Conditional deletion of Cadherin 13 perturbs Golgi cells and disrupts social and cognitive behaviors

Abstract

Inhibitory interneurons mediate the gating of synaptic transmission and modulate the activities of neural circuits. Disruption of the function of inhibitory networks in the forebrain is linked to impairment of social and cognitive behaviors, but the involvement of inhibitory interneurons in the cerebellum has not been assessed. We found that Cadherin 13 (Cdh13), a gene implicated in autism spectrum disorder and attention-deficit hyperactivity disorder, is specifically expressed in Golgi cells within the cerebellar cortex. To assess the function of Cdh13 and utilize the manipulation of Cdh13 expression in Golgi cells as an entry point to examine cerebellar-mediated function, we generated mice carrying Cdh13-floxed alleles and conditionally deleted Cdh13 with GlyT2::Cre mice. Loss of Cdh13 results in a decrease in the expression/localization of GAD67 and reduces spontaneous inhibitory postsynaptic current (IPSC) in cerebellar Golgi cells without disrupting spontaneous excitatory postsynaptic current (EPSC). At the behavioral level, loss of Cdh13 in the cerebellum, piriform cortex and endopiriform claustrum have no impact on gross motor coordination or general locomotor behaviors, but leads to deficits in cognitive and social abilities. Mice lacking Cdh13 exhibit reduced cognitive flexibility and loss of preference for contact region concomitant with increased reciprocal social interactions. Together, our findings show that Cdh13 is critical for inhibitory function of Golgi cells, and that GlyT2::Cre-mediated deletion of Cdh13 in non-executive centers of the brain, such as the cerebellum, may contribute to cognitive and social behavioral deficits linked to neurological disorders.

Keywords: Cadherin 13; GABAergic interneurons; GlyT2; Golgi cells; autism spectrum disorder; cerebellum; cognitive flexibility; reciprocal social interaction.

Figure 1

Cdh13 is expressed by a subpopulation of GABAergic interneurons in the cerebellar cortex. A‐H, Expression of Cdh13 mRNA at embryonic and postnatal stages of development. At E14, Cdh13 is not detected in the VZ (A), where limited GAD67 expression is present (red arrows; B). At E18, the expression of Cdh13 and GAD67 is present in the WM layer of the cerebellum (red arrows; C, D). At postnatal stages P10‐100, Cdh13 and GAD67 are both found in the IGL of the cerebellar cortex (red arrows; E‐H). EGL, external granular layer; PCL, Purkinje cell layer; RL, rhombic lip; WM, white matter; VZ, ventricular zone; ML, molecular layer; IGL, internal granular layer. I‐K, Double fluorescent in situ hybridization shows that Cdh13 and GAD67 mRNA are colocalized in the IGL (white arrows; I‐K). GAD67 signal is detected in inhibitory neurons in all 3 layers (J), but Cdh13 signal is only detected in the IGL (white arrows; I, K). Scale bar = 200 μm (A‐H), 20 μm (I‐K)

Figure 2

GlyT2::Cre mediates Cdh13 deletion in cerebellar Golgi cells. A, Strategy for generation of Cdh13 conditional knockout mice. Exon 3 flanked by loxP is removed by Cre expression in GlyT2::Cre mice. B, Southern blot analysis of WT, Cdh13 ^fl/+, Cdh13 ^fl/fl mice. C, PCR genotyping analysis of WT, Cdh13 ^fl/+, Cdh13 ^fl/fl mice. D‐F, GlyT2::Cre‐mediated recombination occurs in mGluR2⁺ (red arrows; D, E), Neurogranin⁺ (yellow arrows; D, F) and mGluR2⁺Neurogranin⁺ neurons (white arrows; D, E, F) of a P60 GlyT2::Cre; Rosa::lox‐stop‐lox‐eYFP mouse. Soma of Purkinje cells are indicated by asterisks. G‐H, Cdh13 expression is found in the IGL of the cerebellum of a P100 control mouse (I), but not in a P120 GlyT2‐Cdh13 ^−/− mouse (J). Corresponding area (Gi‐Giii and Hi‐Hiii) from control and GlyT2‐Cdh13 ^−/− mouse (red boxes in G, H). Scale bar = 100 μm (D‐E), 1000 μm (G, H), 100 μm (Gi‐Giii, Hi‐Hiii)

Figure 3

Loss of Cdh13 results in reduction of the expression/localization of GAD67. A, B, E, F, Immunohistochemical analysis of the expression of GAD67 in the IGL of Crus2 and lobule IX in the cerebellum of control (A, E) and GlyT2‐Cdh13 ^−/− mice (P75‐90) (B, F). C, G, Quantification of the intensity of GAD67 in Crus 2 and lobule IX (Crus2: control = 19.09 ± 1.14, GlyT2‐Cdh13 ^−/− = 13.7 ± 0.99, U = 99, P = .002, Mann‐Whitney U test, N = 3, n = 20‐22; LIX: control = 17.55 ± 1.08, GlyT2‐Cdh13 ^−/− = 13.2 ± 0.72, U = 103, P = .0003, Mann‐Whitney U test, N = 3, n = 22‐24). D, H, Quantification of the area coverage of GAD67 in Crus 2 and lobule IX (Crus2: control = 202.9 ± 21.87, GlyT2‐Cdh13 ^−/− = 107.5 ± 19.04, U = 103, P = .003, Mann‐Whitney U test, N = 3, n = 20‐22; LIX: control = 147.3 ± 21.42, GlyT2‐Cdh13 ^−/− = 70.3 ± 12.5, U = 100, P = .0002, Mann‐Whitney U test, N = 3, n = 22‐24). Closed circles: control, open circles: GlyT2‐Cdh13 ^−/−. Scale bar = 20 μm

Figure 4

Loss of Cdh13 impairs postsynaptic responses of Golgi cells. A, Whole‐cell patch clamp recording of Golgi cells in the IGL labeled with Neurobiotin. B, Examples of traces of spontaneous postsynaptic currents and miniature postsynaptic currents of Golgi cells in control and GlyT2‐Cdh13 ^−/−mice (P26‐35). C‐F, Analysis of inhibitory postsynaptic current of Golgi cells in the IGL of control and GlyT2‐Cdh13 ^−/− mice shows reduced sIPSC frequency (C, P = .005, Mann‐Whitney U test) and decreased mIPSC frequency of GlyT2‐Cdh13 ^−/− Golgi cells compared to control (D, P = .002, Mann‐Whitney U test). No differences in amplitude of both sIPSC and mIPSC were found between control and GlyT2‐Cdh13 ^−/− Golgi cells (sIPSC amplitude, P = .5 Mann‐Whitney U test; mIPSC amplitude P = 1.0 Mann‐Whitney U test) (E, F). G‐J, Excitatory postsynaptic current in control and GlyT2‐Cdh13 ^−/− Golgi cells. sEPSC frequency and amplitude did not change in GlyT2‐Cdh13 ^−/− Golgi cells compared to control (G, I, sEPSC frequency P = .9, Mann‐Whitney U test; sEPSC amplitude P = .6 Mann‐Whitney U test). In GlyT2‐Cdh13 ^−/− Golgi cells, both amplitude and frequency of mEPSC were not different from that of control Golgi cells (H, J, mEPSC frequency P = 1.0 Mann‐Whitney U test; mEPSC amplitude P = .9 Mann‐Whitney U test). Data presented as mean ± SEM. P26‐35 mice, n = 10‐14/group. Scale bar = 20 μm. See Table S3

Figure 5

Deletion of Cdh13 in the cerebellum has no impact on general activity and motor coordination. A, In multiple trials on a rotating rod, GlyT2‐Cdh13 ^−/− mice display similar performance to that of control mice. B, GlyT2‐Cdh13 ^−/− mice show a tendency of making shorter strides during walking (P = .06, Mann‐Whitney U test). C, Comparison of foot prints between control and GlyT2‐Cdh13 ^−/− mice. D, E, GlyT2‐Cdh13 ^−/− mice show comparable distance traveled (D) and velocity (E) during 7 minutes of free locomotion in an open arena. F, No genotype differences were detected in exploration of the open field as illustrated by the heat map of locomotion tracks. G, Open field anxiety‐related readouts revealed no genotype differences in preference for corner/peripheral zone or center zone. H, Analysis of the number of visits to the center zone of the open field. I, Digging behavior was not influenced by Cdh13 deletion. Data presented as mean ± SEM. Adult male mice, n = 10‐19/group. See Table S4

Figure 6

Overtraining in 2‐choice digging task reveals deficit in cognitive flexibility in GlyT2‐Cdh13 ^−/− mice. A, Experimental design and set of stimuli presented in the 2‐choice digging task: the texture of the outer bowl as the relevant dimension that gave cue for the reward, while the type of the digging media was the irrelevant dimension. B, Both experimental groups were able to complete all 7 stages of the task, however GlyT2‐Cdh13 ^−/− mice need more trials to complete the task (2‐way ANOVA, genotype P = .045, task P = .01; genotype × task P = .7). C, Cumulative performance across all stages before and after overtraining showing higher number of cumulative trials of GlyT2‐Cdh13 ^−/− mice compared to control mice (2‐way ANOVA, genotype P = .0453, task P = .0002; genotype × task P = .5). D, Effects of overtraining on cognitive performance, measured in terms of errors made to reach criterion: average of performance errors of control (left graph: control mice, P = .9, Mann‐Whitney U test) and GlyT2‐Cdh13 ^−/− mice (middle graph: GlyT2‐Cdh13 ^−/−, P = .0005, Mann‐Whitney U test) prior to and after overtraining and GlyT2‐Cdh13 ^−/− mice made more errors at IDS 2 stage, commenced directly after overtraining (right graph: P = .02, Mann‐Whitney U test). Data presented as mean ± SEM. Adult male mice, n = 6‐11 per group. See Table S5

Figure 7

GlyT2‐Cdh13 ^−/− mice exhibit increase in reciprocal social interactions with loss of preference for contact region. A, Pattern of contacts during 10 minutes of reciprocal social interaction in a novel arena of same pair genotype: nose‐nose, nose‐body and nose‐anogenital area. B, During 10 minutes of social interaction, pairs of GlyT2‐Cdh13 ^−/− mice display a cumulative preference to be in contact (P = .002, Mann‐Whitney U test) and took a shorter time to initiate contact (P = .0023, Mann‐Whitney U test). C, Evaluation of the number of contacts to nose, body and anogenital area shows a preference for contact to nose in pairs of control mice (P = .001, Kruskal‐Wallis test), but no preference for nose, body or anogenitcal in GlyT2‐Cdh13 ^−/− mice (P = .82, Mann‐Whitney U test) despite an increase in number of total contacts (see Table S4). D, Analysis of the first 3 of 10 minutes of social interaction revealed increased in the duration (P = .004, Mann‐Whitney U test) and number of contacts (P = .03, Mann‐Whitney U test) made by GlyT2‐Cdh13 ^−/− mice. E, In the resident‐intruder paradigm to assess for inter‐male aggression, no genotype differences was observed in latency to attack and duration of contact, including following and sniffing behavior. Data presented as mean ± SEM. Adult male mice, n = 10‐16 per group. See Table S6