Reversal learning paradigm reveals deficits in cognitive flexibility in the Fmr1 knockout male mouse

Background: Loss of FMR1 is associated with Fragile X syndrome, amongst the most prevalent inherited intellectual disability. Despite extensive research in this area, previous studies have failed to detect consistent evidence of cognitive impairments in the Morris water maze (MWM) task in the Fmr1 knockout (KO) mouse. However, few studies have examined cognitive flexibility in a reversal form of the MWM task, which may illuminate subtle learning deficits. Methods: Adult male Fmr1 wildtype (WT) and KO mice were bred and tested in the MWM reversal paradigm. The testing paradigm consisted of two blocks per day, with 4 trials per block to locate a hidden platform. After the last trials on the fourth day of testing, the animals were given a probe trial with the platform removed. The following week, the location of the platform was switched to the opposite quadrant and the animals received 2 more days of testing, with 4 blocks in total. Results: As expected, Fmr1 KO mice did not display a learning deficit during the acquisition phase, F genotype (1, 24) = 0.034, p = 0.854, and performed similarly on the probe trial, F genotype (1, 23) = 0.024, p = 0.877. However, during the reversal phase of learning, Fmr1 KO mice showed deficits in their ability to learn the new location of the platform, F genotype (1, 23) = 3.93, p = 0.059. Further independent samples t-testing revealed that KO animals displayed significantly higher latency to reach the hidden platform during the third trial, t(23) = -2.96, p < 0.01. Conclusions: While previous studies have not demonstrated deficits in spatial memory in the Fmr1 KO model, it is possible that the acquisition phase of the task is less sensitive to deficits in learning. Future studies using this model to evaluate therapeutic interventions should consider utilizing the MWM reversal paradigm.


Introduction
Fragile X syndrome (FXS) is a neurodevelopmental disorder, caused by a trinucleotide expansion mutation in the FMR1 gene, and is also one of the most prevalent inherited forms of intellectual disability 1 . FXS is often modeled using the Fmr1 knockout (KO) mouse, which can be characterized by several behavioral phenotypes, including alterations in sociability and deficits in fear memory [2][3][4] . Aside from deficits in spatial and non-spatial learning, one understudied facet of intellectual disability is the ability to incorporate new information into existing learning, termed cognitive flexibility. Cognitive flexibility can be studied in rodents using a variant of the Morris water maze (MWM) paradigm 5,6 . In the MWM reversal paradigm, the location of the hidden platform is moved, and the latency to adjust to the new location is measured. As expected, several reports find evidence of impairments in reversal learning in the Fmr1 KO mouse across multiple strains, the C57BL/6J backcrossed strain 7,8 and the albino C57BL/6J background 2 . However, other studies have been unable to replicate these findings, and further investigation points to the possibility of background strain differences 9 . Previous reports have not detected any impairments in reversal learning in the FVB.129 strain 10 . However, it may be that this paradigm is perhaps even more sensitive to methodological differences. The current study adds to this literature by using the FVB.129 strain in a previously utilized paradigm.

Animals
Male Fmr1 +/+ and female Fmr1 +/-FVB.129P2-Pde6b+Tyrc-ch Fmr1tm1Cgr/J (Stock No: 004624, The Jackson Laboratory, Bar Harbor, ME, USA) mice were used as breeders (9 total breeding pairs) to produce the following groups: male WT and male KO pups. Breeding pairs were of the following groupings: WT Female/WT Male (n = 2), KO Female/KO Male (n = 5), WT Female/KO Male (n = 2). Genotype was determined from toe clippings taken prior to age postnatal day (PD) 12 (Mouse Genotype, Escondido, CA, USA). The final sample sizes were as follows: n male WT = 10, n male KO = 16. Target samples sizes (n = 10) were calculated a priori using a power calculation in G*Power 3.1 with the following parameters: f = 0.50 (large effect), α = 0.05, power (1 -β) = 0.80, for the F family of tests with two groups and 8 repeated measures (trials). All pups were housed in individual cages (Allentown Caging PC7115HT, Allentown, PA, USA), filled with sani-chip bedding (7090 Teklad, Envigo, Somerset, NJ, USA). Prior to weaning on PD21, pups were housed with parents (1 male and 2 females) and littermates (up to 12 pups). Following weaning, subjects were housed with mixed genotype littermates, no more than 5 to a cage. The light cycle was kept at 12 hr. light, and the colony room was kept at an ambient temperature of 22° C. Animals had ad libitum access to food and water. All procedures performed were in accordance with Baylor University Institutional Animal Care and Use Committee (Animal Assurance Number A3948-01), as well as the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. All efforts were made to ameliorate any stress and harm to the animals, specifically by habituating animals to the testing apparatus and room prior to trial recordings.

Morris water maze
All behavioral testing was conducted during the light cycle, specifically between 8 am and 5 pm. The methods for the current study were adapted as closely as possible from earlier studies of this behavior in the Fmr1 KO mouse (represented in Figure 1) 9 . Briefly, a 1.3 m diameter white pool was filled with water and made opaque through the addition of non-toxic white paint (Item LT3010, S&S Worldwide, Connecticut). The hidden platform measured 14.5 cm × 14.5 cm and was submerged approximately 2 cm below the water level. The testing paradigm consisted of two blocks per day for 4 days, with 4 trials in each block, for each mouse to test the ability to locate a hidden platform. The mice were habituated to the testing room in their holding cages for 30 minutes prior to the onset of testing. The amount of time spent in each quadrant for each trial was recorded with a ceiling-mounted video camera (Ganz YCH-02, Cary, NC, USA), and analyzed using automated tracking software (Ethovision XT 6, Noldus, Wageningen, Netherlands). After the last trial on the fourth day of testing was completed, the animals were given a probe trial. The probe trial involved removing the platform and allowing the subjects to explore the maze for 60 seconds. During the probe trial, the number of times the animal crossed the location of the hidden platform and the duration of time in each quadrant was calculated. Testing resumed on day 8 after a 3-day rest period. On day 8, the platform was placed in the opposite quadrant from the previous location that housed the hidden platform. Testing progressed as with the initial acquisition phase, with 2 blocks per day for 2 days. On the final day of testing, a visible platform was used to evaluate visual performance as well as swim speed. The visible platform was a two-tiered platform similar to the initial platform, with a second higher tier platform that extended 9.5 cm above the lower platform, allowing the animal to see the platform. The differences in methodology from the cited source were as follows: only four days of acquisition were conducted and the testing paradigm was lengthened to account for a consolidation period between the learning and reversal trials (See Figure 1 for a description of the testing paradigm). One KO animal was excluded from analysis due to a seizure during this task.

Statistical analysis
Statistical analysis was performed in the form of a one-way analysis of variance (ANOVA) with one between-subjects factor (Genotype [wildtype, knockout]) and one within-subjects factor (Trial). All data were analyzed using GraphPad Prism Software 7.0 (San Diego, CA, USA) or IBM SPSS Statistics 23 (Aramonk, NY, USA).

Results
Fmr1 KO mice show no impairment in acquisition of spatial memory To investigate the effect of genotype on hippocampal spatial memory, animals were tested in the MWM paradigm (Dataset 1). During the 8 blocks of learning trials (Figure 2A), there was a significant within-subjects effect of trial for latency to reach the platform, F(3.56, 85.47) = 30.15, p < .0005. Trial results did not interact significantly with genotype, F(3.56, 85.47) = 1.33, p = 0.24, suggesting both groups learned the location of the platform similarly. Between-subjects analyses indicated no effect of genotype, F(1, 24) = 0.03, p = 0.85. Further independent samples t-testing revealed no differences between WT and KO at any of the 8 different trials, p > 0.05.
For the probe trial (Dataset 2), as expected, male KO mice demonstrated similar time spent in the target quadrant, F(1,23) = 0.02, p = 0.88 ( Figure 2B), compared to male wildtype (WT). Further independent samples t-testing revealed no differences in duration in any of the quadrants, p > 0.05.

Loss of Fmr1 impairs ability to update existing learning with new platform location
The week following the initial learning trials, animals were tested in the reversal learning paradigm (Dataset 3). During the 4 blocks of learning trials ( Figure 2C), a one-way ANOVA with repeated measures revealed a significant within-subjects effect of trial, F(3, 69) = 3.8, p < 0.05. Trial results did not interact significantly with genotype, F(3, 69) = 1.28, p = 0.29. However, between-subjects analyses indicated a marginal effect of genotype, F(1, 23) = 3.93, p = 0.059 ( Figure 2C). Further independent samples t-testing revealed that KO animals displayed significantly higher latency to reach the hidden platform during the third trial, t(23) = -2.96, p < 0.01. Together, these results demonstrate that Fmr1 KO males demonstrate decreased learning and altogether a lack of cognitive flexibility across all trials of the MWM reversal task. Impairments were not due to deficits in vision or motor capabilities Visible platform information was also assessed to ensure differences were not due to deficits in vision (Dataset 4). Results were analyzed using a repeated-measures ANOVA across the four visible platform trials on latency to the platform across the two blocks of trials. Results indicated no effect of block, F(1, 24) = 1.341, p = 0.26, nor an interaction of block and genotype, F(1, 24) = 0.0005, p = 0.98. There was not a significant effect of genotype on latency to the platform during these trials either, F(1, 24) = 0.98, p = 0.33 ( Figure 3A). Altogether, these data suggest that differences in latency to the platform could not be attributed to deficits in visual perception in the Fmr1 KO male mouse. Moreover, differences in latency to the platform on the previous trials could not be attributed to impairments in swimming abilities, as no differences in swim speed were detected during the visible trials, t(11.17) = 1.526, p = 0.16 ( Figure 3B). This datasheet contains the raw data exported from the Ethovision program (Columns A -AB) as well as the transformed dataset that was used for analysis for the probe trial (Columns AC -AI)

Discussion
As expected, deletion of Fmr1 did not impact initial spatial learning in the MWM. The current study did, however, demonstrate impairments in cognitive flexibility in the Fmr1 KO. As previously mentioned, other studies have not before detected such changes, and this discrepancy could be attributed to methodological differences 10 . In the aforementioned study, training occurred over 8 days, with only 3 training trials per day, and the reversal paradigm consisted of 4 days, with 3 trials per day. Furthermore, in support of our findings, deficits in long-term potentiation in the prefrontal cortex have been demonstrated in the Fmr1 KO mouse, the area on which the ability to adapt to a new location in this task is dependent on 11-13. Moreover, this ability to adapt to a new location is mediated through multi-synaptic connections between the hippocampus and the prefrontal cortex 13 . This proposed mechanism further supports our findings of no change to the initial spatial learning phase, as lesions to this area did not impact initial learning performance in spatial navigation 11 .
The current study provides preliminary evidence that could be applied to tease apart subtle differences between the male and female Fmr1 KO phenotype, which has been difficult to conclusively evaluate. Future studies should expand upon these findings in females, as many studies have demonstrated a sex-specific effect of loss of Fmr1 on behavior (discussed in a recent review 14 ) 2-4, 15 . Overall, this study corroborates and extends previous evidence of impaired cognitive flexibility in the male Fmr1 KO mouse.

Competing interests
No competing interests were disclosed. The current study examines male Fmr1 knockout mice in the Morris Water maze. This group focuses on the FVB strain as previous work has shown background strain differences can impact the phenotype of Fmr1 knockouts. The authors show that the male Fmr1 knockout mice have no impairment in acquisition of spatial memory but do show differences in the reversal learning paradigm that are not attributed to vision or motor differences. In the reversal learning paradigm, the authors show that during the 4 blocks of reversal learning the Fmr1 KOs show a rather stable latency to find the platform in contrast to the WT mice that show the expected decrease to reach the platform with increasing trials. The authors highlight the finding that male Fmr1 knockout mice have deficits in reversal learning as assessed in the Morris water maze but not in acquisition, demonstrating a link to a specific form of learning in this paradigm. The study is straightforward and the data clearly presented.

Grant information
A couple of comments the authors should further clarify: The authors should elaborate on the breeding scheme as they are not directly comparing littermate controls.
Based on the authors discussion point regarding potential sex differences between male and female Fmr1 KO mice, is there data to suggest differences in cognitive flexibility between the males and female KOs?

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility?

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.

Competing Interests:
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Andre Fenton
Center for Neural Science, New York University, New York, NY, USA I am afraid that I did not find this manuscript and the results to be compelling. The experimental design is straightforward but the analysis is questionable and the report is not as scholarly as it should be.
Statistics are used questionably. 1) t tests are used for post-hoc comparisons, when they are not justified, For example, post-hoc t tests are performed after ANOVA has failed to find relevant differences. 2) Furthermore, the alpha level for post-hoc t tests seem not to be corrected for multiple comparisons. The reported significant difference on trial 3 may not maintain after correcting alpha for multiple comparisons 0.05/4, although it might -a reader can't know. Regardless, what is the justification for the t tests after the ANOVA results were not significant? Dataset 1: Given no effects of genotype or interaction in the ANOVA, what justifies data snooping as follows: "Further independent samples t-testing revealed no differences between WT and KO at any of the 8 different trials, p > 0.05." ? Dataset 2: Given the ANOVA was not significant for a 1-way ANOVA, there is no justification for a t test and in fact it is numerically equivalent to the F-test, so even more so not warranted. "Further independent samples t-testing revealed no differences in duration in any of the quadrants, p > 0.05." The manuscript states that "Fmr1 KO mice showed deficits in their ability to learn the new location of the platform, Fgenotype (1, 23) = 3.93, p = 0.059." Why is p= 0.059 indicative of a deficit?
The Introduction is written as if the water maze is the only way to test spatial memory and cognitive flexibility, when of course it is not.
The Discussion states "The current study did, however, demonstrate impairments in cognitive flexibility in the Fmr1 KO. As previously mentioned, other studies have not before detected such changes" Remarkably, the authors fail to consider the work from the Fenton lab which explicitly has investigated cognitive flexibility in Something is wrong with the initial statement: "Male Fmr1+/+ and 1 Something is wrong with the initial statement: "Male Fmr1+/+ and female Fmr1+/-FVB.129P2-Pde6b+Tyrc-ch Fmr1tm1Cgr/J (Stock No: 004624, The Jackson Laboratory, Bar Harbor, ME, USA) mice were used as breeders (9 total breeding pairs) to produce the following groups:" because WT mice cannot contribute to breeding involving a KO male. Fig 1B should have the days indicated, as well as the rest period. What justification is there for the assertion that "Moreover, this ability to adapt to a new location is mediated through multi-synaptic connections between the hippocampus and the prefrontal cortex13."? What is the mechanism that is referenced by the statement "This proposed mechanism further supports our findings of no change to the initial spatial learning phase, as lesions to this area did not impact initial learning performance in spatial navigation11."?
The final statement seems untrue given that the authors claim to find a reversal deficit but say there had not been one detected previously "Overall, this study corroborates and extends previous evidence of impaired cognitive flexibility in the male Fmr1 KO mouse." What is "t-testing" should this not simply be stated as "t tests" ?

If applicable, is the statistical analysis and its interpretation appropriate? No
Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Consider changing "disability" to "disabilities" in first sentence of the abstract. Please add more detail about the Morris water maze task. For instance, did mice receive consecutive trials within each block? Was entry quadrant varied or static over each mouse's trials within (and across) blocks? Was the temperature of the water controlled? What were the visual cues, where were they positioned with respect to each quadrant, and were they proximal or distal? 6. 7. 8.

9.
10. cues, where were they positioned with respect to each quadrant, and were they proximal or distal? If animals were artificially colored for detection in the maze, please specify. Were animals dried and/or warmed between or after trials? It may be helpful to label days on Fig. 1B.
In the first Results paragraph--in the absence of a significant main effect or interaction involving genotype, further independent samples t-testing between WT and KO should probably not be reported.
In the Discussion, the authors point to deficits in KO prefrontal cortex LTP with regard to the Fmr1 observed reversal learning deficit, which is fine. However, using this same logic to claim support for of observed differences in acquisition, based on findings that PFC is not relevant to lack acquisition, is somewhat flawed. It may be more appropriate (e.g., in the sentence regarding PFC and learning a new location) to simply acknowledge that "PFC is not required for acquisition" and allow the reader to intuit the point (that seems to be intended) more subtly. If this topic will be discussed, other brain regions likely involved in MWM acquisition (e.g., hippocampus), and that also show significant plasticity-related variations in KO compared to WT mice, should be Fmr1 included. The authors' point that this investigation should be made in females is well-received. However, it is not immediately clear how these data in male KO mice provide "preliminary evidence that Fmr1 could be applied to tease apart subtle differences between the male and female KO Fmr1 phenotype." Perhaps this point can be further clarified or removed? In the attached data sets, it is not clear why a few animal IDs are repeated. Specifically, please see the "Learning trial" set, and the "transformed" data on the right-hand side of the screen. Subjects 124 and 134 are listed twice, but all other IDs appear once. Also, a little more explanation, either directly on the data sheets or elsewhere in the manuscript, for the trial names, etc. seen here may be beneficial to users of these data.

Is the work clearly and accurately presented and does it cite the current literature? Partly
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.

Competing Interests:
Referee Expertise: fragile X; autism; addiction; dendritic structure and plasticity I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com