GluD1 knockout mice with a pure C57BL/6N background show impaired fear memory, social interaction, and enhanced depressive-like behavior

The GluD1 gene is associated with susceptibility for schizophrenia, autism, depression, and bipolar disorder. However, the function of GluD1 and how it is involved in these conditions remain elusive. In this study, we generated a Grid1 gene-knockout (GluD1-KO) mouse line with a pure C57BL/6N genetic background and performed several behavioral analyses. Compared to a control group, GluD1-KO mice showed no significant anxiety-related behavioral differences, evaluated using behavior in an open field, elevated plus maze, a light-dark transition test, the resident-intruder test of aggression and sensorimotor gating evaluated by the prepulse inhibition test. However, GluD1-KO mice showed (1) higher locomotor activity in the open field, (2) decreased sociability and social novelty preference in the three-chambered social interaction test, (3) impaired memory in contextual, but not cued fear conditioning tests, and (4) enhanced depressive-like behavior in a forced swim test. Pharmacological studies revealed that enhanced depressive-like behavior in GluD1-KO mice was restored by the serotonin reuptake inhibitors imipramine and fluoxetine, but not the norepinephrine transporter inhibitor desipramine. In addition, biochemical analysis revealed no significant difference in protein expression levels, such as other glutamate receptors in the synaptosome and postsynaptic densities prepared from the frontal cortex and the hippocampus. These results suggest that GluD1 plays critical roles in fear memory, sociability, and depressive-like behavior.

hand, it remains incompletely clear whether and how inconsistent findings were observed between new GluD1 KO and Grid-tm1Jnz mouse lines as detailed below: Major points: 1. Normal anxiety-like behavior (Fig. 1). In the open-field test, GluD1 KO mice showed a tendency to stay in the center (Fig. 1E). In addition, there is a large variation in the results of an elevated plusmaze test (Fig. 1H, I) and a light-dark transition test (Fig. 1P). The authors should increase N and show individual data.
First of all, we really appreciate that Referee 1 agrees with large concern about issues on replication of behavioral phenotype that are possible due to genetic background, remaining selection markers and the experimental condition in each laboratory.
Secondly, we would like to emphasize that our main focus in this study was assessing the physiological function of GluD1 under pure C57BL/6N genetic background instead of comparing the phenotypes between our GluD1 and Grid tm1Jnz mice. To avoid making this kind of misunderstanding, we have re-structured our discussion. As such, our comparison with Grid tm1Jnz is moved to the end of the discussion (page 29, line 637 -page 31, line 683).
Thirdly, according to Referee 1's comment, we added individual data to all the figures.
Fourthly, Referee 1 pointed out that the number of animals were insufficient to resolve the concern on sample size. We would like to explain how to determine the number of animals in this paper. We estimated the sample size and effect size from the literature on Grid tm1Jnz mice [1][2][3] before commencing experiments. From these published results using Grid tm1Jnz mice, we expected the effect size to be high because there was a significant difference even when using less than 10 mice. The effect size was expected approximately d = 1-1.5. Therefore, we used 7-33 mice depending on the tests. We thought that the number of mice we used was in a range from previous reports in ' Table A' below. For these reasons, we think that we used a sufficient number of mice in this paper.
We added in our methods how we decided the number of animals in this paper (page 5, line 114-115). 3 We cannot deny that our statistical power might not have been enough, so we replaced the pharase "normal anxiety-related behavior" to "Locomotor activity and anxiety-related behavior in GluD1-KO mice" in the results (page 15, line 327). We also replaced the phrase in the legend title of Figure 1 "Hyperlocomotor activity but normal anxiety-related behavior in GluD1-KO mice." to "Locomotor activity and anxiety-related behavior in GluD1-KO mice." (page 16, line 335).  2. Normal aggressive behavior (Fig. 2). There is a tendency that GluD1 KO mice showed more aggression. The authors should increase N and show individual data.
We have now added individual data to the figure. In the resident-intruder test, we expected a high effect size based on the Grid tm1Jnz study (WT = 9, Grid tm1Jnz = 12) [2], so we set the sample size (WT = 8, GluD1-KO = 7) to be almost the same as this previous paper. However, as pointed out by Referee 1, our GluD1-KO mice tended to be more aggressive than WT, but there was no statistical difference.
Since we can not deny that our statistical power might not have been enough, we have rephrased the description of our results as follows: "In accordance with these observations, there was no significant difference between groups, but a tendency for aggressive behavior in GluD1 KO mice in the resident-intruder test."(page 17, line 378-380). We also replaced the title of figure 2 "Normal aggression-like behavior in GluD1-KO mice." to "Aggression-like behavior in GluD1- 4. The effect of the serotonin reuptake inhibitors on depression-like behavior (Fig. 6). This is an interesting new finding because lithium and DCS were previously shown to rescue the depression- In this study, we focused on assessing the impact of representative antidepressants on depression-like behavior in GluD1-KO mice. However, we agree that it would be worthwhile to examine whether 8 lithium and DCS could rescue enhanced depressive-like behavior in our GluD1-KO mice. We have added a sentence regarding lithium and DCS in our discussion (page 29, line 634-636).
5. Normal GluA and GluN2B levels (Fig. 7). It remains unclear whether this difference is caused by the genetic background or experimental conditions. The authors claim that previously reported changes may have been caused by normalization with actin levels. I wonder how such normalization procedures differentially caused changes in GluAs and GluN2B levels. At least the authors should directly test the effect of normalization on their samples.
We appreciate a critical comment from Referee 1. We had realised that normalization by actin is less likely to explain the change in the expression level of GluA and GluN specifically in Grid tm1Jnz mice.
We have re-considered the reason for the difference in the protein expression levels between Grid tm1Jnz and our GluD1-KO mice. One possibility is that dramatic changes of synaptic protein expression levels in Grid tm1Jnz may be due to mixed genetic background. On this point, we have rephrased the paragraph in our discussion (page 30, line 663-page 30, line 683).
6. To interpret the three-chambered social interaction test, the authors need to show that GluD1 KO mice had normal olfaction and vision.
We appreciate essential suggestions from Referee 1. The previous study reported that Grid tm1Jnz mice had normal vision and olfaction, so we predicted our GluD1-KO has the same abilities. However, from our behavioral test, we cannot exclude the possibility that decreased social interaction in the We used male mice because our aim is to evaluate the function of the GluD1 gene by using knockout mice. In this study, we prioritized minimizing the number of animals and thus, we only used male mice; female mice have an estrous cycle that influences behavior, such as anxiety-related behavior [19] and depressive-like behavior [20].
However, as pointed out by Referee 2, there is a possibility that the phenotypes of GluD1-KO mice differs between males and females. Therefore, analysis of female GluD1-KO mice, and a comparison with males in warranted in the future. To address this comment, we have reconsidered and replaced the term "hyper-locomotor activity" with "increased locomotor activity" or "higher locomotor activity" in the open-field test (page 2, line 33; page 17, line 373; page 25, line 559).
Regretfully, we did not measure locomotor activity in the home cage which was another suitable parameter to see locomotor activity.
4. The number of subjects varies greatly among experiments (e.g., the number of wild-type animals exceeds 30 in the OF and is less than 10 in the RI). Please explain whether the different number of subjects affects the statistical power of the results.
As we explained to Referee 1, we decided the sample size in this study based on previous studies of Grid tm1Jnz mice [1][2][3]. The effect size was expected to be high because there was a significant difference even when using less than 10 mice in these previous studies. In our behavior tests, such as anxiety, sociability, and depression-like behavior, we increased the cohort of mice group based on our estimated effect size in order to have sufficient statistical power.
5. On page 23, lines 500-507, the pharmacological tests using three drugs seem to be carried out independently. Therefore, the comparison by integrating all the subjects by using one-way ANOVA is not suitable in this case. Please consider comparing each control group with the corresponding drug-treated group by t-test or two-way ANOVA (effects of genotype and drug treatment). In addition, please consider adding the results visualized by graphs as supplemental data.
We completely agree with the comment on the statistical procedure; we actually had done this same analysis that Referee 2 suggested. We apologize for not having clearly presented this statistical analysis in the previous manuscript ( Figure 6B, legend). So, we have rephrased this part to clarify this statistical analysis. We referred to the previous report [19]. We first confirmed there was a significant genotype effect, but no interaction between genotype and treatment in two-way ANOVA.
We then performed one-way ANOVA with Dunnett's post hoc test for each genotype. The Dunnett's test is a method for testing the difference between the mean value of each of the control group and the treated group in the data of one control group (in this case, saline) and two or more treated groups.
" Figure 6B  Due to this comment, we have shared raw data for all the behavioral analysis and western blot on figshare because we also believe that publishing raw data used in the graph is very useful, especially for future experimenters and meta-analysis.