Rapid, automated, and experimenter-free touchscreen testing reveals reciprocal interactions between cognitive flexibility and activity-based anorexia in female rats
- Department of Physiology, Monash University, Australia
- Monash Biomedicine Discovery Institute, Australia
Figures
Figure 1 with 2 supplements
Timeline of experiments.
Rats were acclimated to the reversed light cycle for 7 days before each experiment commenced. In experiment 1, rats underwent touchscreen cognitive testing before undergoing the activity-based anorexia (ABA) paradigm. In experiment 2, rats were exposed to the ABA paradigm prior to cognitive testing in the PhenoSys apparatus. The PhenoSys cognitive testing paradigm consisted of 7–10 days of pre-training, 1–6 days of pairwise discrimination, and 1–8 days of reversal learning. The ABA paradigm consisted of 3 days of habituation with ad libitum food access, 7 days of baseline testing with ad libitum access to food and a maximum of 10 days of ABA conditions with time-limited food access (90 min/day) and unrestricted running wheel access, followed by a minimum of 3 days of body weight recovery with reinstatement of ad libitum access to food.
Figure 1—figure supplement 1
The automated and experimenter-free home cage and touchscreen testing system.
(A) Animals are implanted with unique radiofrequency identification (RFID) transponders. (B) Overview of the entire PhenoSys system, in which a home cage is connected to a touchscreen testing chamber via a series of tunnels and a sorting device positioned over a weight scale. (C) The sorter cage has two RFID readers positioned underneath, and metal guillotine gates control the passage of animals between the home-cage and the touchscreen (D) The touchscreen chamber has an externally mounted pellet dispenser, from which rewards are delivered into a magazine on the opposite side of the chamber from the touchscreen. Images adapted from https://www.phenosys.com/products/touchscreen-chamber/.
Figure 1—figure supplement 2
Schematic overview of touchscreen pre-training and serial reversal learning protocol.
Image-based on Mar et al., 2013; Oomen et al., 2013 and adapted for our protocol. See Supplementary file 1 for specific parameters in each stage for PhenoSys touchscreen protocol. (E) The fan/pinwheel (top) and marble array (bottom) images used for pairwise discriminations and all stages of reversal learning in all experiments.
Figure 2 with 2 supplements
Learning rate over serial reversals and effects of unlimited versus dark-phase only access to touchscreen chambers.
(A–E) Response types over pairwise discrimination (PD), first reversal (R1), second reversal (R2),and third reversal (R3) with unlimited touchscreen access. (A) Number of days to criterion (p=0.0484). (B) Number of sessions to criterion (p=0.0092): R1>R2 (p=0.0099), R1>R3 (p=0.0070). (C) Number of total trials to criterion (p=0.0034): R1>R3 (p=0.0035). Outcome of trials to criterion grouped by outcome (D; p=0.0032) and by phase (E; p<0.0001). (D) Incorrect: R1>PD (p=0.0014), R1>R3 (p=0.0309); Omission: PD>R3 (p=0.0484), R1>R2 (p=0.0092), R1>R3 (p=0.0018). (E) R2: correct>omission (p=0.0045), incorrect>omission (p=0.0059); R3: correct>omission (p=0.0005), incorrect>omission (p=0.0008). (F–K) Effects of unlimited versus dark-phase-only access on cognitive performance. (F) Percentage of correct trials across 15 sessions of each phase of the experiment. (G) Percentage of correct trials (access p=0.0039): PD: Dark phase only>unlimited access (p=0.0371). (H) Percentage of correct trials in the first 100 trials (access p=0.0030): PD: Dark phase only>unlimited access (p=0.0030). (I) Percentage of omission trials across 15 sessions of each phase of the experiment. (J) Percentage of omission trials (access p=0.0737). (K) Percentage of omission trials in the first 100 trials (access p=0.0008): PD: Dark phase only>unlimited access (p=0.0024); R1: Dark phase only>unlimited access (p=0.0332). Bar graphs show group mean ± SEM with individual animals (symbols). Line graphs show group mean ± SEM (shaded bands). *p<0.05, **p<0.01, ***p<0.001.
-
Figure 2—source data 1
Full statistical analysis details and results for Figure 2 and supplements.
- https://cdn.elifesciences.org/articles/84961/elife-84961-fig2-data1-v2.docx
Figure 2—figure supplement 1
Time of day does not influence PhenoSys touchscreen performance.
(A) Start time of each session. (B) Correlation between performance and time of day for all trials and only those that elicited a response (ALL) r=.0465, R2=0.0022, p=0.2088; RESPONDED r=0.0516, R2=0.0027, p=0.1766.
Figure 2—figure supplement 2
Touchscreen pre-training performance measures.
Touchscreen pretraining performance measured by days (A, D, G), sessions (B, E, H), and total trials (C, F, I) to criterion did not systematically differ between any groups. Bar graphs show group mean ± SEM with individual animals (symbols).
Figure 3 with 3 supplements
Impaired cognitive flexibility does not predict susceptibility to ABA.
(A) Schematic of pairwise discrimination (PD) and reversal learning (R1) task and subsequent activity-based anorexia paradigm (ABA). Animals split into two experimental groups determined by body weight loss after exposure to ABA: susceptible or resistant to ABA. Bar graphs show group mean ± SEM with individual animals (symbols). (B) Number of sessions to reach criterion (outcome*phase interaction p=0.0292): R1: ABA resistant>ABA susceptible (p=0.0142). (C) Number of total trials to reach criterion. (D) Number of correct or non-correct trials to reach the PD criterion. (E) Number of correct or non-correct trials to reach R1 criterion (outcome*phase interaction p=0.0389): Non-correct trials: ABA resistant>ABA susceptible (p=0.0401). (F) Progressive performance across the first correct 100 trials in R1 for individual animals: Non-correct response → X+1; correct response → Y+1. *p<0.05.
-
Figure 3—source data 1
Full statistical analysis details and results for Figure 3 and supplements.
- https://cdn.elifesciences.org/articles/84961/elife-84961-fig3-data1-v2.docx
Figure 3—figure supplement 1
Key activity-based anorexia (ABA) parameters that differentiate individuals that are susceptible and resistant to ABA and behavioral profiles during cognitive testing.
Data are from animals that underwent ABA after cognitive testing. (A) Body weight (% of baseline) trajectories for individual animals show that animals split into two subpopulations: ABA susceptible or ABA resistant. (B) Mean daily ABA body weight (BW) % loss (p<0.0001). (C) Mean daily ABA food intake (p<0.0001). (D) Daily running wheel activity (RWA) across both experimental phases. (E) Mean daily RWA (outcome*phase interaction p=0.0232). ABA phase: ABA susceptible >ABA resistant (p=0.0497). (F) Change in mean daily RWA from baseline to ABA (p=0.0232). (G) Mean daily food anticipatory activity (FAA; RWA in the hour before food access; outcome p<0.0001). ABA resistant >ABA susceptible during both baseline (p=0.0011) and ABA (p<0.0001). (H) Number of transitions into the touchscreen testing chamber. (I) Percentage of time moving in the chamber. R1: ABA susceptible >ABA resistant (p=0.0101). (J) Distance (cm) traveled in the chamber. (K) Number of pellet magazine entries. (L) Percentage of time interacting with pellet magazine. (M) Number of total image/screen touches. (N) Percentage of time touching the image/screen. Bar graphs show group mean ± SEM with individual animals (symbols); line graph shows group mean ± SEM. *p<0.05, **p<0.01, ****p<0.0001. PD: pairwise discrimination; R1: reversal learning.
Figure 3—figure supplement 2
Age at onset of activity-based anorexia (ABA) following completion of the Reversal Task.
The age (in weeks; calculated as age in days/7) at the onset of ABA (onset of food restriction) for rats that went on to be Susceptible or Resistant to ABA following completion of the Reversal Task. Independent samples t-test: t(20)=1.061, p=0.3011. ns = not significant.
Figure 3—figure supplement 3
Specific behavioral profiles identified by machine learning tools during touchscreen testing could be used to predict susceptibility or resistance to activity-based anorexia (ABA).
Spider plots and heat maps show the proportion of time spent doing each behavior within each session video during pairwise discrimination (PD; A) or first reversal (R1; B). The spider plots show group mean ± SEM (shaded bands). The time bin heat maps show the change in these proportion values between the groups across time. The values are the log(mean(Susceptible)/mean(Resistant)), where the log is the natural log. The time bins are cumulative, showing e.g., 0-1 min, 0-2 min, etc. (A) Within PD (behavior*outcome interaction p<0.0001), the susceptible rats spent significantly more time inactive (p<0.0001) and significantly less time rearing (p=0.0336) and locomoting (p=0.0190) than the resistant rats. (B) Within R1 (behavior*outcome interaction p<0.0001), the susceptible rats spent more time inactive (p<0.0001) and less time rearing (p=0.0384) than the resistant rats. *p<0.05, ****p<0.0001.
Figure 4 with 5 supplements
Exposure to ABA conditions impairs cognitive performance.
(A) Schematic of experimental paradigm showing activity-based anorexia (ABA) Naïve or Exposed groups and the subsequent pairwise discrimination (PD) and reversal learning (R1) task. There was also a Food restriction-only group that underwent the same time-restricted feeding as the ABA groups but did not have access to running wheels. Animals were split into six experimental groups: Naïve rats that were not exposed to ABA conditions and learned the reversal learning task (ABA Naïve + learned task); ABA Naïve but did not learn the task (ABA Naïve + did not learn); rats previously exposed to ABA condition that learned the subsequent task (ABA Exposed + learned task); rats previously exposed to ABA that did not learn the task (ABA Exposed + did not learn); Food restriction-only rats that learned the task (FR only + learned task); and Food restriction only rats that failed to learn the reversal task (FR only + did not learn). (B) Donut plots of experimental groups: 89% (25/28) of the ABA Naïve rats and 68% (15/22) of the food restriction only rats learned the reversal learning task compared to only 50% (11/22) of the ABA Exposed rats. (C) Number of sessions to reach criterion (main effect of ABA exposure p=0.0068; ABA Exposed + learned task>ABA Naïve + learned task p=0.0051): PD: ABA Exposed + learned task>ABA Naïve + learned task (p=0.0098); R1: ABA Exposed + learned task>ABA Naïve + learned task (p=0.0205). (D) Number of total trials to criterion. (E) Number of correct, incorrect, and omission trials to PD criterion (ABA exposure p=0.0181; ABA Exposed + learned task>ABA Naïve + learned task p=0.0241, ABA Exposed + learned task>Food restriction only + learned task p=0.0412). Correct: ABA Exposed + learned task>ABA Naïve + learned task (p=0.0185), ABA Exposed + learned task>Food restriction only + learned task (p=0.0259). Incorrect: ABA Exposed + learned task>Food restriction only + learned task (p=0.0479). Omission: ABA Exposed + learned task>ABA Naïve + learned task (p=0.0224). (F) Number of correct, incorrect and omission trials to R1 criterion. Number of (G) sessions, (H) total trials, and (I) correct, incorrect and omission trials to PD criterion. Bar graphs show group mean ± SEM with individual animals (symbols). *p<0.05, **p<0.01.
-
Figure 4—source data 1
Full statistical analysis details and results for Figure 4 and supplements.
- https://cdn.elifesciences.org/articles/84961/elife-84961-fig4-data1-v2.docx
Figure 4—figure supplement 1
Comparison of trial types between activity-based anorexia (ABA) naïve rats and ABA exposed or food restriction-only rats that did or did not learn the reversal task.
Number of (J) sessions, (K) total trials and (L) correct, incorrect and omission trials to pairwise discrimination (PD) criterion for food-restricted controls. Number of correct (M), incorrect (N), and omission (O) trials per session during PD and reversal (R1). Bar graphs show group mean ± SEM with individual animals (symbols). Letters above groups indicate whether groups are significantly different or not: b>a p<0.001; b>a p<0.0001.
Figure 4—figure supplement 2
Lone activity-based anorexia (ABA)-exposed Resistant rat highlighted red by request of Reviewers.
(Cs and Ds) Replicates of Figure 4C and D with the only ABA exposed rat that was Resistant to ABA highlighted red.
Figure 4—figure supplement 3
Behavioral differences during touchscreen testing due to order effects of cognitive task and activity-based anorexia (ABA) exposure.
Spider plots and heat maps show the proportion of time spent doing each behavior within each session video during pairwise discrimination (PD; A) or first reversal (R1; B). The spider plots show group mean ± SEM (shaded bands). The time bin heat maps show the change in these proportion values between the groups across time. The values are the log(mean(ABA Naïve)/mean(ABA Exposed)), where the log is the natural log. The time bins are cumulative, showing e.g., 0-1 min, 0-2 min, etc. (A) Within PD (behavior*ABA timing interaction p<0.0001), the ABA Exposed rats spent significantly more time inactive (p<0.0001), and significantly less time rotating their body (p=0.0002) and investigating (p=0.0078) than the ABA Naïve group. (B) Within R1 (behavior*ABA timing interaction p=0.0028), the ABA Exposed rats spent significantly less time rotating their body than the ABA Naïve rats (p=0.0039). (C–I) Bar graphs show group mean ± SEM with individual sessions (symbols); ABA Naïve = Reversal Task Before ABA; ABA Exposed = Reversal Task After ABA. (C) Number of transitions in the chamber (ABA exposure p<0.0001). ABA Exposed >ABA Naïve during both PD (p=0.0001) and R1 (p<0.0001). (D) Percentage of time moving in the chamber. (E) Distance (cm) traveled in the chamber (ABA exposure p=0.0014). PD: ABA Naïve>ABA Exposed (p=0.0277). (F) Number of pellet magazine entries (ABA exposure p=0.0053). PD: ABA Naïve>ABA Exposed (p=0.0454). (G) Percentage of time interacting with pellet magazine (ABA exposure p<0.0001). PD: ABA Naïve>ABA Exposed (p=0.0014). (H) Number of total image/screen touches. (I) Percentage of time touching the image/screen. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
Figure 4—figure supplement 4
Behavioral differences during touchscreen testing due to whether rats learned or did not learn first reversal after prior exposure to activity-based anorexia (ABA).
Spider plots and heat maps show the proportion of time spent doing each behavior within each session video during pairwise discrimination (PD; A) or first reversal (R1; B). The spider plots show group mean ± SEM (shaded bands). The time bin heat maps show the change in these proportion values between the groups across time. The values are the log(mean(Learners)/mean(Non-learners)), where the log is the natural log. The time bins are cumulative, showing e.g., 0-1 min, 0-2 min, etc. (A) Within PD (behavior*outcome interaction p=0.0045), the non-learners spent significantly more time inactive than the learners (p<0.0059). (B) Within R1 (behavior*outcome interaction p<0.0001), the non-learners spent significantly more time inactive (p<0.0001) and significantly less time investigating (p=0.0006) than the learners. (C–I) Bar graphs show group mean ± SEM with individual sessions (symbols). (C) Number of transitions in the chamber. (D) Percentage of time moving in the chamber (outcome p=0.0031). R1: ABA Exposed learned>did not learn (p=0.0100). (E) Distance (cm) traveled in the chamber (stage*outcome interaction p=0.0185). R1: ABA Exposed learned>did not learn (p=0.0021). (F) Number of pellet magazine entries (outcome p=0.0557). R1: ABA Exposed learned>did not learn (p=0.0199). (G) Percentage of time interacting with pellet magazine (outcome p=0.0448). R1: ABA Exposed learned>did not learn (p=0.0172). (H) Number of total image/screen touches (stage*outcome interaction p=0.0025). R1: ABA Exposed learned>did not learn (p=0.0012). (I) Percentage of time touching the image/screen (outcome p=0.0136). R1: ABA Exposed learned>did not learn (p=0.0075). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
Figure 4—figure supplement 5
Performance during the first and last pairwise discrimination (PD) and first reversal (R1) sessions.
Performance during the first 10 min (1) or the full (2) session for each critical session (see below) comparing between ABA Naïve animals that were susceptible or resistant to ABA (A), ABA Exposed animals that learned or did not learn R1 (B), and ABA Naïve versus ABA Exposed animals (C). Critical sessions are: First PD, First session of PD, this is the animal’s first exposure to the two novel stimuli and the pairwise discrimination task; Criterion PD, The session of PD in which an animal reached progression criterion (i.e. Made 30 correct responses with >80% accuracy); First R1, The first session of R1, this is each animal’s first exposure to the reversed reward contingencies; and Last R1, The last session of R1, for animals that successfully learned R1 within 20 sessions this is the session in which they reached progression criterion (i.e. made 30 correct responses with <80% accuracy), for animals that did not reach R1 progression criterion this is their last session (i.e. session 20). (A2) Correct trials, ABA outcome p=0.0486. (B1) Correct trials, all ps<0.0001, Last R1 Learned>Did not learn p<0.0001. Omission trials, all ps <.0192, Last R1 Learned<Did not learn p=0.0120. Percent correct, all ps<0.0001, Last R1 Learned>Did not learn p<0.0001. (B2) Correct trials, all ps<0.0443, Last R1 Learned>Did not learn p=0.0312. Incorrect trials, outcome*session interaction p=0.0090. Omission trials, outcome*session interaction p=0.0006, Last R1 Learned<Did not learn p=0.0610. Percent correct, all ps<0.0001, Last R1 Learned>Did not learn p<0.0001. (C1) Correct trials, ABA exposure p=0.0024, First PD ABA Naïve>ABA Exposed p=0.0001. Incorrect trials, all ps<0.0159, ABA Naïve>ABA Exposed p=0.0005. Omission trials, ABA exposure p=0.0168, ABA Naïve<ABA Exposed p=0.0537. Percent correct, all ps<0.0007, First PD ABA Naïve>ABA Exposed p<0.0001. (C2) Correct trials, all ps<0.0003, First PD ABA Naïve>ABA exposed p<0.0001. Incorrect trials, all ps<0.0001, First PD ABA Naïve>ABA Exposed p<0.0001, First R1 ABA Naïve>ABA Exposed p=0.0024. Percent correct, all ps<0.0002, First PD ABA Naïve>ABA Exposed p<0.0001. Data are group mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
Figure 5 with 1 supplement
Prior training on the reversal task changes the development of the ABA phenotype.
(A) Schematic of activity-based anorexia (ABA) paradigm and the prior or subsequent pairwise discrimination (PD) and reversal learning (RL) task in Reversal Task. (B) Survival plot comparing order effects: ABA resistance was 56% (14/25) for rats that were exposed to ABA after Reversal Task compared to 5% (1/22) for rats that underwent ABA before Reversal Task (p<0.0001). (C) Body weight (% of baseline) trajectories for individual animals across a maximum of 10 days of ABA or until they reached <80%. Data shown are from ALL animals that underwent ABA (D, E, G, H) or ONLY ABA susceptible animals (I, J, L, M). (D) Mean daily ABA body weight (BW) % loss, Before Reversal Task>After Reversal Task (p<0.0001). (E) Mean daily ABA food intake, After Reversal Task>Before Reversal Task (p=0.0009). (F) Daily running wheel activity (RWA) across both experimental phases. Baseline, all ps<0.0001: Before Reversal Task>After Reversal Task (Day 3, p=0.0440; Day 4, p=0.0105; Days 5–7, all ps<0.0001). (G) Mean daily RWA (ABA timing p=0.0002): Before Reversal Task>After Reversal Task during both baseline (p=0.0160) and ABA (p<0.0001). (H) Mean daily food anticipatory activity (FAA; RWA in the hour before food access; ABA timing p<0.0001). After Reversal Task>Before Reversal Task during both baseline (p=0.0010) and ABA (p<0.0001). Mean daily ABA body weight % loss (I) and food intake (J). (K) Daily RWA across both experimental phases. Baseline, all ps<0.0002: Susceptible Before Reversal Task>Susceptible After Reversal Task (Day 5: p=0.0001; Days 6–7: ps<0.0001). (L) Mean daily RWA (ABA timing p=0.0065): Susceptible Before Reversal Task>Susceptible After Reversal Task during both baseline (p=0.0426) and ABA (p=0.0165). (M) Mean daily FAA (ABA timing p=0.0157). Susceptible After Reversal Task>Susceptible Before Reversal Task during ABA (p=0.0357). Bar graphs show group mean ± SEM with all individual animals (symbols); line graphs show group mean ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.
-
Figure 5—source data 1
Full statistical analysis details and results for Figure 5 and supplements.
- https://cdn.elifesciences.org/articles/84961/elife-84961-fig5-data1-v2.docx
Figure 5—figure supplement 1
Restricting food access without a running wheel does not lead to significant weight loss.
(A) Body weight (% of baseline) trajectories for individual animals across a maximum of 10 days of activity-based anorexia (ABA) or until they reached <80% that were either exposed to ABA conditions (Before Reversal Task) or experienced the same time-restricted feeding schedule but did not have access to a running wheel (Food Restriction Only). (B) Mean daily ABA body weight (BW) % loss, Before Reversal Task>Food Restriction Only (p<0.0001). (C) Mean daily ABA food intake, Food Restriction Only >Before Reversal Task (p=0.0045). Bar graphs show group mean ± SEM with all individual animals (symbols). **p<0.01, ****p<0.0001.
Tables
Key resources table
| Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
|---|---|---|---|---|
| Strain, strain background (Rattus norvegicus, female) | Sprague-Dawley | Monash Animal Research Platform | n/a | |
| Software, algorithm | GraphPad Prism 9.1.1 | GraphPad Software | RRID:SCR_002798 | |
| Software, algorithm | Scurry Activity Monitoring Software | Lafayette Instruments, IN | Model 80859 S | https://lafayetteneuroscience.com/products/scurry-activity-software/ |
| Software, algorithm | DeepLabCut | Mathis et al., 2018 Nath et al., 2019 | Version 2.2.1.1 | https://github.com/DeepLabCut/DeepLabCut |
| Software, algorithm | B-SOiD | Hsu and Yttri, 2021 | Version 2.0 | https://github.com/YttriLab/B-SOID |
| Software, algorithm | MultiCam Capture | Pinnacle Studio | https://www.pinnaclesys.com/en/products/multicam-capture/ | |
| Software, algorithm | PhenoSoft | PhenoSys, GmbH | ||
| Other | 2.5 mm diameter sucrose pills | Homeopathic Supply Company | Batch number: 08753017 | https://www.hsconline.co.uk/collections/tablets/products/2-5mm-diameter-sucrose-pillules |
| Other | Lab Bedding Products | The Andersons | Pure-o-cell | https://andersonsplantnutrient.com/engineered-products/lab-and-enrichment/products/pure-ocel |
| Other | Laboratory Enrichment | The Andersons | Enrich-n’Nest | https://andersonsplantnutrient.com/engineered-products/lab-and-enrichment/products/enrich-nnest |
Additional files
-
MDAR checklist
- https://cdn.elifesciences.org/articles/84961/elife-84961-mdarchecklist1-v2.docx
-
Supplementary file 1
Touchscreen parameters.
Parameters for pretraining and 2VDLR in the novel touchscreen apparatus.
- https://cdn.elifesciences.org/articles/84961/elife-84961-supp1-v2.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Rapid, automated, and experimenter-free touchscreen testing reveals reciprocal interactions between cognitive flexibility and activity-based anorexia in female rats
eLife 12:e84961.
https://doi.org/10.7554/eLife.84961
