it without attempting any interpretation. Even when an attempt was made, not all mechanisms have been clarified. In some cases (cholinesterase inhibitors) a pharmacological mechanism has been proposed but most of the interpretations rest on the effects of emotional arousal modifications.

Arousal has been invoked as an essential endogenous modulator of memory processes (Gold and Zornetzer 1983; McGaugh 1989a; Lupien and McEwen 1997). The “arousal hypothesis” of IUSDEC is based on the finding that the arousal state can interact with exogenously supplied mnemoactive or presumptive mnemoactive agents to alter their effectiveness at any given dosage. On the other hand, there is considerable evidence that retention can be modulated by the administration of hormones and neuromodulators that are normally released during experiences comparable to those observed during training, as well as evidence that retention can be influenced by treatments that alter the functioning of these systems (McGaugh 1989a). Moreover it is of interest that IUSDEC has been reported after both pre-training and post-training treatments. Some doubts on the importance of active compounds (and the related IUSDEC) in specific mnemonic processing can be advanced in the case of pre-training administration. In fact, it is sufficient to surmise that the active compound lowers pain thresholds and improves olfactory and/or visual acuity: if so, learning would be affected, without any direct interference on memory processing. On the other hand, in post-training administration the possibility of alterating sensory or motor function is avoided, and thus are avoided their possible, but not assessable, effects on mnemonic processing. Indeed, since IUSDEC has been reported many times following post-training administration, the dose-response effects measured in these experimental conditions can be much more closely related to specific actions on mnemonic processing (McGaugh 1989b). Nevertheless, a caution must be expressed before proposing shorp divisions. In fact, if it may be accepted that arousal level influences learning by acting on how the animal acquires the information to be memorized, it cannot be excluded that also during consolidation this same factor may still act on memorization (Gold and Zornetzer, 1983). Thus, even during consolidation, when the animal is neither called to perform any overt behavior nor receives further sensorial inputs, trace processing cannot be thought to be completely unrelated to the emotive/motivational condition of the experimental subject. In fact it is accepted that what may be called an “appropriate” level of neural activation, lasting some time after the training events, is a necessary condition to achieve viable engram elaboration (McGaugh, 1989a). In other words, if the post-training treatment entails an alteration of the background matrix of neural activity currently referred as arousal, engram elaboration may nevertheless be influenced, even if the treatment, per se, is not directly mnemoactive. Presently the experimental findings on IUSDEC in modulation of arousal and consolidation will be reported, starting from the well documented relationships between stress, peripheral epinephrine levels, cerebral norepinephrine levels and learning. All of these show that the combined effect of training and drugs on mnemonic processing follows a biphasic nonlinear trend.

THE INVERTED U-SHAPED RELATIONSHIP BETWEEN LEARNING AND AROUSAL LEVEL: ROLE OF SYSTEMIC EPINEPHRINE AND BRAIN NOREPINEPHRINE

Early investigations showed that retention is modulated by post-training peripheral administration of the adrenal medullary hormone epinephrine (Gold and Van Buskirk, 1975, 1976; Gold, 1984; McGaugh and Gold, 1988). Typically, retention is enhanced by low to moderate dosages and impaired by higher ones. The IUSDEC suggests that plasma epinephrine might be involved in amnesia as well as in memory enhancement (Gold and van Buskirk, 1975). Stressful stimulations, including the mildly stressful ones typically used in studies of learning and memory in animals, cause release of epinephrine from the adrenal medulla; therefore, findings indicating that memory is influenced by post-training injections of epinephrine are consistent with the view that memory storage is modulated by the release of endogenous epinephrine (Gold and McGaugh, 1975). On this point it has been repeatedly shown that plasma epinephrine concentrations are altered by various stressors including the footshocks which are part of the training paradigm. In a study concerning the relationships between footshock intensity, stress amine levels and memory, animals received sham, low, or high footshock training and plasma samples were taken immediately after training. Low footshock training caused an increase in plasma epinephrine similar to that shown by handled, not-footshocked rats; furthermore, low footshock intensity did not produce an optimal inhibitory avoidance retention performance. In contrast, high footshock training, at an intensity that produced very good retention performance, resulted in higher plasma epinephrine levels (McCarty and Gold, 1981). It was also found that in experiments where rats are trained with a mild footshock punishment and given post-training systemic epinephrine, the dose that yields optimal enhancement of memory produces (shortly after training) plasma epinephrine levels comparable to those found in untreated animals, trained with a high footshock intensity that yields good retention (Gold and McCarty, 1981). Gold and van Buskirk (1976) found that the same epinephrine dose that enhanced retention performance after low footshock training could produce amnesia if administered after high footshock training. All these findings support the view that treatments or conditions which alter the endogenous plasma epinephrine levels in response to training, either by increasing or decreasing them, interfere with memory processing, according to an IUSDEC, probably by modifying the emotional arousal level. Therefore, under weak training conditions, those treatments which appear to mimic the adrenergic response to more severe training procedures enhance memory retention.

Stress footshock and other stressors result in a transient decrease in rat's brain norepinephrine content, due to the release and metabolization of this amine prior to its resynthesis. Not surprisingly concentration variations of brain norepinephrine after inhibitory avoidance training with low and high footshocks were found to be analogous to those described above for plasma epinephrine (Gold and van Buskirk, 1978). Low footshocks had no significant effect on brain norepinephrine concentration. High training footshocks were followed by a transient 20% decrease in brain norepinephrine concentration. It should be stressed that the magnitude of this transient post-training decrease is closely correlated with later retention performance over a wide range of experimental conditions. Moreover, it was convincingly shown that epinephrine plasma level modulates brain norepinephrine concentration, thus proving a direct correlation between peripheral and central amine concentrations. The animals which receive a subcutaneous epinephrine injection (0.1 mg/kg) after low footshock training exhibit enhanced retention performance compared to saline-injected rats trained with high footshocks. In addition, both these groups of animals exhibit a 20% decrease in brain norepinephrine concentration. On the other hand, a higher amnesia-producing epinephrine dose (0.5 mg/kg) elicits a 40% decrease in brain norepinephrine concentration (Gold and Van Buskirk, 1978). Thus, not only modulation of plasma epinephrine levels but also modulation of brain norepinephrine levels, due to arousal and pharmacological treatments, directly influence memory retention, and this influence is expressed as an inverted U-shaped relationship.

More recent studies have advanced our understanding of how epinephrine and norepinephrine influence memory consolidation. It has been shown that there is a peripheral-central neuronal pathway mediating the effects of epinephrine on memory consolidation. The nucleus of the solitary tract appears to act as an interface between the peripheral endocrine/autonomic milieu and the neural mechanisms regulating memory consolidation. Epinephrine activates β-adrenoceptors on vagal afferents terminating on brainstem noradrenergic cell groups in the nucleus of the solitary tract. Temporary inactivation of this nucleus with a local infusion of lidocaine blocks epinephrine effects on memory consolidation (Williams and McGaugh 1993). Post-training electrical stimulation of the ascending vagus nerve induces memory enhancement similar to that produced by epinephrine (Clark et al., 1998). Noradrenergic projections from the nucleus of the solitary tract innervate forebrain structures involved in learning and memory and influence norepinephrine release via descending projections to the nucleus paragigantocellularis in the lower medulla, which projects to the locus coeruleus (Van Bockstaele et al., 1998). Noradrenergic neurons in this structure project to the amygdala. In fact, epinephrine affects memory consolidation by activating noradrenergic projection to the amygdala. Infusions of norepinephrine or β-adrenoceptor agonists into the amygdala after training enhance memory consolidation (Ferry and McGaugh 1999) and the activation of both β- and_α-adrenoceptors in the basolateral nucleus of the amygdala appears to be necessary to mediate these noradrenergic influences on memory consolidation (Ferry et al., 1999a, 1999b; Hatfield and McGaugh 1999). In vivo microdialysis and high-performance liquid chromatography studies indicate that an increase of central epinephrine, due to systemic administration of this hormone (Williams et al., 1998) or consequent to emotionally arousing training experiences (Clark et al., 1998; Quirarte et al., 1998), induces the release of norepinephrine in the amygdala. Interestingly, footshock stimulation induces a release of norepinephrine in the amygdala which varies according to the intensity of the stimulus (Galvez et al., 1996; Quirarte et al., 1998). Finally, epinephrine infusion into the nucleus of the solitary tract enhances training-induced norepinephrine release in the amygdala and improves retention performance (Clayton and Williams 2000). These findings support the hypothesis that norepinephrine release in the amygdala plays a critical role in mediating emotional arousal effects on memory consolidation.

IUSDEC AND THE MODULATION OF AROUSAL AND CONSOLIDATION

Other compounds, besides amines, may modify memory retention following a biphasic non- linear dose-effect relationship, probably acting on arousal emotional levels interacting on the plasma epinephrine—brain norepinephrine system. For instance, a close relationship between amine and endorphins or glucose levels has been repeatedly reported (Gold and Zornetzer, 1983; McGaugh, 1989a). Not surprisingly, it has been suggested that there is a relationship between the levels of these compounds and arousal. It is quite interesting that the inversion of the effect on memory retention at high epinephrine dosages was due to the fact that epinephrine caused the liberation of β-endorphins at these dosages (Introini-Collison and McGaugh 1987). Endorphins, systematically administered in the rat, caused disruption of passive avoidance retention, and facilitated active avoidance extinction, following an IUSDEC (Gaffori and De Wied 1982). A quite clear-cut proof that the memory impairment induced by high doses of epinephrine is due to opioid peptide release was the demonstration that this effect was blocked by naloxone (Introini-Collison and McGaugh 1987). On the other hand, the systemic administration of opioid antagonists, by itself, caused memory enhancement, either in a passive avoidance test in the mouse (Introini-Collison and McGaugh 1987) or on recognition memory in the monkey (Aigner and Mishkin 1988). The procognitive effect followed a IUSDEC.

It has been suggested that epinephrine effects on memory might be mediated, at least in part, by the release of glucose (Gold, 1988). Post-training systemic injections of glucose produce non-linear, dose-dependent effects on inhibitory avoidance retention, similar to those reported after epinephrine administration (Gold, 1986). Moreover, plasma levels of glucose measured shortly after training vary according to the footshock intensity used in training. Interestingly, the systemic, post-training glucose administration enhanced the retention performance of a habituation response in the open-field in the mouse, whereas insulin administration acted in the opposite way. The effects of both compounds followed a IUSDEC (Kopf and Baratti 1999). Since glucose readily enters the brain, it may be that glucose affects memory by directly affecting brain glucoreceptors (Oomura et al., 1988). Moreover post-training cerebroventricular glucose administration produces dose-dependent effects on retention (Lee et al., 1988).

Facilitation of memory processes is reported when amphetamine is administered shortly after a training experience. In a manner similar to several other adrenergic agents, systemic amphetamine administration exerts a IUSDEC action on memory processes. In early studies memory facilitation was reported after post-training amphetamine administration at low dosages: higher dosages proved ineffective in altering an appetitive discrimination response (Krivanek and McGaugh 1969). Similarly, high doses of post-training amphetamine resulted in memory disruption in a single-trial inhibitory avoidance paradigm (Weissman, 1967). Amphetamine acts through peripheral catecholamine mechanisms: central administration of amphetamine did not affect retention (Martinez et al., 1980).

It is well known that emotional arousal also activates the hypothalamic-pituitary-adrenocortical axis, elevating plasma levels of corticosterone. Ample evidence indicates that glucocorticoids influence long-term memory consolidation (De Kloet et al., 1999; Roozendaal 2000). It has been shown that their effects on memory follow an inverted-U shape relationship. Acute corticosterone administration influences the spatial memory deficit induced by adrenalectomy in adult rats in a biphasic way (McCormick et al., 1997). Acute post-training administration of low doses of glucocorticoids enhances memory consolidation, in a manner highly similar to that of epinephrine on spatial memory (Sandi et al., 1997) and on fear conditioning (Pugh et al., 1997; Cordero and Sandi 1998). On the other hand, it appears that adrenergic and glucocorticoid hormonal systems interact, influencing memory consolidation. In fact, blockade of the corticosterone stress response, by means of the corticosterone synthesis inhibitor metyrapone, prevents the inhibitory avoidance retention enhancement induced by post-training epinephrine injections or exposure to psychological stress (Roozendaal et al., 1996; Liu et al., 1999).

In humans, the IUSDEC relationship reported between glucocorticoid levels and cognitive function was explained as due to increased arousal. Circadian variations of the effect of corticosterone oral administration on a free recall test in young humans were measured (Fehm-Wolfsdorf et al., 1993). Corticosterone administration suppressed the increased cognitive performance in the morning when endogenous corticosterone levels are at their peak, while it had no effect on cognitive performance when administered at night, when corticosterone is at the lowest concentration. Probably the high endogenous corticosterone levels in the morning corresponded to the peak of the inverted-U shape function between corticosterone levels and cognitive performance, and the corticosterone administration at that time shifted the performance towards a decrease. On the contrary, the corticosterone administration in the evening (at low endogenous corticosterone levels) may not have been sufficient to increase cognitive performance toward the peak of the inverted-U shape function influencing the processes of arousal and selective attention. The inverted-U shape relationship between corticosteroids and memory led to the question of whether this process involves opposing or synergic processes that could be mediated by the two types of adrenal steroid receptors reported to exist in the brain: mineralcorticoid receptors (Type I) and glucocorticoid receptors (Type II). When the performance in the Y-maze of rats administered with either Type I or Type II receptor antagonists was measured, only the Type II antagonist-treated group showed impaired spatial memory performance (Conrad et al., 1996). Successively the authors showed that if a IUSDEC explains the results obtained with memory performance at different corticosterone doses, it may only be related to Type II receptor activation (Conrad et al., 1999).

The report that glucocorticoid effects on memory consolidation enhancement depend on the emotionally arousing content of the administered stimulation (Sandi, 1998; Buchanan and Lovallo 2001), is consistent with extensive evidence indicating that noradrenergic activation in the amygdala is involved in mediating glucocorticoid effects on memory consolidation (De Quervain et al., 1998; Roozendaal 2000, 2002). The infusion into the basolateral amygdala, immediately after training, of the specific Type II agonist RU28362 enhances retention performance while the infusion of the Type II antagonist RU38486 impairs retention performance (Roozendaal and McGaugh 1997). Selective lesions of this nucleus block retention enhancement induced by post-training systemic injections of dexamethasone (Roozendaal and McGaugh 1996). Thus glucocorticoid effects on memory consolidation depend on basolateral amygdala function. Moreover, noradrenergic cell groups of the nucleus of the solitary tract and of the locus coeruleus express high densities of Type II receptors (Harfstrand et al., 1987). Post-training activation of these receptors on noradrenergic cell groups in the nucleus of the solitary tract induces memory enhancement (Roozendaal et al., 1999). As recalled above, this nucleus projects directly to the amygdala and infusion of a β-adrenoceptor antagonist into the basolateral nucleus blocks this glucocorticoid-induced memory enhancement (Roozendaal et al., 1999).

Not all agents that influence memory, presumably acting on arousal levels, act through peripheral adrenergic mechanisms. Post-training subcutaneous injections of ACTH affect later avoidance retention performance. The effects on memory are dose dependent; immediate post-training, systemic administrations of moderate doses of ACTH enhance, and higher doses impair memory storage in a passive avoidance paradigm in the rat (Gold and van Buskirk 1976). It has been shown that ACTH interaction with the level of training-related stress is quite similar to that of amines: a single post-trial administration of ACTH will enhance retention after training with a weak footshock and will impair retention of training with a strong footshock (Gold and Zornetzer 1983). But, on the other hand, systemic ACTH injections do not produce reliable changes in epinephrine and norepinephrine plasma levels. ACTH, then, does not initiate adrenomedullary or sympathetic activity which would normally follow a footshock, and this hormone must therefore act through other (probably central) mechanisms (McCarty and Gold, 1981). ACTH cerebroventricular administration either post-training or 1 h before retention testing enhanced or disrupted the passive avoidance response in the rat according to the dosage-arousal levels (Sahgal et al., 1983).

Similarly, vasopressin effects on learning and memory were discussed as due to emotional arousal level modulation (Sahgal 1984; Ambrogi Lorenzini et al., 1991). Indeed, the initially reported results showing that post-training vasopressin administration facilitated memory processes in a dose-dependent manner, were presented as proof that vasopressin peculiarly enhanced mnemonic capacity (De Wied et al., 1976). Later investigations showed that this effect presumably was due to arousal modifications. Sahgal et al., (1983) found that post-trial cerebroventricular administration of vasopressin improved the performance of some rats in a passive avoidance task, while impairing that of others, and argued that exogenous vasopressin may increase the rats' state of arousal. Thus the amine levels-footshock intensity relationship suggests that if an animal is in a state of low arousal before vasopressin treatment, then an increase in arousal will facilitate performance. However, if the animal is in an optimal or high arousal state, a further increase in arousal will impair performance. It was proposed that vasopressin may be involved in the selection of a high arousal state, or in the regulation of arousal by the noradrenergic dorsal bundle (Sahgal 1984). Finally, oxytocin, another neurosecretory product of the hypothalamo-neurohypophyseal system, appears to have effects opposite to those of vasopressin. Oxytocin impairs passive avoidance performance after post-trial systemic administration and this effect is dose-dependent in a biphasic manner (Bohus et al., 1978; Boccia et al., 1998).

As stated in the Introduction, in several papers the hypothesis of a relationship between IUSDEC and the emotional arousal state is not discussed or presented. For instance, it was found that the administration of D-cycloserine (an NMDA agonist) enhances recognition memory in monkeys, after systemic pre-test administration (Matsuoka and Aigner 1996) and that γ-L-glutamyl-L-aspartate and D-2-amino-5-phosphonovalerate (both NMDA antagonists) after intracerebroventricular post-training administration disrupt the retention of an active avoidance response in the mouse (Mathis et al., 1991). The post-training intracerebroventricular administration of 2-deoxy-D-galactose (a compound antagonizing glycoprotein fucosylation) disrupts the retention of a passive avoidance response in the rat (Ambrogi Lorenzini et al., 1997). The pre-training systemic administration of the nootrope minaprine enhances an active avoidance response in the rat (Ambrogi Lorenzini et al., 1993). The intracerebroventricular administration of the neuropeptide PACAP-38 enhances the passive avoidance response in the rat (Sacchetti et al., 2001). The same compound elicites a similar dose-response effect on the excitability of an in vitro rat hippocampal slice preparation (Roberto et al., 2001). The post-training intrahippocampal administration of nifedipine (a Ca⁺⁺ channel blocker of the class of dihydropyridines) enhances retention of inhibitory step-down avoidance in the rat (Lee and Lin 1991; Quevedo et al., 1998). In all these instances the Authors described a IUSDEC, but did not discuss its possible mechanisms. In some instances that finding was explained by simply suggesting down-regulation or tolerance. In the case of cholinesterase inhibitors, the hypothesis that the activation of presynaptic autoreceptors may play a role in reducing the activity of these compounds was advanced (Braida et al., 1998). In an early study it was reported that systemic post-training administration of physostigmine affects memory processing of an appetitive maze learning task in rats, again following a IUSDEC trend (Stratton and Petrinovich 1963). Similarly, more recent acetylcholinesterase inhibitors (MF201, MF268) were found to antagonize scopolamine-induced amnesia of spatial memory tasks in the rat at low but not at high dosages (pre-trial oral administration) (Braida et al., 1996, 1998), as was found to occur with other cholinergic agonists and cholinesterase inhibitors, which improve performance at low doses but are ineffective at higher ones (Flood et al., 1981; Wanibuchi et al., 1994; Waite and Thal 1995).

CONCLUSION

As initially stated, the IUSDEC is a widely described and poorly understood phenomenon. Thus, in the present paper it is not possible to clarify all the unanswered questions, possibly related to facets and mechanisms not yet adequately investigated. This non-linear relationship has been reported for many active compounds, in several learning paradigms, in some animal species and does not depend on either administration route (systemic or endocerebral) or on administration time (before or after training). The IUSDEC response is possibly a multifactorial phenomenon, and the single components may not be easy to isolate experimentally. Nevertheless some mechanisms, at least, have been well studied. Therefore, for some compounds the “arousal hypothesis” is worth discussing in some depth. On the other hand in many instances the IUSDEC in learning and memory was not attributed to emotional arousal modifications. Although this hypothesis was not discussed in these papers, arousal levels may be involved in even these circumstances, since there is no experimental evidence which completely excludes it. However, the IUSDEC cannot be exhaustively explained by modification of the emotional state. The fact that the IUSDEC has been reported equally for compounds exhibiting quite opposite (positive or negative) effects on learning and memory is not of secondary importance. It may be more difficult to propose this hypothesis for compounds which cause amnesia at low dosages, the effect disappearing at higher ones. Nevertheless, the “arousal hypothesis” appears to be quite interesting and possibly heuristic, and may help to explain at least some of the IUSDEC evidence in learning and memory consolidation. But it must be remembered, when considering this hypothesis, that there are several mechanisms, simultaneous or otherwise, which concur to elevate or decrease emotional arousal levels (adrenal output, pituitary hormones, interoceptive and exteroceptive environmental stimulation and activation of the reticular formation, emotional commutations of the sensorial input). Thus memory processes may be modified by modulating the non specific physiological response to a training experience. So when their net action produces the optimal emotional arousal level, a generalized mnemonic facilitation may result. Conversely, where their net effect decreases arousal beneath a certain level, memorization may be disrupted. Indeed this is true both for pre- and post-training administration because, as recalled in the Introduction, the arousal level may be critical not only during acquisition but also during the early stage of the consolidation process (Gold and Zornetzer, 1983). In a more generalized way, we may surmise that the IUSDEC can be the expression of an “escape” mechanism, which is actuated when sufficiently high concentrations of a given active compound are reached. Sustaining the “escape” mechanism could be a hierarchic organization of mnemonic processing organization, whose components would be progressively more impervious to external influences. Anyhow, since the IUSDEC is a well-documented phenomenon in memorization, the present scarcity of interpretations must be an encouragement to further investigations in this field. Indeed, at least within the “arousal hypothesis” of IUSDEC, there are some findings which may point the way for future research, like the recent ones on how epinephrine and glucocorticoids modulate long-term memory consolidation in animals and human subjects: release of norepinephrine and activation of β-adrenoceptors within the basolateral amygdala appear to be critical in mediating adrenal stress hormone regulation of memory consolidation.