Altered CSMD1 Expression Alters Cocaine-Conditioned Place Preference: Mutual Support for a Complex Locus from Human and Mouse Models.

Abstract

The CUB and sushi multiple domains 1 (CSMD1) gene harbors signals provided by clusters of nearby SNPs with 10-2 > p > 10-8 associations in genome wide association (GWAS) studies of addiction-related phenotypes. A CSMD1 intron 3 SNP displays p < 10-8 association with schizophrenia and more modest associations with individual differences in performance on tests of cognitive abilities. CSDM1 encodes a cell adhesion molecule likely to influence development, connections and plasticity of brain circuits in which it is expressed. We tested association between CSMD1 genotypes and expression of its mRNA in postmortem human brains (n = 181). Expression of CSMD1 mRNA in human postmortem cerebral cortical samples differs 15-25%, in individuals with different alleles of simple sequence length and SNP polymorphisms located in the gene's third/fifth introns, providing nominal though not Bonferroni-corrected significance. These data support mice with altered CSMD1 expression as models for common human CSMD1 allelic variation. We tested baseline and/or cocaine-evoked addiction, emotion, motor and memory-related behaviors in +/- and -/- csmd1 knockout mice on mixed and on C57-backcrossed genetic backgrounds. Initial csmd1 knockout mice on mixed genetic backgrounds displayed a variety of coat colors and sizable individual differences in responses during behavioral testing. Backcrossed mice displayed uniform black coat colors. Cocaine conditioned place preference testing revealed significant influences of genotype (p = 0.02). Homozygote knockouts displayed poorer performance on aspects of the Morris water maze task. They displayed increased locomotion in some, though not all, environments. The combined data thus support roles for common level-of-expression CSMD1 variation in a drug reward phenotype relevant to addiction and in cognitive differences that might be relevant to schizophrenia. Mouse model results can complement data from human association findings of modest magnitude that identify likely polygenic influences.