Background
Schizophrenia (SCZ) is a severe, highly heritable and heterogeneous disease characterized by positive symptoms (e.g., hallucinations, delusions), negative symptoms (e.g., apathy, lack of emotion, poor social functioning), and cognitive deficits [
1]. It affects approximately 0.5–1% of the world population and accounts for 1.1% of the total disability-adjusted life years [
2]. Moreover, onset of SCZ typically manifests during late adolescent and early adulthood [
3]. So far, the pathophysiology of SCZ has not been fully understood due to the phenotypic and psychopathological complexity and heterogeneity [
4]. The neurodevelopmental model of schizophrenia posits that early neurodevelopmental abnormalities of brain development may have a role in SCZ [
5‐
8]. According to the neurodevelopmental hypothesis, schizophrenia is mainly caused by genetic components, which affects prenatal and postnatal neurodevelopment. Environmental events increase the risk for phenotype expression of schizophrenia among these genetically susceptible individuals [
5]. For example, fetal hypoxia was associated with significant decreases in gray matter density among schizophrenia patients and their healthy siblings, but not non-familial controls [
9]. Moreover, substantial evidence has linked neurodevelopmental insults to a series of substantial risk genes for schizophrenia [
10]. For example, the schizophrenia risk gene,
NRGN, bidirectionally modulates synaptic plasticity via regulating the neuronal phosphoproteome [
11]. Evidence points to the role of
DISC1 in regulation of intracellular trafficking of a wide range of neuronal cargoes [
12]. Furthermore, the
C4A expression and structural variation have been found associated with neurodevelopment in schizophrenia [
13].
Given the importance of genetic influences on schizophrenia susceptibility, there are increasingly well-powered genomic studies on identifying disease-related variants and loci [
14,
15]. For example, the Schizophrenia Working Group of the Psychiatric Genomics Consortium reported 108 independent risk loci based on a multi-stage genome-wide association study (GWAS) in ~150,000 individuals [
16]. More recently, using GWAS, Lam and colleagues compiled the largest East Asian genetics cohort and identified 208 significant associations in 176 genetic loci between East Asian and European ancestries, suggesting consistency of schizophrenia risk alleles across ethnicities and cultures [
17]. As accumulating genomic variations are found to be associated with SCZ, genetic risk factors likely explain the common clinical and etiological features of schizophrenia (e.g., brain expression changes and morphological impairment in gray matter) [
18]. Therefore, identifying potential risk loci and elucidating how they affect schizophrenia pathogenesis will provide important knowledge about the pathophysiology of SCZ. On the other hand, approximately 90% of the single-nucleotide polymorphisms (SNPs) or variants identified by GWAS were located in a noncoding region [
19]. One explanation is that noncoding variants may exert functional impacts through modulating mRNA expression of nearby or distal genes [
20]. For example, through integrating expression quantitative trait loci (eQTL) and GWAS, Zhang and colleagues [
21] showed that SCZ risk variant rs2071287 might confer SCZ risk by modulating
NOTCH4 expression. Consistently,
NOTCH4 was significantly downregulated in schizophrenia patients compared with controls [
21]. Integrative approaches (such as Sherlock integrative analysis, a Bayesian approach that combining genetic associations from GWAS and human brain eQTL data) provide valuable insights into the gene regulatory mechanisms of schizophrenia [
22]. For instance, a recent report identified
LRP8 as a schizophrenia risk gene by integrative analysis [
23]. In addition, network analysis showed that LRP8 directly participates in a highly interconnected protein-protein interaction network built by top risk genes for SCZ [
24].
Previous studies have linked tRNA modification and metabolic abnormalities to neurodevelopmental disorder [
25,
26].
TYW5 (tRNA-YW Synthesizing Protein 5) is a major tRNA hydroxylase involved in epigenetic modification in brain [
27‐
29]. Some researchers have shown a link between
TYW5 and mental illnesses such schizophrenia [
15,
30], whereas others have been unable to corroborate this association [
31,
32]. More studies are needed to determine whether
TYW5 is a risk gene for schizophrenia. More importantly, little is known about how
TYW5 genetic variations confer schizophrenia susceptibility or the role of
TYW5 in schizophrenia pathophysiology. Here, we employed Sherlock and Summary data-based Mendelian Randomization (SMR) integrative analysis integrate disease associations and eQTL signatures in GWAS loci, to discover
TYW5 as a SCZ risk gene, which is likely to play a role in SCZ pathogenesis. We also examined the expression level of
TYW5 in the dorsolateral prefrontal cortex of schizophrenia cases and controls using expression data. Furthermore, we explored the potential role of
TYW5 in schizophrenia pathogenesis using induced pluripotent stem cells. Our findings support that
TYW5 is a schizophrenia risk gene whose expression may be regulated by schizophrenia GWAS SNPs. Finally, we also provided evidence for an association between
TYW5 and gray matter volume abnormalities in the frontal-partial regions, suggesting the potential pathophysiological role of
TYW5 in SCZ.
Discussion
Considering that most of the identified risk variants are located in the noncoding region, it is likely that these risk variants impart SCZ risk through altering gene expression. As the GWAS findings alone cannot predict whether the discovered SCZ risk variants have functional repercussions, a statistical method to combine data from all disease associations and independent expression QTL data is required [
55]. In the present study, we utilized the largest GWAS of SCZ (PGC EAS+EUR) to date and conducted the genome-wide integrative analyses through combining brain eQTL, followed by independent replications in differential expression analysis in DLPFC and hiPSC neurons. Through this stepwise analysis, we found that
TYW5 is a new risk gene for SCZ. Besides, the risk allele of rs203772 (which predicts higher
TYW5 mRNA expression in the DLPFC) was also associated with the SCZ-relevant middle frontal gyrus and precuneus volumes in first-episode untreated samples. Our independent integrative analyses results provide convergent evidence to support the potential role of
TYW5 in SCZ.
TYW5 is an essential tRNA hydroxylase, and previous studies have found that tRNA alteration defects are linked to many neurodevelopmental disorders [
25,
56]. Accumulating evidence shows that
TYW5 is one of the best replicated SCZ risk genes [
15,
57‐
61]. However, due to the high level of linkage disequilibrium, it is challenging to precisely locate the disease-causing gene in a small sample [
15,
30]. In our study, the rs203772 in
TYW5 showed strong association with susceptibility to SCZ (
P = 1.83 × 10
−8), supporting that it is a true risk gene for SCZ. In addition,
TYW5 may also be involved in the genetic susceptibility of mental illness that shares the common risk factor with SCZ [
62,
63], such as neurodevelopment-related disorders autism spectrum disorder [
63] and bipolar disorder [
62]. These lines of evidence provide convergent evidence to support that
TYW5 may represent an authentic susceptibility gene for SCZ [
64]. In order to explore whether the
TYW5 identified by Sherlock integrative analysis were dysregulated in patients with SCZ, we also examined the brain expression level of
TYW5 in SCZ cases. We found that
TYW5 was significantly upregulated in DLPFC in patients with schizophrenia, which also suggests that
TYW5 may act as a potential therapeutic target of SCZ.
Although the pathophysiology of neuronal cell-specific damage in SCZ remains unclear, cortical neuronal abnormalities in SCZ have received extensive attention [
65,
66]. The potential effects of
TYW5 on neuronal or synaptic function remain unclear [
62,
67]. However,
TYW5 protein was likely expressed in cortical neurons during the process of synapse formation [
67]. Our study found that
TYW5 is expressed in a variety of cell types in the human cerebral cortex [
68], with the highest expression level in neurons and astrocytes. Our results indicate that
TYW5–associated tRNA lesions in neurons and astrocytes may be one of the potential pathological changes of SCZ [
69‐
71]. Next, we explored the differential expression of
TYW5 in induced pluripotent stem cells (iPSCs) and cortical interneurons in SCZ patients and healthy controls. We found that
TYW5 is significantly increased in cortical interneurons [
72,
73], which is consistent with the dysregulated pattern in the DLPFC, suggesting that
TYW5 plays an important regulatory role in the development of cortical neuronal in SCZ [
74]. In addition, our study also provides an opportunity to study the specific role of
TYW5 in the development of neural stem cells [
75,
76].
Earlier studies have revealed abnormal neuronal differentiation, reduced synapse density, and abnormal expression of synaptic markers in the frontal lobe of SCZ patients [
77]. Besides, in vitro and in vivo studies also discovered that genetic risk factors of SCZ usually result in disruption of synaptic morphology and function as well as brain circuits that are essential for positive symptoms and cognition, and thereby eventually lead to the onset of SCZ [
78,
79]. Therefore, it is widely accepted that genetic factor in frontal dysfunctions play pivotal roles in the pathogenesis of SCZ [
80]. The recent SCZ GWAS also supports this view, because genes involved in frontal eQTL and differential gene expression have been repeatedly emphasized [
22,
81]. Our results indicate that the risk allele of rs203772 was associated with higher
TYW5 expression in the DLPFC. These results suggested the idea that the SCZ GWAS locus near rs203772 may confer risk of SCZ by regulating the expression level of the
TYW5 gene in frontal lobe brain tissue [
82]. Furthermore, for the first time, by using neuroimaging results obtained from human subjects and ruled out the influence of drug confounding factors, we explored the genetic effects of
TYW5 on the entire brain gray matter with data-driven strategy. We found that risk allele of rs203772 (G) was associated with two special frontal sub-regions: the right middle frontal gyrus and left precuneus gray matter in first-episode schizophrenia. One of the possible functional mechanisms is that
TYW5 acts as a downstream regulator of the iron distribution pathway during normal and oncogenic neurodevelopment and may regulate the dopamine transporter by regulating the distribution of iron in the frontal lobe [
83,
84]. From the perspective of neurodevelopmental function,
TYW5 may affect the development and function of the prefrontal cortex involved in the abnormal cognitive process of SCZ [
15,
30,
85]. In addition,
TYW5 participates in mitochondrial biogenesis through interaction with methionyl-tRNA synthase 2, which may also be related to mitochondrial abnormalities related to frontal lobe development in the pathogenesis of SCZ [
86]. Although the exact function of
TYW5 in this brain function is not yet clear, more functional studies are still urgently needed to gain insights on whether and how it affects brain circuits and behavior in disease-specific ways [
87‐
89].
In addition to identify
TYW5 as a SCZ risk gene, other evidence also support that
TYW5 may play important roles in the central nervous system. Recent studies have also shown expression dysregulation of
TYW5 in cancer [
90], including testicular germ cell tumors [
91]. Also, studies have shown that
TYW5 regulates migration, invasion, and tumor cell proliferation [
92]. These studies demonstrated the essential role of
TYW5 in cancer. Fascinatingly, schizophrenia has been reported to be a risk factor in cancer prognosis [
93]. Also, earlier research suggested that SCZ prevalence was higher relative to cancer patients’ general population [
94,
95].
Our study has several advantages. First, we used different integrative methods (by integrating eQTL and GWAS data) to discover and verify
TYW5 as a potential SCZ risk gene. In addition, protein integrative analysis using a large and comprehensive human proteome [
96] and summary statistics from the most recent SCZ GWAS supported
TYW5 as a SCZ potential therapeutic targets. Finally, we found that rs203772 is also associated with gray matter abnormalities of the right middle frontal gyrus and left precuneus. The whole-brain imaging approach used in this work prevents empirical pre-selection of brain areas, and the first-episode, treatment-naive individuals avoid pharmaceutical confounding effects [
97].
While this study offers some interesting observations, it should be noted that the present evidence is limited, and we interpret the results cautiously. First of all, due to the complexity of linkage disequilibrium and gene regulation, the causal (or functional) variants that regulate the expression of
TYW5 and the exact regulatory mechanism remain elusive. Second, different eQTL datasets might offer different results and refined single-cell expression data will promote the identified risk gene’s authenticity [
98]. Third, our Sherlock analysis identified multiple genes whose expression disruption could play a role in SCZ; however, in this study we only focused on the top significant
TYW5 after multiple corrections. More work is required to illuminate whether other risk genes identified by Sherlock integrative analysis also have a role in SCZ. Furthermore, we would replicate the association between the rs203772 genotypes and grey matter abnormalities in an independent patient cohort in future research. Finally,
TYW5 expression modulation in related (eventually patient) cell lines or animal models will provide further evidence for the potential role of
TYW5 in schizophrenia.
Acknowledgements
Data were generated as part of the CommonMind Consortium supported by funding from Takeda Pharmaceuticals Company Limited, F. Hoffman-La Roche Ltd and NIH grants R01MH085542, R01MH093725, P50MH066392, P50MH080405, R01MH097276, RO1-MH-075916, P50M096891, P50MH084053S1, R37MH057881 and R37MH057881S1, HHSN271201300031C, AG02219, AG05138, and MH06692. Brain tissue for the study was obtained from the following brain bank collections: the Mount Sinai NIH Brain and Tissue Repository, the University of Pennsylvania Alzheimer’s Disease Core Center, the University of Pittsburgh NeuroBioBank and Brain and Tissue Repositories and the NIMH Human Brain Collection Core. CMC Leadership: Pamela Sklar, Joseph Buxbaum (Icahn School of Medicine at Mount Sinai), Bernie Devlin, David Lewis (University of Pittsburgh), Raquel Gur, Chang-Gyu Hahn (University of Pennsylvania), Keisuke Hirai, Hiroyoshi Toyoshiba (Takeda Pharmaceuticals Company Limited), Enrico Domenici, Laurent Essioux (F. Hoffman-La Roche Ltd), Lara Mangravite, Mette Peters (Sage Bionetworks), Thomas Lehner, Barbara Lipska (NIMH). The data available in the AD Knowledge Portal would not be possible without the participation of research volunteers and the contribution of data by collaborating researchers. We thank the participants of the ROS, MAP, Mayo, Mount Sinai Brain Bank, and Banner Sun Health Research Institute Brain and Body Donation Program for their time and participation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.