Parallel learning and cognitive flexibility impairments between Fmr1 knockout mice and individuals with fragile X syndrome

Abstract

Introduction: Fragile X Syndrome (FXS) is a monogenic condition that leads to intellectual disability along with behavioral and learning difficulties. Among behavioral and learning difficulties, cognitive flexibility impairments are among the most commonly reported in FXS, which significantly impacts daily living. Despite the extensive use of the Fmr1 knockout (KO) mouse to understand molecular, synaptic and behavioral alterations related to FXS, there has been limited development of translational paradigms to understand cognitive flexibility that can be employed in both animal models and individuals with FXS to facilitate treatment development.

Methods: To begin addressing this limitation, a parallel set of studies were carried out that investigated probabilistic reversal learning along with other behavioral and cognitive tests in individuals with FXS and Fmr1 KO mice. Fifty-five adolescents and adults with FXS (67% male) and 34 age- and sex-matched typically developing controls (62% male) completed an initial probabilistic learning training task and a probabilistic reversal learning task.

Results: In males with FXS, both initial probabilistic learning and reversal learning deficits were found. However, in females with FXS, we only observed reversal learning deficits. Reversal learning deficits related to more severe psychiatric features in females with FXS, whereas increased sensitivity to negative feedback (lose:shift errors) unexpectedly appear to be adaptive in males with FXS. Male Fmr1 KO mice exhibited both an initial probabilistic learning and reversal learning deficit compared to that of wildtype (WT) mice. Female Fmr1 KO mice were selectively impaired on probabilistic reversal learning. In a prepotent response inhibition test, both male and female Fmr1 KO mice were impaired in learning to choose a non-preferred spatial location to receive a food reward compared to that of WT mice. Neither male nor female Fmr1 KO mice exhibited a change in anxiety compared to that of WT mice.

Discussion: Together, our findings demonstrate strikingly similar sex-dependent learning disturbances across individuals with FXS and Fmr1 KO mice. This suggests the promise of using analogous paradigms of cognitive flexibility across species that may speed treatment development to improve lives of individuals with FXS.

Keywords: FMR1; autism; cognitive flexibility; executive function; fragile X syndrome.

FIGURE 1

Schematic of the human probabilistic reversal learning task (A) and error types (B). Participants were instructed to select the animal in the correct stimulus location on the screen (Acquisition). Once participants reached performance criterion, the correct stimulus location changed without warning to the other location on the screen (Reversal). Across both acquisition and reversal phases, reinforcement was provided on an 80:20 probabilistic schedule, and correct responses were reinforced with coins. A red “X” appeared over the animal if the incorrect stimulus location was chosen by the participant. Perseverative errors occurred when a participant chose the previously reinforced location prior to choosing the newly reinforced (correct) location. Regressive errors occurred when a participant chose the previously reinforced location after choosing the newly reinforced location. Lose:shift errors are a specific type of regressive error when participants chose the previously reinforced location immediately following inaccurate feedback (e.g., 20% of trials) that their choice was incorrect. Animal and location on the screen were randomized for each participant with a total of four possible combinations of animal and location. Permission to use image from author (Schmitt et al., 2022).

FIGURE 2

Mouse probabilistic reversal learning task. Photograph of the apparatus used for testing mice in the acquisition and reversal learning of a spatial discrimination with probabilistic reinforcement. Food-restricted mice were first trained to recover sweetened cereal pieces from food wells in choice areas. After achieving criterion, mice received an acquisition test the following day. The day after acquisition, mice first received a retention test followed 30 min later by a reversal learning test. In all tests, a 80/20 probabilistic reinforcement schedule was used such that the “correct” choice was reinforced with 80% probability.

FIGURE 3

Trials to reach criterion on a spatial discrimination task (80/20 probabilistic learning procedure). (A) Mean (± SEM) trials to criterion on training task (no reversal). Males with fragile X syndrome (FXS), but not females with FXS required significantly more trials to criterion on acquisition than their TDC counterparts. Males with FXS also required more trials than females with FXS. (B) Mean (± SEM) trials to criterion on reversal learning. Males and females with FXS required more trials to reach criterion compared to that of control males and females. *p < 0.05 vs. TDC female; ***p < 0.001 vs. TDC male; ^∧p < 0.05 vs. FXS male. Solid dot indicates individual participant data point.

FIGURE 4

Error committed during reversal learning of a spatial discrimination (80/20 probabilistic learning procedure). (A) Mean (± SEM) perseverative errors during reversal learning. Fragile X syndrome (FXS) and TDC subjects did not differ in the number of perseverative errors. However, males with FXS made more perseverative errors than their TDC counterparts. (B) Mean (± SEM) regressive errors during reversal learning. FXS participants committed more regressive errors compared to controls during reversal learning. This finding was significant comparing male groups, but only trending when comparing female groups. (C) Mean (± SEM) lose-shift errors during acquisition. FXS participants made more lost-shift errors, which was driven by females. (D) Mean (± SEM) lose-shift errors during reversal learning. Males with FXS committed more lose-shift errors than TDC males, a finding that was only trending among females. *p < 0.05 vs. TDC males; **p < 0.05 vs. TDC females; ***p < 0.001 vs. TDC males; ∼p < 0.10 vs. TDC females. Solid dot indicates individual participant data point.

FIGURE 5

Correlations with clinical data for males and females with fragile X syndrome (FXS). (A) Increased lose:shift errors during the reversal phase related to fewer anticipation errors during KiTAP Alert task in FXS males (black square). (B) Likewise, lose:shift errors during the reversal phase related to fewer errors during KiTAP Distractibility task in FXS males. (C) In contrast, increased total trials needed to reach criterion related to more severe ABC irritability ratings in FXS females (black triangle). (D) Increased perseverative errors related to more severe ADAMS ratings in FXS females as well. Spearman correlation and p-values provided for each graph.

FIGURE 6

Acquisition and reversal learning of a spatial discrimination task (80/20 probabilistic learning). (A) Mean (± SEM) trials to criterion on acquisition. Male Fmr1-knockout (KO) mice, but not female Fmr1-KO mice required significantly more trials to criterion on acquisition than wildtype (WT) male and female mice. (B) Mean (± SEM) trials to criterion on retention. WT and Fmr1-KO mice required similar number of trials to criterion in retention. (C) Mean (± SEM) trials to criterion on reversal learning. Male and female Fmr1-KO mice required more trials to reach criterion compared to that of WT male and female mice. *p < 0.05 vs. WT male and female mice; **p < 0.01 vs. WT male and female mice. In the female Fmr1- KO group, •, heterozygous mice and ∘, homozygous mice.

FIGURE 7

Error committed during reversal learning of a spatial discrimination (80/20 probabilistic learning procedure). (A) Mean (± SEM) perseverative errors during reversal learning. Wildtype (WT) and Fmr1-knockout (KO) mice did not differ in the number of perseverative errors. (B) Mean (± SEM) regressive errors during reversal learning. Male and female Fmr1-KO mice committed more regressive errors compared to that of WT male and female during reversal learning. (C) Mean (± SEM) percentage win-stay probabilities during reversal learning. Fmr1-KO mice had lower win-stay probabilities than WT mice. (D) Mean (± SEM) percentage lose-shift probabilities during reversal learning. Both male and female Fmr1-KO mice had higher lose-shift probabilities than WT male and female mice. *p < 0.05 vs. WT male and female mice; **p < 0.01 vs. WT male and female mice. In the female Fmr1- KO group, •, heterozygous mice and ∘, homozygous mice.

FIGURE 8

Performance in elevated plus maze and reward conflict test. (A) Mean (± SEM) duration in the open arms in the elevated plus maze test. Wildtype (WT) and Fmr1-knockout (KO) mice did not differ in open arm duration. (B) Mean (± SEM) percent correct across three 10 min blocks in reward conflict test. WT and Fmr1-KO mice exhibited similar performance during the first block of testing with male and female WT mice exhibiting greater performance than Fmr1-KO mice during the second and third blocks. *p < 0.05 vs. WT male and female mice. (C) Mean (± SEM) percent of open arm trials cereal reward was consumed. WT and Fmr1-KO mice did not differ in percent of open arm trials cereal reward was consumed. In the female Fmr1- KO group, •, heterozygous mice and ∘, homozygous mice.