RNA-seq analysis of amygdala tissue reveals characteristic expression profiles in schizophrenia

The amygdala brain region has been implicated in the pathophysiology of schizophrenia through emotion processing. However, transcriptome messages in the amygdala of schizophrenia patients have not been well studied. We used RNA sequencing to investigate gene-expression profiling in the amygdala tissues, and identified 569 upregulated and 192 downregulated genes from 22 schizophrenia patients and 24 non-psychiatric controls. Gene functional enrichment analysis demonstrated that the downregulated genes were enriched in pathways such as 'synaptic transmission' and 'behavior', whereas the upregulated genes were significantly over-represented in gene ontology pathways such as 'immune response' and 'blood vessel development'. Co-expression-based gene network analysis identified seven modules including four modules significantly associated with 'synaptic transmission', 'blood vessel development' or 'immune responses'. Taken together, our study provides novel insights into the molecular mechanism of schizophrenia, suggesting that precision-tailored therapeutic approaches aimed at normalizing the expression/function of specific gene networks could be a promising option in schizophrenia.


Analysis of undifferentiated, disorganized, and paranoid subtypes of schizophrenia
We also calculated the differentially expressed genes between three schizophrenia subtypes (undifferentiated, disorganized and paranoid), and controls respectively. A total of 41 genes were significant after multiple testing corrections by all the three methods in undifferentiated schizophrenia, including GBP1, HBA2, IFITM1, IFITM3. In comparison, 12 genes were identified in disorganized schizophrenia including HBA1 and HBA2.
However, no gene was consistently significant after multiple testing corrections across different methods in paranoid schizophrenia, which may be due to the small number of cases.
Similarly, we also analyzed nominally significant genes identified by the three methods. This analysis revealed 798, 616 and 454 differentially expressed genes in the undifferentiated, disorganized and paranoid schizophrenia, respectively (Table S8-10). The similarity between the groups of differentially expressed genes was evaluated by Jaccard index, demonstrating that the overlap of differentially expressed genes was higher between undifferentiated and disorganized schizophrenia (Table S11).

Differentially expressed genes among subtypes of schizophrenia
We also analyzed the differentially expressed genes between each of the schizophrenia subtypes. No gene was significant after multiple testing corrections likely due to the small number of cases. However, 325, 293 and 98 nominally significant genes were identified from comparisons between undifferentiated-paranoid, disorganized-paranoid and undifferentiateddisorganized analyses, respectively (Table S15-17). Among them were genes such as ERBB4, GRIN2D, GRM3 and NTRK2 that have been previously shown to be associated with the pathogenesis of schizophrenia.
Interestingly, ERBB4, GRIN2D and NTRK2 are up-regulated in paranoid schizophrenia, while down-regulated in undifferentiated and disorganized schizophrenia. GRM3 is up-regulated in disorganized and paranoid schizophrenia, while down-regulated in undifferentiated schizophrenia.
CNTN2 is highly up-regulated in disorganized schizophrenia, while downregulated in paranoid schizophrenia ( Figure S7). Tables   Table S1. Description of postmortem samples used in this study (included as a separate excel file). "+++": significant by all the three methods after multiple testing. "++": significant by all the three methods without multiple testing. "+":significant by at least one method after multiple testing. Reference: (1-13). Table S4. 761 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 22 schizophrenia cases and 24 controls (included as a separate excel file).    Table S8. 798 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 9 undifferentiated schizophrenia cases and 24 controls (included as a separate excel file). Table S9. 616 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 7 disorganized schizophrenia cases and 24 controls (included as a separate excel file). Table S10. 454 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 5 paranoid schizophrenia cases and 24 controls (included as a separate excel file).  Table S12. Functional enrichment of differentially expressed genes in differentiated schizophrenia.

Expression
Term Fold Enrichment P-value Benjamini Q-value Table S15. 325 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 9 undifferentiated schizophrenia and 5 paranoid schizophrenia patients (included as a separate excel file). Table S16. 293 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 7 disorganized schizophrenia and 5 paranoid schizophrenia patients (included as a separate excel file). Table S17. 98 genes showed P-value < 0.05 (without multiple testing correction) by all three methods in the analysis of 9 undifferentiated schizophrenia and 7 disorganized schizophrenia patients (included as a separate excel file). Figure S1. Venn diagram of differentially expressed genes among three methods. a. significant genes after multiple testing correction.

Supplementary Figures
b. significant genes without multiple testing correction.