Acetylcholine plays a major role in mediating attention processes. We investigated the muscarinic antagonist effect of scopolamine on functional neuro-anatomy of attention and cognition. We assessed 12 healthy volunteers while performing the Attention Network Task on 0.4 mg scopolamine and placebo in a single-blind randomized trial in a 1.5 T magnetic resonance scanner. Neurocognitive measures included verbal learning, verbal memory, verbal fluency, trail making, digit span, a continuous performance task and a planning task (Tower of London). When compared to placebo, scopolamine increased reaction times for conflicting stimulus processing, together with decreasing brain activation in the anterior cingulate cortex (a brain region involved in conflict processing) suggestive of a muscarinic antagonist effect on executive control of attention. Contrary to the notion of a predominantly right-hemispheric lateralization of cognitive processes associated with orienting attention, scopolamine reduced brain activity in left superior and left middle frontal brain areas. Our neuropsychological test data revealed a selective effect of scopolamine on verbal learning and memory while other cognitive domains, such as planning and working memory, were unaffected. These findings are consistent with muscarinic modulation of dopaminergic neurotransmission in frontal attention networks when processing conflicting information.
Acetylcholine (ACh) plays a major role in mediating attention (Callejas et al.2004; Fan et al.2002; Freedman et al.1988, 1999, 2000; Ochoa & Lasalde-Dominicci, 2007; Sabri et al.2008). ACh exerts its diverse physiological actions by binding to and activating two structurally and functionally distinct families of cell-surface receptors: (1) the nicotinic ACh receptors, which form ligand-gated ion channels, and (2) the G protein-coupled muscarinic ACh receptors. Both muscarinic and nicotinic receptors have been implicated in cognition, particularly in attention and memory function (Friedman, 2004; Green et al.2005; Sarter et al.2005) and have also been linked to conditions like Alzheimer's disease and schizophrenia (Buchanan et al.2007; Langmead et al.2008; Sarter et al.2005). There is also some evidence suggesting potentially beneficial effects of muscarine agonists when targeting cognitive impairment and psychotic symptoms (Shannon et al.1994; Shekhar et al.2008).
Most experimental studies have investigated the effects of nicotinic and muscarinic antagonists on cognition. For instance, the muscarinic antagonist scopolamine has detrimental effects on working and declarative memory, sustained visual attention (Furey et al.2008), and psychomotor speed learning when tested in healthy subjects (Ellis et al.2006; Sitaram et al.1978). Sherman et al. (2003) reported impaired recollection and familiarity of memory performance in a set of complex visual images in response to a single dose of 0.4 mg scopolamine. Koller et al. (2003) reported impaired memory and vigilance performance irrespective of dose (i.e. 0.3 mg and 0.6 mg). Green et al. (2005) further demonstrated that scopolamine significantly impaired both object and spatial N-back working-memory performance. These effects were not present with the nicotinic antagonist mecamylamine, thus suggesting some specificity of the muscarinergic ACh receptors in mediating the reported effects on cognition.
Other reports also suggest that scopolamine interferes with sensory information processing. Pekkonen et al. (2001) reported decreased mismatch negativity amplitudes with scopolamine when assessing pre-attentive auditory sensory memory function with frequency deviants. In contrast, middle-latency magnetic fields in response to auditory stimuli increase with scopolamine (Jaaskelainen et al.1999).
These findings in humans are largely supported by animal research in non-human primates. For instance, scopolamine disrupts auditory short-term memory in a delayed matching to sample task (Plakke et al.2008) while intracranial infusion of scopolamine into the intraparietal cortex slowed covert orienting performance (Davidson & Marrocco, 2000). Scopolamine also impairs visual short-term memory (Bartus & Johnson, 1976) and reduces attention modulation in primary visual cortex (Herrero et al.2008). The latter effect was not present with mecamylamine, thus supporting the notion of some specificity of the muscarinic ACh receptors in modulating attention and working-memory processes.
Functional magnetic resonance imaging (fMRI) studies on healthy volunteers indicate reduction of extrastriate, middle frontal, and inferior frontal brain activation associated with repetition suppression in left neocortex in response to scopolamine (Thiel et al.2001). Scopolamine also disrupts experience-dependent plasticity in auditory cortex (Thiel et al.2002) and de-activates hippocampal, fusiform, and inferior prefrontal regions when performing a face-name association task (Sperling et al.2002). However, functional brain-imaging studies specifically investigating scopolamine effects on attention processes are lacking.
The present study investigated the effects of scopolamine on cognition and, specifically, assessed the functional neuro-anatomy of attention processes in healthy subjects when performing the Attention Network Task (ANT; Fan et al.2002; Thienel et al.2009 for mecamylamine effects). This task allows for the discrimination of alerting, orienting and executive control processes of attention.
Each of these processes has been linked to different brain regions and neuromodulators. Alerting is predominantly associated with thalamic, anterior and posterior cortical activation (Fan et al.2005) and involves noradrenergic up-modulation of frontal and parietal regions in the right hemisphere (Witte & Marrocco, 1997). Executive control appears to be mainly modulated by dopamine and engages the anterior cingulate cortex (ACC) and the lateral prefrontal cortex (Bush et al.2000; MacDonald et al.2000; Marrocco & Davidson, 1998).
Most relevant to this study, ACh plays a major role in orienting (see Ochoa & Lasalde-Dominicci, 2007 for review; Posner & Petersen, 1990) which engages the temporo-parietal junction and the inferior frontal gyrus in the right hemisphere (Corbetta et al.2000).
Based on these findings, we predicted reduced frontal and parietal activation for orienting when challenged by scopolamine. However, effects of muscarinic signalling on mid-brain dopamine release via M4 receptors (Langmead et al.2008) suggest changes in mesocortical neurotransmission further downstream, thus also affecting executive control performance. Hence we predicted decreased activation in the ACC and the lateral prefrontal cortex with scopolamine when performing executive control along with increased response times when processing conflicting information. Muscarinic antagonist effects on alerting should not be present.
Following approval by the RWTH Aachen Medical Research Ethics Committee, 12 male right-handed (Edinburgh Handedness Inventory Score 9±0.9) subjects with a mean age of 26±2.1 yr (range 23–29 yr), were recruited into the study which also investigated nicotinic effects on ANT performance in a parallel functional brain-imaging study as reported in a companion paper (Thienel et al.2009). Subjects achieved a mean verbal IQ of 118±13.8 [Mehrfachwahl-Wortschatz-Intelligenztest (Multiple choice vocabulary test); Lehrl, 2005] and did not meet criteria for a current or lifetime psychiatric diagnosis [including current nicotine or other substance abuse or lifetime diagnosis of substance addiction (SCID-I)], or any other relevant past or present medical or neurological condition, including standard MRI exclusion criteria. Subjects gave informed, written consent and received a small honorarium for participation.
Subjects took part in three successive MRI sessions at least 1 wk apart and were randomized into a single-blind, crossover design. Active drugs and placebo were administered in a double-dummy procedure as described by Thienel et al. (2009) for mecamylamine effects on ANT performance: (1) A single dose of 0.4 mg scopolamine was injected intravenously and placebo given orally; (2) a single dose of 15 mg mecamylamine was given orally with a saline injection intravenously; or (3) placebo was administered orally together with an intravenous saline injection. Mecamylamine findings are reported elsewhere (Thienel et al.2009).
Stimuli and tasks
A modified version (Thienel et al.2009) of the ANT (Fan et al.2002, 2005; Konrad et al.2005) served as cognitive challenge in the MRI scanner. Three cueing conditions were used: (1) no cue (central fixation cross presented), (2) a central cue, or (3) a spatial cue which predicted the location of a target arrow 4° to the right or 4° to left of the central fixation cross. Left- or right-pointing target arrows were presented with four flanker arrows (two above and two below) either pointing in the same direction (congruent condition) or in the opposite direction (incongruent condition) to the target arrow.
Ten repetitions of the 24 possible stimulus combinations (i.e. three cue conditions, two flanker conditions, two target presentation sides, and two target-pointing directions) were presented in a randomized and counterbalanced sequence of 240 trials. Study participants were asked to respond by pressing a button with their right index or middle finger depending on the direction the target arrow was pointing (see Thienel et al.2009 for more detail). Reaction times were analysed for congruency, cueing and drug conditions using a full-factorial linear mixed-effects model (Pinheiro & Bates, 2000).
Following each MRI session, participants underwent neuropsychological testing consisting of (1) the Verbal Learning Memory Test [VLMT; Helmstaedter et al.1990, the German adaptation of the Rey Auditory Verbal learning Test (RAVLT; Schmidt, 1996)]; (2) the Tower of London task (ToL; Shallice, 1982; adapted by Tucha & Lange, 2004); (3) Trail Making Test (TMT A/B; Reitan, 1958); (4) Verbal Fluency [Regensburger Wortfluessigkeitstest (RWT); Aschenbrenner et al.2000]; (5) Digit Span (Haerting et al.1999); and (6) d2 test (d2; Brickenkamp, 1994). Analyses of variance (ANOVA) for repeated measures were employed to compare scopolamine vs. placebo test performance at p<0.0016 (Bonferroni-corrected for multiple comparisons).
MRI data acquisition
Echo-planar imaging (EPI) took place at the Jülich Research Centre (Germany) using a 1.5 T Siemens Symphony MRI scanner (90° flip angle, 3 s repetition time, and 60 ms echo time). Images consisted of 30 slices of 4-mm thickness (gap between slices was 0.4 mm) resulting in an in-plane resolution of 64×64. The field of view was 200×200 mm2 with voxel dimensions of 3.125×3.125×4 mm3. A series of 286 images was acquired for each subject discarding the initial three images to account for T1 stabilization effects.
MRI data analyses
MRI data were analysed using SPM2 (http://www.fil.ion.ucl.ac.uk/spm/). Images were aligned to the first volume using rigid-body transformations. Acquisition delays of individual slices were corrected by the 15th slice of each volume as reference in time before transforming into standard MNI space by normalizing to an EPI template. Data were spatially interpolated to a voxel size of 2×2×2 mm3 and spatially smoothed using an isotropic Gaussian kernel of 10 mm full width at half maximum.
Target stimulus onset determined event onset for haemodynamic modelling. Non-sphericity correction was performed to control for heteroscedasticity and covarying conditions. T contrasts were calculated at p<0.01 (voxel level) and clusters of >54 continuous voxels (equivalent to Monte Carlo-corrected threshold of p<0.05, 1000 iterations; Slotnick et al.2003) for correctly performed trials for scopolamine vs. placebo for (1) central vs. no cue (alerting), (2) spatial vs. central cue (orienting) and (3) incongruent vs. congruent (executive control), respectively, by repeated-measures ANOVA with reaction time as covariate of no interest (see Thienel et al.2009 for more details). Inclusive masking with main placebo effects at p<0.05 were conducted to reveal which areas with drug interaction were also activated with placebo alone. Automatic anatomic labelling was performed in SPM (automatic anatomic labelling, http://www.fil.ion.ucl.ac.uk/spm/ext/#AAL; Tzourio-Mazoyer et al.2002).
Adverse drug effects and behavioural data
None of the participants reported subjective drug effects when tested with placebo. Scopolamine induced modest side-effects in all participants (i.e. hyposalivation and mild sedation). However, one participant was excluded from our study due to excessive sedation.
The results for the placebo condition are reported in detail in Thienel et al. (2009). Briefly, with placebo the overall correct response rate was 98%. Across all ANT conditions, mean reaction time (±s.e.m.) was 638±11 ms. Alerting (central cue vs. no cue) and orienting (spatial cue vs. central cue) significantly facilitated response time by 31±7 ms and 75±11 ms [F(2, 55)>40.8, p<0.0001] and incongruent vs. congruent flanker conditions, assessing for executive control, slowed response times by 77±8 ms. [F(1, 55)=245.2, p<0.0001]. Cue interacted with flanker type [F(2, 55)=5.4, p<0.008; Fig. 1].
Rate of correct responses was not affected by scopolamine when compared to placebo (F<1.0). In contrast, scopolamine slowed mean reaction times (±s.e.m.) across all task conditions by 81±22 ms vs. placebo [F(1, 121)=39.8, p<0.0001] with a significant reaction time increase of 91±9 ms for the incongruent condition [congruency×drug interaction: F(1, 121)=5.7, p=0.02] when assessing for executive control and comparing to placebo. Scopolamine also abolished the cue×flanker type interaction seen in the placebo condition. However, no specific scopolamine effects were found for alerting and orienting conditions with comparable facilitations of response times [F(2, 55)>511.0, p<0.0001] by 30±7 ms and 87±9 ms, respectively; when comparing with placebo performance.
Functional imaging data
Alerting was predominantly associated with increased blood oxygenation level dependent (BOLD) effect in the left interior temporal gyrus with placebo. When contrasted with the placebo condition, scopolamine increased the alerting effect in the right middle temporal gyrus and the left middle occipital gyrus but reduced the alerting effect in the left hippocampus (Table 1, Fig. 2). Covarying reaction times did not change the results except for hippocampus BOLD activation becoming non-significant.
Bilateral prefrontal cortex, right precuneus, and left caudate activation was present with orienting with placebo while the orienting effect was significantly reduced with scopolamine in the left prefrontal and right precentral gyrus when compared to placebo. The prefrontal area was confirmed as a region activated with scopolamine vs. placebo and placebo only as well (Table 2, Fig. 3). Covarying reaction times did not change the results except for precentral BOLD activation becoming non-significant.
Brain areas co-activated with placebo and drug challenge.
Executive control was associated with increased BOLD bilaterally in the ACC, right superior frontal and parietal gyri, left gyrus rectus, right angular and left inferior occipital gyrus and bilateral precuneus in the placebo condition. The executive control effect was significantly reduced for scopolamine vs. placebo in regions also activated by placebo alone, e.g. the right superior and middle orbito-frontal gyrus, precuneus and the left gyrus rectus. Other areas affected by scopolamine included the right ACC, right middle orbicular frontal gyrus, lingual, inferior temporal, and precuneus as well as left lingual and inferior parietal gyrus with reduced executive control effect, while left inferior parietal gyrus showed increased executive control effect (Table 3, Fig. 4). Covarying reaction times did not change the results for clusters with increased BOLD (i.e. left inferior parietal lobe) or reduced BOLD (i.e. left gyrus rectus and right lingual gyrus) in response to scopolamine. However, peak activity of some clusters slightly shifted within the same anatomical areas in right superior frontal, precuneus, middle orbicular frontal, ACC, and inferior temporal gyrus.
Brain areas co-activated with placebo and drug challenge.
Scopolamine significantly impaired verbal learning memory performance (VLMT) by reducing the number of words reproduced in the free recall sessions 1–5 from 56±6 (values are ±s.d.) to 49±8 compared to 56±6 for placebo [F(1, 11)=12.9, p=0.004]. Delayed recall of word after 30 min was also affected by scopolamine from 14±2 to 12±4 [F(1, 11)=10.7, p=0.007]. No further effects of scopolamine on neuropsychological test performance was confirmed (F<1.0).
Our functional imaging findings support our predictions that muscarinic antagonism significantly reduced brain activation in parts of the orienting and executive control network. The latter finding is supported by our behavioural data showing a selective drug effect on processing incongruency. This association is further supported by parameter plots showing that muscarinic antagonism reverses brain activation when compared to placebo.
The most pronounced scopolamine effect was found with executive control when increasing reaction times were assessed after incongruent flanker conditions. This finding is consistent with our functional imaging data which demonstrate that brain activation associated with executive control in the ACC – a brain region subserving conflict processing as shown in previous studies (Fan et al.2005) – is abolished by scopolamine. In particular, the ACC is engaged in conflict resolution (Botvinick et al.2001; Bush et al.2000) as induced by incongruent conditions, while executive function tasks generally activate fronto-parietal cortical areas (Perfetti et al.2007; Specht et al.2008). Since executive control is hypothesized to be associated with dopaminergic neurotransmission, this finding can also be interpreted on the basis of the modulatory role of muscarinic receptors in mesocortical dopamine pathways (Langmead et al.2008).
When controlling for reaction time, the main results were largely confirmed with a few exceptions (i.e. hippocampus activation in alerting contrast; precentral activation in orienting contrast). While some peak activation slightly shifted when controlling for reaction time, they largely centred within the same areas as those identified for the placebo main effects. However, ACC differed in this respect. Scopolamine challenge activated a more dorsal portion of ACC compared to placebo. Nicotinic challenge, on the other hand, resulted in a general slowing of reaction times, irrespective of attention components along with no confounding effects on BOLD activation contrasts (Thienel et al.2009).
Our neuropsychological data further demonstrate that muscarinic antagonism selectively impairs verbal learning memory performance. Largely consistent with our results, Sherman et al. (2003) reported that scopolamine impaired the recollection and familiarity of items that had been learned previously. Ellis et al. (2006) also found detrimental effects of scopolamine on immediate and delayed recall. Surprisingly little or no effect was recorded for any other cognitive domain in our study which may be explained by a ceiling effect in our well-performing cohort of healthy volunteers or lack of test specificity in relation to executive control of attention (Riedel et al.1997).
On the other hand, potentially relevant to our neurocognitive findings are reports of beneficial effects of the selective M1/M4 mAChR agonist xanomeline on verbal learning and short-term memory functions in patients with schizophrenia (Shekhar et al.2008). Verbal memory deficits are also one of the few cognitive markers known to be sensitive to prodromal states and transition to psychosis, including schizophrenia (Lencz et al.2006), thus potentially providing a therapeutic target for selective M1/M4 mAChR agonists in ‘ultra high-risk’ populations by increasing ACh and dopamine release in the medial prefrontal cortex and hippocampus (Li et al.2007; Weiner et al.2004). Other preclinical studies provide supportive data for M1 agonists when treating cognitive deficits (Harries et al.1998; Loudon et al.1997; Schwarz et al.1999; Shannon et al.1994), while M4 agonists are more likely to act as an antipsychotic by inhibiting dopamine release (Langmead et al.2008).
Following placebo administration the orienting contrast (spatial vs. central cue) activated, as predicted, the predominantly right-lateralized neural network comprising inferior and superior frontal cortical areas, precuneus and caudate. This largely confirms previous reports (Corbetta et al.2000; Fan et al.2005); however, our findings differ in parietal activation by involvement of the right precuneus rather than the intraparietal sulcus (Corbetta et al.2000). Scopolamine disrupted the orienting effect in the left superior and left middle frontal brain areas, contrary to the notion of a predominantly right-hemispheric orienting network (see also Thienel et al.2009).
Our alerting (central vs. no cue) findings also have some inconsistencies with previous reports. In contrast to our data, Corbetta et al. (2000) and Fan et al. (2005) described right-hemispheric activation in the temporo-parietal junction and the inferior frontal gyrus. Our results showed increased alerting effects in the left inferior temporal gyrus with placebo and reduced brain activation in response to scopolamine vs. placebo in the left hippocampus. Consistent with previous studies (Fan et al.2005), this region did not show a placebo response for the alerting contrast. Furthermore, drug effects on hippocampus activation were not confirmed when controlling for reaction times.
Inconsistencies with other studies may also reflect differences in our study design when presenting the central cue as a fixation aid rather than a fixation cross, thus diminishing the ‘alerting salience’ of our cueing stimulus. This may also explain inconsistent findings for the alerting contrast with mecamylamine challenge, showing increased BOLD response in the left superior orbito-frontal and the right precentral, middle temporal, and middle occipital gyri (Thienel et al.2009) that is not supported by the literature (Witte & Marrocco, 1997). Nevertheless, scopolamine did increase brain activation in the right middle temporal gyrus, although this brain region was silent in the placebo condition, thus potentially indicating compensatory mechanisms in order to execute the task successfully. Given that the alerting network is proposed to be predominantly modulated by noradrenaline, our findings may also reflect a non-specific effect of scopolamine.
Cerebrovascular coupling is defined as the global effect of a drug on local blood flow independent of changes in neuronal metabolic activity. However, this mechanism is unlikely to explain the specificity of our other findings in relation to neural networks and associated attention processes as operationalized by specific contrasts (e.g. spatial vs. central cue, etc.). This approach also controls for generalized drug effects, such as sedation, but may still play a role in non-specific compensatory mechanisms in response to the drug challenge in areas otherwise silent with placebo.
In summary, our study is a first attempt to uncover muscarinic antagonist effects on functional and behavioural correlates of attention. Consistent with the notion of muscarinic modulation of dopaminergic neurotransmission in frontal attention networks (Langmead et al.2008; Levin et al.1990) our findings provide evidence for disruptive effects of scopolamine on executive control processes. Our findings are only partly consistent with the notion of muscarinic modulation of orienting. While muscarinic blockade did not affect the critical brain areas subserving orienting processes in the right hemisphere, our data on nicotinic antagonism illustrate a down-regulation of this network (Thienel et al.2009). The findings from both studies together suggest a larger degree of nicotinic modulation of orienting and a stronger muscarinic effect on executive control. Our neuropsychological data further demonstrated a selective scopolamine effect on verbal learning and memory while other cognitive domains, such as planning and working memory, were not affected by scopolamine.
We thank the radiographers Cordula Kemper and Dorothe Krug at the Jülich Research Centre for help with fMRI scanning, Helen Stain and Brian Kelly from the Centre for Rural & Remote Mental Health in Orange for encouraging data analysis and presentation. We gratefully acknowledge the participation of our subjects. This work was supported by the Deutsche Forschungsgemeinschaft (KFO 112).
Statement of Interest
- nuclear magnetic resonance
- memory, short-term
- mental processes
- neuropsychological tests
- reaction time
- single-blind method
- synaptic transmission
- verbal learning
- brain activity
- tnf-related apoptosis-inducing ligand
- verbal fluency
- anterior cingulate gyrus