Procholinergic and Memory Enhancing Properties of The

Home , Arecoline

Molecular Psychiatry (2006) 11, 187–195 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp ORIGINAL ARTICLE Procholinergic and memory enhancing properties of the selective norepinephrine uptake inhibitor atomoxetine ET Tzavara, FP Bymaster, CD Overshiner, RJ Davis, KW Perry, M Wolff, DL McKinzie, JM Witkin, and GG Nomikos Eli Lilly and Company, Lilly Corporate Center, Neuroscience Discovery Research, Indianapolis, IN, USA

Atomoxetine has been approved by the FDA as the first new drug in 30 years for the treatment of attention deficit/hyperactivity disorder (ADHD). As a selective norepinephrine uptake inhibitor and a nonstimulant, atomoxetine has a different mechanism of action from the stimulant drugs used up to now for the treatment of ADHD. Since brain acetylcholine (ACh) has been associated with memory, attention and motivation, processes dysregulated in ADHD, we investigated the effects of atomoxetine on cholinergic neurotransmission. We showed here that, in rats, atomoxetine (0.3–3 mg/kg, i.p.), – increases in vivo extracellular levels of ACh in cortical but not subcortical brain regions. The marked increase of cortical ACh induced by atomoxetine was dependent upon norepinephrine a-1 and/or dopamine D1 receptor activation. We observed similar increases in cortical and hippocampal ACh release with methylphenidate (1 and 3 mg/kg, i.p.) – currently the most commonly prescribed medication for the treatment of ADHD – and with the norepinephrine uptake inhibitor reboxetine (3–30 mg/kg, i.p.). Since drugs that increase cholinergic neurotransmission are used in the treatment of cognitive dysfunction and dementias, we also investigated the effects of atomoxetine on memory tasks. We showed that, consistent with its cortical procholinergic and catecholamine-enhancing profile, atomoxetine (1–3 mg/kg, p.o.) significantly ameliorated performance in the object recognition test and the radial arm-maze test. Molecular Psychiatry (2006) 11, 187–195. doi:10.1038/sj.mp.4001763; published online 18 October 2005 Keywords: ADHD; atomoxetine; acetylcholine; microdialysis; reboxetine; methylphenidate

Introduction attention and working memory depend upon the integral function of cortical and hippocampal The norepinephrine (NE) uptake inhibitor cholinergic afferents (reviewed in Everitt and Rob- atomoxetine constitutes today the only first-line bins3). Procholinergic drugs such as cholinesterase alternative to the psychostimulant drugs, methylphe- inhibitors are used in the therapeutics of Alzheimer’s nidate and D-amphetamine for pharmacotherapy disease and other neurodegenerative diseases with of attention deficit/hyperactivity disorder (ADHD). cognitive impairment. Recently, the ability of atypical The psychostimulants are currently classified as antipsychotics to reduce negative symptoms and to schedule II drugs of the Controlled Substance improve performance in cognitive tasks has been Act and have been used for half a century in associated with the stimulatory effects of these agents the treatment of ADHD, the most common psychiatric on neocortical and hippocampal acetylcholine (ACh) disorder in children (3–10% prevalence) that release.4,5 often (60% of the cases) persists into adulthood. Although the effects of atomoxetine on DA and ADHD is characterized by attention deficits, hyper- NE efflux in the brain have been reported,6 its effects activity and impulsivity, and atomoxetine signifi- on cholinergic neurotransmission have not been cantly improves these symptoms in children and studied. Therefore, we assessed the ability adults with ADHD.1,2 of the compound to enhance ACh efflux in cortical In the central nervous system, cholinergic (medial prefrontal cortex and hippocampus) neurons modulate information flow in cortical and subcortical (nucleus accumbens) regions by and subcortical regions implicated in vigilance and in vivo microdialysis in rats. We also investigated cognition. In particular, sustained and selective in this species the procognitive potential of the drug in two animal cognition models, the radial-arm Correspondence: Dr GG Nomikos, Eli Lilly and Company, Lilly maze and the object-recognition tests. We report Corporate Center-DC0510, Indianapolis, 46285, IN, USA. E-mail: [email protected] increases in cortical ACh efflux engendered by Received 29 March 2005; revised 19 August 2005; accepted 10 atomoxetine and parallel enhancement in memory September 2005; published online 18 October 2005 performance. Atomoxetine increases cortical acetylcholine ET Tzavara et al 188 Materials and methods were used were based on those previously used in in vivo microdialysis studies and are relevant to clini- Animals cally used doses.6 All studies were performed according to the guide- Data (n ¼ 4–7 rats per group) were expressed as lines set forth by the National Institutes of Health and multifold change from baseline, which is the average implemented by the Animal Care and Use Committee of the five basal values before any manipulation and of Eli Lilly and Company. Male Wistar or Sprague– were analyzed either with one-way, that is, treatment Dawley rats (250–300 g, purchased from Harlan (between subjects variable), two-way, that is, treat- Sprague–Dawley, Indianapolis, IN, USA) were used ment (between subjects variable) Â time (within sub- for experiments. jects variable) or three-way (treatment 1 Â treatment In vivo microdialysis 2 Â time) ANOVA followed by Duncan’s test. The effects of each of the drugs are presented both over a Surgical procedures. At 2 weeks prior to the course of time every 15 min after the injection of the microdialysis experiments, the rats were drug as well as overall average effects during the 3-h anesthetized with a mixture of chloral hydrate and observation period after the injection of the drug pentobarbital (170 and 36 mg kgÀ1 in 30% propylene (index of area under curve). glycol and 14% ethanol), placed in a stereotaxic apparatus and implanted with a guide cannula Object recognition test (Bioanalytical Systems, West Lafayette, IN, USA Each rat was placed in a clear 25 Â 25 cm2 Plexiglas (BAS)) in the hippocampus (coordinates AP: À5.2, observation box with two identical objects, desig- ML: 5.2, DV: À3.8), medial prefrontal cortex (AP: 3.2, nated A. The rat was allowed to explore for 2 min and ML: 0.6, DV: À2.2) or nucleus accumbens (AP: 1.6, the time interacting with the objects (sniffing, gnaw- ML: 1.2, DV: À6.3) according to Paxinos and Watson.7 ing and behavior oriented to an object) was recorded. At 24 h before testing, a 4 mm (hippocampus, Behavior oriented to the object distinguishes between prefrontal cortex) or a 2 mm (nucleus accumbens) accidental sitting, standing on the object or touching concentric microdialysis probe (BAS, model BR-4 or the object when passing by, and active interaction/ BR-2) was inserted through the guide cannula. The exploration of the object. After a 3-h delay, the rat was actual location of the probes was verified returned to the observation box for the test trial. The histologically at the end of the experiment. test box contained a familiar object (object A) and a novel (object B) object. In the test trial, the objects ACh measurements. ACh determination in were placed at the exact same position as in dialysates from the different brain regions was the learning trial. The amount of time spent interact- performed as described8 with some modifications.9 ing with each object during the 2-min test was On the day of the experiment, a modified Ringer’s recorded. Atomoxetine was administered orally in

solution (147.0 mM. NaCl, 3.0 mM KCl, 1.3 mM CaCl2, 5% acacia over a dose range of 0.3, 1, 3 and 10 mg/kg, 1.0 mM MgCl2, 1.0 mM Na2HPO4 Â 7H2O, 0.2 mM 1 h before the first trial. Results are expressed as NaH2PO4 Â H2OpH¼ 7.25) supplemented with percent time exploring the novel object during the 0.1 mM neostigmine was perfused at a rate of 2.4 ml/ retention test (i.e., tB Â 100/(tA þ tB), where tA and tB min in the hippocampus, prefrontal cortex or nucleus are the time spent during test trial with familiar object accumbens. Samples were collected every 15 min and A and novel object B, respectively), and were analyzed immediately, on-line, with HPLC coupled to analyzed with an one-way ANOVA followed by electrochemical detection, with a 150 Â 3 mm ACH-3 Dunnett’s test. column (Environmental Sciences Associates (ESA), Inc., MA, USA) maintained at 351C. The mobile phase Radial arm-maze (100 mM di-sodium hydrogen phosphate, 2 mM 1- The effects of atomoxetine on memory retention were octanesulfonic acid and 50 ml/l of a microbicide examined in a delayed nonmatch to sample task reagent (MB, ESA, Inc.); pH 8.0, adjusted with conducted in an eight-arm radial maze. Well-trained phosphoric acid) was delivered by an HPLC pump rats were required to recall, after a 7-h delay period, (ESA, Inc.) at 0.4 ml/min. The potentiostat used for where they received food during the information electrochemical detection (ESA, Inc.; model phase in order to obtain the remaining rewards during Coulochem II) was connected with a solid phase the retention phase conducted after a delay period. reactor for ACh (ESA, Inc.; model ACH-SPR) and with The apparatus, the training of the rats as well as an analytical cell with platinum target (ESA, Inc.; the information and retention sessions of the test were model 5041). as previously described.10 Atomoxetine (in 5% aca- Atomoxetine, reboxetine and methylphenidate cia) over a dose range of 1, 3 and 10 mg/kg or vehicle (synthesized at Eli Lilly and Company) were dis- was administered orally immediately after the infor- solved in saline (0.9% NaCl) and injected i.p. at a mation phase. During the retention phase, an entry volume of 1 ml/kg, at the doses indicated. Prazosin into a nonbaited arm or a re-entry into an and SCH 23390 (Sigma) (dissolved in saline) were arm previously visited during this phase of testing administered s.c. at a volume of 1 ml/kg. The doses of was counted as an error. Significance (Po0.05) was atomoxetine, reboxetine and methylphenidate that determined using a repeated measure ANOVA

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 189 followed by Dunnett’s test. It should be noted that for were analyzed with one-way ANOVA followed by the behavioral experiments, the oral route of admin- Duncan’s post hoc test. istration was selected for reasons of consistency with previous studies assessing the cognitive profile of diverse reference compounds. Results In vivo microdialysis studies Locomotor activity measures The effects of a single i.p. administration of atomox- The effects of atomoxetine, as compared to the etine (0.3, 1 and 3 mg/kg), reboxetine (1, 3, 10 and stimulant methylphenidate, on locomotor activity 30 mg/kg), and methylphenidate (1 and 3 mg/kg) on were measured with a 20 station Photobeam Activity ACh efflux were assessed by in vivo microdialysis in Systems (San Diego Instruments, San Diego, CA, the medial prefrontal cortex, the hippocampus and USA) with seven photocells per station. Rats were the nucleus accumbens in rats. There were no placed in the locomotor activity boxes statistically significant differences in basal ACh (40.6 Â 20.3 Â 15.2 cm3) for a habituation period of values among groups receiving vehicle or any of the 20 min. Immediately after habituation, drugs (atomox- drugs in any of the regions tested. Basal ACh values etine 1, 3, 10 mg/kg or methylphenidate 3 mg/kg (in pmol/15 min sample) were 1.4870.12 for the dissolved in saline) or saline were administered i.p. medial prefrontal cortex, 1.2970.2 for the hippocam- at a volume of 1 ml/kg. Locomotion was assessed for a pus and 0.4870.02 for the nucleus accumbens. 60-min period following the injection. Data (n ¼ 8 rats per group) are expressed as total ambulations (where Effects of atomoxetine on ACh efflux. Atomoxetine ambulation was defined as the sequential breaking of (0.3, 1 and 3 mg/kg) dose-dependently increased ACh adjacent photobeams) for the entire 60 min period and efflux in the prefrontal cortex and the hippocampus

atomoxetine 3 mg/kg atomoxetine 3 mg/kg vehicle atomoxetine 1 mg/kg atomoxetine 1 mg/kg atomoxetine 3 mg/kg atomoxetine 0.3 mg/kg atomoxetine 0.3 mg/kg acvehicle vehicle e

3.5 3.5 2.0

3.0 3.0 1.5 2.5 2.5

2.0 2.0 1.0

1.5 1.5

(fold over baseline) 0.5 (fold over baseline) (fold over baseline) 1.0 1.0 ACh efflux in the n.Acc. ACh efflux in the prefrontal cortex

ACh efflux in the prefrontal cortex 0.5 0.5 0.0 -90 -60 -30 0 30 60 90 120 150 180 -90 -60 -30 0 30 60 90 120 150 180 -90 -60 -30 0 30 60 90 120 150 180 time (min) time (min) time (min) b d f

2.4 2.4 2.4 2.2 2.2 2.2 2.0 2.0 2.0 1.8 1.8 1.8 1.6 1.6 1.6 1.4 1.4 1.4 1.2 1.2 1.2 1.0 (fold over baseline) 1.0 1.0 (fold over baseline)

0.8 0.8 ACh efflux in the n.Acc. (fold over baseline) 0.8 0.6 0.6 ACh efflux in the prefrontal cortex 0.6 ACh efflux in the prefrontal cortex

g/kg vehicle vehicle 1 mg/kg 3 mg/kg vehicle 1 mg/kg 3 mg/kg 3 m 0.3 mg/kg 0.3 mg/kg atomoxetine atomoxetine atomoxetine Figure 1 Time course (a, c, e) and overall (b, d, f) effects of atomoxetine on ACh efflux in the prefrontal cortex (a, b), the hippocampus (c, d) and the nucleus accumbens (e, f) of the rat. Atomoxetine dose-dependently increases ACh levels in the prefrontal cortex (F(1,12) ¼ 11.33, Po0.001; two-way ANOVA for time course effects at the highest dose and F(3,15) ¼ 35.56, Po0.001; one-way ANOVA for overall effects) as well as the hippocampus (F(1,70) ¼ 2,18, Po0.05; two-way ANOVA for time course effects at the highest dose and F(3,17) ¼ 6.61, Po0.01; one-way ANOVA for overall effects) but not in the nucleus accumbens (F(1,48) ¼ 0.51, NS; two-way ANOVA for time course effects and F(2,6) ¼ 0.839, NS; one-way ANOVA for overall effects). Data (expressed as fold over baseline established prior to any treatment) represent mean7s.e.m. of n ¼ 4–6 animals per group. *Po0.05; **Po0.01; ***Po0.001 as compared to the vehicle-treated group. For time course effects (a, c, e) statistical significance as compared to vehicle for each time point is indicated only for the highest dose of the drug for reasons of clarity of the presentation.

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 190 but not in the nucleus accumbens as assessed by in (Figure 3). In the medial prefrontal cortex at the vivo microdialysis (Figure 1). In the medial prefrontal highest dose tested (30 mg/kg for reboxetine and cortex (Figure 1a, b), the atomoxetine-induced 3 mg/kg for methylphenidate), the two drugs increases of extracellular ACh were statistically increased ACh release by three fold (i.e., with the significant at all doses tested. At the highest dose same peak amplitude as atomoxetine). In the hippo- depicted (3 mg/kg), atomoxetine produced a marked campus, both reboxetine and methylphenidate dose- (three-fold) and sustained (2 h) peak increase of dependently enhanced ACh release, but as for cortical ACh. A higher dose of atomoxetine (10 mg/ atomoxetine, increases in the hippocampus were less kg) did not produce further increases in cortical ACh pronounced than those observed in the cortex. In the efflux indicating that maximal efficacy is achieved nucleus accumbens, the highest dose of methyl- with the 3 mg/kg dose (data not shown). Atomoxetine phenidate but not of reboxetine induced a transient also augmented hippocampal extracellular ACh and small, but statistically significant, increase in ACh (Figure 1c, d) in a dose-dependent manner, although concentration. this increase was significant only for the 1 and 3 mg/ kg doses and much less pronounced as compared to the cortex. In contrast, atomoxetine had no effect on The procholinergic effects of atomoxetine are ACh concentrations in the nucleus accumbens even at dependent upon a1-NE and D1-DA receptor activa- the highest dose tested (Figure 1c). tion. As a selective NE uptake inhibitor, atomoxetine increases brain NE levels; atomoxetine has also been shown to increase DA release in the medial prefrontal Effects of reboxetine and methylphenidate on ACh cortex of the rat.6 We investigated whether NE efflux. A similar dose-dependent increase of cortical or DA systems mediate the procholinergic effects and hippocampal ACh was also observed after of atomoxetine in the medial prefrontal cortex. For administration of either the selective NE uptake this, the a1-NE antagonist prazosin or the D1-DA inhibitor reboxetine (Figure 2) or methylphenidate receptor antagonist SCH 23390 were administered

reboxetine 30 mg/kg reboxetine 30 mg/kg vehicle reboxetine 10 mg/kg reboxetine 10 mg/kg reboxetine 10 mg/kg reboxetine 3 mg/kg vehicle reboxetine 1 mg/kg a vehicle c e 3.5 3.5 2.0 3.0 3.0 1.5 2.5 2.5

2.0 2.0 1.0

1.5 1.5 0.5 1.0

1.0 baseline) over (fold ACh efflux in the n.Acc. the in efflux ACh cortex (fold over baseline) ACh efflux in the prefrontal 0.5 0.5 0.0

-90-60 -300 30 60 90 120 150 180 ACh efflux in the hippocampus (fold over baseline) -90-60 -300 30 60 90 120 150 180 -90-60 -300 30 60 90 120 150 180 time (min) time (min) time (min) b d f 2.4 2.4 2.4

2.2 2.2 2.2 ) 2.0 2.0 2.0 1.8 1.8 1.8 1.6 1.6 1.6 1.4 1.4 1.4 1.2 1.2 1.2 1.0 1.0 1.0

0.8 0.8 baseline over (fold 0.8 ACh efflux in the n.Acc. the in efflux ACh

ACh efflux in the prefrontal cortex (fold over baseline) 0.6 0.6 0.6 ACh efflux in the hippocampus (fold over baseline) g/kg vehicle1 mg/kg3 mg /kg vehicle vehicle 10 m 30 mg/kg 10mg/kg 30 mg/kg 3 mg/kg reboxetine reboxetine reboxetine Figure 2 Time course (a, c, e) and overall (b, d, f) effects of reboxetine on ACh efflux in the prefrontal cortex (a, b), the hippocampus (c, d) and the nucleus accumbens (e, f) of the rat. Reboxetine dose-dependently increases ACh levels in the prefrontal cortex (F(1,12) ¼ 2.55, Po0.05; two-way ANOVA for time course effects at the highest dose and F(3,20) ¼ 15.61, Po0.001; one-way ANOVA for overall effects) as well as the hippocampus (F(1,70) ¼ 2.53, Po0.05; two-way ANOVA for time course effects at the highest dose and F(2,11) ¼ 14.02, Po0.001; one-way ANOVA for overall effects) but not in the nucleus accumbens (F(1,48) ¼ 1.54, NS; two-way ANOVA for time course effects and F(2,6) ¼ 0.086, NS; one-way ANOVA for overall effects). Symbols and figure details as in Figure 1.

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 191 methylphenidate 3 mg/kg methylphenidate 3 mg/kg methylphenidate 1 mg/kg methylphenidate 1 mg/kg vehicle vehicle vehicle methylphenidate 3 mg/kg ace 3.5 3.5 2.0 3.0 3.0 2.5 2.5 1.5 2.0 2.0 1.5

(fold over baseline) 1.5 1.0 1.0

1.0 baseline) over (fold (fold over baseline) over (fold

ACh efflux in the n.Acc. ACh efflux in the prefrontal cortex 0.5 0.5 0.5 -90 -60 -300 30 60 90 120 150 180 -90 -60 -300 30 60 90 120 150 180

ACh efflux in the hippocampus -90 -60 -300 30 60 90 120 150 180 time (min) time (min) time (min) b df 2.4 2.4 2.4 2.2 2.2 2.2 2.0 2.0 2.0 1.8 1.8 1.8 1.6 1.6 1.6 1.4 1.4 1.4 1.2 1.2 1.2

(fold over baseline) 1.0 1.0 1.0 (fold over baseline) 0.8 0.8 baseline) over (fold 0.8

ACh efflux in the n.Acc.

ACh efflux in the prefrontal cortex 0.6 0.6 0.6 ACh efflux in the hippocampus

vehicle vehicle 1 mg/kg 3 mg/kg 1 mg/kg 3 mg/kg vehicle 3 mg/kg methylphenidate methylphenidate methylphenidate Figure 3 Time course (a, c, e) and overall (b, d, f) effects of methylphenidate on ACh efflux in the prefrontal cortex (a, b), the hippocampus (c, d) and the nucleus accumbens (e, f) of the rat. Methylphenidate dose-dependently increases ACh levels in the prefrontal cortex (F(1,12 ¼ 8.94, Po0.001; two-way ANOVA for time course effects at the highest dose and F(2,14) ¼ 11.09, Po0.01; one-way ANOVA for overall effects) as well as the hippocampus (F(1,70) ¼ 6.28, Po0.001; two- way ANOVA for time course effects at the highest dose and F(2,10) ¼ 16.05, Po0.001; one-way ANOVA for overall effects) and in the nucleus accumbens (F(1,60) ¼ 2.45, Po0.05; two-way ANOVA for time course effects and F(2,7) ¼ 5.76, Po0.05; one-way ANOVA for overall effects). Symbols and figure details as in Figure 1. systemically 15 min before the injection of putative cognitive enhancers.10–12 When injected atomoxetine (3 mg/kg). SCH 23390 at 0.3 mg/kg, a immediately after the information phase, atomo- dose that by itself did not affect cortical ACh xetine at 3 mg/kg decreased the number of errors concentrations, completely prevented the increase in that occurred during the retention phase conducted ACh efflux in the medial prefrontal cortex induced by 7 h later (Figure 6). However, atomoxetine at 1 and atomoxetine (Figure 4). Prazosin (1 mg/kg) that by 10 mg/kg did not siginificantly alter performance in itself had no effect on cortical ACh concentration, this task. also completely abolished the stimulatory effects of atomoxetine on ACh efflux in the prefrontal cortex (Figure 5). Object recognition The two-object recognition task is an animal model for testing memory performance Behavioral tasks and is based on the rat’s natural differential exploration of new and familiar objects.13,14 Given a Memory tasks. The procholinergic profile of choice a rat will spend more time interacting with a atomoxetine that we characterize here indicates a new object than with a familiar object. The preference potential for this compound to enhance cognition. We for novelty is observed across species and used as an therefore tested the ability of atomoxetine to improve indicator of recognition memory. During the learning memory retention in two distinct tasks that involve trial, there was no difference in the behavior of the neocortical and hippocampal circuits, a spatial delay animals in the different groups (vehicle- versus task performed in a 8-arm radial maze and the object atomoxetine-treated) and the time spent exploring recognition test. the objects was very similar (between 10 and 15 s, for a 2-min trial) among all groups. In the test trial, 8-arm radial maze Performance in this task is atomoxetine (1 and 3 mg/kg) significantly improved dependent upon the length of time the information performance, as evidenced by an increase in must be retained and is sensitive to amnesics and preference for the novel object at testing (Figure 7),

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 192 SCH 23390 0.3 mg/kg prazosin 1 mg/kg atomoxetine 3 mg/kg atomoxetine 3 mg/kg SCH 23390 0.3 mg/kg + prazosin 1 mg/kg+ atomoxetine 3 mg/kg atomoxetine 3 mg/kg vehicle vehicle

atomoxetine atomoxetine 3 kg/mg a 3.5 3 mg/kg a 3.5 SCH 23390 3.0 0.3 mg/kg 3.0 prazosin 1 mg/kg 2.5 2.5

2.0 2.0

1.5 the prefrontal cortex cortex prefrontal the 1.5

1.0 (fold over baseline) over (fold 1.0

0.5 0.5

ACh efflux in the prefrontal cortex (fold over baseline) ACh efflux in in efflux ACh

0.0 0.0 -90-60-300 30 60 90 120 150 -90 -60 -30 0 30 6090 120 150 time (min) time (min)

b 3.0 b 3.0

2.5 2.5

2.0 2.0

1.5 1.5

1.0

1.0 (fold over baseline) over (fold

ACh efflux in the prefrontal cortex (fold over baseline) 0.5

0.5 ACh efflux in the prefrontal cortex prefrontal the in efflux ACh 0.0 0.0 prazosin 1 mg/kg 1 mg/kg SCH 23390 0.3 mg/kg 0.3 mg/kg atomoxetine 3 mg/kg 3 mg/kg atomoxetine 3 mg/kg 3 mg/kg Figure 5 The a1 antagonist prazosin abolishes the stimu- Figure 4 The D1 antagonist SCH 23390 abolishes the latory effects of atomoxetine on ACh efflux in the medial stimulatory effects of atomoxetine on ACh efflux in the prefrontal cortex. Prazosin administered (1 mg/kg, s.c.) medial prefrontal cortex. SCH 23390 (0.3 mg/kg, s.c.) 15 min 15 min before atomoxetine (3 mg/kg) prevented the atomox- before atomoxetine (3 mg/kg) prevented the atomoxetine- etine-induced ACh release (F(1,72) ¼ 2.5, Po0.01; three- induced ACh release (F(1,81) ¼ 4.09, Po0.001; three-way way ANOVA for time course effects; F(1,15) ¼ 46.21, ANOVA for time course effects; F(1,18) ¼ 41.6, Po0.001; Po0.001; two-way ANOVA for overall effects). Data two-way ANOVA for overall effects). Data (expressed as fold (expressed as fold over baseline established prior to any increase over baseline established prior to any treatment) 7 7 treatment) represent mean s.e.m. of n ¼ 4–6 animals per represent mean s.e.m. of n ¼ 4–6 animals per group. group. *Po0.05, *Po0.01 and ***Po0.001 as compared to *Po0.05, *Po0.01 and ***Po0.001 as compared to vehicle treated animals (filled stars) and versus animals vehicle-treated animals (filled stars) and versus animals treated with vehicle prior to atomoxetine (open stars), at treated with vehicle prior to atomoxetine (open stars), at each time point. each time point.

whereas 0.3 and 10 mg/kg did not significantly alter any of the doses tested (1, 3 and 10 mg/kg), whereas, performance in this task. as expected because of its profile as a stimulant, methylphenidate (3 mg/kg) robustly (by three-fold) Locomotor activity increased locomotion (Figure 8). The lack of effect of We next assessed the effects of atomoxetine, as atomoxetine on locomotor activity suggests that compared to the stimulant methylphenidate, on basal nonspecific motor effects of atomoxetine do not locomotor activity in the rat. We show that atomox- confound our results in the radial arm maze and the etine had no effect on horizontal locomotor activity at object recognition test.

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 193 5 200 4 150

3 100 2 ambulations 50 number of errors number (retention session) 1

0 0

vehicle vehicle 1 mg/kg 3 mg/kg 10 mg/kg

atomoxetine atomoxetineatomoxetine 1 mg/kgatomoxetine 3 mg/kg 10 mg/kg Figure 6 Atomoxetine improves performance in the 8- methylphenidate 3 mg/kg radial-arm maze test. Atomoxetine administered immedi- Figure 8 Atomoxetine does not affect spontaneous loco- ately after the information phase reduced the average motor activity. Rats that received 1, 3 and 10 mg/kg of number of errors that occurred during the retention phase atomoxetine did not show alterations in spontaneous conducted after a 7-h delay (F(3,16) ¼ 7.518 Po0.05; one- locomotor activity. On the contrary, methylphenidate 7 way ANOVA for treatment). Data represent mean s.e.m. of (3 mg/kg) induced a marked hyperlocomotion n ¼ 7 animals per group. *Po0.05 for animals treated with (F(4,35) ¼ 7.4, Po0.001; one-way ANOVA for treatment). atomoxetine versus animals treated with vehicle. Data (expressed as the average number of ambulations during the entire test period, i.e., total number of ambulations/60 min) represent mean7s.e.m. of n ¼ 8 animals per group. *Po0.05 for animals treated with methylphenidate 100 versus animals treated with vehicle. 80

60 We indeed show here by in vivo microdialysis that atomoxetine increases ACh efflux and presumably 40 increases cholinergic neurotransmission in cortical regions, the medial prefrontal cortex, in particular. (% session time) 20 Furthermore, we provide evidence that in addition to

time exploring new object new time exploring atomoxetine, the NE-uptake inhibitor reboxetine, and 15 0 (as previously reported in Acquas and Fibiger ) methylphenidate stimulate ACh release in this cor- vehicle tical region. In our model, although methylphenidate 1 mg/kg 3 mg/kg10 mg/kg 0.3 mg/kg increases cortical ACh release to about the same atomoxetine extent as pharmacologically equivalent doses of atomoxetine and reboxetine, methylphenidate’s sti- Figure 7 Atomoxetine improves performance in the object mulatory effects upon ACh efflux were more transi- recognition test. Rats that received 1 and 3 mg/kg of ent. These results, when compared to the effects of atomoxetine spent significantly more time exploring the atomoxetine, reboxetine and methylphenidate on novel object, following the 3-h delay, than did vehicle- cortical DA and NE release that we reported pre- treated rats (F(4,39) ¼ 5.06, Po0.01; one-way ANOVA for 6 treatment). Data (expressed as percent time exploring the viously, show that for each of the above drugs the novel object during the retention test) represent mean7 time course curves for ACh, DA and NE increases s.e.m. of n ¼ 8–10 animals per group. *Po0.05 for animals overlap to a great extent. Thus, similarly to our treated with atomoxetine versus animals treated with observations for ACh efflux, the methylphenidate- vehicle prior to the acquisition trial. induced rise of cortical NE and DA was apparent for only about 1 h while those of atomoxetine and reboxetine were maintained for twice as long.6 The Discussion short-lived, methylphenidate-induced DA and ACh increases are predicted by the relatively brief half-life Catecholaminergic activity in the cortex and subcor- of the compound and may account for the constraint tex is central to current pathophysiological models of of multiple daily administrations for its efficacy in ADHD. The present results indicate that enhancing ADHD patients. Furthermore, methylphenidate but cortical ACh efflux could also contribute to the neither atomoxetine nor reboxetine increased ACh mechanism of action of drugs that effectively treat efflux in the nucleus accumbens, consistent with and ADHD. similar to previous findings on DA release in this

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 194 region.6 It should be noted that D-amphetamine, the two distinct behavioral assays, the 8-arm radial arm other stimulant drug widely used in ADHD therapeu- maze and the object recognition test, classically tics, also increases ACh efflux in the striatal complex used in rodents to screen for compounds that alter including the nucleus accumbens.16 Thus, atomox- learning and memory. Although these behavioral etine and reboxetine are differentiated from the tasks are not specific indicators of cholinergic system psychostimulants in their lack of increase of striatal tone and their performance measures can be and nucleus accumbens DA and ACh levels. The altered by many manipulations, both tests involve enhancement of striatal and nucleus accumbens DA cortical (prefrontal and hippocampal) pathways and and ACh efflux by the psychostimulants may con- depend upon cholinergic activity. Thus, in general, tribute to their abuse liability. cholinergic antagonists (e.g., scopolamine) impair, We show that the atomoxetine-induced ACh release while procholinergic compounds (in particular in the prefrontal cortex is increased in the same acetylcholinesterase inhibitors) improve performance dose range as NE and DA.6 Furthermore, our data, in the object recognition test and in the radial-maze along with data from other studies, are consistent (see, e.g., Ennaceur and Meliani14 and Myhrer26). with a synchronous a1-NE and D1-DA receptor Consistent to its neurochemical profile and at activation mechanism implicated in cortical ACh doses relevant to those that increase cortical mono- release. Thus, there is ample evidence that, in amine release, atomoxetine increased the time this region, where NE- and DA-afferents converge, rats spent interacting with the novel object in the NE activity can regulate extracellular DA efflux in object recognition test and decreased errors in multiple ways. These could involve either the NE spatial pattern recognition in the radial arm maze, transporter that removes NE and DA from the suggesting improved memory performance in both synaptic cleft with equipotent affinity17 or NE- tests. The two tasks differ in that the object recogni- heterorereceptors, which could directly control tion task depends on the animal’s natural exploratory terminal DA activity.18 However, it is quite unlikely behavior and is nonspatially mediated, whereas the that atomoxetine increases cortical ACh release 8-arm radial maze test is a spatial task driven by through a serial NE-a1-DA-D1 trans-synaptic circuit. food reward. In addition, in the object recognition First, a2- rather than a1-NE receptors were shown task, atomoxetine was administered prior to the task to modulate cortical DA release18 and antagonism of and thus may affect acquisition, consolidation and a1-NE receptors blocked atomoxetine-induced in- retrieval processes. Atomoxetine was administered creases in ACh efflux. Second, D1-mediated DA after the information phase in the 8-arm radial maze control of cortical ACh release involves distant task and more likely would affect memory consolida- but not local activation of D1 receptors.19,20 Further- tion and less likely retrieval processes due to its more, in our hands blockade of a1-NE receptors relatively short half-life in rats. It should be noted by prazosin (1 mg/kg) effectively reversed cortical nevertheless that while the increases in monoamine ACh, but not cortical DA efflux elicited by atomox- efflux were linear over the dose spectrum studied (see etine (data not shown). Alternatively, the atomoxe- also Bymaster et al.6), the improved cognitive perfor- tine-induced ACh release could involve an integrated mance was characteristically nonlinear with high mechanism operating only when parallel D1-DA doses not showing any activity. Bell-shaped efficacy and a1-NE signals temporally and spatially coincide. curves in animal cognitive models have been noted Thus, converging DA and NE systems could with a number of mechanistically diverse procogni- regulate cortical ACh release remotely by modulating tive agents27–32 and in clinical studies as well.33–35 the firing of cholinergic neurons at the somatoden- This genre of dose–response relationship between dritic level. Indeed, in the nucleus basalis, where neurochemistry and behavior suggests that a fine the ACh neurons projecting to the cortex reside, permissive tuning of cortical monoaminergic neuro- DA and NE afferents have been shown to synapse transmission probably accounts for the expression of on cholinergic somata.21 The role of nucleus cognitive amelioration. basalis D1-DA and a1-NE receptors in regulating In conclusion, the evidence we present here cortical ACh release and in mediating the procholi- indicates a procholinergic profile and a procognitive nergic effects of atomoxetine warrants further inves- potential for the novel nonstimulant ADHD medica- tigation. tion atomoxetine. In view of the clinically validated As mentioned, a procholinergic neurochemical value of atomoxetine in the treatment of ADHD, profile has been related to the ability of a number of further studies addressing the underlying molecular compounds (marketed or under development) to mechanisms of the procholinergic and procognitive ameliorate cognitive function in patients, or to effects of the compound that we demonstrate here are improve attention and memory in animal models.14,22 warranted. Specifically, studies with knockout mice have shown that the integrity of cholinergic neurotransmission in particular is necessary for habituation and memory References functions,23,24 as well as for behavioral shifting and 25 1 Caballero J, Nahata MC. Atomoxetine hydrochloride for the temporal organization of action. Thus, we investi- treatment of attention-deficit/hyperactivity disorder. Clin Ther gated the procognitive properties of atomoxetine in 2003; 25: 3065–3083.

Molecular Psychiatry Atomoxetine increases cortical acetylcholine ET Tzavara et al 195 2 Spencer T, Biederman J, Wilens T. Nonstimulant treatment of adult 19 Day JC, Fibiger HC. Dopaminergic regulation of septohippocampal attention-deficit/hyperactivity disorder. Psychiatr Clin North Am cholinergic neurons. J Neurochem 1994; 63: 2086–2092. 2004; 27: 373–383. 20 Day J, Fibiger HC. Dopaminergic regulation of cortical acetylcho- 3 Everitt BJ, Robbins TW. Central cholinergic systems and cognition. line release. Synapse 1992; 12: 281–286. Annu Rev Psychol 1997; 48: 649–684. 21 Zaborszky L, Cullinan WE. Direct catecholaminergic-cholinergic 4 Ichikawa J, Dai J, O’Laughlin IA, Fowler WL, Meltzer HY. Atypical, interactions in the basal forebrain. I. Dopamine-beta-hydroxylase- but not typical, antipsychotic drugs increase cortical acetylcholine and tyrosine hydroxylase input to cholinergic neurons. J Comp release without an effect in the nucleus accumbens or striatum. Neurol 1996; 374: 535–554. Neuropsychopharmacology 2002; 26: 325–339. 22 Auld DS, Kornecook TJ, Bastianetto S, Quirion R. Alzheimer’s 5 Shirazi-Southall S, Rodriguez DE, Nomikos GG. Effects of typical disease and the basal forebrain cholinergic system: relations to and atypical antipsychotics and receptor selective compounds on beta-amyloid peptides, cognition, and treatment strategies. Prog acetylcholine efflux in the hippocampus of the rat. Neuropsycho- Neurobiol 2002; 68: 209–245. pharmacology 2002; 26: 583–594. 23 Anagnostaras SG, Murphy GG, Hamilton SE, Mitchell SL, 6 Bymaster FP, Katner JS, Nelson DL, Hemrick-Luecke SK, Threlkeld Rahnama NP, Nathanson NM et al. Selective cognitive dysfunction PG, Heiligenstein JH et al. Atomoxetine increases extracellular in acetylcholine M1 muscarinic receptor mutant mice. Nat levels of norepinephrine and dopamine in prefrontal cortex of rat: Neurosci 2003; 6: 51–58. a potential mechanism for efficacy in attention deficit/hyperacti- 24 Tzavara ET, Bymaster FP, Felder CC, Wade M, Gomeza J, Wess J vity disorder. Neuropsychopharmacology 2002; 27: 699–711. et al. Dysregulated hippocampal acetylcholine neurotransmis- 7 Paxinos G, Watson C. The Rat Brain in Stereotaxic Coordinates. sion and impaired cognition in M2, M4 and M2/M4 Academic Press: New York, 1982. muscarinic receptor knockout mice. Mol Psychiatry 2003; 8: 8 Damsma G, Westerink BH, de Boer P, de Vries JB, Horn AS. Basal 673–679. acetylcholine release in freely moving rats detected by on-line 25 Granon S, Faure P, Changeux JP. Executive and social behaviors trans-striatal dialysis: pharmacological aspects. Life Sci 1988; 43: under nicotinic receptor regulation. Proc Natl Acad Sci USA 2003; 1161–1168. 100: 9596–9601. 9 Tzavara ET, Davis RJ, Perry KW, Li X, Salhoff C, Bymaster FP et al. 26 Myhrer T. Neurotransmitter systems involved in learning and The CB1 receptor antagonist SR141716A selectively increases memory in the rat: a meta-analysis based on studies of four monoaminergic neurotransmission in the medial prefrontal cortex: behavioral tasks. Brain Res Brain Res Rev 2003; 41: 268–287. implications for therapeutic actions. Br J Pharmacol 2003; 138: 27 Sweeney JE, Bachman ES, Coyle JT. Effects of different doses 544–553. of galanthamine, a long-acting acetylcholinesterase inhibitor, 10 Wolff MC, Leander JD. SR141716A, a cannabinoid CB1 receptor on memory in mice. Psychopharmacology (Berlin) 1990; 102: antagonist, improves memory in a delayed radial maze task. Eur J 191–200. Pharmacol 2003; 477: 213–217. 28 Cai JX, Arnsten AFT. Dose-dependent effects of the dopamine D1 11 Staubli U, Rogers G, Lynch G. Facilitation of glutamate receptors receptor agonists A77636 or SKF81297 on spatial working memory enhances memory. Proc Natl Acad Sci USA 1994; 91: 777–781. in aged monkeys. J Pharmacol Exp Ther 1997; 282: 1–7. 12 Pussinen R, Sirvio J. Effects of cycloserine, a positive modulator of 29 Flood JF, Uezu K, Morley JE. Effect of histamine H2 and H3 N-methyl-aspartate receptors, and ST 587, a putative alpha-1 receptor modulation in the septum on post-training memory adrenergic agonist, individually and in combination, on the non- processing. Psychopharmacology (Berlin) 1998; 140: 279–284. delayed and delayed foraging behaviour of rats assessed in the 30 Popke EF, Mayorga AJ, Fogle CM, Paule MG. Effects of acute radial arm maze. J Psychopharmacology 1999; 13: 171–179. nicotine on several operant behaviors in rats. Pharmacol Biochem 13 Ennaceur A, Delacour J. A new one-trial test for neurobiological Behav 2000; 65: 247–254. studies of memory in rats. 1: behavioral data. Behav Brain Res 31 Andersen JM, Lindberg V, Myhrer T. Effects of scopolamine and D- 1988; 31: 47–59. cycloserine on non-spatial reference memory in rats. Behav Brain 14 Ennaceur A, Meliani K. Effects of physostigmine and scopolamine Res 2002; 129: 211–216. on rats’ performances in object-recognition and radial-maze tests. 32 Lidow MS, Koh P-O, Arnsten AFT. D1 dopamine receptors Psychopharmacology (Berlin) 1992; 109: 321–330. in the mouse prefrontal cortex: immunocytochemical and 15 Acquas E, Fibiger HC. Chronic lithium attenuates dopamine D1- cognitive neuropharmacological analyses. Synapse 2003; 47: receptor mediated increases in acetylcholine release in rat frontal 101–108. cortex. Psychopharmacology (Berlin) 1996; 125: 162–167. 33 Soncrant TT, Raffaele KC, Asthana S, Berardi A, Morris PP, Haxby 16 Acquas E, Wilson C, Fibiger HC. Nonstriatal dopamine D1 JV. Memory improvement without toxicity during chronic, low receptors regulate striatal acetylcholine release in vivo. J Pharma- dose intravenous arecoline in Alzheimer’s disease. Psychophar- col Exp Ther 1997; 281: 360–368. macology (Berlin) 1993; 112: 421–427. 17 Raiteri M, Del Carmine R, Bertollini A, Levi G. Effect of 34 Canal N, Imbimbo BP. Relationship between pharmacodynamic sympathomimetic amines on the synaptosomal transport of activity and cognitive effects of eptastigmine in patients with noradrenaline, dopamine and 5-hydroxytryptamine. Eur J Phar- Alzheimer’s disease. Eptastigmine Study Group. Clin Pharmacol macol 1977; 41: 133–143. Ther 1996; 60: 218–228. 18 Gresch PJ, Sved AF, Zigmond MJ, Finlay JM. Local influence of 35 Braida D, Sala M. Eptastigmine: ten years of pharmacology, endogenous norepinephrine on extracellular dopamine in rat toxicology, pharmacokinetic, and clinical studies. CNS Drug Rev medial prefrontal cortex. J Neurochem 1995; 65: 111–116. 2001; 7: 369–386.

Molecular Psychiatry