Layman Summary
Whenever an individual is waiting for a traffic light, that individual has to judge, based on what the traffic light signals, what next action is appropriate. A red light signals the individual not to cross the street, and therefore, the appropriate choice of action is waiting, or inhibiting action. On the other hand, a green light signals the individual to cross the street, and therefore, the appropriate choice of action is crossing the street, or initiating action. We would like to understand how this decision making process is executed by the brain. From previous research, two particular brain regions, located in the front of the brain (or prefrontal cortex), were proposed as being involved in this decision making process: the infralimbic cortex and the prelimbic cortex. Therefore, in this research project, we wanted to investigate the role of these two brain regions in the decision making process.
In order to do this, we tried to mimic the traffic light analogy in a model using rats. The animals had to either initiate action or inhibit action, depending on what behavioural response was signalled, just like the traffic light. We call this the Go/No-Go paradigm. The animals were rewarded with sugar after a correct response in order to encourage them to keep responding. In order to investigate what the roles of the two brain regions are, we planned on temporarily lowering the activity of these two regions. By lowering their activity, we hoped that the ‘normal’ function of these regions would be impaired or stopped, which could emerge as a worse performance in our Go/No-Go paradigm. After the experiment, we indeed found that inhibition of the infralimbic and prelimbic cortex resulted in a change of behaviour in our paradigm. Both regions seem to affect the Go part of our paradigm, but not the No-Go part. However, we also found that the animals seemed to move less after injection of the drugs. This could also be an explanation for the decreased Go performance, as the animals need to move to complete a Go trial, but not a No-Go trial. Therefore, this study can not claim to have found the role of the two brain regions in the selection of action, and further investigation is needed.
Bart Kok
MSc Neuroscience & Cognition Clinical and Experimental Track
Research project
The role of the rat prelimbic (PL) and infralimbic (IL) in action selection and extinction
Daily Supervisors Examinator Assessor Aishwarya Parthasaranthy Frank J. Meye Ingo Willuhn
& Eugenia Poh
Abstract
The prefrontal cortex is thought to be highly involved in the process of selecting situation appropriate actions. In the rat PFC, a functional dichotomy has been proposed, where the infralimbic cortex (IL) inhibits actions, and the prelimbic cortex (PL) promotes action. However, this dichotomy has been challenged by multiple studies suggesting that both IL and PL encode context appropriate behaviour and inhibit situation inappropriate behaviour. Further research is needed to understand the role of IL and PL in guiding action selection. Therefore, in this study, we use a Go/No-Go rat model to
investigate the role of the IL and PL in action selection. The Go/No-Go task was designed so animals had to either initiate an or inhibit action, based on cues that signal situation appropriate action to earn a reward. We perturbated IL, PL and IL + PL using two Designer Receptor Exclusively Activated by Designer Drugs (DREADDs), in order to understand their role in the Go/No-Go behavioural paradigm.
This experimental set-up allowed us to investigate the role of the IL and PL in selecting situation appropriate actions and inhibiting situation inappropriate actions. Additionally, it allowed us to investigate the role of the IL in extinction learning, by extinction of the Go response. Following the hypothesized functional dichotomy, one would expect inhibition of PL to result in a worse Go performance, and inhibition of IL to result in a worse No-Go performance. Thereby, given the role of the IL in inhibition of inappropriate action, IL inhibition was expected to negatively affect extinction learning. We observed that Go performance was significantly affected, while sparing No-Go
performance, especially after inhibition of both IL and PL. Decreased Go performance seemed to be caused by an impaired ability of the animals to flexibly change behaviour while making a nose-poke response, suggesting that action initiation was impaired due to perturbation of these two regions.
However, several limitations in this study complicated interpretation of our results: The effects on our behavioural measurements of the two drugs used to inhibit the two regions differed, general
movement deficits due to inhibition of IL and PL seem to confound our observations and histological analysis is not completed yet. These limitations should be tackled in a follow-up study in order to find out what the effects of IL and PL inhibition are on behaviour in the Go/No-Go task.
Introduction
To achieve the most rewarding outcome, animals will select a certain action expected to result in the best outcome for (future) survival. For example, a food source close by is more rewarding compared to a similar food source further away, as the latter source will cost more energy to obtain. In the moment, the animal has to recollect and compare reward- (or threat)- related cues and predict which action will lead to the most rewarding outcome and initiate that action (Cisek & Kalaska, 2010;
Gallivan et al., 2018). Thus, selection of a situation appropriate behaviour requires both learning cue- reward associations and motor control to initiate or inhibit goal-directed actions. Therefore, the action selection mechanism needs high level processing, and is mediated, in part, by the prefrontal cortex (PFC) (Dalley et al., 2004). The PFC, part of the executive control system, is thought to be highly involved in the process of selecting situation appropriate actions (Norman & Shallice, 1986a). The executive control system monitors external sensory information, and internal physiological states, called interoception (Schnider, 2003). Therefore, this system seems well suited to choose an action based on the sensory information of the environment, and based on the bodily needs from
interoceptive signals, while inhibiting less salient alternative actions (Norman & Shallice, 1986b).
The primate PFC can be divided into several subregions: medial PFC, dorsolateral PFC, ventrolateral PFC and orbitofrontal cortex. No homologue for the primate PFC has been found in rodents, as the rodent frontal regions lack a granular cell layer (Brown & Bowman, 2002). Projections of the mediodorsal nucleus in the thalamus could be used as a criterion to define the PFC in rodents (Preuss, 1995). These efferent terminals can be found in the following regions: secondary motor cortex (M2), cingulate cortex, Prelimbic cortex (PL), infralimbic cortex (IL), and orbitofrontal cortex (Groenewegen et al., 1987; Krettek & Price, 1977; Kuramoto et al., 2017; Ray & Price, 1992).
In a reward-seeking context, a functional dichotomy has been hypothesized between PL and IL: PL contributes to promoting reward-related actions, whereas IL inhibits reward-related actions (Peters et al., 2009). This functional dichotomy has been especially observed during fear conditioning and drug- seeking paradigms. For example, inhibiting PL results in a reduced reinstatement of drugs of abuse, and decreases fear expression (Fuchs et al., 2005; McFarland & Kalivas, 2001; McLaughlin & See, 2003; Peters et al., 2008a). Conversely, IL inhibition resulted in increased cocaine seeking during extinction (Peters et al., 2008b), and impairs fear extinction learning (Peters et al., 2009; Sierra- Mercado et al., 2011). Several studies have reported the functional dichotomy during natural reward seeking, where PL inactivation decreases and IL inactivation increases natural reward seeking (Akinori Ishikawa et al., 2008; S. E.V. Rhodes & Killcross, 2007; Sarah E.V. Rhodes & Killcross, 2004; Sangha et al., 2014; Shipman et al., 2018). These studies support the idea of a Go/No-Go dichotomy between the PL and IL: PL promotes initiation of behaviour, while IL suppresses behaviour.
The functional dichotomy hypothesis is supported by the differential projections from PL and IL. Even though IL and PL both have afferent projections to the nucleus accumbens (NAc; Berendse et al., 1992; Brog et al., 1993; Vertes, 2004), PL neurons project primarily to the NAc core (NAcc), while IL neurons project to the NAc shell (NAcs) (Schmidt et al., 2005). A similar dichotomy, as discussed in the aforementioned paragraphs between IL and PL, is also suggested for these two NAc subregions:
NAcc is thought to be critical for action initiation, while NAcs contributes to response-suppression (Ambroggi et al., 2011; Floresco, 2015; A. Ishikawa et al., 2008; Peters et al., 2008b; Piantadosi et al., 2018). The anatomy therefore, suggest that the PL contributes to the initiation of action via its projection to the NAcc, while IL contributes to action suppression via its projections to the NAcs. The NAc receives dopaminergic input from the VTA. These dopaminergic neurons encode reward
prediction error (RPE), representing the difference between predicted and received reward size
(Schultz et al., 1997). The predicted value of a reward is shown to not be encoded by these
dopaminergic neurons themselves, but by efferent projections to dopamine neurons, including those from the PFC (Maes et al., 2020; Sharpe et al., 2020). Dopamine signalling is important for learning cue-reward associations, as novel cues evoke dopamine release in contrast to familiar cues (Morrens et al., 2020). Additionally, impairing dopamine release while presenting a novel cue, prevents learning about that cue. Moreover, representation of movement has been shown to be represented in VTA dopamine neurons; using a Go/No-Go task, an attenuation of dopamine release was observed to a No-Go cue compared to a Go cue, predicting the same reward (Syed et al., 2015). Thus, the reward prediction error encoded by VTA dopaminergic neurons projecting to the NAc are important in the decision-making process, as it is used to learn cue-reward associations and helps updating any changes in this association. Thereby, these neurons seem to be involved in movement initiation as well. Given the observed effects of perturbation of IL and PL projections to the NAc and the role of dopamine in the NAc, the NAc seems to be a suitable region to be involved in action selection.
Although, the functional dichotomy between IL and PL and their projections to the NAc is convincing to this point, multiple studies have found contradicting results, creating debate on its validity
(Moorman et al., 2015), especially when using a natural reward for reinforcement (Bari et al., 2011;
Bossert et al., 2012; Hamlin et al., 2007; A. Ishikawa et al., 2008; Mihindou et al., 2013; Weissenborn et al., 1997). For example, PL inactivation increases nonspecific (A. Ishikawa et al., 2008) and
premature lever pressing (Jonkman et al., 2009). Additionally, PL has a role in the inhibition of responses during a stop-signal reaction time task (Bari et al., 2011). Another study found that inhibition of PL or IL resulted in a decrease of performance in inhibitory trials, while leaving active trials intact. Additionally, recorded PL and IL activity during a sucrose-seeking paradigm revealed no functional dissociations between PL and IL neurons (Moorman & Aston-Jones, 2015a). More specifically, PL and IL neurons both increased in activity while making a lever-press (Reward-seeking or Go), and after extinction, both PL and IL were more active compared to baseline while withholding from making a lever-press (No-Go). These studies show that the previously proposed dichotomy is more complex and suggests that different neuronal ensembles in the IL and PL encode different cognitive elements to optimize behaviour.
In line with the suppression of behaviour it has been proposed that IL has a prominent role in extinction learning. Several studies suggested that IL does not play a role in the acquisition of extinction, but rather in the consolidation and retrieval of long-term extinction memory (Burgos- Robles et al., 2007; Laurent & Westbrook, 2008; Sotres-Bayon et al., 2009). On the other hand, PL only affects fear expression but does not affect extinction learning or learning (Choi et al., 2012;
Sierra-Mercado et al., 2011). Conversely, one study found that both IL and PL were strongly
modulated during extinction. Additionally, inactivation of either IL or PL resulted in deficits in acute behaviour and extinction learning (Moorman et al., 2015). Thus, the role of IL and PL in extinction learning remains to be clarified.
The valence of the context seems to determine which cognitive elements IL and PL encode, as type of responses, type of reinforcer and flexibility to change behaviour influence IL and PL activity
(Moorman & Aston-Jones, 2015b). Thus, both IL and PL seem to be encoding contextually appropriate behavioural information, irrespective of whether this involves initiation or suppression of behaviour.
This suggests that IL and PL are important in analysing context in order to maximize energy
management, by promoting activity when behaviours are rewarded and conservation of energy when not. This could explain the opposing results found in the field; within some contexts, a functional dichotomy between the two prefrontal brain regions might appear, whereas in other contexts, it might not.
In short, a lot of opposing observations have been made about the functional dichotomy of the IL and PL in action selection. Therefore, more research on the role of IL and PL in action initiation and inhibition is needed. Here, we will investigate the role of the IL and PL in action selection (initiation vs. inhibition) using a rat model Go/No-Go task. The Go/No-Go task was developed to motivate animals to select a situation appropriate action. The freely moving animals received one of either two distinct auditory cues every trial: one instructing the animal to initiate action (lever presses: Go) and one instructing the animal to withhold from action (stay in the nosepoke hole: No-Go). After
completing a trial successfully, the animals were rewarded with a sucrose pellet, reinforcing situation appropriate responses. This experimental setup creates a way to measure the ability of the animals to initiate situation appropriate actions, and inhibit situation inappropriate actions. . IL and PL were temporarily inhibited before test sessions, using chemogenetic Designer Receptors Exclusively Activated by Designer Drugs (DREADDs). Therefore, this experiment may give insight in the role of the IL and/or PL in selecting situation appropriate action and or initiation situation appropriate action.
Finally, we will investigate the role of the IL in extinction learning by changing the Go trial, where no lever press response is required to obtain a reward. Extinction acquisition can only be measured once per animal, and therefore, in this experiment we chose to inhibit IL, given its proposed role in
inhibition of action.
We expect that inhibition of IL and PL alone will lead to a decrease in task performance. Following the functional dichotomy hypothesis, it is possible that IL inhibition leads to decreased No-Go
performance and PL inhibition leads to decreased Go performance. However, as IL and PL encode contextual appropriate actions for action initiation and inhibition, another hypothesis may be that inhibition of either region could also lead to a decrease in both trial types. Simultaneous IL and PL inhibition is expected to decrease performance in both Go and No-Go trials, as the ability of the animals to select a context appropriate action is impaired. Finally, IL inhibition is expected to negatively affect extinction learning compared to control animals, due to its proposed role in the inhibition of inappropriate actions.
Materials & Methods
Animals
Seven female and six male inhouse bred Long-Evans rats were obtained, at an age of eight weeks.
They were socially housed divided by sexes, under a reversed 12-hour day/night cycle and controlled temperature. The animals were food restricted to 85% of their free-feeding body weight and their diet consisted of sucrose pellets, obtained during performing the task in the operant chamber, and chow three hours after training/experiments in their home cage. They had ab libitum access to water.
The body weight, health and behaviour of the animals was checked at least 5 days a week. The experiments performed on these animals, and the housing facility were in accordance with Dutch (Wet op de Dierproeven, 2014) and European legislation (Guideline 86/609/EEC; Directive 2010/63/EU) and approved by the Animal Experimentation Committee of the Royal Netherlands Academy of Arts and Sciences.
Surgical procedure
Animals were anaesthetized under isoflurane (flow rate: 0.3 ml/min O2 ), induced with 3% and maintained at 1.5%. The analgesia Metacam (0.2 mg meloxicam/kg) was injected subcutaneously.
During surgery, the animals were placed in a stereotactic apparatus and their body temperature was kept at 37˚C by a heating pad. The eyes were covered with ophthalmic ointment, to prevent them from drying up. The scalp of the animals was shaved, and their skin disinfected with 70% ethanol, whereafter the scalp and periosteum were removed. The exposed skull and surrounding tissue was treated with lidocaine and kept moist with saline during surgery. The target coordinates for PL were ML ± 0.7mm AP 3.2mm DV -3.4mm located relative to bregma and the target coordinates for IL were ML ± 0.7mm AP 2.9 DV -5mm (all coordinates are relative to bregma; The Rat Brain in Stereotaxic Coordinates - 7th Edition, n.d.), as shown in figure 1. A craniotomy was made at the target location after which a Hamilton syringe was slowly lowered to the correct DV coordinate. Intracranial infusion of the viruses was performed at the target coordinates with a rate of 100nl/min. Animals were individually housed in a temperature-controlled cabinet post-operatively until they recovered from anaesthesia. Animals were then transferred to the rodent housing room and housed individually for ten days after surgery, and welfare checks were performed daily. Thereafter, the animals were given two more week to recover before training was started and to get sufficient virus expression.
Chemogenetics
IL PL
Figure 1: Schematic overview of virus injection targeting. This figure is adapted from the Rat brain in Stereotaxic Coordinates – 7th edition. This figure is representative of a coronal section at 3.24 DV relative to bregma. PL targeting is illustrated by the bigger stars (ML ± 0.7mm AP 3.2mm DV -3.4mm) and IL by the smaller stars ML ± 0.7mm AP +2.9mm DV -5mm.
DREADDs are a tool that can be used to control cell activity; DREADDs are made up of genetically engineered G-protein coupled receptors (GPCRs) or kappa-opioid receptors coupled to an inhibitory GPCR that can be activated by engineered drugs (Armbruster et al., 2007). During surgery, nine animals received an intracranial microinjection of two viral vectors AAV-8/2-mCaMKIIa-
hM4D(Gi)_mCherry-WPREhGHp(A), referred to as human M4 muscarinic acetylcholine inhibitory receptor (hM4Di), and AAV5/2-mCaMKIIa- HA_hKORD_IRES_mCitrine-WPRE-hGHp(A), referred to as kappa-opioid receptor (KORD), in either the PL or IL. Injection of these viral vectors results in
expression of the DREADD receptors at the local infusion side. Viruses were randomised between brain regions. Four control animals were infused in both IL and PL with AAV8/2-mCaMKIIa-mCherry- WPRE-hGHp(A), referred to as mCherry, which did not lead to the expression of DREADDs.
The designer drug Salvinorin B (Sal-B) was used to antagonize KORD and clozapine-N-oxide (CNO) to antagonize hM4Di. Four drug injections were done during this experiment. Table 1 shows how each drug should affect the three groups of animals. During the test days, the animals received injections, ten minutes before the start of a session of vehicle (Dimethyl sulfoxide (DMSO, 100%)), Sal-B (0.1 mg/kg), CNO (0.1 mg/kg) and Sal-B + CNO (0.1 mg/kg + 0.1 mg/kg), respectively on different injection days. DSMO injections functioned as control injections for the drug injections. During Sal-B or CNO injection days, either IL or PL would be inhibited depending on the group. During the Sal-B + CNO injection day, both IL and PL were inhibited in experimental groups. mCherry control animals, do not express the receptors, and therefore are not affected by drug injections.
Apparatus
Table 1: Drug - receptor interactions. In this experiment 3 types of animals were used KORD in IL, hM4Di in PL and vice versa and mCherry animals. Four different injections were done during this experiment (DMSO, Sal-B, CNO and Sal-B + CNO). This table shows the effect the injections have on the three groups: not affected (x) or inhibited (-).
Behavioural testing and training were conducted in operant conditioning chambers (32 x 30 x 29 cm, Med Associated Inc., VT, USA), equipped with a metal grid floor, food magazine connected to an automated food-pellet dispenser, a house-light, a speaker connected to a multiple tone generator, two nose-poke (NP) holes, and two extendable levers. A schematic overview of the operant chambers can be found in figure 2. The operant conditioning chambers were placed inside a sound insulated cabinet, ventilated by a fan. A video camera was located above the operant chamber inside the cabinet. MEDPC software was used for acquisition of behavioural data with a sampling rate of 100Hz.
Figure 2: Schematic overview operant chamber: The animals were trained and tested in an operant conditioning chambers and the most important parts are depicted here. On one side of the box, two nose-poke holes and two levers were present.
During this experiment, only the right nose-poke hole and right lever were active. The left lever was always retracted. On the other side of the box, a food magazine, connected to a pellet dispenser was present. A house light was present in the box, and functioned to signal an erroneous response. The speaker functioned cued which response (Go/No-Go) was needed to obtain a reward.
Go/No-Go
The animals were trained in the operant conditioning chambers for a Go/No-Go task. This task was developed to measure execution and suppression behaviour within the same context, schematically shown in figure 3. During this task, the animal could earn a reward, in the form of a sucrose pellet, for either correctly initiating an action (Go) or refraining from action (No-Go). Before a trial starts, NP- light will turn on. The animals can initiate the trial by holding a NP for 0.5 seconds, after which the animal will be presented with either of two tones: Go-tone (3 kHz) or No-Go-tone (1 kHz). After presentation of the Go-tone, the animal is required to leave the NP-hole and press an extended lever twice in a timeframe of 2.75 seconds in order to receive a reward. After No-Go-tone presentation, the animal is required to hold the NP for 3 seconds in order to receive a reward. The NP-light and tone presentation will last until a correct response is given or the animal fails to respond correctly within the trial time. Go trials were failed, if the animal does not reach the two lever press threshold within the trial time. No-Go trials were failed if the animal fails to hold the NP hold for the required 3 seconds. Additionally, failure to hold the initial 0.5 second NP response is also considered incorrect responding. Incorrect responding resulted in a time-out of 5 seconds and the house light turned on, signalling to the animals they made a mistake. In between trials there was a random inter-trial- interval (ITI) of 5-15 seconds.
Training
Animals started training after they recovered from surgery (approximately 3 weeks). First, animals started with magazine training. During magazine training, animals received random pellets to create an association between the magazine and sucrose pellet delivery. No-Go training started after that, in which the animals had to make gradually longer nose poke responses, until they reached the 3s threshold. Thereafter, Go-trials were added to the training sessions. The ratio of Go trials to No-Go trials changed during training and was dependent on the performance of either trial type, for example, poor performance in Go trials warranted a higher proportion of Go trials in subsequent sessions, relative to No-Go trials. Go-training started with a fixed ratio 1 (FR1) paradigm, which was increased gradually, until the animals consistently pressed multiple times. Once animals established proper Go and No-Go responding, training was complete. After the behaviour stabilized across sessions, the animals continued to test sessions. Four test sessions were conducted: after DMSO injection, after Sal-B injection, after CNO injection and after Sal-B + CNO injection in order. 10 minutes after injection, the animals were placed in their respective box, the Go/No-Go script was started and behaviour was recorded.
Figure 3: Experimental set-up. In between trials, the animals were exposed to an inter trial interval (ITI), during which the NP light is off, lever is inactive, so no rewards could be earned. At the end of the ITI, the nose poke light turns on, signalling the animal a trial can be started. After making a NP- response of 0.5s, either a Go or a No-Go tone starts playing. Go tone, signalling a Go-response should be made. The animal has to press 3 times in order to earn a reward. No-Go tone, signalling a No-Go response. The animal has to hold the NP for 3s in order to earn a reward. Incorrect responses lead to a time-out of 5 seconds, during which the house-light turns on and no new trial could be started.
Data Analysis
MATLAB (version: 2021b) was used to analyse and visualize the behavioural data collected with MEDPC (MED associates). The behavioural measures we analysed are shown in figure 4. We measured the success rate (successful trials/failed trials) to assess the overall performance of the animals during the session. Latency to make the first lever press (LP) during Go-trials was measured as time from tone onset until the first lever press, and was a measure for action initiation. LP rate represents the number of lever presses per second during the trial after the first LP. We chose to measure LP rate after the first press, as this first press represents action initiation, and is already represented in LP latency. Therefore, LP rate functions as an indicator for general movement. Nose poke hold was calculated as the time from the initial nose poke after the nose poke light turned on, until the animal left the NP hole, or when a trial ended. Nose poke hold functions as a measure of action inhibition in No-Go trials, and potential impaired action initiation during in Go trials. Finally, we measured the latency to collect the reward after the animal completed a trial correctly. This was measured as the time of reward presentation until the first IR beam break present in the food magazine. Reward pickup latency gives information about task comprehension. Additionally, we can compare pickup latencies after Go trials, where the animals already left the NP hole and pickup latencies after No-Go trials, where the animal still has to leave the NP hole. This could give
information about the flexibility, as the animal has to switch from inhibition to picking up the reward after No-Go trials, whereas during Go trials, the animals are already in action. During the ITI we calculated the head entry rate (HE-rate), NP hold time, and LP rate, and all functioned as control for task comprehension. HE-rate corresponds to the number of food magazine beam breaks per second during the ITI. NP hold time is calculated as the fraction of ITI time spend in the NP hole. For all measurements, the mean was calculated for every animal over trials (or ITIs) for the four injection days, resulting in one datapoint per measurement per animal per session. These means were used for statistical analysis.
After the variables were calculated in MATLAB, statistical analyses were performed in SPSS (version:
SPSS statistics 28). For success rate of Go and No-Go trials, LP rate during successful and failed Go trials, NP hold during successful and failed Go trials and pickup latency after Go and No-Go trials, we performed a one-way MANOVA (main effects: Successful Go trials, failed Go trials/Go trials, No-Go trials). All other variables were analysed with one-way ANOVAs. ANOVAs were conducted to compare differences in behavioural measurements for control, IL inhibited, PL inhibited and IL+PL inhibited animals. This was repeated, but now with drug injected (DMSO x CNO x Sal-B x Sal-B + CNO) as an independent factor. We chose not to do a multifactorial set-up due to region inhibited and drug injected not being independent from each other. Sidak post hoc tests were conducted to compare individual groups, if the main effect was significant. mCherry animals were included in the ANOVAs using drug injected as independent factor. Preferably, the mCherry animals would be analysed separately, so they could be used as a control for potential confounding effects of the drugs. However, a group size of four animals is not sufficient for statistical analysis. A alpha of 0.05 was used as threshold of significance.
To control for potential changes in general movement due to the drug injection or inactivation of the target brain region, we used Bonsai (version: Bonsai 2.7.2) to analyse the distance travelled during a session. Bonsai output is smoothened and quantified using MATLAB. Average distance travelled, corrected for time (distance/min), was used to quantify movement during a session. A one-way ANOVA was used, with independent factor drug injected, to compare movement after different injections.
Histology
At the end of the experiment the animals were perfused in order to obtain their brains for histology.
The animals were anaesthetised with pentobarbital, whereafter cardiac perfusion with cold 4% PFA in 1x PBS fixative was performed. The brains were removed and kept in 30% sucrose in PBS over a few nights and snap frozen in isopentane cooled by dry ice. Later, the brain was sliced using a cryostat (LEICA CM3050 S) into 40 µm coronal sections. The slices were stained using Rabbit anti-GFP and Chicken anti-RFP as primary antibodies overnight at 4 degrees. Donkey anti-rabbit Alexa 488 and Donkey anti-Chicken Alexa594 were used as secondary antibody and was left in for 1 hour at room temperature. DAPI was added during the last washing steps. Blocking medium consisted of 5% BSA, 5% NDS and 0.2 TritonX and PBS, and was added for 1 hour at room temperature. Slices were mounted on glass slides using Mowiol mounting medium. The slices were imaged using a Zeiss Axio Scan.Z1, and monitored with ZEN 3.1.
Histology was not completed by the submission of this thesis due to insufficient time and difficulties in acquiring images. Therefore, all animals were assumed to have accurate targeting of viruses, and included for analysis.
Figure 4: Schematic overview of the Go/No-Go task. The variables we investigated are shown at what point in the trial they are measured.
Over the timespan of the ITI, we measured the HE rate (number of head entries into the magazine per second), NP hole time (ratio of ITI- time that the animal spend in the nose-poke hole), and LP rate (number of lever press responses per second). At the end of the ITI, the NP light turns on. From that timepoint we measured the latency for the animal to make a NP response (NP latency). After the animal made a NP response, either of the two tones starts playing after the animal holds for 0.5s. The latency for the animal to make the first lever press during Go trials will be measured from the moment the Go tone starts playing (LP latency). The lever press rate will be calculated by taking the number of lever presses after the first press, divided by the remaining Go time after the first press. We calculated the NP hold for both Go trials and No-Go trials, what represents the time form the initial NP response after the ITI until the moment the animal leaves the NP hole. After correct completion of either trial, the animal will receive a reward. The latency to pickup the reward is measured from the moment of trial completion until the first IR beam break in the food magazine.
Results
Animals received injections on four different days, with at least one day of baseline training in between. During the first injection day, the animals received a vehicle injection (DMSO). During subsequent injection days, animals received a Sal-B injection, CNO injection and Sal-B + CNO injection respectively. After injection, the animals were exposed to the Go/No-Go task. For analysis animals were sorted on the brain region(s) inhibited (no region inhibited (Control), IL inhibited, PL inhibited and IL + PL inhibited) and drug injected (DMSO, Sal-B, CNO and Sal-B + CNO). Then one-way (M)ANOVAs and Sidak post hoc analyses were conducted. The same analysis was done using drug injected as independent variable, to investigate a potential effect of the drug itself.
Performance
First, we investigated whether our intervention had an effect on overall performance in the Go/No-Go task, shown in figure 5. A one-way ANOVA revealed a significant effect of region inhibited on success rate (F(3,31) = 9.318, p < 0.001; Fig 5A). Sidak’s post hoc test revealed significantly higher success rate for control injections compared to IL + PL inhibition (p < 0.001), but not compared to IL inhibition (p = 0.984) and PL inhibition (p = 0.987). Additionally, success rate after IL + PL inhibition was significantly lower compared to IL inhibition (p = 0.002) and PL inhibition (p = 0.002). IL and PL were not
significantly different from each other (p = 1.000).
A one-way ANOVA revealed a significant effect of injection on the overall performance (F(3,47) = 12.094, p < 0.001; Fig 5B). Sidak’s post hoc test revealed a significantly lower success rate after Sal-B + CNO injection compared with DMSO injection (p < 0.001) and CNO injection (p < 0.001). Additionally, success rate after Sal-B injection was significantly lower compared to CNO injection (p = 0.004). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.052), DMSO vs. CNO (p = 0.927), Sal-B vs. Sal-B + CNO (p = 0.471)).
Thus, our interventions seemed to drop overall performance when both IL and PL were inhibited.
However, we found that our two injections Sal-B and CNO differed in effect on overall performance.
Figure 5: Performance Go/No-Go. Success rate for the whole session after region inhibited (A) and Drug injected (B). Success rate Go trials after region inhibited (C) and Drug injected (D). Success rate No-Go trials after region inhibited (E) and Drug injected (F). ANOVA’s revealed a significant effect of region inhibited and drug injected on the success rate of the overall task, and Go trials specifically.
Next, we split up the trials in Go and No-Go trials and we conducted a one-way MANOVA on the success rates for the two trials types. Using Pillai’s trace, there was a effect region inhibited on success rate (V = 0.555, F(3,31) = 4.095, p = 0.002). The success rate in Go trials was affected by region inhibited (F(3,31) = 9.350, p < 0.001; Fig 5C), whereas performance on No-Go trials was not affected by region inhibited (F(3,31) = 1.481, p = 0.238; Fig 5E). Pairwise comparisons revealed that the success rate for Go trials was significantly higher for control injections compared to IL + PL inhibited (p < 0.001), but not compared to IL inhibited (p = 0.510) nor PL inhibited (p = 0.885). Success rate for Go trials was significantly decreased for IL + PL inhibition compared with IL inhibition (p = 0.012) and PL inhibition (p = 0.002). No significant difference between IL inhibition and PL inhibition (p = 0.993) on Go success rate was found. No significant difference between the four groups was revealed in No-Go trials.
This was repeated using drug injected as independent variable. Pillai’s Trace revealed a significant effect of injection on performance on the two trials (V = 0.483, F(3,47) = 5.090, p < 0.001). Go success rate was significantly affected by injection (F(3,47) = 11.704, p < 0.001; Fig 5D), whereas No- Go performance was not significantly affected by injection (F(3,47) = 1.776, p = 0.164; Fig 5F). Sidak’s post hoc test revealed a significantly better Go performance after DMSO injection compared to Sal-B injection (p = 0.010) and Sal-B + CNO injection (p < 0.001), but not CNO injection (p = 1.000).
Additionally, the effect on Go performance was significantly higher for Sal-B injection compared to CNO injection (p = 0.004). Sal-B + CNO injection resulted in a significantly higher effect on Go performance compared to CNO injection (p < 0.001), but not Sal-B injection (p= 0.089). Sidak’s post hoc analysis did not reveal any differences in No-Go performance between different injections.
In short, inhibition of both IL + PL resulted in a decrease in Go performance, but had no effect on No- Go performance. Inhibition of IL or PL alone did not affect Go performance significantly. However, Sal- B injections seem to drive the effects we have seen, as Sal-B injection alone significantly affected Go performance compared to both control injections and CNO injections. CNO injections alone did not affect Go performance significantly compared to DMSO injections, suggesting that CNO does not affect behaviour. These results suggest that both IL and PL are important for Go performance.
NP latency
We investigated the latency to initial NP response, shown in figure 6. A one-way ANOVA revealed a significant effect of region inhibited on NP latency (F(3,31) = 14.331, p < 0.001; Fig 6A). Sidak’s post hoc analysis revealed a lower NP latency for control injections compared to IL + PL inhibition (p <
0.001), but not compared to IL inhibition (p = 0.952) and PL inhibition (p = 0.447). Additionally, NP latency after IL + PL inhibition was significantly higher compared to IL inhibition (p < 0.001) and PL inhibition (p < 0.001). No significant difference was found between IL inhibition and PL inhibition (p = 0.950).
Similarly, injection significantly affected the NP latency (F(3,47) = 14.367, p < 0.001; Fig 6B). Sidak’s post hoc test revealed a significantly lower effect on NP latency of DMSO injection compared to Sal-B injection (p = 0.028) and Sal-B + CNO injection (p < 0.001), but not compared to CNO injection (p = 1.000). Sal-B injection resulted in a greater increase in NP latency compared to CNO injection (p = 0.018). Sal-B + CNO injection also resulted in a bigger increase in NP latency compared to CNO injection (p < 0.001), but not compared to Sal-B injection (p = 0.094).
Figure 6: NP latency sorted on region inhibited (A) and sorted on drug injected (B). ANOVAs revealed an effect of region inhibited and drug injected.
Inhibition of both IL and PL significantly increased latency to make the NP response to initiate a new trial. Again, we found Sal-B injections to have a larger effect on behaviour compared to CNO
injections. As the IL inhibited and PL inhibited groups contain animals with only CNO injections, this tends to drive the mean down, resulting in a significant effect of IL + PL inhibited compared to IL inhibited and PL inhibited. From these results we can conclude that IL and PL seem to be involved in the latency to make the initial NP response.
Go trials
The perturbation affected performance of Go trials, but leaves No-Go performance unaffected, suggesting that action initiation could be affected. To further investigate whether our perturbations affected action initiation, a one-way ANOVA was conducted on the LP latency to make the first lever press in Go trials, shown in figure 7. This revealed no significant effect of region inhibited (F(3,31) = 2.518, p = 0.076; Fig 7A). Sidak’s post hoc test did not reveal any significant differences in LP latency between the four groups. A one-way ANOVA revealed a significant effect of injection on LP latency (F(3,47) = 3.557, p = 0.021; Fig 7B). Sidak’s post hoc test revealed a significantly higher LP latency after Sal-B + CNO injection compared to DMSO injection (p = 0.042). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.263), DMSO vs. CNO (p = 0.999), Sal-B vs. CNO (p = 0.497), Sal-B vs.
Sal-B + CNO (p = 0.965), CNO vs. Sal-B + CNO (p = 0.107)).
No effect on the latency to make the first lever press was found after inhibition, suggesting that action initiation was not affected and not the reason for a worse Go performance. Surprisingly, a significant effect of Sal-B + CNO injection was found compared to DMSO injection. The control group and the IL + PL groups and the DMSO and Sal-B + CNO groups are the same apart from the added mCherry animals in the drug injected ANOVA. Therefore, this difference is likely to be caused by a bigger power for the drug injection comparison. In short, our interventions seem to affect LP latency only slightly.
Figure 7: LP latency sorted on region inhibited (A) and drug injected (B). ANOVA’s revealed a significant effect of region inhibited and drug injected on the latency for the first press.
Next, we conducted a one-way ANOVA on LP rate after the first lever press during Go trials, shown in figure 8. This revealed a significant effect of region inhibited (F(3,31) = 3.903, p = 0.018; Fig 8A).
Sidak’s post hoc test revealed a significant decrease of LP rate after IL + PL inhibition, compared with control injections (p = 0.021), but not compared to IL inhibition (p = 0.086) and PL inhibition (p = 0.115). All other comparisons were not significant (Control vs. IL (p = 0.993), Control vs. PL (p = 0.981), IL vs. PL (p = 1.000)). A one-way ANOVA revealed a significant effect of drug injected on the LP rate after the first lever press (F(3,47) = 3.950, p = 0.013; Fig 8B). Sidak’s post hoc test revealed a significantly lower LP rate after Sal-B + CNO injection compared to CNO injection (p = 0.044). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.263), DMSO vs. CNO (p = 0.999), DMSO vs.
Sal-B + CNO (p = 0.116), Sal-B vs. CNO (p = 0.114), Sal-B vs. Sal-B + CNO (p = 0.999)).
LP rate seems to contribute to the worse Go performance after IL + PL inhibition/Sal-B + CNO injection compared to control injections, but not after Sal-B injection only. This decrease in LP rate could be an indicator for impaired general movement.
Figure 8: LP rate sorted on region inhibited (A, C and E) and drug injected (B, D and F). ANOVA’s revealed a significant effect of region inhibited and drug injected on the LP rate after the first press during over all Go trials. The effect of region inhibited disappeared after splitting the Go trials in successful and failed Go trials. But ANOVA still revealed a significant effect of drug injected over failed Go trials, but not over successful Go trials.
Next, we split the Go trials in successful and failed Go trials. However, using Pillai’s trace, a one-way MANOVA revealed no significant effect of Go trial outcome on region inhibited (V = 0.133, F(3,31) = 0.737, p = 0.622). No significant effect of region inhibited on LP rate was revealed on successful Go trials (F(3,31 = 0.118, p = 0.949; Fig 8C) and failed Go trials (F(3,31) = 1.449, p = 0.248; Fig 8E). No effect of region inhibited after splitting the Go trials in failed and successful trials seems contradictory to the effect found in all Go trials. However, this can be explained by the increased number of failed trials after IL + PL inhibition, dragging the overall LP rate down, as LP rate is lower in failed Go trials.
A one-way MANOVA revealed a significant effect of injection on the LP rate during successful and failed Go trials (V = 0.309, F(3,47) = 2.866, p = 0.013). Injection did not affect LP rate during successful Go trials (F(3,47) = 0.612, p = 0.611; Fig 8D), but did affect LP rate during failed Go trials (F(3,47) = 6.001, p = 0.002; Fig 8E). LP rate during failed Go trials was significantly higher after CNO injection compared to Sal-B injection (p = 0.006) and Sal-B + CNO injection (p = 0.006). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.197), DMSO vs. CNO (p = 0.714), DMSO vs. Sal-B + CNO (p
= 0.175), Sal-B vs. Sal-B + CNO (p = 1.000)).
In short, splitting the Go trials in successful and failed Go trials revealed, once again, that the effect of injection is leading in the observed effects, as an effect of drug injected was found, but not an effect of region inhibited. Although we cannot tell the relative effect size of IL and PL on LP rate, these results suggest that a decrease in LP rate contributes to the decrease in Go performance in the affected groups.
Next we investigated the NP hold from the initial NP response during Go trials, shown In figure 9. A one-way ANOVA revealed a significant effect of region inhibited on the NP hold during Go trials F(3,31) = 8.829, p < 0.001; Fig 9A). Sidak’s post hoc test revealed a significantly higher NP hold time after IL + PL inhibition compared to Control injections (p < 0.001), IL inhibition (p = 0.009) and PL inhibition (p = 0.005). No other comparisons were significant (Control vs. IL (p = 0.681), Control vs. PL (p = 0.838), IL vs. PL (p = 1.000)). A one-way ANOVA revealed a significant effect of injection on the NP hold (F(3,47) = 14.702, p < 0.001; Fig 9B). Sidak’s post hoc test revealed a significantly lower NP hold after DMSO injection compared to Sal-B injection (p = 0.005) and Sal-B + CNO injection (p < 0.001), but not compared to CNO injection (p = 1.000). NP hold was significantly higher after Sal-B injection compared to CNO injection (p = 0.003), but not compared to Sal-B + CNO injection (p = 0.428). NP hold was significantly higher after Sal-B + CNO injection compared to CNO injection (p < 0.001).
In short, an increased NP hold contributes to the poorer Go performance. Again, Sal-B injections had a bigger effect size compared to CNO injections. The increase in NP hold suggests that animals had trouble leaving the NP hole during a NP response. Additionally, this explains the decrease in LP rate and it also explains the small effect we observed on the LP latency, as animals were to slow to leave the NP hole, that they never pressed the lever.
Thereafter, we split the NP hold in failed and successful Go trials and performed a one-way
MANOVA.. Following Pillai’s trace, region inhibited (V = 0.364, F(3,31) = 2.297, p = 0.046) does affect NP hold. Region inhibited did affect NP hold time during successful Go trials (F(3,31) = 3.452, p = 0.028; Fig 9C) and failed Go trials (F(3,31) = 3.062, p = 0.043; Fig 9E). Pairwise comparisons with Sidak’s post hoc analysis revealed a significant increase in NP hold during successful Go trials for IL + PL inhibited compared to Control injections (p = 0.025). All other comparisons were not significant (Control vs. IL (p = 0.983), Control vs. PL (p = 0.814), IL vs. PL (p = 0.997), IL vs. IL + PL (p = 0.125), PL vs. IL + PL (p = 0.321)). The analysis revealed no significant differences between the four groups
during failed Go trials.
Following Pillai’s trace, injection has a significant effect on NP hold (V = 0.353, F(3,47) = 3.359, p = 0.005). NP hold was significantly affected by injection during both successful Go trials (F(3,47) = 4.898, p = 0.005; Fig 9D) and failed Go trials (F(3,47) = 6.146, p = 0.001; Fig 9F). Sidak’s post hoc test revealed a significant increase in NP hold time during successful Go trials after Sal-B + CNO injection compared to DMSO injection (p = 0.006) and CNO injection (p = 0.020). No other comparisons were significant (DMSO vs. Sal-B (p = 0.459), DMSO vs. CNO (p = 0.999), Sal-B vs. CNO (p = 0.750), Sal-B vs.
Sal-B + CNO (p =0.365)). NP hold time during failed Go trials was significantly higher after Sal-B + CNO injection compared to CNO injection (p = 0.006) and DMSO injection (p = 0.009). All other
comparisons were not significant (DMSO vs. Sal-B (p = 0.118), DMSO vs. CNO (p = 1.000), Sal-B vs.
CNO (p = 0.085), Sal-B vs. Sal-B + CNO (p = 0.886)).
Splitting the Go trials confirmed again that the effect size of Sal-B is bigger compared to the effect size of CNO. Additionally, it suggests that trouble to leave the NP hole during a Go trial is the main reason for the observed decrease in Go performance.
In short, the worse performance we observed during Go trials seems to be mostly affected by Sal-B injections, or Sal-B + CNO injections. As the effect sizes of Sal-B and Sal-B + CNO injections are often similar, and differ significantly from DMSO and CNO injections, we can conclude that the observed results are generally caused by Sal-B. The Sal-B injections caused a decrease in LP rate and a increase in NP hold. We observed a significant increase of LP latency after Sal-B + CNO injections compared to DMSO injections, although this effect seems to be caused by the increase of group size after addition of the mCherry animals. Surprising are the differences between the clear trends we see in NP hold, where these trends are not so clear in the LP latency. Potentially, animals were to slow to press before the end of the Go trial, leading to no data point entry for the LP latency. This lost datapoint is represented in the NP hold, in the form of a longer NP hold. Therefore, Sal-B injection probably resulted in a longer NP hold, resulting in a lower LP rate as a result as well.
Figure 9: NP hold during Go trials sorted on region inhibited (A, C and E) and drug injected (B,D and F). ANOVA’s revealed a significant effect or region inhibited and drug injected on the NP hold duration over all Go trials, successful Go trials and failed Go trials.
No-Go trials
To confirm No-Go trials were not affected by our interventions, we investigated the NP-hold during No-Go trials, shown in figure 10. A one-way ANOVA was performed on NP hold time and revealed no significant effect of region inhibited on the NP hold time (F(3,31) = 0.333, p = 0.802; Fig 10A).
A one-way ANOVA revealed no significant effect of injection on NP hold during No-Go trials (F(3,47) = 0.263, p = 0.852; Fig 10B).
Figure 10: NP hold during No-Go trials sorted on region inhibited (A and C) and drug injected (B and D). ANOVA’s revealed no significant effect of region inhibited or drug injected on the NP hold duration during all No-Go trials and failed No-Go trials.
Then we split up the No-Go trials, and analysed the failed No-Go trials. A one-way ANOVA revealed a significant effect of region inhibited on the NP hold time (F(3,31) = 3.317, p = 0.032; Fig 10C). Sidak’s
post hoc test revealed no significant differences in NP hold time for all comparisons (Control vs. IL (p = 0.818), Control vs. PL (p = 1.000), Control vs. IL + PL (p = 0.057), IL vs. PL (p = 0.848), IL vs. IL + PL (p = 0.557), PL vs. IL + PL (p = 0.065)). Significant F tests in ANOVAs only guaranties that one of the possible contrasts is statistically significant, but not that the contrast is significant after correction for multiple comparisons. This explains why the ANOVA turned out significant, but post hoc pairwise comparisons did not. Additionally, the effect seemed to be caused by one outlier in the PL inhibited group. A one-way ANOVA revealed no significant effect of injection on NP hold during failed No-Go trials (F(3,47) = 0.426, p = 0.735; Fig 10D). In short, these results confirm that No-Go trials were not affected by our perturbations.
Reward pickup latency
We then investigated the latency to pickup the reward after correct completion of a trial, shown in figure 11. A one-way ANOVA on the reward pickup latency after Go and No-Go trials revealed a significant effect of region inhibited (F(3,31) = 7.252, p < 0.001; Fig 11A). Sidak’s post hoc analysis revealed a significant increase in latency to pickup the reward after IL + PL inhibition compared with IL inhibition (p = 0.018) and PL inhibition (p = 0.012) and control injections (p < 0.001). No other
comparisons were significant (Control vs. IL (p = 0.804), Control vs. PL (p = 0.884), IL vs. PL (p = 1.000).
A one way ANOVA revealed a significant effect of injection on the latency to pickup the reward after successful trials (F(3,47) = 8.124, p < 0.001; Fig 11B). Sidak’s post hoc test revealed a significantly higher pickup latency after Sal-B + CNO injection compared to DMSO injection (p < 0.001) and CNO injection (p = 0.001). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.182), DMSO vs. CNO (p = 0.999), Sal-B vs. CNO (p = 0.391) and Sal-B vs. Sal-B + CNO (p = 0173).
These results suggest that inhibition of IL and PL results in an increased latency to pickup the reward.
This could be an indication of decreased task comprehension or action initiation.
Next we split up the pickup latency in latency after Go trials and latency after No-Go trials. A one-way MANOVA revealed a significant effect of region inhibited (V = 0.363, F(3,31) = 2.295, p = 0.046).
Region inhibited did not affect pickup latency after Go trials (F(3,31) = 2.111, p = 0.119; Fig 11C), but did after No-Go trials (F(3,3a) = 5.438, p = 0.004; Fig 11E). Regarding pickup latency after No-Go trials, Sidak’s post hoc test revealed a significantly higher pickup latency after IL + PL inhibition compared to PL inhibition (p = 0.031) and control injections (p = 0.003). All other comparisons were not significant (Control vs. IL (p = 0.846), Control vs. PL (p = 0.951), IL vs. PL (p = 1.000), IL vs. IL + PL (p = 0.058)).
A one-way MANOVA revealed a significant effect of injection on pickup latency (V = 0.352, F(3,47) = 3.346, p = 0.005). Injection affected the pickup latency after No-Go trials significantly (F(3,47) = 8.222, p < 0.001; Fig 11E), but did not significantly affect the pickup latency after Go trials (F(3,47) = 2.339, p
= 0.085; Fig 11D). This did reveal a significantly higher pickup latency after No-Go trials after Sal-B + CNO injection compared to DMSO injection (p < 0.001) and CNO injection (p = 0.002). No other comparisons were significant (DMSO vs. Sal-B (p = 0.108), DMSO vs. CNO (p = 0.987), Sal-B vs. CNO (p
= 0.404), Sal-B vs. Sal-B + CNO (p = 0.198)). If task comprehension is decreased, we would observe an increase in latency after both trials. However, we only found a increased latency after No-Go trials, but not after Go trials. This is in line with behaviour observed during Go trials, where animals seemed to have trouble leaving the NP hole. This could mean that action initiation was impaired due to inhibition of IL and PL or impaired general movement.
Figure 11: Pickup latency sorted on region inhibited (A,C and E) and drug injected (B,D and F). ANOVA’s revealed a significant effect of region inhibited and drug injected on the overall pickup latency and pickup latency after No-Go trials, but not after Go trials.
Behaviour during the ITI
We analysed behaviour during the ITI, to control whether the animals understood the contingencies of the Go/No-Go task and to control for movement, as shown in figure 12. We investigated the head entry rate, LP rate and time spend in the NP hole during ITI’s. Head entry rate should have a baseline level, as animals pickup their reward during the ITI, but should not be affected by our interventions.
LP rate should be low, as the lever is inactive during the ITI. However, animals keep pressing at a individually specific amount, and therefore, LP can also be used as an indicator for general
movement. Fraction of time spend in the NP hole should be high with high task comprehension, as staying in the NP hole during ITI is the quickest way to start a new trial, and is not punished. One-way ANOVA’s revealed no significant effect of region inhibited on HE rate (F(3,31) = 0.265, p = 0.850; Fig 12A) and time spend in the NP hole (F(3,31) = 0.037, p = 0.990; Fig 12C). A one-way ANOVA revealed no significant effect of injection on HE rate (F(3,47) = 0.193, p = 0.901; Fig 12B) and NP hold time (F(3,47) = 0.117, p = 0.950; Fig 12D).
No significant effect of region inhibited was found on the LP rate during ITI (F(3,31) = 2.078, p = 0.123;
Fig 12E). A one-way ANOVA revealed a significant effect of injection on the LP rate during ITI (F(3,47)
= 11.765, p < 0.001; Fig 12F). Sidak’s post hoc test revealed a significantly higher LP rate after CNO injection compared to Sal-B injection (p < 0.001) and Sal-B + CNO injection (p < 0.001). All other comparisons were not significant (DMSO vs. Sal-B (p = 0.239), DMSO vs. CNO (p = 0.053), DMSO vs.
Sal-B + CNO (p = 0.065), Sal-B vs. Sal-B + CNO (p = 0.993)).
In short, our interventions did not alter the amount of magazine checks the animal did during the ITI.
Additionally, the animal did not spend more or less time in the NP hole. From that, it seems like that our interventions did not affect task comprehension. LP rate was not affected by region inhibited, but was affected by drug injected. This, additionally to the different effect sizes we saw between drugs, points towards a possible movement deficit, especially after Sal-B injection.
Figure 12: Quantified behaviour during ITI: head entry rate (head entries per second) sorted on region inhibited (A) and sorted on drug injected (B). The ratio of time the animal spend in the NP hole during the ITI sorted on region inhibited (C) and sorted on Drug injected (D). The LP rate (LP responses per second) during the ITI sorted on region inhibited (E) and sorted on drug injected (F). One-way ANOVA’s revealed a significant effect of drug injected on the LP rate during ITI. All others were not significant.
Movement during the Go/No-Go sessions To control for potential impaired general
movement we analysed the video recordings of all four test sessions. Movement was corrected for session length, resulting in a measure of distance travelled per minute (cm/min). We conducted a repeated measures ANOVA with injection (DMSO x Sal-B x CNO x Sal-B + CNO) and virus (DREADD x mCherry) as variables, shown in figure 13. Pillai’s trace revealed a significant effect of injection on distance travelled per minute (V = 0.844, F(3,31) = 16.253, p < 0.001), but no interaction between injection and virus (V = 0.310, F(3,31) = 1.347, p <
0.320). Sidak’s post hoc analysis revealed a significantly higher distance travelled per minute after DMSO injection compared to Sal-B injection (p = 0.008) and Sal-B + CNO injection (p = 0.005).
Additionally, animals travelled more after CNO injection compared to Sal-B injection (p < 0.001) and Sal-B + CNO injection (p = 0.001).
Thus, we observed a decrease in movement after Sal-B and Sal-B + CNO injection, suggesting an effect of Sal-B
on movement, independent on the brain region inhibited. These observations suggest that the behavioural effects observed are likely caused by general movement deficits. Especially due to the statistically significant decrease in movement after Sal-B injection compared to movement after CNO injection, and behavioural effects were specifically expressed after Sal-B injection, The effects on general movement are specific of Sal-B, supported by the recovery of general movement over time during the trial, shown in supplementary figure 1.
Extinction
Finally, we inhibited IL and exposed the animals to a Go-extinction session. During this session, to earn a reward during a Go-trial, the animals had to avoid pressing the lever. When no lever presses were made, the animal would earn a reward. No-Go trials were not changed. This way, we examine the role of the IL in the acquisition of extinction.
We found no effect of IL inhibition on acquisition during the Go extinction trials compared to control animals (p = 0.8850; Fig 14A). No-Go trials were also unaffected by IL perturbation (p = 0.6215; Fig 14B). Additionally, we found no significant differences in LP latency (p = 0.3929; Fig 14C), LP rate (p = 0.2622; Fig 14D) and NP hold (p = 0.7685; Fig 14E, p= 0.8239; Fig 13F) between IL Finally, we found no differences between the test and control animals in the speed at which the animals adapted to the newly introduced extinction contingency (K-S test: p = 0.999), shown in figure 15. Thus, IL inhibition did not affect the acquisition of the extinction response.
Figure 13: Distance per minute after the four injections. One- way ANOVA revealed a significant effect of drug injected on the distance travelled per minute. Post hoc test revealed that movement is significantly different after Sal-B and Sal- B + CNO injection compared to movement after DMSO and CNO injection.
Figure 14: Performance during extinction session comparing the mCherry control animals (N = 4) and IL inhibited animals (N
= 9).. Performance on the extinction Go trials (A) and No-Go trials (B) is shown. Additionally, the performance during the extinction Go trial is represented in LP latency (C), LP rate (D), NP hold during correct Go trials (E) and NP hold during all Go trials (F).
Figure 15: The mean time spend in minutes by the animals to reach thresholds of 5 percentiles of the maximum number of completed Go trials. One datapoint, therefore, contains the mean of time of either DREADD.
Discussion
In this study, we investigated the role of the IL and PL in action selection, using a rat-model where animals had to either initiate action or inhibit action in order to earn a reward. A previously proposed dichotomy, where IL is involved in action inhibition and PL is involved in action initiation is becoming more and more controversial due to a growing number of studies finding opposing results. Our results suggested that inhibition of IL and PL may have resulted in as impaired ability to initiate action. However, there are several limitations of this study that should be taken into account.
Inhibition of IL and PL on Go/No-go performance
In this experiment, we found a decrease in overall performance in the Go/No-Go task due to IL and PL inhibition. Specifically, success rate was decreased in Go trials, whereas No-Go trials were unaffected.
This effect was specific after inhibition of both IL and PL . Inhibition of only one region was not sufficient to affect Go trials, but inhibiting both seemed to result in an additive effect. Thus, at first glance, our results show that IL and PL together contribute to initiating action, but not to inhibiting action, in order to obtain a reward.
A decrease in Go trial performance after inhibition of IL + PL, suggests that these regions have a role in action initiation, but not in action inhibition, as No-Go performance was unaffected. This is
supported by the observed increase in NP latency and increase in NP hold during Go trials. LP latency is a more suitable measure for action initiation compared to NP latency, due to the fact that the animals are all located in the NP hole and are not active (0.5s NP hold response). Therefore for the LP latency measurement, the animals have to make the same response from the same location and are in the same inactive state. Whereas animals during the ITI, when the NP light turns on, could be anywhere in the cage and facing in different directions, confounding the NP latency. We observed a small effect of our interventions on the LP latency. Specifically, after sorting the data per injection, we found an effect of Sal-B + CNO injection. Surprisingly, the trends we saw for the LP latency were more clearly expressed in the NP hold in Go trials. These two variables seem to be strongly related, as a longer NP hold will automatically lead to a longer latency to make the first press. However, if the animals holds the NP for too long, and never makes a lever press, no datapoint for LP latency will be registered. Therefore, the ‘slowest’ responses will not be captured by the LP latency variable, but is represented in the NP hold. It seems like the animals had trouble leaving the NP hole after inhibition of IL and PL, which may indicate that IL and PL are involved in action initiation. The trouble the animals had with leaving the NP hole after inhibition of both IL and PL is also represented in the reward pickup latency. Pickup latency after Go trials was not affected by this perturbation, where the animals had already left the NP hole. Pickup latency after No-Go trials was significantly increased after IL and PL inhibition, where the animals had to leave the NP hole in order to collect their reward.
Thus, the animals performed worse on Go trials due to holding the NP response for too long. This increase in NP hold also explains why No-Go trials are not affected, as this is the exact response needed to correctly complete a No-Go trial. These observations suggest an impairment in action initiation after inhibition of IL and PL together.
Potentially, inhibition of both IL and PL resulted in an inability of the animals to flexibly instigate instrumental action in a conflicting situation. It has been proposed that, PL may be involved after a task is well learned, to detect changes in or update rule/context representations (Moorman et al., 2015). The different roles of IL and PL in reward seeking and avoidance tasks could be explained by opposing prepotent responses, triggered in threatening and rewarding contexts. For example, aversive contexts can induce behavioural suppression, but animals have to learn to override this tendency in order to correctly perform active avoidance responses (Cain & Kline, 2019). Conversely,
appetitive contexts can promote approach behaviour, that the animals have to learn to override in order to correctly perform inhibitory appetitive responses. Therefore, these responses require greater cognitive control to override prepotent behaviours. In our paradigm, every trial, Go or No-Go is initiated by a NP response. Potentially, the NP response in our Go/No-Go design represents a prepotent response that has to be overwritten, which requires great cognitive control, in order to change behaviour (Cain & Kline, 2019). Inhibition of IL and/or PL might have impaired this ability of the animal to override this prepotent response. This would explain why No-Go trials were unaffected in this experiment, because a correct No-Go response requires no overriding of the prepotent NP response, rather a continuation of this response. Additionally, this would explain why pickup latency after No-Go trials is affected by inhibition of IL and PL, but pickup latency after Go trials is not. This theory is complemented by studies proposing a role of the PL in integrating contextual information to mediate response and cue conflicts (Haddon & Killcross, 2006; Marquis et al., 2007). Additionally, the human homologous of the PL, anterior cingulate, mediates cognitive control over instrumental responses and inhibits prepotent responses during decision making (Kolling et al., 2016; Shenhav et al., 2016). Overriding prepotent responses require greater cognitive control compared to continuing this response, and therefore, effects we found exclusively during Go trials, could represent a role of the IL and PL in regulating the overriding of prepotent responses.
Thus, solely looking at the behavioural measurements, these results suggest that inhibiting IL or PL does not affect action initiation. However, inhibiting both results in an impairment of action initiation, suggesting that IL and PL work together in order to facilitate action initiation. However, we discovered that the effects of Sal-B and CNO on our behavioural measurements were different, which could change the interpretation of these results.
Effect of drug injected on Go/No-Go performance
When investigating the effects of injections on our behavioural measurements, we found that Sal-B seems to contribute to the observed effects the most, as most measurements were affected by Sal-B injection and Sal-B + CNO injection. Some measurements however, only Sal-B + CNO injection
resulted in a significant effect, indicating that CNO injections did affect behaviour. Although the effect sizes between CNO and Sal-B differ, the effect sizes seem to be constant between animals. Thereby, because of the counterbalanced DREADD receptors, should not affect the differences between IL inhibited and PL inhibited, and any differences there could be informative. However, this is more difficult when comparing IL/PL inhibited with IL + PL inhibited, due to the fact that the means of IL inhibited and PL inhibited groups will be dragged down because they consist of a mix of CNO injected animals and Sal-B injected animals. This makes it hard to interpret the relative contribution of IL and PL inhibition on observed behaviour, and if inhibition of both regions results in bigger effects on our behavioural measurements compared to only one of the regions. Future analysis should divide the IL inhibited and PL inhibited groups should be divided on what receptor is expressed in the
corresponding region. This makes it possible to make fair comparisons between IL inhibition and PL inhibition. We chose not to make this split in this analysis, as group sizes would be to small for statistical analysis.
We should also keep in mind that the pharmacokinetic profiles of Sal-B and CNO are different and differentially affects the temporal dynamics of neuronal activity (Vardy et al., 2015): Sal-B mediates rapid short-lasting effects and CNO mediated delayed long-lasting effects. Therefore, Sal-B-activated DREADDs may be more informative of how transient inactivation of neurons affects behaviour, whereas CNO-activated DREADDs may be informative of how tonic inactivation of neurons influences behaviour. Differences in behaviour between acute and tonic activation of neurons has been shown (Urban et al., 2016), and could be the cause of the differences in effect size between Sal-B and CNO.
