Introduction

Top Summary Introduction Materials & Methods Results Discussion References

LC norepinephrine neurons send noradrenergic fibers into different forebrain structures (Fillenz, 1990) and modulate different cognitive functions, such as arousal, attention, and planning (Crow, 1968; Kety, 1970; Arnsten and Goldman-Rakic, 1985; Arnsten and Leslie, 1991; Riekkinen et al., 1992; Sahakian et al., 1994; Arnsten et al., 1996; Coull et al., 1996). Because ARs are located both presynaptically and postsynaptically, it is not surprising that pharmacological studies have found that noradrenergic ₂-AR agonists and antagonists can affect many behaviors mediated by different neural systems. ₂-ARs are divided into three different subtypes, termed _2A-, _2B-, and _2C-ARs (Kobilka et al., 1987; Regan et al., 1988; Lomasney et al., 1990; Link et al., 1992), and all these subtypes have distinct anatomic distributions in brain areas involved in separate functional systems (Nicholas et al., 1993, 1996; Aoki et al., 1994; Scheinin et al., 1994; MacDonald and Scheinin, 1995; Rosin et al., 1996; Talley et al., 1996). _2A-ARs are located in the LC, elsewhere in the brainstem, and throughout the cerebral cortex and many deeper forebrain structures; _2B-ARs are found nearly exclusively in the thalamic nuclei; and _2C-ARs are located in the hippocampus, cerebral cortex, and striatum. Importantly, in the caudate and accumbens nuclei, _2C-ARs predominate, suggesting that this receptor subtype may mediate effects of ₂ agonists and antagonists on modulation of cognitive functions via frontostriatothalamic feedback loops.

The behavioral functions of different subtypes of ₂-ARs have been difficult to study because there are no ligands that selectively activate or block only one of the three subtypes. Therefore, to study the role of different ₂-ARs in behavioral functions and to better evaluate the potential for cognition-enhancement by subtype-selective ₂-AR active drugs, we have started to investigate the effects of overexpression and knockout of ₂-AR subtype genes on different behaviors in mice and how these manipulations affect reactions to subtype-nonselective ₂-AR agonists and antagonists (MacDonald et al., 1997). Sallinen et al. (1997) found that spontaneous locomotor activity and the effects of an ₂-agonist or -antagonist on brain monoamine turnover and on motor activity were not influenced by altered _2C-AR expression. However, mice with tissue-specific overexpression of _2C-AR were slightly more sensitive and KO mice were less sensitive to the hypothermic effects of the agonist DEX than their controls. The OE mice have ~3-fold elevated densities of _2C-AR in regions that normally express this subtype (the caudate-putamen and CA1 region of the hippocampus) (Sallinen et al., 1997). In KO mice, the total ₂-AR density is diminished in brain areas normally expressing the _2C subtype (Link et al., 1995). In this study, cue and spatial WM navigation were evaluated in OE mice and control (WT) mice and spatial navigation was evaluated in KO mice and CKO because our previous studies have shown that ₂-AR agonists may modulate performance in these tests (Sirviö et al., 1991, 1992). Furthermore, to study the specificity of the WM navigation changes observed in OE mice, we also compared behaviors of WT and OE mice in OF and PA tests (Riekkinen et al., 1992). In addition, cortical EEGs were investigated. All these tests are modulated by ₂-adrenergic drugs (Riekkinen et al., 1990, 1993). The OF test is a measure of anxiety; thus if animals are anxious, they will spend more time in the peripheral annulus of the OF arena. In the WM task, the escape platform is located in the middle annulus, and anxiety-induced swimming in the peripheral annulus reduces the chances to find the platform. The PA test measures stimulus-response learning. We aimed to test whether OE mice have a learning defect in the PA paradigm. Cortical EEG monitoring was performed to detect differences in general arousal. An appropriate state of arousal is needed to perform accurately in the WM test. We also studied whether OE mice have an altered response to the sedative action of ₂-adrenergic drugs. The effects of _2C-AR overexpression may be due to an increase in basal and agonist-activated signal transduction mediated through this receptor subtype. Elevated levels of receptor expression can lead to an increase in basal, constitutive signal transduction, which can mimic agonist activation (Samama et al., 1997). To examine the role of enhanced signaling of _2C-AR on the defect in WM performance in OE and WT mice, we examined the effect of ATI, a highly ₂-AR-specific but not subtype-selective antagonist. Furthermore, to assess the role of _2C-ARs in the sedative actions of ₂-AR agonists in WT and OE mice, we also compared the effect of DEX, a highly specific ₂-AR agonist that does not differentiate between ₂-AR subtypes. There has been no rigorous evaluation of ectopic _2C expression in brain regions outside those normally expressing _2C-AR. Such ectopic expression could account for the differences seen in OE and WT mice in their WM behavior. To study the problem of ectopic expression, we used KO mice in a control WM study.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Animals

Adult (3-5 months old;: weight, 20-28 g) female FVB/N (control mice for OE mice, WT) and FVB/N-TgNAdra2C (OE) (both OE and WT mice were from Stanford University Medical Center, Stanford, CA) were used in the study (total, 109 OE and 106 WT mice). This strain of genetically engineered mice, which has tissue-specific overexpression of _2C-ARs, was generated at Stanford University by pronuclear injection using standard methods. OE mice have ~3-fold elevated densities of _2C-AR in their caudate-putamen and CA1 region of the hippocampus (Sallinen et al., 1997).

Adult (3-5 months old; weight, 20-28 g) female C57BL/6J for CKO and KO mice also were used in the study (total, 34 KO and 33 CKO mice). A strain of genetically engineered mice that had targeted inactivation of the _2C-AR gene was generated at Stanford University (both KO and CKO mice were from Stanford University Medical Center). KO mice do not express a functional _2C transcript (Link et al., 1995).

All the mice used in the current experiments were bred in the Central Laboratory Animal Facility of the University of Turku, Finland, according to standard procedures. The mice were housed five per cage. The housing conditions (National Animal Center, Kuopio, Finland) were controlled; constant temperature (21 ± 1°), humidity (50-60%), and light period (lights on from 7:00 a.m. to 7:00 p.m.) were maintained; and food and fresh water were freely available. All experiments were performed during the light period. The study was approved by the Animal Welfare Committee of the University of Kuopio.

Drugs

ATI (3-300 µg/kg; Orion Corporation Farmos, R & D Pharmaceuticals, Turku, Finland), a selective ₂ antagonist (Scheinin et al., 1988; Virtanen et al., 1989), and DEX (0.5-300 µg/kg; Orion Corporation Farmos, R & D Pharmaceuticals), a selective ₂ agonist (MacDonald et al., 1991; Savola and Virtanen, 1991), were dissolved in saline and injected subcutaneously (5 ml/kg). ATI and DEX were injected 20 min before daily behavioral testing, and for four WM groups, ATI (30-300 µg/kg) was administered immediately after daily training.

WM

Cue and spatial navigation were evaluated in a WM pool (black; diameter, 59 cm). Four starting points (north, south, east, and west) were located at the pool rims. A black 3.5- × 3.5-cm platform was located in the middle annulus, 0.5 cm above (visible) or 1 cm below (hidden) the water line. The visible platform had a 10-cm-high white mast. The location of the visible platform was changed daily during the visible platform training (cue navigation), but the hidden platform (place navigation) was kept in a fixed place. The mice were facing the wall and were gently released to begin the first daily trial from the starting position farthest from the platform. The other four trials were started in a semirandom order. Five trials with a cutoff value of 50 sec were tested every day (5-sec reinforcement on the platform). The mice that did not find the platform were placed on it for 5 sec. A 30-sec recovery period was allowed between the trials. A computerized video monitoring system (HVS Image, Hampton, UK) calculated the number of animals that failed to find the platform, swimming speed, and swimming in three annuli of equal surface area. Our preliminary data showed that the KO mice and their controls performed better than OE mice and their controls in the WM task. The KO and CKO mice completed the task (i.e., they found the platform) almost every time. Therefore, the parameter "platform finding" (completing the task) was too insensitive to detect any difference between CKO and KO mice, so we applied here a widely used measure, swimming distance (i.e., the distance the animals swam before completing the task or reaching the cutoff time value).

Experiment 1. We used the following treatments for both the WT and the OE mice: saline, ATI 30 and 300 µg/kg (15 OE or WT mice/group). First, we trained the mice to find a visible platform (cue navigation) that was moved every day to a novel position. Training continued for 5 consequent days (five trials/day as described above).

Experiment 2. The following treatments were used for both the WT and the OE mice: saline, ATI 30 and 300 µg/kg immediately after daily training (11-13 OE or WT mice/group). These mice had only 5 days of hidden platform training.

Experiment 3. The following treatments were used for both the CKO and the KO mice: saline, DEX 0.5 and 10 µg/kg (10-13 KO or CKO mice/group). These mice had only 3 days of hidden platform training.

Cortical EEG measurements

Two stainless steel screws acting as epidural recording electrodes were bilaterally implanted 1.0 mm anterior and 2.0 mm lateral to bregma. Two additional screws acting as indifferent and ground electrodes were implanted in the nasal bone and above the cerebellum. The effects of DEX (3-300 µg/kg) and ATI (3-300 µg/kg) on cortical EEG activity were measured by recording five 4-sec-long artifact-free EEG episodes with relaxed waking-immobile animals (24 OE and 22 WT). EEG spectra were divided into the frequency bands of 1-4 Hz , 4-8 Hz , 8-12 Hz , 12-20 Hz , and two upper frequency bands of 20-30 and 30-60 Hz.

OF

A dry Morris WM pool was used to perform an OF task. The mice were placed on the bottom of the pool with the nose pointing toward the wall. A computer connected to an image analyzer calculated the total walking distance, walking speed, and time spent in the three annuli. Effects of SAL and DEX and ATI were tested. One 50-sec trial of free exploration was given to all mice.

PA

A single training trial step through PA test was conducted after the WM study. In the training trial, the experimenter injected the mice with SAL, DEX (2.0, 5.0, or 20.0 µg/kg), or ATI (30 or 300 µg/kg); 20 min later, the mice were placed at the far end of the bright side of the PA box and the guillotine door was opened 30 sec later. The latency to enter the dark side was measured. Ten seconds after the entry (all four paws inside), an electric shock (DC current 0.20 mA, 3-sec duration; Shock Source 521 C, Campden Instruments, Leicestershire, UK) was given. After 24 hr, the mouse was placed again on the bright side, and 30 sec later, the guillotine door opened (testing trial; maximum testing time of 360 sec as a cutoff value). The reentry latency was measured.

Statistics

The effects of strain, drug treatment, and their interaction (strain × treatment) were evaluated using analysis of variance for repeated measurements and simple factorial analysis of variance.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

WM experiment 1: ATI 30 and 300 µg/kg before daily training

Effects of _2C overexpression on WM escape behavior. Visible platform days 1-5. Compared with WT mice, OE mice did not find the visible platform as accurately [strain: F(1,28) = 19.251, p < 0.001] (Fig. 1A). During visible platform training, the swim pattern of OE mice was not different from WT mice as they swam equally in the peripheral, middle, and inner annuli [strain: F(1,28)  4.11, p  0.052, for all].

View larger version (38K): [in this window] [in a new window]   Fig. 1.   Mice overexpressing (OE) 2C-AR were impaired compared with the control (WT) mice in WM escape performance: fewer OE mice found the platform (percent failures to find platform), and the time spent in the peripheral annulus, which did not contain the platform, was higher (percent swimming in peripheral annulus). The escape platform was located in the middle annulus. The effects of ATI 30 µg/kg are also shown. Left, percentage of mice that failed to find the platform (y-axis), Right, time spent in the peripheral annulus (y-axis). The x-axis shows the training day. Daily group mean values are shown. The p values of significant strain and treatment effects and strain × treatment (S × T) interactions are shown. , WT saline; , OE saline; , WT ATI 30 µg/kg; , OE ATI 30 µg/kg. A, Atipamezole (ATI; 30 µg/kg subcutaneous) stimulated accuracy of WM visible platform (place navigation) navigation and decreased swimming in the peripheral annulus in WT and OE mice. B, During drug-free training days, control and previously ATI 30 µg/kg-treated mice performed equally. C, ATI 30 µg/kg was not as effective at improving platform finding in OE mice during the hidden platform (spatial navigation) test but stimulated accuracy of WT mice. However, ATI 30 µg/kg decreased swimming in the peripheral annulus equally effectively in both strains.

Effects of ATI 30 µg/kg on WM escape behavior. Visible platform. ATI 30 µg/kg only tended to improve platform finding in WT and OE mice [treatment: F(1,56) = 3.737, p = 0.058] (Fig. 1A). ATI 30 µg/kg decreased swimming in the peripheral annulus [treatment: F(1,56) = 14.99, p = 0.001]. ATI 30 µg/kg increased swimming in the middle annulus, which contained the platform, more effectively in OE mice than in WT mice [treatment: F(1,56) = 14.14, p = 0.001, strain × treatment: F(1,56) = 7.40, p = 0.009). ATI 30 µg/kg did not affect swimming in the inner annulus [treatment: F(1,56) = 2.97, p > 0.05].

Effects of ATI 300 µg/kg on WM escape behavior. Visible platform. ATI 300 µg/kg improved visible platform finding in OE mice more effectively than in WT mice [treatment: F(1,56) = 14.022, p = 0.001; strain × treatment: F(1,56) = 5.548, p = 0.022] (Fig. 2A). ATI 300 µg/kg decreased swimming in the peripheral annulus equally in WT and OE mice [treatment: F(1,56) = 32.16, p < 0.001; strain × treatment: F(1,56) = 0.61, p > 0.1]. ATI 300 µg/kg did not affect swimming in the middle annulus [treatment: F(1,56) = 3.58, p = 0.064] but increased swimming in the inner annulus equally in OE and WT mice [treatment: F(1,56) = 18.47, p < 0.001].

View larger version (39K): [in this window] [in a new window]   Fig. 2.   Platform finding (percent failures to find platform) and the time spent in the peripheral annulus (percent swimming in peripheral annulus) of OE and WT mice treated with ATI 300 µg/kg. Left, percentage of mice that failed to find the platform (y-axis). Right, time spent in the peripheral incorrect annulus that did not contain the platform (y-axis). The x-axis shows the training day. Daily group means are shown. The p values of significant strain and treatment effects and strain × treatment (S × T) interactions are shown. , WT saline; , OE saline; , WT ATI 300 µg/kg; , OE ATI 300 µg/kg. A, Atipamezole (ATI; 300 µg/kg subcutaneous) stimulated OE mice WM visible platform finding more than WT mice and decreased swimming in the peripheral annulus to the same extent in both strains. B, During drug-free training days, the WT and OE mice previously treated with ATI 300 µg/kg found the visible platform more effectively than their saline controls. C, ATI 300 µg/kg improved hidden platform finding equally in OE and WT mice, but it decreased swimming in the peripheral annulus more effectively in OE mice.

WM experiment 2: ATI 30-300 µg/kg after daily training

Effects of ATI 30-300 µg/kg after daily training on WM escape behavior. Hidden platform days 1-5. ATI 30-300 µg/kg after daily training had no effect on platform finding, swimming in the peripheral, middle or inner annuli, or swim speed [treatment: F(1,44)  1.79, p > 0.1, for all] (data not shown).

WM experiment 3: KO DEX 0.5-10 µg/kg before daily training

Effects of KO on WM escape behavior. KO mice found the hidden platform as accurately as CKO mice [strain: F(1,23) = 2.065, p > 0.1]. The swimming distance of KO and CKO mice was equal [strain: F(1,23) = 2.53, p > 0.1] (Fig. 3). During hidden platform training, the swim pattern of KO mice was not different from CKO mice as they swam equally in the peripheral, middle, and inner annuli [strain: F(1,23)  1.42, p > 0.1, for all].

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Swimming distance of KO mice and their CKO treated with DEX 10 µg/kg (y-axis). The x-axis shows the training day. Daily group means are shown. The p values of significant strain and treatment effects and strain × treatment (S × T) interactions are shown. , CKO saline; , KO saline; , CKO DEX 10 µg/kg; , KO DEX 10 µg/kg. Saline-treated CKO and KO mice had equal swimming distances. KO mice were less sensitive to the DEX 10 µg/kg-induced increase in swimming distance than were CKO mice.

Effects of DEX 0.5-10 µg/kg on WM escape behavior. DEX 10 µg/kg increased swimming distance more effectively in CKO than in KO mice [treatment: F(1,43) = 22.47, p < 0.001; strain × treatment: F(1,43) = 4.18, p = 0.047] (Fig. 3). DEX 10 µg/kg had no effect on the other parameters measured [treatment: F(1,43)  1.86, p > 0.1, for all]. DEX 0.5 µg/kg had no effect on any of the parameters measured [treatment: F(1,41)  1.93, p > 0.1, for all].

Cortical EEG measurements

WT and OE mice showed no difference in delta amplitudes measured during base-line recordings [strain: F(1,44) = 1.82, p > 0.1]. DEX 3-300 µg/kg increased delta amplitude in WT and OE mice [treatment: F(1,44)  5.22, p < 0.05, for all] (Fig. 4A). ATI 3 µg/kg decreased delta amplitude [treatment: F(1,44) = 5.15, p = 0.028]. ATI 30-300 µg/kg had no effect on delta amplitude [treatment: F(1,44)  2.77, p > 0.05].

View larger version (46K): [in this window] [in a new window]   Fig. 4.   A, DEX 3-300 µg/kg increased relative (vehicle = 100%)  (1-4 Hz) amplitude in WT and OE mice measured from frontal epidural EEG electrodes. ATI 3 µg/kg decreased  amplitude, but ATI 30-300 µg/kg had no effect. B, In an OF test, DEX 0.5-5 µg/kg and ATI 30-300 µg/kg had no effects on ambulation in the peripheral annulus in WT and OE mice. C, Passive avoidance training latencies of WT and OE mice did not differ. DEX 2-20 µg/kg increased training latency equally effectively in both strains. D, Passive avoidance testing latencies of WT and OE mice did not differ. DEX 2-5 µg/kg had no effect on testing latency, whereas DEX 20 µg/kg decreased training latency as effectively in both strains. *, p < 0.05 versus saline-treated control mice.

OF

OE and WT mice had similar walking speed and explored equally the peripheral, middle, and inner annuli [strain: F(1,18)  3.706, p  0.071, for all]. DEX 0.5-5 µg/kg and ATI 30-300 µg/kg had no effect on walking speed or ambulation in the peripheral, middle, and inner annuli [treatment: F(1,32)  3.1, p > 0.05, for all] (Fig. 4B).

Passive avoidance, DEX 2-20 µg/kg and ATI 30-300 µg/kg

The training and testing latencies of WT and OE mice did not differ [strain: F(1,27)  3.8, p > 0.05, for both] (Fig. 4, C and D).

Training. ATI 30-300 µg/kg had no effect on training latency [treatment: F(1,53)  1.598, p > 0.1] (data not shown). DEX 2-20 µg/kg increased training latency [treatment: F(1,34)  6.080, p  0.002] (Fig. 4C).

Testing. ATI 30-300 µg/kg had no effect on testing latency [treatment: F(1,53)  0.889, p > 0.1] (data not shown). DEX 2-5 µg/kg had no effect on testing latency [treatment: F(1,34/1,35)  2.038, p > 0.1]. DEX 20 µg/kg decreased testing latency in both WT and OE mice [treatment: F(1,34) = 23.319, p < 0.001] (Fig. 4D).

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

OE mice developed an abnormal escape pattern (Simon et al., 1994), characterized by increased swimming in the peripheral annulus of the pool (near the walls), and could not find the visible or hidden platform as accurately as the WT mice. This defective exploration pattern is likely to result from _2C-AR overexpression because a dose-response relationship for ATI was observed. First, we observed that ATI 30 µg/kg slightly decreased swimming near pool walls during hidden and visible platform finding in both groups, but the OE mice were impaired compared with the WT mice at this dose. In contrast, ATI 300 µg/kg fully blocked the abnormal search pattern in OE mice. Second, the treatment with ATI 30 µg/kg did not stimulate accuracy to find the platform during the visible platform stage in OE or WT mice, but during the hidden platform stage, ATI 30 µg/kg clearly and exclusively stimulated the accuracy of WT mice. This suggests that the low dose of ATI 30 µg/kg is not sufficient to overcome the increased _2C-AR function in OE mice, but in WT mice it is able to effectively antagonize _2C-AR function and improve accuracy of WM escape behavior. In contrast, a higher dose of the ₂-AR antagonist ATI 300 µg/kg may block _2C AR function in both WT and OE mice and thus improve WM navigation. The high dose of ATI 300 µg/kg improved visible platform finding of OE mice to equal accuracy as that of WT mice but had no further effect in WT mice. This finding that ATI stimulates visible platform navigation only in OE mice indicates that the _2C-AR may play a more crucial role in the control of those brain areas important for cue navigation than the other ₂-AR subtypes.

We evaluated the WM performance using different versions of WM training: visible platform, visible platform without drug treatment, and hidden platform. The strategies used to locate a visible platform are different from those needed for hidden platform navigation. Visible platform training is a nonspatial version of the WM task. During visible platform training, animals are required to seek the clearly visible escape platform without the need to use distal extra-maze cues. The use of extra-maze cues is disruptive during visible platform training because the platform location is changed daily. During the hidden platform training, the use of extra-maze cues is unavoidable because no intra-maze cues are present. The withdrawal of drug treatment at the end of the visible platform training phase was aimed at investigating whether the drug treatments had any long lasting effects on memory of WM performance. The present findings suggest that the _2C-AR overexpression does not impair and ATI does not stimulate cue and spatial navigation by affecting memory processes per se. First, the OE mice were impaired already during the first day of training, the learning curves of WT and OE mice remained parallel during the experiment, and ATI failed to modulate the slope of the learning curves. Traditionally, parallel learning curves have been related to defects in processes other than learning and memory. Furthermore, we observed that administration of ATI 30 or 300 µg/kg immediately after daily training trials had no effect on performance, thereby ruling out any beneficial effects on memory consolidation processes. Finally, during the drug withdrawal training period on days 6 and 7, the performance of mice treated with ATI during the training days 1-5 was superior to that of saline-treated WT and OE mice. During days 1-5, no difference existed in the escape ability of WT and OE mice receiving daily ATI 300 µg/kg treatment, but discontinuation of ATI on days 6 and 7 released functioning of overexpressed _2C-ARs and disturbed the accuracy of OE mice. Therefore, it is possible that _2C-AR overexpression disrupts the accuracy of WM navigation by inhibiting the execution of the normal escape behavior required for success in the WM paradigm and releases abnormal exploration patterns. However, because OE mice treated with ATI 300 µg/kg performed better during the drug withdrawal stage than the vehicle-treated OE mice, it is possible that _2C-ARs may also, to some extent, control memory acquisition of cue navigation performance.

Previous studies describing the distribution of _2C-ARs and the anatomic pattern of _2C-AR overexpression (Sallinen et al., 1997) suggest that the striatum and CA1 region of the hippocampus may represent regions that are under increased _2C-AR modulation in our OE mouse strain. Therefore, it is relevant to note that behavioral studies have characterized the consequences of brain lesions restricted to hippocampus and ventral or dorsal striatum. Hippocampal lesions do impair spatial navigation performance in the WM test (Marston et al., 1993), and this defect is believed to result from impaired formation of new relational memory engrams and from defects in spatial processing. A lesion of the nucleus accumbens, a region of the ventral striatum connected with the hippocampus and the prefrontal cortex, slightly impairs accuracy of WM navigation (Annett et al., 1989) by disrupting the execution of complex movement patterns. In contrast, destruction of the dorsal striatum has no effect on spatial navigation and selectively impairs cue navigation, indicating that this region is involved in a brain memory system that processes a different kind of information than the hippocampal system. Therefore, these previous studies suggest that _2C-ARs may modulate those brain areas involved in nonspatial and spatial memory and also affect execution of learned escape behaviors. Indeed, we found that KO mice were less sensitive than CKO mice to the defect in spatial navigation induced by DEX 10 µg/kg. This finding with KO mice supports our contention that _2C-overexpression in anatomically relevant brain areas is responsible for the performance defect in OE mice.

Our control studies indicated that the OE mice were not different in other behavioral and physiological measures, providing evidence for the specificity of the WM abnormality. We analyzed PA and OF behaviors and cortical electrical arousal, but no differences between the OE and WT mice were detected. The lack of a strain difference in the OF test indicates that differences in anxiety do not play a role in the abnormal WM search pattern of OE mice. The PA performance of OE mice was normal, suggesting that it is not a simple defect in stimulus-response learning, which is the foundation of the impaired WM accuracy of OE mice (Thomas, 1996). In addition, we tested the action of DEX to increase and ATI to suppress cortical slow waves and observed that _2C-AR OE did not modulate the effects of ₂-AR active drugs on cortical EEG arousal (Riekkinen et al., 1990, 1993). EEG measurements revealed no differences between OE and WT mice, thereby ruling out altered arousal state as the cause of the WM defect of OE mice. Also, the defects in performance after DEX during PA training and testing trials and those during the visible platform navigation stage were of equal magnitude in WT and OE mice. These defects in WT and OE mice are likely to result from the sedative effects of DEX. The need for a greater DEX dose to disrupt WM navigation than that needed to interfere with PA behavior may be simply related to the high arousal and pronounced LC firing rate occurring during the WM testing.

The decreased swim speed in OE mice may not be related to overexpression of _2C-AR because a high dose of ATI (300 µg/kg) retarded swim speed in OE and WT mice to the same extent. Therefore, this difference between the WT and OE mice is the only one that may not be specifically related to overexpression of _2C-AR.

In summary, our results revealed that OE mice were impaired only in the WM test, which requires organization of a complex escape behavior. We also observed an alteration in the dose-response relationship of the beneficial effect of an ₂ antagonist on WM performance in _2C-AR OE mice. The poor WM performance of OE mice and the improving effect of ATI on navigation performance are difficult to attribute simply to disrupted learning and memory. It is likely that _2C-AR overexpression may not impair learning per se because the OE and WT mice had parallel learning curves, ATI 300 µg/kg treatment did not affect the slope of the learning curves, and ATI did not improve memory consolidation. Therefore, _2C-ARs may modulate frontostriatothalamic feedback loops (Scheel-Krüger and Willner, 1991; Coull et al., 1995) and influence complex spatial navigation behaviors. These results raise the possibility that antagonists selective for the _2C-AR subtype and agonists devoid of any _2C-AR affinity could modulate cognition more favorably than subtype-nonselective drugs. For example, _2C-AR subtype-selective antagonists could be effective in the treatment of some of the cognitive dysfunctions that are characterized by impaired frontostriatothalamic functions, such as planning and controlling of complex escape behavior, and are susceptible to ₂-AR subtype-nonselective antagonists. Furthermore, ₂-AR agonists devoid of any _2C-AR affinity may be also more effective than the currently available ₂-AR agonists in enhancing memory and impulse control functions of prefrontal cortex (Steere and Arnsten, 1994).