Neuropsychological deficits in mice depleted of the schizophrenia susceptibility gene CSMD1

Abstract

Recent meta-analyses of schizophrenia genome-wide association studies (GWASs) have identified the CUB and SUSHI multiple domains 1 (CSMD1) gene as a statistically strong risk factor. CSMD1 is a complement control-related protein suggested to inhibit the classical complement pathway, being expressed in developing neurons. However, expression of CSMD1 is largely uncharacterized and relevance for behavioral phenotypes is not previously demonstrated. Here, we assess neuropsychological behaviors of a Csmd1 knockout (KO) mouse in a selection of standard behavioral tests. Deregulation of neuropsychological responses were observed in both the open field and the elevated plus maze tests, in which KO mice spent 55% and 33% less time than WT littermate mice in open areas, respectively. Altered behaviors were also observed in tail suspension and to higher acoustic stimuli, for which Csmd1 KO mice showed helplessness and moderate increase in startle amplitude, respectively. Furthermore, Csmd1 KO mice also displayed increased weight-gain and glucose tolerance, similar to a major phenotype of the metabolic syndrome that also has been associated to the human CSMD1 locus. Consistent with a role in the control of behaviors, Csmd1 was found highly expressed in the central nervous system (CNS), and with some expression in visceral fat and ovary, under tissue-specific control by a novel promoter-associated lncRNA. In summary, disruption of Csmd1 induces behaviors reminiscent of blunted emotional responses, anxiety and depression. These observations suggest an influence of the CSMD1 schizophrenia susceptibility gene on psychopathology and endophenotypes of the negative symptom spectra.

Figure 1. Csmd1 RNA and protein expression in Csmd1 knock-out and wild-type mice.

(A) Schematic representation of the KO-strategy. A 1 kb genomic region (white lines) of exon1/intron1 was replaced with a selection cassette (grey box). (B) Expression of Csmd1 mRNA measured by QPCR in an adult mouse tissue panel. Csmd1 is predominantly expressed in brain tissues as compared to peripheral tissues. The highest expression level was identified in areas of the cortex. (C) Depletion of Csmd1 mRNA in the cortex was documented by two exon-exon specific QPCR assays. Transcription of exon 1–2 was depleted, while about 20% residual expression could be observed when amplifying exon 32–33. KO mice lacked a protein band of expected size (389 KDa, arrow), as demonstrated by immunoblotting. Signals of lower molecular weight are indicated (a and b). (D) Mapping of RNA-seq reads to the Csmd1 locus. RNA sequencing of cortex is shown for 4wild-type (green) and 4 Csmd1 KO (red) mice (transcript scale: 0–150 reads). Coverage signals of modified nucleosomes (H3K4me3, H3K4me1 and H3K27Ac) and polymerase-2 binding profiles are shown for the mouse cortex. The 1 kb deleted sequence of Csmd1 is highlighted in yellow (upper panel) and blue (lower panel). No RNA reads were mapped to the deleted genomic region in the KO mice. Abbreviations: Cx, cortex; VCx, visual cortex; FCx, frontal cortex; Hipp, hippocampus; Hyp, hypothalamus; Ob, olfactory bulb; Cer, cerebellum; Visc. Fat, visceral fat.

Figure 2. Transcriptome analyses by RNA sequencing of cortex from Csmd1 KO and WT mice.

(A) Heat map of RNA-seq expression levels, as measured by rpkm, mapping to individual exons and introns of Csmd1. The arrow indicates the orientation of exons and introns in their increasing number. (B) Analysis of differentially expressed RNAs. Significantly up-regulated and down-regulated transcripts are represented by red dots. Y-axis indicates the log2 (fold change).

Figure 3. Expression of Csmd1 promoter-associated long non-coding RNA (pas-lncRNA) in mouse tissues.

(A) Map of RNA sequencing reads aligned to the promoter region of Csmd1. Data from individual Csmd1 KO (N = 4; red lines) and WT (N = 4; green lines) mice are shown. The novel pas-lncRNA RNA (black thick line) is expressed antisense to the Csmd1 promoter sequence (transcript scale: 0–30 reads). (B) Heat map representing the relative expression level of pas-lncRNA and Csmd1 mRNA in peripheral and CNS tissues of Csmd1 KO and WT mice, respectively. Expression values of each transcript are calculated relative to their respective expression level in cortex of WT mice. pas-lncRNA expression was induced in the CNS but not in peripheral tissues of Csmd1 KO mice. Fold change values (FC) and statistical significances are listed for each tissue. (C) Co-regulated expression of Csmd1 and pas-lncRNA in the developing cortex and cerebellum (Csmd1:pas-lncRNA expression correlation coefficient, cortex: r² = 0.92). Y-axis represents relative expression level of RNA. X-axis indicates the postnatal day. Abbreviations: asterisk, t-test P-value<0.05; n.s.; not significant.

Figure 4. Behavior of Csmd1 KO and WT mice in the open field arena.

(A) Total time spent in the center of the arena is shown for KO and WT mice, respectively. (B) Sequential time bin analysis demonstrates that WT mice adapt to the test arena after the first time bin, while KO mice avoid the center throughout the test period. (C and D) Comparison of total path length and sequential bin analysis of path length demonstrate no statistically significant difference between KO and WT mice. Abbreviations: n.s., not significant; asterisk, statistically significant P-value<0.05; s, seconds.

Figure 5. Behavior of Csmd1 KO and WT mice in the elevated plus maze.

(A) Analysis of total time spent on open arms demonstrate significant less time for KO mice (N = 13) as compared to WT mice (N = 8). (B) Compiled tracks from all mice show that Csmd1 KO mice avoid entering open arms, as opposed to WT mice traveling over the entire test arena. Abbreviations: asterisk, statistically significant (P-value<0.05); s, seconds.

Figure 6. Behavior of Csmd1 KO and WT mice to acoustic stimuli and tail suspension.

(A) Startle responses of Csmd1 KO mice in response to acoustic stimuli in the range of 80–120 dB, as compared to WT mice. The startle baseline was similar in male and female WT mice. A marginal increase in startle responses could be observed for higher acoustic stimuli in both genders of Csmd1 KO mice, reaching statistical significance when analyzing all mice together (genotype-group interaction P-value<0.05). (B) There was no difference in amplitude response between KO and WT mice in the degree of inhibition after pre-pulse inhibition. (C) Tail suspension test demonstrated longer accumulated time immobile for KO mice as compared WT mice (P-value<0.05). Borderline statistical significance was observed for male Csmd1 KO mice as compared to WT mice (P-value = 0.1).

Figure 7. Analysis of object recognition memory in Csmd1 KO and WT mice.

(A and B) Object contacts in the first and second trial of exposure to novel and familiar objects demonstrated increased contact counts for Csmd1 KO mice (P-value<0.05). (B and C) Discrimination and preference analysis demonstrated no effect of Csmd1 on object recognition. Abbreviation: n.s., not significant; asterisk, statistically significant (P-value<0.05).