Atomoxetine improves memory and other components of executive function in young-adult rats and aged rhesus monkeys

Abstract

Atomoxetine is a norepinephrine reuptake inhibitor and FDA-approved treatment for attention deficit/hyperactivity disorder (ADHD) in children, adolescents, and adults. While there is some evidence that atomoxetine may improve additional domains of cognition beyond attention in both young adults and aged individuals, this subject has not been extensively investigated. Here, we evaluated atomoxetine (in low mg/kg doses) in a variable stimulus duration (vSD) and a variable intertrial interval (vITI) version of the five choice-serial reaction time task (5C-SRTT), and an eight-arm radial arm maze (RAM) procedure in young-adult rats. The compound was further evaluated (in μg/kg-low mg/kg doses) along with nicotine (as a reference compound) and the Alzheimer’s disease treatment donepezil in a distractor version of a delayed match to sample task (DMTS-D) in aged monkeys (mean age = 21.8 years). Atomoxetine (depending on the dose) improved accuracy (sustained attention) as well as behaviors related to impulsivity, compulsivity and cognitive inflexibility in both the vSD and vITI tasks and it improved spatial reference memory in the RAM. In the DMTS-D task, both nicotine and atomoxetine, but not donepezil attenuated the effects of the distractor on accuracy at short delays (non-spatial working/short term memory). However, combining sub-effective doses of atomoxetine and donepezil did enhance DMTS-D accuracy indicating the potential of using atomoxetine as an adjunctive treatment with donepezil. Collectively, these animal studies support the further evaluation of atomoxetine as a repurposed drug for younger adults as well older individuals who suffer from deficits in attention, memory and other components of executive function.

1. Introduction

Atomoxetine is a norepinephrine reuptake inhibitor and FDA-approved treatment for attention deficit/hyperactivity disorder (ADHD) in children, adolescents, and adults. Multiple double-blind, placebo-controlled clinical trials (Michelson et al., 2003; Vaughan et al., 2009) have shown that atomoxetine improves the core symptoms of ADHD (inattention, impulsivity, and hyperactivity) as well as quality of life and emotional lability (reviewed, Childress 2015). However, due to its generally smaller effect size, atomoxetine is typically considered a second line therapy for ADHD when compared to stimulants such as methylphenidate and amphetamine (Michelson et al. 2003; Cheng et al., 2007; Schwartz and Correll, 2014). Nonetheless, atomoxetine offers the advantages of reduced abuse liability, decreased risk of motor side effects, and it can serve as an alternative medication for patients who are not responsive to stimulants (Newcorn et al, 2008).

The therapeutic effects of atomoxetine in ADHD are primarily attributed to its potent and relatively selective action as an inhibitor of the norepinephrine transporter (NET), which results in elevations of synaptic levels of norepinephrine in the central nervous system (Bymaster et al. 2002; Robbins and Arnsten 2009; Childress and Sallee 2014). Atomoxetine is also known to modulate cortical synaptic dopamine uptake via the NET to specifically increase extracellular levels of dopamine in the prefrontal cortex, but not in motor or reward-related areas of the striatum (Tatsumi et al. 1997; Bymaster et al. 2002; Swanson et at. 2006; Robbins and Arnsten 2009). These effects of atomoxetine are thought to underlie its reduced abuse potential and lower tendency to produce stereotypies and tics when compared to classical stimulants (Barton, 2005).

There is some evidence to suggest that the mechanism of the therapeutic actions of atomoxetine may extend beyond what is described above. For example, while older microdialysis studies did not show an increase in extracellular serotonin in the prefrontal cortex of rodents following atomoxetine treatment, (Bymaster et al., 2002; Koda et al., 2010), in a recent positron emission tomography (PET) study, atmoxetine at clinically relevant doses significantly occupied both the serotonin transporter (SERT) as well as the NET (as expected) in rhesus monkeys (Ding et al., 2014). In addition, in cultured rat brain neurons in whole-cell patch-clamp studies, clinically relevant concentrations of atomoxetine inhibited NMDA receptors via a voltage- and magnesium-dependent open channel blocking mechanism (Barygin et al., 2017). This action was similar to some older antidepressants and antipsychotics (e.g., amitriptyline, chlorpromazine) and is particularly interesting in light of the recently discovered (acute) antidepressant effects of the NMDA receptor antagonist, ketamine (Fava et al., 2018). Here it is important to note that the while atomoxetine was initially evaluated in phase II clinical trials for depression in the 1980s and 1990s with some positive results, it was not developed further for this indication (see Preti, 2002). Atomoxetine has also been shown to increase extracellular levels of acetylcholine in cortical and hippocampal, but not subcortical brain regions in rodents, an effect that appears to be dependent on norepinephrine α-1 and/or dopamine D1 receptor activation. These secondary cholinergic effects have been hypothesized to contribute to the atomoxetine-related improvements in object recognition and working memory observed in rodents (Tzavara et al., 2006). Collectively, the studies described here indicate that there are potentially useful neuropharmacological effects of atomoxetine beyond NET inhibition and further, that it may have the potential to improve domains of cognition beyond attention. To date, the effects of atomoxetine on memory and other components of executive function have not been systematically investigated. Moreover, atomoxetine has not been evaluated extensively in aged subjects, an important issue since deficits in attention, working memory, and other components of executive function are common in age-related disorders of cognition. Positive effects of atomoxetine on executive function in aged subjects might indicate its potential as a repurposed drug for Alzheimer’s disease (AD) and other forms of dementia.

In the studies described here, we thus evaluated atomoxetine in several behavioral tasks in animals including a five choice-serial reaction time task (5C-SRTT) and an eight-arm radial arm maze (RAM) procedure in young-adult male rats, and a distractor version of a delayed match to sample task (DMTS-D) in aged male and female rhesus monkeys. In the 5C-SRTT, we sought to expand upon previous reports where atomoxetine was shown in adult rodents in improve sustained attention and to decrease impulsivity-like behaviors (Navarra et al., 2008; Paterson et al., 2011; Robinson, 2012), by extending the studies to further evaluate additional outcome measures related to compulsivity-like behaviors and cognitive inflexibility. In the RAM procedure, we evaluated both spatial working and reference memory in young-adult rats, while in the DMTS-D studies; we evaluated non-spatial working memory and susceptibility to distraction in aged monkeys. Finally, based on reports of indirect cholinergic effects of atomoxetine noted above, we also evaluated atomoxetine as part of an adjunctive treatment strategy in combination with the AD treatment and cholinesterase inhibitor, donepezil, in aged monkeys in the DMTS-D task.

2. Materials and Methods

All procedures employed during this study were reviewed and approved by the Augusta University Institutional Animal Care and Use Committee and are consistent with AAALAC guidelines. Measures were taken to minimize pain and discomfort in accordance with the National Institutes of Health, Guide for the Care and Use of Laboratory Animals (NRC, 2010). Significant efforts were also made to minimize the total number of animals used while maintaining statistically valid group numbers.

2.1. Rodent Behavioral Studies

2.1.1. Subjects

Experimentally naive, male Wistar or Long-Evans rats (3 months old; Envigo Sprague-Dawley, Inc, Indianapolis, IN) were housed individually in polycarbonate cages (45 × 30 × 18 cm) with corncob bedding in a vivarium of constant temperature (21-23°C) and humidity (40-50%). Lighting was maintained on a 12-hr light-dark cycle (7:00 a.m.−7:00 p.m.) with free access to water and food during the first week (see subsequent food restriction procedures below). All behavioral testing was performed during the light portion (9 a.m.−5 p.m.) of the light/dark cycle (Monday thru Friday).

2.1.2. 5-Choice Serial Reaction Time (5C-SRTT) Procedure

One week prior to 5C-SRTT training and throughout testing, male Wistar rats were food restricted to approximately 85% of their age-dependent, free-feeding weights based upon Envigo Laboratories growth rate curves. Animals were trained in eight automated 5C-SRTT operant chambers (Med Associates, St. Albans, VT, USA), controlled by MedPC software (Med Associates), as described previously (Terry et al., 2014). Briefly, each operant chamber was equipped with five apertures containing a photocell beam to detect nose pokes and a lamp (2.8W) that could be illuminated randomly at varying durations. Food pellets (45 mg chow pellet, BioServ, Frenchtown, NJ, USA) were delivered automatically to a magazine, located on the opposite wall to the nose pokes, that was also equipped with a light that turned on to indicate that a pellet had been dispensed. The house-light remained on for the entire session unless an error or omission occurred. Training sessions began with the delivery of a food reward and retrieval triggered the first trial. After a 5 sec inter-trial-interval (ITI), a stimulus light within one of the five apertures was illuminated for a fixed duration (see below) and a single nose-poke into this opening during the signal illumination period or during the 5 sec limited hold period delivered a reward (correct response); a nose-poke into a non-illuminated aperture (incorrect response) resulted in a 5 sec time-out period and no food reward. Failure to respond within the 5 sec limited hold period (omission) also resulted in a time-out. Training began with the stimulus duration set at 10 sec, a limited hold period of 5 sec and an ITI of 5 sec. The stimulus duration (e.g., 5, 2.5, 2.0, 1.5 and 1.25) was gradually reduced, maintaining stable performance, until a final duration of 1.0 sec was achieved. Sessions ended when 30 minutes had lapsed or 100 trials had been completed. Sixteen animals were trained 5 days per week until they reached a stable performance criterion of 70-75% accuracy, <20% omissions and completion of all 100 trials for 5 consecutive days. Upon meeting the performance criteria animals were separated into two groups (N=8 per group). Group 1 subjects were assessed using a randomized variable stimulus duration (vSD) version of the 5C-SRTT with 0.5, 1.0 and 2.0 sec SDs and a 5 sec ITI, and Group 2 subjects were assessed is a variable inter-trial-interval (vITI) version of the task with 1.0, 5.0 and 10.0 sec ITIs and a 1 sec SD. The vSD and vITI tasks were conducted no more than twice in any given week and in the remaining test sessions; a standard 1.0 sec stimulus duration with a 5 sec intertrial interval was used.

A within subjects design was used such that all animals received all treatments (and drug doses) in a pseudorandom fashion. Performance parameters measured were: % correct ((# correct /(# correct + # incorrect))×100), premature responses (total # of nose-pokes into any aperture after trial initiation but before onset of the stimulus light), timeout responses (total # of nose pokes into any aperture during a timeout period), perseverative responses (total # of nose pokes occurring after the correct response had been made but before reward collection), total number of food magazine head entries, omissions and total trials completed. Animals were administered atomoxetine or vehicle intraperitoneally (i.p.) 60 min prior to test in a dose volume of 1 ml/kg.

2.1.3. Eight-Arm Radial Arm Maze (8-RAM) Procedure

One week prior to 8-RAM training and throughout testing, male Long-Evans rats were food restricted to approximately 85% of their age-dependent, free-feeding weights based upon Envigo Laboratories growth rate curves. Animals were trained in a computer-automated eight-arm RAM apparatus (Med Associates, St. Albans, VT, USA) using a modification of a method previously described (Terry et al., 2012). The maze consisted of a central octagonal hub (arena) with automatic guillotine doors connected to aluminum arms (8.9 cm wide) radiating distally (45.7 cm long). IR-photo beam sensors were positioned at the entrance to each runway and a food pellet receptacle and head entry detector was positioned at the end of each runway. The maze was positioned approximately 90 cm above the floor in a testing room with a number of extra-maze cues (composed of large geometrical shapes). Initially, subjects were given two 15 min free exploration habituation sessions in which reinforcement food pellets (45 mg Dustless Precision Pellets, Bioserve, Frenchtown, NJ.) were scattered randomly around the entire maze area. After this habituation phase, subjects were trained in a win-shift procedure. A trial began when the experimenter placed the test subject into the central octagonal arena. After a 60 sec delay, all guillotine doors raised allowing access to all of the eight arms. When the animal broke a photo-beam in the pellet receptacle at the end of the runway a single reward was delivered. When the rat moved back into the central arena all doors closed for 5 sec and then reopened. Working memory errors were scored and defined as all reentries into an arm that had previously delivered a food reward. All animals were trained in the win-shift task for a minimum of 10 days and performance criterion was achieved when the subject completed the task within the 15 min time limit on four consecutive days with ≤ two working memory errors. Once stable win-shift performance was acquired the subjects advanced to the delayed non-match to position (DNMTP) phase. DNMTP testing began with an information (forced-4) session in which four of the eight arms were blocked. The session ended when all four arms were visited or when the trial timed out (15 min.). The animal remained in the testing room for the delay period (4 hrs). In the “free-8” (retention) test session, all eight arms were accessible however, food reinforcement occurred only at the ends of the arms not visited in the previous information session. The test session continued until all four of the previously blocked arms were visited, or until 15 min elapsed. The number of arm entries was recorded, along with two types of errors: reference memory errors and working memory errors. Reference memory errors were defined as entries into an arm that was open and baited during the forced-4 information session and working memory errors were defined as reentries into an arm that had previously delivered a food reward in the same free-8 retention session. At the completion of the second test session in each trial block, the animal was returned to its home cage in the housing facility, until the next day’s information session. Animals were trained for 10 days at a 4 hr delay between the forced-4 and the free-8 sessions. A between subject design (N=13 rats/treatment) was used and test doses of atomoxetine or vehicle (1 ml/kg dose volume) were administered subcutaneously (s.c.) 30 min prior to the forced-4 information session in a pseudorandom order.

2.2. Non-human Primate Behavioral Studies

2.2.1. Subjects

The nonhuman primate subjects used in this study included nine rhesus macaques (Macaca mulatta), 2 males and 7 females, with an average age of 21.8 years old (see Table 1 for additional details). All of the subjects except one (a 12-year-old female) were designated as aged (≥ 19 years old). This designation (“aged”) is based on previous studies in rhesus monkeys where cognitive impairments (and accompanying neuroanatomical changes) began to emerge late in the second decade of life (see Price and Sisodia 1994 for review). Subjects were individually housed in double stainless steel cages composed of two 127 × 71 × 66 cm units in rooms with up to five other monkeys. The room was maintained at a constant temperature (21-23°C) and humidity (45-50%). Lighting was maintained on a 12-hr light-dark cycle (7:00 a.m.−7:00 p.m.) with behavioral testing occurring during the light portion (9 a.m.−5 p.m.) of the cycle Monday thru Friday. Animals had access to filtered tap water (unlimited) and monkey chow (Envigo Teklad Laboratory monkey diet, Madison, WI) supplemented with fruits and vegetables. Enrichment (e.g., toys, foraging tubes and television viewing) was also provided except during behavioral testing.

Table 1:

Monkey Subject Information

				DMTS Delay Intervals (sec)		Distractor Stimulus
Subject ID	Gender	Age	Wt (kg)	Short	Long	Duration (sec)
E1V	F	24	7.3	5	90	3
EXO	F	24	8.6	5	90	3
HPE	F	20	10.0	7	100	5
P73	F	26	7.0	4	50	3
590	F	23	9.7	7	40	5
689	F	24	6.1	10	110	8
77F	F	12	10.1	5	55	3
618	M	23	10.5	4	40	3
Na8	M	20	9.0	10	100	8
	Mean	21.8	8.7	6.3	75.0	4.6
	SEM	1.7	0.6	1.0	11.5	0.9

2.2.2. Delayed Match to Sample (DMTS) Procedure

The DMTS task was conducted as previously described (Callahan et al., 2013) and all animals were well trained (>100 individual sessions). A touch-sensitive screen (15 in. AccuTouch LCD Panelmount Touch Monitor)/pellet dispenser units (Med Associates) mounted in light-weight aluminum chasses was attached to the home cage. The DMTS task was presented via a computer-automated system. A 300 mg banana flavored food pellet (Bio-Serv, Flemington, NJ) was provided as the reward. The stimuli included rectangles of various colors (e.g., red, blue, yellow). A trial was initiated by presentation of a sample composed of one of three colors which remained in view until the monkey touched it thereby initiating a pre-programmed delay (retention) interval. After the delay interval, two choice colors were presented with one of the two choice colors matching the sample stimulus. Choosing the correct previously observed sample color resulted in a reward (correct choice) whereas selecting the non-sample color (incorrect) had no consequences (i.e., neither reinforced nor punished). The inter-trial interval was 5 sec and each session consisted of 96 trials. Presentations of stimulus color, choice colors, and choice positions were counterbalanced over the 96 trials. Delay intervals were established during non-drug or vehicle sessions prior to initiating the study. During training sessions, the duration of each delay interval was adjusted for each subject until three levels of group performance accuracy were approximated: zero delay (85-100% of trials answered correctly), short delay interval (70-84% correct) and long delay interval (50-65% correct).

2.2.3. DMTS-Distractor Stimulus (DMTS-D) Procedure

In these experiments, monkeys were tested using a modification of the DMTS-D procedure previously described (see Terry et al. 2002; Terry et al., 2016). During the DMTS-D test session, 24 of the 96 trials contained a distractor stimulus consisting of a random array of flashing colored lights (i.e., the same three hues selected for the sample and choice stimuli) that appeared on the test panels. The total duration for a given colored distractor stimulus was 0.33 sec with each of the three different colored stimuli being presented to the subject. The onset of the distractor stimulus began 1 sec after the monkey touched the sample color and remained active for approximately 70% of the time established by the monkeys’ standard short delay duration (see Table 1). The duration of the distractor was chosen based on our observation that distractor lengths of shorter duration were not consistently effective in disrupting DMTS short delay performance. Animals received twelve distractor short and long delay trials (total 24 trials) that were interspersed among the standard (non-distractor) DMTS delay intervals (i.e., zero, short and long). DMTS-D sessions were conducted 1 to 2 times per week with intervening standard DMTS maintenance sessions conducted between test drug (or vehicle) administration. Test doses of atomoxetine (60 min pretreatment), donepezil (30 min pretreatment), nicotine (15 min pretreatment) or saline (control; 60 min pretreatment) were administered by the intramuscular (i.m.) route prior to DMTS-D test in a volume of 0.035 ml/kg. In addition, as a potential adjunctive treatment strategy, we combined sub-effective doses of atomoxetine and donepezil and then evaluated DMTS-D performance. Each test compound’s dose response curve was completed prior to initiating testing of the next compound; saline controls were presented throughout DMTS-D testing to ensure that the negative effects of the distractor on short delay performance were maintained. Test doses of atomoxetine and nicotine were administered at least three times in a pseudo-random order and reported as the average of these three evaluations. Doses of donepezil (alone) and donepezil plus atomoxetine were administered once to the monkeys. DMTS-D results are presented as the mean (± S.E.M.) values of the difference (delta) from the DMTS distractor short delay % correct (accuracy) and the DMTS non-distractor short delay % correct and depicted as the “change from non-distractor control”. The non-distractor control value represents the delta between the DMTS-standard short delay accuracy and the DMTS non-distractor short delay response. Using this calculation and presentation of the results emphasizes the magnitude of effect of the distractor stimulus on DMTS short delay performance following vehicle control and drug test treatments.

2.3. Drugs

Unless otherwise noted, all compounds were prepared in physiological saline (0.9% NaCl). Doses refer to the weight of the salt, except where noted. Drugs used and suppliers were: atomoxetine HCl (OChem Inc, Des Plaines, IL), donepezil HCl (Memory Pharmaceuticals, Montvale, NJ) and (−) nicotine hydrogen tartrate salt (base weight; Sigma-Aldrich, St Louis, MO). Saline served as the vehicle control in all behavioral tests.

2.4. Data Analysis

Behavioral data were analyzed using a one or two factor analysis of variance (ANOVA), with repeated measures when indicated followed by the Student Newman Keuls or Dunnett’s (for comparisons to vehicle controls only) post-hoc test (SigmaPlot 11.2). All results are expressed as the mean (± S.E.M.). Differences between means were considered significant at the p<0.05 level.

3. Results

3.1. Rodent Behavioral Studies

3.1.1. Atomoxetine effects in the variable stimulus duration (vSD) version of the 5C-SRTT

Statistical analysis of the effects of atomoxetine on overall vSD accuracy (% correct collapsed across all SDs) indicated that the 3.0 mg/kg dose improved performance ([F(3,7)=3.70, p=0.027], Fig 1A). Analysis of performance associated with the individual stimulus durations also indicated a significant effect of dose [F(3,6)=4.27, p=0.008], stimulus duration [F(2,21)=31.13, p<0.0001] and a significant dose by stimulus duration interaction [F(6,63)=3.54, p=0.004]. Post-hoc analysis indicated significant improvements in accuracy at the 1 sec stimulus duration following atomoxetine (1 mg/kg) and at the 2 sec stimulus duration following atomoxetine (3 mg/kg). (Fig. 1A) (Fig. 1A). Assessment of vSD inhibitory control measures indicated that atomoxetine significantly decreased the number of premature responses [F(3,7)=18.36, p<0.0001, Fig 1B], timeout responses [F(3,7)=5.04, p=0.008, Fig 1C] and food magazine head entries [F(3,7)=6.75, p=0.002, Fig 1D]. Notably, atomoxetine improved inhibitory control across several stimulus durations at doses of 1.0 and 3.0 mg/kg. Conversely, vSD perseverative responses (Table 2) following atomoxetine administration were not significantly different from vehicle performance [F(3,7)=1.77, p=0.18]. Atomoxetine did increase trial omissions and decrease trial completions (Table 2), [F(3,7)= 20.73, p<0.0001] and [F(3,7)=7.76, p=0.001], respectively, although the magnitude of the effects was modest. Similarly, atomoxetine was also associated with modest increases in response latencies in the vSD version of the 5C-SRTT (Table 2). Analysis of the overall correct response latency revealed a significant effect of dose [F(3,7)=20.23, p<0.0001] with post-hoc analysis indicating increased response latencies following the 1.0 and 3.0 mg/kg test dose. Analysis of the individual stimulus durations revealed a significant effect of dose [F(3,6)=11.25, p=0.0001] and a significant dose by stimulus duration interaction [F(6,63)=2.32, p=0.043]. Post-hoc analysis revealed significant increase in correct response latencies at the 2 sec stimulus duration associated with atomoxetine 1.0 and 3.0 mg/kg. Significant effects of atomoxetine were also observed for the vSD incorrect response latency [F(3,7)= 5.04, p=0.008] at the 2 sec stimulus duration following the 1.0 mg/kg test dose. Latencies to retrieve the reward were not significantly altered (p>0.05).

Fig 1.

Effects of atomoxetine (0.3-3.0 mg/kg, i.p.) in a variable stimulus duration (vSD) version of the 5C-SRTT in adult Wistar rats. (A) Percent correct (accuracy); (B) total number of premature responses (impulsive behavior); (C) total number timeout responses (compulsive behavior/cognitive inflexibility); D. total number of magazine head entries (compulsive behavior/cognitive inflexibility). Each bar represents the mean ± SEM for each test group; (N=7 rats). * = significantly different (p<0.05) from vehicle (saline)-related response. ATX = atomoxetine.

Table 2.

Effects of Atomoxetine on response latencies and trial completions in the 5C-SRTT task in young Sprague-Dawley rats.

Treatment	Test Measure	Latency Correct (s)	Latency Incorrect (s)	Latency Reward (s)	Omissions	Perseverative Responses	Trial Completions
Variable Stimulus Duration (vSD)
Saline	Overall vSD	0.98 ± 0.07	2.05 ± 0.19	1.91 ± 0.16	10.12 ± 1.32	5.37 ± 1.33	96.5 ± 2.41
	0.5 sec	1.00 ± 0.12	2.05 ± 0.23	2.28 ± 0.39	5.12 ± 0.58	2.62 ± 1.03	33.6 ± 0.18
	1.0 sec	0.90 ± 0.06	2.26 ± 0.19	1.68 ± 0.15	3.62 ± 0.68	1.25 ± 0.37	32.0 ± 0.82
	2.0 sec	1.00 ± 0.05	1.30 ± 0.26	1.80 ± 0.17	1.37 ± 0.32	1.75 ± 0.55	32.1 ± 0.85
Atomoxetine	Overall vSD	0.91 ± 0.06	2.01 ± 0.18	1.70 ± 0.14	4.75 ± 0.84	3.00 ± 0.70	100 ± 0.00
0.3 mg/kg	0.5 sec	0.85 ± 0.07	2.16 ± 0.26	1.72 ± 0.13	1.62 ± 0.59*	1.25 ± 0.59	30.0 ± 1.50
	1.0 sec	0.90 ± 0.05	2.02 ± 0.28	1.78 ± 0.23	2.00 ± 0.62	1.25 ± 0.36	33.2 ± 0.16
	2.0 sec	0.96 ± 0.07	1.55 ± 0.14	1.60 ± 0.12	1.12 ± 0.39	0.50 ± 0.18	33.1 ± 0.12
Atomoxetine	Overall vSD	1.15 ± 0.06*	2.80 ± 0.27	2.26 ± 0.19	15.00 ± 2.05	2.62 ± 0.84	90.0 ± 4.82
1 mg/kg	0.5 sec	1.08 ± 0.15	2.52 ± 0.26	2.32 ± 0.49	6.37± 1.03	0.75 ± 0.31	30.0 ± 1.50
	1.0 sec	1.05 ± 0.06	2.87 ± 0.39	2.11 ± 0.32	4.87 ± 0.78	0.62 ± 0.32	30.1 ± 1.74
	2.0 sec	1.31 ± 0.07*	2.81 ± 0.19*	2.28 ± 0.28	3.75 ± 0.75*	1.25 ± 0.59	29.8 ± 1.59
Atomoxetine	Overall vSD	1.17 ± 0.08*	2.54 ± 0.15	2.80 ± 0.52	20.87 ± 1.70*	2.75 ± 1.09	69.0 ± 8.67*
3 mg/kg	0.5 sec	0.93 ± 0.09	2.32 ± 0.21	3.68 ± 1.27	8.62 ± 0.59*	0.75 ± 0.36	23.0 ± 2.92*
	1.0 sec	1.11 ± 0.12	3.04 ± 0.20	1.64 ± 0.11	7.12 ± 0.91*	0.25± 0.16	22.7 ± 2.88*
	2.0 sec	1.34 ± 0.09*	2.70 ± 0.67	3.07 ± 0.65	5.12± 0.89*	1.75 ± 1.08	23.2 ± 2.87*
Variable Inter-trial Interval (vITI)
Saline	Overall vITI	1.02 ± 0.08	2.41 ± 0.16	1.83 ± 0.21	11.25 ± 1.31	3.87 ± 1.67	100 ± 0.00
	2.5 sec	1.21 ± 0.12	2.97 ± 0.22	1.92 ± 0.31	3.00 ± 0.65	1.87 ± 1.04	33.6 ± 0.18
	5.0 sec	0.92 ± 0.08	2.22 ± 0.27	1.73 ± 0.19	1.00 ± 0.26	1.37 ± 0.42	33.2 ± 0.16
	10.0 sec	0.97 ± 0.07	1.45 ± 0.14	1.83 ± 0.23	7.25 ± 1.03	0.62 ± 0.37	33.1 ± 0.12
Atomoxetine	Overall vITI	1.04 ± 0.07	2.44 ± 0.13	1.87 ± 0.26	16.37 ± 1.65	3.00 ± 1.03	100 ± 0.00
0.3 mg/kg	2.5 sec	1.18 ± 0.11	3.06 ± 0.15	2.11 ± 0.37	5.62 ± 0.68	1.50 ± 0.70	33.2 ± 0.16
	5.0 sec	1.00 ± 0.12	1.99 ± 0.22	1.70 ± 0.17	3.37 ± 0.80	0.87 ± 0.29	33.3 ± 0.18
	10.0 sec	0.88 ± 0.05	1.45 ± 0.44	1.81 ± 0.29	7.12 ± 0.61	0.50 ± 0.26	33.0 ± 0.00
Atomoxetine	Overall vITI	1.09 ± 0.09	2.86 ± 0.23	2.75 ± 0.39	35.50 ± 3.52*	3.50 ± 1.23	97.7 ± 2.25
1 mg/kg	2.5 sec	1.28 ± 0.12	3.39 ± 0.44	2.29 ± 0.31	15.00 ± 2.06*	1.00 ± 0.46	32.7 ± 0.70
	5.0 sec	1.03 ± 0.09	3.19 ± 0.52	1.96 ± 0.16	10.25 ± 1.56*	1.25 ± 0.64	32.6 ± 0.82
	10.0 sec	1.00 ± 0.08	1.25 ± 0.33	3.46 ± 1.00	10.12 ± 0.78	3.37 ± 2.04	32.2 ± 0.75
Atomoxetine	Overall vITI	1.09 ± 0.06	2.54 ± 0.32	3.78 ± 0.90	27.62 ± 3.68*	3.5 ± 0.96	76.7 ± 9.77*
3 mg/kg	2.5 sec	1.25 ± 0.09	3.48 ± 0.42	5.41 ± 2.17	10.75 ± 1.48*	1.25 ± 0.41	25.1 ± 3.16*
	5.0 sec	1.09 ± 0.11	2.46 ± 0.34	4.08 ± 0.97*	8.25 ± 1.41*	0.87 ± 0.29	25.2 ± 3.16*
	10.0 sec	1.00 ± 0.06	1.50 ± 0.31	2.97 ± 0.69	8.50 ± 1.62	1.62 ± 0.68	25.0 ± 3.00*

Data represent the mean (± S.E.M.) values for each treatment.

^*= significantly different (p<0.05) from vehicle control performance.

3.1.2. Atomoxetine effects in the variable intertrial interval (vITI) version of the 5C-SRTT

Statistical analysis of the effects of atomoxetine on overall vITI accuracy (% correct) did not indicate a significant effect of dose [F(3,7)=0.694, p=0.56], although there was a significant dose by inter-trial interval interaction with post-hoc analysis indicating improvements in accuracy at the 2.5 sec interval following the 3 mg/kg test dose (Fig. 2A). Similar to the effect indicated that atomoxetine significantly decreased premature responses [F(3,7)=14.34, p<0.0001], timeout responses [F(3,7)=10.26, p=0.0002] and magazine head entries [F(3,7)=9.54, p=0.0004] in the vITI version of the 5C-SRTT. As shown in Fig 2, atomoxetine improved inhibitory control across several time intervals following doses of 1.0 and 3.0 mg/kg. Perseverative responses (Table 2) during the vITI test were not significantly altered [F(3,7)=0.108, p=0.95], although trial omissions and trial completions (Table 2) (similar to the vSD task) were modestly altered by atomoxetine (i.e., increased [F(3,7)= 15.14, p<0.0001] and decreased, [F(3,7)=4.84, p=0.010], respectively. Analysis of the individual ITIs also indicated modest, but significant atomoxetine-related increases in trial omissions and decreases in trial completions (p<0.05). Analysis of the effects of atomoxetine on response latencies in the vITI task are shown in Table 2. Correct and incorrect response latencies following atomoxetine were not significantly different from that observed after vehicle control administration (p>0.05) whereas the latency to retrieve the reward was slightly increased (p<0.05) following atomoxetine.3.0 mg/kg administration at the 5 sec ITI.

Fig 2.

Effects of atomoxetine (0.3-3.0 mg/kg, i.p.) in a variable intertrial interval (vITI) version of the 5C-SRTT in adult Wistar rats. (A) Percent correct (accuracy); (B) total number of premature responses (impulsive behavior); (C) total number timeout responses (compulsive behavior/cognitive inflexibility); D. total number of magazine head entries (compulsive behavior/cognitive inflexibility). Each bar represents the mean ± SEM for each test group; (N=7 rats). * = significantly different (p<0.05) from vehicle (saline)-related response. ATX = atomoxetine.

3.1.3. Atomoxetine effects on RAM performance

The effects of atomoxetine administration in the rat DNMTP version of the RAM task at a 4 hr delay interval are presented in Fig 3. The rats trained in this version of the RAM task committed very few working memory errors (Fig 3A) in the free-eight (retention) session and none of the doses of atomoxetine affected the number of working memory errors. In contrast, the rats did commit a significant number of reference memory errors and these errors were reduced by atomoxetine. There was a significant main effect of dose [F(3,48)=4.63, p=0.006] and a significant day effect [F(9,48)=7.88, p<0.001], while the dose by day interaction was not significant [F(27,431)=0.881, p=0.64]. Post-hoc analysis indicated significant (p<0.05) improvements in performance associated with atomoxetine at the 5.6 mg/kg on day 7 and with both the 3.0 mg and 5.6 mg/kg dose on day 10 (Fig 3B). In addition, in an AUC analysis for all 10 days of testing in the free-eight session, the 5.6 mg/kg dose was associated with improved performance (p=0.01, see Fig 3B inset).

Fig 3.

Effects of atomoxetine (1.0-5.6 mg/kg, s.c.) in a delayed non-match to position (DNMTP) version of a RAM task in young adult Long Evans rats. (A) Effects of atomoxetine on spatial working memory errors across the 10 days of treatment. (B) Effects of atomoxetine on spatial reference memory errors across the 10 days of treatment. The Inset to Fig 3B depicts the effects of atomoxetine on the Area Under the Curve (AUC) for spatial reference memory errors across the 10 days of testing, Data points represent mean (± S.E.M) for each treatment (N=13 rats/treatment) across the 10 days of testing. * = significantly different (p<0.05) from vehicle when compared to saline (vehicle) control. ATOM = atomoxetine.

3.2. Non-Human Primate Behavioral Studies

3.2.1. DMTS-standard and DMTS-D baseline performance

As noted above, the effects of atomoxetine on distractibility in a working/short-term memory task were assessed in male (N=2) and female (N=7) rhesus monkeys with an average group age of 21.8 ± 1.7 yrs old. Task durations for the short and long delay intervals averaged 6.3 ± 1.0 and 75.0 ± 11.5 sec, respectively, with no significant duration differences observed between the sexes (Table 1). Baseline performance following saline administration (using the DMTS-standard protocol without distractors) resulted in accuracies (% correct) that approximated our established performance criteria (see Methods); 96.7 ± 0.82 (zero delay), 76.9± 1.1 (short delay) and 54.1 ± 1.0 (long delay) with a delay dependent reduction in DMTS accuracy [F(2,16)=488.8, p<0.0001] (Fig. 4A). Post-hoc analysis indicated that DMTS-standard accuracy was significantly reduced with each increase in delay duration (i.e., zero>short>long delay accuracy).

Fig 4.

Effects of nicotine and atomoxetine in a distractor version of a delayed match to sample task (DMTS) in aged rhesus monkeys. (A). Baseline performance on standard DMTS trials and DMTS trials with distractor (DMTS-D). The DMTS-standard task consisted of 96 trials divided equally across the three delays (i.e., zero, short and long) whereas the DMTS-distractor task had 96 trials, 72 trials were non-distractor trials (presented equally across the three delays) and 24 trials were distractor trials for which an interfering visual stimulus was presented prior to the short and long delays. *= significantly different (p<0.05) from DMTS non-distractor-related accuracy (B). Effects of nicotine (0.0125-0.05 mg/kg, i.m.) on short delay DMTS-distractor performance in aged rhesus monkeys. (C). Effects of atomoxetine (0.03-1.0 mg/kg, i.m.) on short delay DMTS-distractor performance in aged rhesus monkeys. Histograms presented in Fig 4B and 4C represent the mean (± S.E.M.) values of the difference (delta) from DMTS distractor short delay accuracy and DMTS non-distractor short delay accuracy and are illustrated as the “change from non-distractor control”. The DMTS non-distractor control value represents the delta between the DMTS-standard short delay accuracy and the DMTS non-distractor response. * = significantly different (p<0.05) from DMTS non-distractor saline response. # = significantly different (p<0.05) from the DMTS distractor response. Nic = nicotine; ATX = atomoxetine. N=9

In comparison to the accuracy associated with the standard version of the DMTS-task, testing subjects under the DMTS distractor method (DMTS-D) following saline revealed a significant main effect of the task [F(2,24)=334.1, p<0.0001], delay [F(1,24)=191.2, p<0.0001] and task by delay interaction [F(2,24)=53.6, p< 0.0001] (Fig. 1). Presenting the subject with a distractor stimulus lasting approximately 70% of the short delay duration significantly reduced accuracy by an average of 28% (range: 25.4-30.0%) compared to the accuracy obtained during the non-distractor trials within the same session as well as when compared to the accuracy observed during the DMTS-standard short delay trials (Fig. 4A). This DMTS-D short delay decrement in accuracy remained consistent throughout all pharmacological testing (see Figures 3-4). Accuracy following the DMTS-D long delay trials was not significantly different from accuracies obtained during the non-distractor or the DMTS-standard long delay trials.

3.2.2. Nicotine effects on DMTS-D performance

Nicotine administration was associated with improvements in accuracy in the DMTS-D task at short delays (Fig. 4B). Assessment of the delta (DMTS distractor minus non-distractor accuracy) indicated a significant effect of atomoxetine on accuracy [F(4,32)=19.25, p<0.0001]. The distractor-induced reduction in short delay accuracy was attenuated with increasing doses of nicotine. Post-hoc analysis indicated that the 0.025 and 0.05 mg/kg doses of nicotine significantly attenuated the effects of the distractor stimulus on short delay accuracy and the accuracy following the 0.05 mg/kg dose of nicotine was not significantly different from that observed during non-distractor saline administration. Nicotine did not significantly alter the accuracy of the DMTS non-distractor short delay trials or the non-distractor or distractor long delay trials (data not shown). Response latencies following nicotine were also not significantly different from vehicle control (data not shown).

3.2.3. Atomoxetine effect on DMTS-D performance

Administration of atomoxetine also produced a significant improvement in DMTS-D short delay accuracy in aged rhesus monkeys (Fig. 4C). Statistical analysis of the distractor-non-distractor delta revealed a significant effect of atomoxetine on accuracy [F(5,40)=5.27, p=0.0008]. Post-hoc analysis indicated that atomoxetine completely reversed the distractor-induced decrement in accuracy following doses of 0.1, 0.3 and 1.0 mg/kg. The effects of atomoxetine on the accuracy of the DMTS non-distractor short delay trials, the non-distractor long delay and distractor long delay trials were not significantly different from saline control (data not shown). Response latencies following atomoxetine were also not significantly different from vehicle control (data not shown).

3.2.4. Donepezil effect on DMTS-D performance

While there appeared to be a trend toward an attenuation of the effect of the distractor by the 0.05 mg/kg dose of donepezil, none of the doses resulted in a statistically significant change in accuracy compared to saline (Fig. 5A). Moreover, none of the doses of donepezil had a significant effect on non-distractor short delay accuracy or distractor or non-distractor long delay accuracy (data not shown). Response latencies following donepezil were also not significantly different from vehicle control (data not shown).

Fig. 5.

(A) Effects of donepezil on short delay performance in a distractor version of a delayed match to sample task (DMTS) in aged rhesus monkeys. (B). Effects of combing sub-effective test doses of atomoxetine (0.03 mg/kg, i.m.) and donepezil (0.05 mg/kg, i.m.) on short delay DMTS-distractor performance in aged rhesus monkeys. Data represent the mean (± S.E.M.) values of the delta change from non-distractor control performance (see Fig 4 legend for details). * = significantly different (p<0.05) from DMTS non-distractor saline response. # = significantly different (p<0.05) from the DMTS distractor response. Don = donepezil; ATX = atomoxetine. N=9

3.2.5. Combined sub-effective doses of atomoxetine and donepezil on DMTS-D performance

The combination of non-efficacious doses of atomoxetine (0.03 mg/kg, i.m.) and donepezil (0.05 mg/kg, i.m.) significantly improved DMTS-D short delay accuracy (Fig. 5B). There was a main effect of treatment [F(4,32)=10.34, p < 0.0001] and post-hoc analysis indicated that the drug-combination improved accuracy at the short delay in the DMTS-D to baseline non-distractor control levels.

4. Discussion

The results of this study can be summarized as follows: 1) in the both the vSD and vITI versions of the 5C-SRTT, atomoxetine (depending on the dose) improved both accuracy and inhibitory response control (i.e., it decreased premature responses, timeout responses and excessive food magazine head entries) in adults rats, 2) in the RAM task in adults rats, atomoxetine, in a dose dependent manner, improved spatial reference memory after a 4 hr delay interval, and 3) in aged monkeys both nicotine and atomoxetine, but not donepezil, attenuated the effects of the distractor on accuracy at the short delays in the DMTS-D task. Moreover, combining sub-effective doses of atomoxetine and donepezil was also effective at enhancing DMTS-D accuracy.

The objective of the 5C-SRTT experiments was to determine if atomoxetine administration might result in a decrease in vulnerability to alterations of attention and inhibitory response control when the demands of the task were increased. This was accomplished by varying the stimulus durations and intertrial intervals. The vSD version of the 5C-SRTT is used to increase attentional load, thus it often results in decreased accuracy (% correct), the main measure of attentional performance in the task (Higgins and Breysse, 2008; Bari et al., 2008). As expected, when tested using vSDs, the experimental subjects exhibited a significant (stimulus-dependent) decrease in accuracy. The vITI version of the 5C-SRTT is often used to increase the demand on the inhibition of inappropriate responding by making the appearance of the stimuli unpredictable. This version of the task often results in increased premature responses which are generally interpreted as a form of impulsive behavior (see review, Robbins 2002; Bari et al., 2008; Amitai and Markou, 2011). Predictably, when the test subjects were exposed to a pseudorandom presentation of different ITIs, there was a significant increase in premature responses at the longest (10 sec) ITI.

Depending on the stimulus duration in the vSD task and the intertrial interval in the vITI task, atomoxetine improved accuracy, the primary measure of sustained attention in the 5C-SRTT. Interestingly, in both versions of the 5C-SRTT, atomoxetine also decreased premature responses (i.e., inappropriate nose pokes during the intertrial interval before the target stimulus has been presented) timeout responses (i.e., nose pokes made during the timeout interval which occurred after an incorrect response or premature response) and food magazine head entries. The positive effects of atomoxetine on premature responses are in agreement with several previously published rodent studies (e.g., Baarendse and Vanderschuren, 2012; Fernando et al., 2012; Robinson et al., 2008) and are indicative of decreases in impulsive-like behavior. The atomoxetine-related effects on the number of timeout responses and food magazine head entries indicate an attenuation of compulsive-like behaviors as well as improvements of cognitive flexibility (i.e., the ability to alter behavior in reaction to changing situational demands, in this case, disorganized responses that are not tied to the stimulus presentation, see Amitai and Markou, 2010). There are several observations that led us to conclude that the effects of atomoxetine were primarily related to improvements of attention/executive function as opposed to alterations in motivation, locomotor activity, etc. We only noted one case where a statistically significant effect of atomoxetine on the magazine latency (i.e., the latency to collect food rewards) was observed (i.e., a minor increase associated with the 3.0 mg/kg dose of atomoxetine at the 5 sec ITI in the vITI task). Food magazine latencies have been described as an index of motivation in the 5C-SRTT (see review, Robbins, 2002). Likewise, we only observed a few, minor increases in the response latencies following atomoxetine administration, all in the vSD version of the 5C-SRTT. An increase in the total number of trial omissions was observed in both versions of the task, but associated only with the highest (3.0 mg/kg) dose of atomoxetine.. At the 3.0 mg/kg dose of atomoxetine the number of total trials completed was reduced from ~96.5 (saline) to ~69 in the vSD task and from 100.0 (saline) to ~77 in the vITI task, respectively. In the 5C-SRTT, subjects must nose poke the food magazine after an incorrect response (i.e., selection of incorrect stimulus aperture), premature response, or an omission (i.e., when 5 sec expired after the stimulus presentation without a nose poke) in order to initiate the next trial. Therefore, significant delays in initiation (and thus reduced motivation and/or negative drug effects on locomotor activity) would likely be exemplified by a more robust decrease in the total number of trials completed. An additional argument against motivational deficits underlying the increased omissions and response latencies is the observation that norepinephrine reuptake inhibitors increase reinforcement rates in differential reinforcement (O'Donnell et al. 2005).

Modest, atomoxetine-related increases in omissions and reaction times have been observed previously in various versions of the 5C-SRTT as well as other rodent tasks of attention (Jentsch et al. 2009; Paterson et al., 2011; Baarendse and Vanderschuren, 2012; Fernando et al., 2012). Interestingly, Jentsch and colleagues (Jentsch et al. 2009) speculated that the tendency of atomoxetine to increase response latencies and omissions while at the same time improving task performance (accuracy, inhibitory control) indicates a more conservative strategy of waiting for target delivery during the observing response. In other words, by generally slowing the behavioral output, atomoxetine may allow the subject to take additional time for more accurate processing of the stimulus light location.

In our RAM studies, we employed a modification of a version of the task that was originally developed to evaluate both working and reference memory at the same time (Jarrard, 1983, see also review, Dudchenko, 2004). In the Jarrard version of the task, only four maze arms are baited and the same arms are baited each day, thus across sessions, the rats learn to ignore the remaining four arms which never contain a food reward. Entries into a never-baited arm are considered reference memory errors, while re-entries into one of the four recently baited arms are considered working memory errors. In our version of the task (see Terry et al., 2012), rats are first trained in a simple win-shift version of the RAM task for two weeks and then exposed to a delayed non-match to position task with a significant delay between the forced 4 (information) and free-eight (retention) session. This increases task difficulty and emphasizes the spatial recall elements of the procedure. The task difficulty is further increased by changing the location of the baited arms in the information (forced 4) session each day. In the current study, using a four-hour delay between the information and retention session, the subjects made very few spatial working memory errors and this error rate was not affected by atomoxetine treatment. In contrast, the subjects (under vehicle conditions) committed more spatial reference memory errors during the retention sessions, an effect that was improved by atomoxetine administration. Specifically, an AUC analysis of the spatial reference memory errors across the 10 days of testing indicated improvements in overall performance of the task associated with the 5.6 mg/kg dose of atomoxetine. Moreover, the 3.0 and 5.6 mg/kg doses of atomoxetine reduced the number of these errors on specific days of testing.

In the final series of experiments, we evaluated atomoxetine for effects on non-spatial working memory and susceptibility to distraction in aged monkeys using a distractor version of a delayed match to sample task (DMTS-D). Monkeys (specifically macaques) have unique translational value due to their genetic homology with humans and similar brain anatomy, as well as their more complex behavioral repertoire. Moreover, studies in old monkeys may have particular relevance to human cognitive disorders (e.g., AD, mild cognitive impairment) where old age is a major risk factor. Like humans, performance across multiple domains of cognition declines in macaques and they exhibit a variety of age-related changes in the brain that are similar (e.g., neuronal loss, amyloid plaques, reactive glia, decreased neurotransmitters, etc.). Please see a more detailed discussion of the translational value of macaques and the effects of aging in our previous manuscript, Callahan et al. 2013. Distractors in delayed response tasks in monkeys have been employed on a number of occasions to increase task difficulty (e.g., Arnsten and Contant, 1992; Prendergast et al., 1998a, Terry et al., 2016). It is believed that the presentation of distractors during the delay or recall interval, a time during which selective attention and rehearsing are thought to be important, disrupts the cognitive processes subserving working memory (reviewed, Rodriguez and Paule, 2009). Here it is important to note that the inability to maintain attention in the presence of distracting stimuli is a key feature of many age-related neurodegenerative disorders including AD (Broks et al. 1988; Davis et al. 1990; Jones et al. 1992). Thus, the use of delayed recall tasks that employ distracting stimuli may provide a model for assessing the potential efficacy of pharmacologic agents in reducing distractibility in humans.

We began this series of studies by evaluating nicotine as a reference compound since we have found it to be efficacious in the DMTS-D task in macaques in previous studies (Prendergast et al., 1998a). As expected, nicotine in a dose-dependent fashion attenuated the negative effects of the distractors at the short delay interval across a similar range of μg/kg doses as used in the Prendergast study. In the Prendergast study, the subjects were young-adults, whereas in the current study all but one of the subjects was aged (>20 years old). Here it is important to note that the positive effects of nicotine observed in some studies specifically on delayed match to sample performance in monkeys have been challenged due to the reliance on an individualized optimal (best) dose analysis which in the view of Kangas and Branch, 2012 can inflate effect size. In our current study, each of the three doses of nicotine evaluated was administered at least three times in a pseudo-random order and reported as the average of these three evaluations. We did not rely on a best (individualized) dose of nicotine.

In the next series of DMTS-D experiments, we evaluated atomoxetine. Here we also observed a dose dependent attenuation of the effects of the distractors at the short delay interval. These two positive observations with nicotine and atomoxetine may be important in the context of age-related disorders of cognition, since we have not found classic, stimulant-based, treatments for ADHD like methylphenidate to be particularly effective at improving distractibility in aged monkeys (see Prendergast 1998b). Likewise, Bhattacharya and colleagues found that while methylphenidate was effective in young-adult rats in a sustained attention task that included a visual distractor, it was not effective in aged rats (Bhattacharya et al., 2015). Here it is also important to note that in one of our previous studies (Bain et al., 2003) in a younger cohort of monkeys (a mixture of rhesus and pigtail macaques, average age~ 9.8), we did not observe robust effects of atomoxetine in a DMTS-D task. While the basis for these contrasting results is unclear, the mixed species and younger age may have contributed. Moreover, in the previous study we used different task parameters in the DMTS-D procedure (e.g., four delays such that fewer distracters were administered per delay in a given session, and a different distractor duration).

In the final phase of DMTS-D experiments, we evaluated the commonly prescribed AD treatment and cholinesterase inhibitor, donepezil. The results indicated that clinically relevant doses of donepezil were not effective at reducing the impairing effect of the distractor in the DMTS-D task. This observation contrasts with the positive effects we observed previously in the standard version of the DMTS procedure in old monkeys, where long delay performance was improved by clinically relevant doses of donepezil (see Callahan et al., 2013). Interestingly, when we combined sub-effective doses of donepezil and atomoxetine in the DMTS-D task, the negative effects of the distractors at the short delay interval were attenuated. Given that we only evaluated one combination of ineffective doses of the two compounds, we are unable to determine if this interaction reflects an additive or synergistic effect (see Tallarida 2016). Nevertheless, this observation would appear to suggest that each compound (donepezil and atomoxetine) can increase the effective dose range of the other and thus raises the question of whether such an adjunctive treatment strategy might be effective for attention-related impairments in older human patients who suffer from cognitive disorders. This adjunctive treatment strategy may be particularly important in AD since patients enrolled in future clinical trials will likely already be taking a cholinesterase inhibitor. The mechanism of positive effects of the donepezil-atomoxetine combination on DMTS-D performance is unclear, but could be related to the ability of atomoxetine to increase cortical acetylcholine release (as indicated in the aforementioned Tzavara et al., 2006 rodent study) and the ability of donepezil to sustain synaptic acetylcholine levels via cholinesterase inhibition. Since both donepezil and atomoxetine are substrates of the cytochrome P450 (CYP) 2D6 enzyme (see Coin et al., 2016; Sauer et al., 2005), the positive effect on cognition could also be related to decreases in drug clearance and consequent increases in drug levels in the brain.

Our data in aged monkeys would appear to suggest that atomoxetine might have potential as a therapeutic agent in several diseases where deficits in executive function are prevalent. Interestingly, a recent review of multiple randomized controlled trials in Parkinson’s disease patients indicated that atomoxetine could improve several markers of executive dysfunction including impulsivity, risk taking, and global cognition (Warner et al., 2018). In our old monkey experiments, each of the individual doses of atomoxetine was administered at least three times over a multi-week period (indicating the reproducibility of the cognitive effect), however, future studies with repeated drug dosing regimens would be necessary to determine how efficacious the regimens are likely to be in the clinical realm over time.

It should be noted that some relatively small (but randomized and controlled) trials in patients with attention deficits resulting from Huntington’s disease (Beglinger et al., 2009) schizophrenia (Kelly et al., 2009) or traumatic brain injury (Ripley et al., 2014), did not yield any significant improvement in cognition with atomoxetine. Moreover, there is one published study to date where the addition of atomoxetine to ongoing cholinesterase-inhibitor therapy in AD patients (N=92) did not significantly improve cognitive function (Mohs et al., 2009). Interestingly, the Mini-Mental State Examination (MMSE) scores used to evaluate cognitive function in this study ranged between 10 and 26 at baseline. This wide range of test scores is indicative of a very heterogeneous group of test subjects, which most likely had very wide differences in the level of neurodegeneration. It has been argued previously that for therapeutic interventions to be effective in AD, they most likely need to be applied much earlier in the course of AD than has been done to date (reviewed, Sperling et al., 2011). The very low MMSE scores in some of the patients in the Mohs study are indicative of advanced levels of brain degeneration, which may not be responsive to any type of pharmacologic intervention.

In conclusion, the results of the experiments described in this report indicate that atomoxetine improves several components of executive function in young-adult and aged animals (e.g., attention, cognitive flexibility, working/short term memory, susceptibility to distraction). Collectively, these observations in animals support the further evaluation of atomoxetine as a repurposed drug for younger adults with deficits in attention, memory and other components of executive function, as well as for age-related disorders where these domains of cognition are impaired (e.g., mild cognitive impairment, Parkinson’s disease, and AD.

Highlights.

• Atomoxetine improved attention and inhibitory response control in adult rats.

• Atomoxetine improved spatial reference memory in adult rats.

• Atomoxetine, but not donepezil, improved distractibility in aged monkeys.

• Atomoxetine plus donepezil improved distractibility in aged monkeys.