br Conclusion br Funding source br

Conclusion
Funding source
Ethical approval
Conflict of interest statement
Introduction Traditionally, schizophrenia (SZ) was thought to be associated with neuronal dysfunction; however, the current hypothesis that myelin and, specifically, oligodendrocytes are also involved in the development of SZ is gaining traction (Ren et al., 2013; Roussos and Haroutunian, 2014). In this regard, after screening the discoidin domain receptor 1 (DDR1) locus in DNA from 100 patients diagnosed with SZ, we identified an association of several genetic variants with the disease (Roig et al., 2007). The haplotype encompassing rs1049623, rs2267641 and rs2239518 was associated with SZ after adjusting for multiple testing (Roig et al., 2007). DDR1 is a membrane-anchored tyrosine kinase receptor whose unique ligand identified to date is collagen (Leitinger, 2014; Ullrich and Schlessinger, 1990; Vogel, 1999). The DDR1 gene is located at 6p21.3, near the major histocompatibility complex (MHC) region. This chromosomal region has been associated with SZ (Ripke et al., 2013) but most GWAS arrays used in SZ studies did not contain rare variants at the DDR1 locus such as rs2267641. Alternative splicing of DDR1 produces at least 5 isoforms, known as DDR1a to DDR1e (Alves et al., 2001). Isoforms DDR1a and DDR1b are the most abundant in several tissues, including blood (Leitinger, 2014). Detailed analysis of DDR1 protein in the adult human Cyclic di-GMP revealed that its expression pattern parallels that of the MBP, a classic white matter (WM) protein (Bàrbara Roig et al., 2010). DDR1c isoform expression is highly correlated with the myelin genes MAG and OLIG2 in the human brain (Barbar Roig et al., 2012b). In rodents, high oligodendrocyte expression of DDR1 parallels developmental myelination (Franco-Pons et al., 2006) and remyelination after experimental demyelination (Franco-Pons et al., 2009). SNP rs2267641, located in a heterogeneous nuclear ribonucleoprotein A2 response element (A2RE) sequence in DDR1 exon 14, influences alternative splicing and the concentrations of the isoforms DDR1b and DDR1c (Barbara Roig et al., 2012b). In the same study, SNP rs1264323, which is in linkage disequilibrium (LD) with rs1049623, was also found to influence the expression of isoforms DDR1b and DDR1c (Barbara Roig et al., 2012b). Moreover, the abundance of the DDR1c isoform was significantly elevated in brain dorsolateral prefrontal cortex (DLPFC) tissue from patients with SZ compared to controls (Bàrbara Roig et al., 2012a). Diffusion tensor imaging (DTI) has made it possible to investigate microstructural WM abnormalities in vivo (Garin-Muga and Borro, 2014). On the basis of DTI data, several studies reported abnormal WM microstructure in patients with chronic and first-episode psychosis (Andreasen et al., 2011; Bora et al., 2011; Pomarol-Clotet et al., 2010; Tamnes and Agartz, 2016) and even in those who were naïve to antipsychotic drugs (Alvarado-Alanis et al., 2015; Zeng et al., 2016; Zhang et al., 2016). Differential expression levels of myelin-related genes, such as NRG1, ERBB4, DISC1, RTN4R, OLIG2, CNP and MAG, in brain tissue from subjects with SZ also revealed molecular WM abnormalities, matching the results found by DTI (Roussos and Haroutunian, 2014; Takahashi et al., 2011; Voineskos, 2015). The abnormal expression of myelin-related genes in SZ was most pronounced in the DLPFC, hippocampus, superior temporal cortex and cingulate gyrus (Höistad et al., 2009; Katsel et al., 2005). Only a few reports have demonstrated a relationship among gene variants, WM tract integrity, and cognitive performance. For instance, genetic variability at MAG, OLIG2, and CNP influenced cognitive performance in a manner mediated by the integrity of WM fiber tracts in patients with SZ (Voineskos et al., 2013); more recently, Poletti et al. showed the influence of the COMT Val158Met polymorphism on the association between cognitive function and WM microstructure (Poletti et al., 2016). Our present hypothesis is that SNP variants in the DDR1 locus confer susceptibility to SZ in association with WM microstructure variation and neurocognitive deficits such as processing speed (PS) (Karbasforoushan et al., 2015). To test this hypothesis, we designed 3 different but interrelated studies to achieve three main aims: in Study 1, we aimed to replicate the association between DDR1 variants and SZ; in Study 2, we aimed to determine whether these DDR1 variants influence PS; and in Study 3, we assessed whether there is a link between WM microstructure, PS and DDR1 genotype.