Introduction

Schizophrenia, a severe neuropsychiatric disorder with complex etiology, has a worldwide lifetime prevalence of ~1% and remains a significant public health problem.1 Among its multiple etiological hypotheses, the importance of genetic components in addition to other developmental and environmental influences has been consistently demonstrated by family, twin and adoption studies.2, 3, 4 From early linkage and association analyses5, 6, 7, 8 to recent large-scale genome-wide association studies (GWAS),9, 10, 11, 12, 13, 14, 15, 16, 17, 18 multiple risk candidates with strong evidence have been identified. Particularly, given that the GWAS approach provides opportunities to study common genetic variations across the entire genome without any a priori hypotheses, many novel but seemingly less relevant risk variants with functions beyond the known disease biology have been discovered.19 One example is the first genetic variant that achieved genome-wide level of statistical significance in schizophrenia GWAS: rs1344706 in zinc-finger protein 804A (ZNF804A).9 This gene was subsequently studied with replicative and explorative analyses, which have enhanced our understanding of its involvement in schizophrenia.20, 21, 22

However, despite the substantial genetic evidence and preliminary understanding of the molecular and neuronal mechanisms of the ZNF804A risk allele in schizophrenia susceptibility, a systemic synthesis of the current evidence to depict the potential use of this gene in future prevention and therapies is lacking. For this purpose, we have examined recent progresses on the roles of ZNF804A in schizophrenia biology revealed by genetic, molecular biology and neuroscience research. Potential mechanisms for disease progression and recommendations for future research are proposed.

Identification of a genome-wide significant variant rs1344706 in ZNF804A for schizophrenia in European populations

In 2008, O’Donovan et al.9 performed a GWAS in a UK sample of 479 schizophrenia patients and 2937 controls followed by replication analyses of top risk loci in 16 726 additional subjects. This GWAS study found that a single-nucleotide polymorphism (SNP) rs1344706 (in the intron 2 of ZNF804A) exhibited the strongest statistical association with schizophrenia in a meta-analysis of all samples (P=1.61 × 10–7 in 7308 cases and 12 834 controls). The significance was further enhanced when both schizophrenia and bipolar disorder were considered as the associated phenotypes (P=9.96 × 10–9 in 9173 cases and 12 834 controls).9 In addition to this GWAS, several following replication studies each provided independent support for rs1344706 in populations of European ancestry (Table 1). For example, Steinberg et al.21 observed a nominal significant association of rs1344706 with schizophrenia and bipolar disorder in large-scale samples mainly from Europe (P=0.00065 in 5164 schizophrenia patients, 609 cases with bipolar disorder and 20 709 controls); however, they showed a relatively smaller effect size compared with that of O’Donovan’s GWAS,9 probably due to the ‘winner’s curse’ effect.23, 24, 25 In addition, Riley et al.20 replicated the association in an Irish sample of 1021 schizophrenia cases and 626 controls (P=0.0227). The association between rs1344706 and schizophrenia was further confirmed by Williams et al.22 through a meta-analysis in 21 274 cases (schizophrenia and bipolar disorder) and 38 675 controls (P=4.10 × 10–13). Finally, in the latest PGC2 GWAS of schizophrenia (including 34 241 cases and 45 604 controls at the discovery stage), rs1344706 was genome-wide significantly associated with the disease (P=1.27 × 10–10).10 These cumulative data provided strong and consistent evidence that rs1344706 is an authentic risk variant for schizophrenia in European populations, making it a highly promising candidate for future studies.26

Table 1 Published association results for rs1344706 with schizophrenia in Europeans
Full size table

Cis-acting effects of the genome-wide significant variant rs1344706 in ZNF804A

The discovery of the schizophrenia risk SNP rs1344706 in ZNF804A is intriguing. However, the molecular mechanism by which the risk allele contributes to disease etiology remains to be determined. It was hypothesized that schizophrenia risk alleles were enriched in brain gene expression quantitative trait loci (eQTL).27, 28 As such, rs1344706 might affect the expression of ZNF804A or even other genes. Based on this idea, researchers have conducted eQTL analyses to investigate the risk mechanisms of rs1344706.29, 30 In a small sample set, Hill et al.29 used allelic expression assays and found that the rs1344706 genotype had a significant effect on ZNF804A allelic expression in second trimester fetal brain, with the schizophrenia risk (A) allele predicting reduced ZNF804A gene-level expression. In the subsequent independent study with a much larger sample, including healthy controls and psychiatric patients, Tao et al.30 identified a previously uncharacterized, brain abundantly expressed and developmentally regulated truncated ZNF804A transcript (ZNF804AE3E4) using next-generation sequencing and PCR-based methods (Figure 1a). Interestingly, they found that rs1344706 influenced expression of ZNF804AE3E4 mRNA only in fetal brain, and the risk (A) allele indicated reduced expression.30 ZNF804AE3E4 expression was also significantly reduced in patients with schizophrenia compared with healthy controls,30 suggesting that lower expression of this transcript was likely a risk factor. These studies converge on the conclusion that schizophrenia risk SNP rs1344706 has a cis-acting effect on ZNF804A expression in the brain during the fetal age, a critical period in neurodevelopment and probably schizophrenia onset.

Figure 1
figure 1

(a) Schematic of mRNA structure of ZNF804A isoform. (b) Schematic of domain structure of predicted ZNF804A protein isoforms. NLS, nuclear localization sequence.

PowerPoint slide

Full size image

Although various studies in addition to these two have also demonstrated cis-effects of rs1344706 on ZNF804A expression in brains,22, 29, 30, 31, 32, 33 the direction of allelic effects remains controversial. In addition, given that rs1344706 is an intronic SNP, the molecular mechanisms causing altered ZNF804A expression by the risk allele remain poorly understood. Although Hill et al.34 proposed that rs1344706 was a functional polymorphism that affected DNA–protein interaction through an electrophoretic mobility shift assay, further validation studies using ChIP-Seq and reporter gene assays are still lacking. In addition, other possibilities, that is, rs1344706 altered the secondary structure of ZNF804A mRNA, or was in linkage disequilibrium (LD) with the causative mutations, cannot be excluded.

Biological functions of ZNF804A

Given that the strong association between rs1344706 in ZNF804A and schizophrenia is reproducible and compelling, the possible roles of ZNF804A in schizophrenia pathogenesis have been intensively studied to provide novel insights into disease biology. The progress to date remains limited, but researchers have gained some hints through recent efforts. Consisting of four exons and transcribing a protein of 1210 amino acids, ZNF804A is expressed in the brain and contains a C2H2-type domain associated with the zinc-finger protein family (Figure 1b). However, the exact function of its protein product is currently unknown. Previous studies of proteins with this zinc-finger domain originally identified these proteins as DNA-binding molecules with a role in transcription. However, these proteins exhibit diverse interactions with numerous other molecules, including RNA and proteins.35, 36, 37, 38, 39 As a result, ZNF804A likely has pivotal roles in cell physiology. Indeed, knockdown of ZNF804A in human neural progenitor cells (hNPCs) or in developing neurons derived from human-induced pluripotent stem cells resulted in altered expression of genes involved in cell adhesion, neurite outgrowth, synapse formation and cytokine signaling.40, 41 On the other hand, overexpression of ZNF804A in rat neural progenitor cells affects the expression of several genes associated with schizophrenia.42 This latter study further localized the rat homolog of ZNF804A, zfp804A, to the nucleus of rat neural progenitor cells, supporting its potential roles in gene expression regualtion.42

Recent studies on zfp804a have provided additional information regarding its function. For example, zfp804a has recently emerged as a target for HOXC8,43 a major transcription factor known to promote nerve growth, suggesting its involvement in the regulation of early neurodevelopment. Further, Hinna et al.44 demonstrated that zfp804a expression was increased in the rat brain at the time of birth, coinciding with neuronal differentiation. They also showed that zfp804a was localized to growth cones of growing neurites, implicating the role of zfp804a in growth cone function and neurite elongation.44 Moreover, Chang et al.45 identified changes in zfp804a expression in the rat hippocampus, frontal cortex, and thalamus across postnatal neurodevelopment, implying its possible roles in the development of these brain regions that are substantially affected in schizophrenics. These lines of evidence implicate the potential importance of ZNF804A (and zfp804a) in neurodevelopment, although its impact on neuronal phenotypes and mechanisms in psychiatric disease development were unclear until a recently published study.46

This landmark study was conducted to characterize the neuronal function of ZNF804A by Deans et al.46 They first examined the subcellular localization of ZNF804A protein in diverse neurons derived from hNPCs or human-induced pluripotent stem cells or in primary rat cortical neurons. They showed that endogenous ZNF804A protein localized to somatodendritic compartments and colocalized with the putative synaptic markers (PSD-95) in developing neurons derived from hNPCs and human-induced pluripotent stem cells. In mature rat neurons, zfp804a was present in a subset of dendritic spines and colocalized with synaptic proteins in specific nanodomains. These data suggested that ZNF804A affected local signaling during neural development rather than having general gene expression regulatory roles in neurons. They then analyzed its functions in neurite formation, maintenance of dendritic spine morphology and responses to activity-dependent stimulations. Interestingly, young neurons with suppressed expression of ZNF804A exhibited attenuated neurite outgrowth potentially through reducing neuroligin-4 expression. In mature rat neurons, knockdown of zfp804a resulted in loss of dendritic spine density and impaired responses to activity-dependent stimulation.46 Obviously, ZNF804A (zfp804a) has a functional role in synaptic development, probably via influencing local signaling of neuron synapses. To the best of our knowledge, this study for the first time described detailed localization of ZNF804A, and proposed the compelling hypothesis that this protein contributed to psychiatric disorders though alteration of neuronal and synaptic structures.

Overall, these studies found that the risk allele at rs1344706 is associated with reduced ZNF804A mRNA in human brain, indicating its decreased expression as a likely risk factor for schizophrenia. Indeed, reduced ZNF804A expression in neurons resulted in aberrant neurite growth and loss of dendritic spine density, which was consistent with the clinical observations in the brains of schizophrenia patients.47, 48 The defined effects of ZNF804A on neurite growth and dendritic spines were also consistent with the previous findings that schizophrenia susceptibility genes often affect synaptic development and functions.49, 50, 51 The hypothesized potential biological mechanism underlying the genetic risk of rs1344706 in schizophrenia and the logic of evidence are summarized in Figure 2.

Figure 2
figure 2

A hypothesized biological mechanism for rs1344706 in schizophrenia pathogenesis, hNPC, human neural progenitor cell; hiPSC, human-induced pluripotent stem cell; NLGN4, neuroligin-4; ZNF804A, zinc-finger protein 804A.

PowerPoint slide

Full size image

Neuroimaging studies of ZNF804A

In addition to investigating the physiological outcome of genetic modification of ZNF804A in cell cultures and animals, analyzing changes in regional brain structures and functions may provide more information regarding the biological roles of ZNF804A in schizophrenia. In fact, these brain structural alterations, which are named intermediate phenotypes, are believed to better represent underlying pathophysiology than clinical diagnostic categories.52, 53 These intermediate phenotypes are assumed to involve the same biological pathways as the illness but exhibit a simpler etiologic background and are more closely related to relevant gene effects.54, 55 Although measuring these readouts may or may not directly increase power to detect schizophrenia risk genes,54 they allow stratified delineation of the effects of particular risk alleles on brain structures and functions rather than that of simply the diagnostic phenotypes. This approach was previously validated in the studies of several schizophrenia candidate genes.56, 57, 58, 59

In 2009, Esslinger et al.60 investigated the influence of ZNF804A rs1344706 on functional connectivity between brain regions that were often associated with dysfunction in schizophrenia during working memory (N-back task) performance in a healthy sample. Notably, the authors observed reduced connectivity in the dorsolateral prefrontal cortex (DLPFC) between hemispheres and increased connectivity between the right DLPFC and left hippocampal formation (HF) during the working memory task.60 In a follow-up study, the same imaging sample was reanalyzed to understand how the effects of rs1344706 generalize across different experimental settings.61 This study revealed that the reduced interhemispheric DLPFC connectivity at higher rs1344706 risk status persisted in both resting state and cognitive states induced by an emotion recognition.61 Contrarily, the increase in prefrontal–hippocampal connectivity was exclusively observed during working memory engagement.61 Such changes in functional coupling between the right DLPFC and HF in the context of the working memory task were later replicated by Paulus et al.62 in an independent healthy sample, albeit with weaker effects compared with original study.60 These studies served as the initial evidence that rs1344706 affected brain function and thus memory formation.

This contention was further supported by Walter et al.63 using a sample that overlapped with that reported in Esslinger et al.60 They investigated cortical activation and connectivity associated with rs1344706 during performance on a theory of mind task (which measured participant’s ability to infer mental state) in healthy subjects. A significant risk allele dose effect was observed for activation of regions implicated in theory of mind function: the dorsomedial prefrontal cortex and the temporoparietal cortex.63 The authors also observed differences in activation of the left inferior prefrontal associated with the ZNF804A risk allele, and this brain region is attributed to general social information processing difficulties.63 Later, in an independent healthy sample, Mohnke et al.64 observed reduced activity of the left temporoparietal junction, dorsomedial prefrontal cortex and the posterior cingulate cortex with increasing numbers of rs1344706 risk alleles during theory of mind tasks, confirming the results in the study by Walter et al.63 Mohnke et al.64 further reported negative genotypic effects in the left dorsomedial prefrontal cortex and the temporal and parietal regions, highlighting the importance of ZNF804A in social cognition.

Given that ZNF804A is important in cognitive function in healthy individuals, researchers then analyzed its roles in such traits in schizophrenia patients. In an independent sample including normal controls, schizophrenia patients and their unaffected siblings, Rasetti et al.65 found that siblings and patients exhibited abnormal DLPFC functional coupling with the HF during working memory tasks. ZNF804A rs1344706 significantly modulated right DLPFC coupling with the HF in the control group, which was further confirmed in siblings and patients.65 Later, Thurin et al.66 explored the role of rs1344706 on functional connectivity measures related to cognitive control (using the modified Flanker task that includes response inhibition). The rs1344706 allele loaded the effect on right DLPFC and anterior cingulate cortex functional coupling, with risk allele carriers exhibiting increased coupling. Taken together, these data provided further evidence that ZNF804A modulated cortical network connectivity during executive cognition.

In addition to brain function, efforts were also made to understand the effects of rs1344706 on brain structure. In contrast to the fruitful and relatively reproducible functional magnetic resonance imaging results for rs1344706 across samples, this schizophrenia risk SNP exhibited more variable effects on brain structures. Cousijn et al.67 showed that genetic variations in ZNF804A, including the genome-wide significant risk SNP rs1344706, did not affect either total or regional brain volumes in 892 healthy young adults. In another study including 335 individuals with schizophrenia spectrum disorders and 198 healthy volunteers, Wassink et al.68 observed significant effects of rs1344706 on total and frontal white matter volumes in the patient group, and the pattern of effects was concordant with risk allele carriers having larger volumes than non-risk homozygotes. In the healthy control group, risk allele homozygotes exhibited increased total white matter volume compared with non-risk allele carriers, replicating a previously reported association.69 Further, in an independent sample, Voineskos et al.70 found that individuals who were homozygous for the risk allele had reduced cortical gray matter thickness in the superior temporal gyrus and the anterior and posterior cingulate cortices compared with non-risk allele carriers. Those risk allele homozygotes also demonstrated reduced attention control, aligning with findings in the anterior cingulate cortex. These data of ZNF804A in brain structure modification were less consistent compared with its role in brain function. However, the data are not completely contradictory, as no difference in total brain volume could be attributed to simultaneous increases and decreases of certain brain regions. In addition, the effects of rs1344706 could differ in schizophrenic individuals with different behavioral defects. As a result, although we can tentatively conclude that ZNF804A SNP rs1344706 probably confers schizophrenia risk via augmenting white matter volume but reducing the bulk of matter volume, further stratified research analyzing its effects on specific structures of the brain according to the detailed behavioral phenotypes are needed.

Neurocognitive studies of ZNF804A

As previously mentioned, one important characteristic of schizophrenia patients is the impairment of cognitive function. Hence, the neurocognitive analyses of ZNF804A rs1344706 were critical and compelling. One study by Walters et al.71 on the neuropsychological effects of rs1344706 offered a likely explanation of ZNF804A’s effect on cognition. This study sought to investigate neuropsychological performance in patients and healthy controls on cognitive functions typically impaired in schizophrenia, that is, intelligence quotient (IQ), episodic memory, working memory and attention control. Carriers of the risk allele exhibited significantly ‘better’ cognitive performance (e.g., working memory and episodic memory) compared with non-risk allele carriers only in patients but not healthy controls.71 Of note, the affected cognitive functions implicated precisely those cortical regions affected by rs134706: DLPFC and HF.60, 65 In addition, Walters et al.71 also found that the association between rs1344706 and schizophrenia was strengthened in patients with high IQ, indicating potential interaction effects between the risk variant rs1344706 and IQ. This interaction among cognitive functions (i.e., working memory and executive functions) was later confirmed in a Chinese sample by Chen et al.72 but only in schizophrenia patients. Specifically, the schizophrenia risk allele was associated with poorer cognitive function in patients with high IQ but better cognitive function in patients with low IQ.72 Chen et al.72 reported that the association between rs1344706 and schizophrenia was modulated by IQ, with a stronger association among individuals with relatively high IQ. These converging lines of evidence suggest a possible modulating effect of IQ on the roles of rs1344706 in the etiology of schizophrenia. Further, the counterintuitive response, that is, rs1344706, affected cognition only in patients and not in healthy participants, suggesting that ZNF804A may be a risk factor for a subgroup of schizophrenia in which cognitive performance was relatively less impaired.

Following the recent ‘theory of mind’ study,63 Hargreaves et al.73 tested the hypothesis that the risk of rs1344706 was partially mediated through affecting social cognition. They tested this hypothesis based on behavioral measures of two constructs (The ‘Eyes of the Mind’ task, which indexes mental state decoding, and the ‘Hinting task’, which measures mental state reasoning) widely investigated in schizophrenia. They indexed attribution style when interpreting positive and negative events using the ‘Interpersonal social attributions questionnaire’. They found that rs1344706 was significantly associated with variation in interpersonal attributions in healthy participants but not in patients.73 Combining these results with those from the Walter et al.63 imaging study, the ZNF804A risk allele was clearly associated with negatively altered responsiveness in social cognition in healthy participants.

Despite the consistency between the studies discussed above, the effect of ZNF804A was not always consistent, and negative results have also been observed. In 1507 healthy young men undergoing induction to military training, Stefanis et al.74 sought to investigate whether common ZNF804A variants (e.g., rs1344706) affect psychosis-related intermediate phenotypes, such as cognitive performance dependent on prefrontal and frontotemporal brain function, schizotypal traits and attenuated psychotic experiences. However, they did not observe significant associations through central indexes of sustained attention or working memory performance. Instead, the psychosis risk variants were associated with a refined positive schizotypy phenotype characterized primarily by self-rated paranoia/ideas of reference. This phenotype was previously suggested to reflect genetic tendency to psychosis (the high likelihood to 'misinterpret otherwise neutral social cues and perceptual experiences in one’s immediate environment, as personally relevant and significant information').74 As a result, the fact that this study did not arrive at a significant observation did not reject the previous conclusion, as clues of ZNF804A’s impact on cognition were presented. In addition, the negative results could be attributed to the biased sample selection, as all participants were young males experiencing military activities.

Overall, these neuroimaging and neurocognitive studies implied that the ZNF804A rs1344706, as a genome-wide significant variant, may display pleiotropic effects on the intermediate phenotypes of schizophrenia, and a summary of the results is presented in Figure 3.

Figure 3
figure 3

A summary of rs1344706’s results in neuroimaging and neurocognitive studies. ACC, anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; HF, hippocampal formation.

PowerPoint slide

Full size image

Expanding ZNF804A common variants in schizophrenia among Europeans

Although demonstrating genome-wide significant association in GWAS,9 no direct evidence suggested that rs1344706 was the strongest genetic variant within ZNF804A. Although Williams et al.22 performed fine-mapping analyses in the UK sample and suggested that rs1344706 was among the strongest associations signals, phenotypic and genetic heterogeneity still exist among different populations. For example, Riley et al.20 replicated the association at rs1344706 and extended the analyses to 11 additional LD-tagging SNPs that captured common variations across ZNF804A in an Irish sample. Intriguingly, the GWAS SNP rs1344706 was not exclusively significant in this study.20 An additional SNP, rs7597593, which was in moderate LD with rs1344706 in Europeans (r2=0.40), exhibited a stronger association with schizophrenia. Similarly, Zhang et al.75 performed an association analysis of rs7597593 in four GWAS cohorts of European ancestry and found that rs7597593 was significantly associated with schizophrenia in the combined samples. Rs1344706 was also significant in the study by Zhang et al.75 but was less consistent than rs7597593 across samples. More recently, in the PGC2 schizophrenia GWAS,10 a newly identified SNP rs11693094 exhibited the strongest association within the ZNF804A region (P=1.53 × 10–12; Figure 1), although rs1344706 and rs7597593 were also genome-wide significantly associated with schizophrenia in this GWAS (P=1.27 × 10–10 and 1.47 × 10–9, respectively). These results suggest that there are (at least partially) independent association signals for schizophrenia within ZNF804A in Europeans.

Rare genetic variants at ZNF804A and risk of schizophrenia

In addition to the risk SNPs identified by GWAS, there are other types of rare genetic variants accounting for the disease risk of specific genes. In the case of schizophrenia, both common and rare variants could contribute to its genetic risk.76 Based on this hypothesis, Steinberg et al.21 investigated copy number variants (CNVs) at the ZNF804A locus in psychiatric samples, including schizophrenia, bipolar disorder, depression and anxiety. They identified three CNVs spanning at least part of ZNF804A after screening 5408 psychiatric patients, and none of these CNVs were present in the 39 481 controls (P=0.0016).21 These CNVs included a deletion in an individual with schizophrenia, a deletion in an individual with anxiety and a duplication in an individual with bipolar disorder. Although this result provided important evidence regarding the involvement of CNVs in the disease risk conferred by ZNF804A, caution is also required as other genome-wide CNV studies of schizophrenia and bipolar disorder have failed to identify additional carriers of CNVs affecting ZNF804A.77, 78, 79, 80, 81, 82, 83, 84, 85

To understand whether rare (frequency ~0.001%) coding variants in ZNF804A are associated with schizophrenia, Dwyer et al.86 screened the coding regions of the gene in 517 schizophrenic cases and 501 controls and further genotyped rare nonsynonymous variants in an independent sample (692 cases and 1456 controls). This sample had sufficient power to detect associations with rare alleles with an effect size (odds ratio) of 5.00. However, no single rare nonsynonymous variant was associated with schizophrenia in their study or the burden test. The negative results were then supported by the current exome-wide or genome-wide sequencing analyses of rare exonic mutations in schizophrenia.87, 88, 89, 90, 91 As such, rare nonsynonymous variants at the ZNF804A locus are unlikely involved in schizophrenia susceptibility.

Relevance to clinical intervention

With all the evidence for its association with schizophrenia risk, rs1344706 influences the response of positive symptoms to antipsychotics in schizophrenia patients.92 Specifically, the risk genotype AA of rs1344706 led to a poorer response of positive symptoms in patients.92 In another study, Zhang et al.93 found that rs1344706 was associated with a positive response to atypical antipsychotic treatment in a group of first-episode schizophrenia patients, and the risk allele carriers exhibited significantly less improvement in total Positive and Negative Syndrome Scale scores and positive subscores after treatment compared with G homozygotes. These studies indicated that ZNF804A rs1344706 does not only influence the development of schizophrenia but also prognostic response to therapies, making it a potential target for future intervention development. However, this proposal is still under debate, as the OR (~1.1) conferred by rs1344706 is only considered modest.

Common variants in ZNF804A and risk of schizophrenia in Asians

To date, converging data suggest that ZNF804A is undoubtedly a risk gene for schizophrenia in populations of European ancestry. However, in the genetically divergent Asian population, a significant association of rs1344706 with schizophrenia was only replicated in several cohorts,72, 94, 95 whereas negative results were observed in most additional samples (Table 2).96, 97, 98 In addition, rs1344706 was not associated with schizophrenia in recent Han Chinese GWASs, suggesting that it is likely not a susceptibility variant for schizophrenia in some Asian populations.15, 17, 99, 100 Researchers have attempted to prove this hypothesis through larger meta-analyses but could not obtain consistent results.101, 102, 103 We previously compared the LD patterns of the genomic region covering ZNF804A between Asians and Europeans and observed sharp differences,104 supporting the hypothesis that rs1344706 is unlikely a causative-linked variant for schizophrenia in Asians. On the other hand, the differences of rs1344706 in association with schizophrenia between Europeans and Asians likely reflect the genetic heterogeneity often observed in the genetic association analyses for complex diseases, probably as a result of differential population histories. Other population specific factors, such as diet, culture or environmental exposure, may also contribute to this inconsistency.

Table 2 Published association results for rs1344706 with schizophrenia in Asians
Full size table

In addition to rs1344706, several studies have also analyzed the associations of rs7597593 with schizophrenia in Asian populations. However, only Schwab et al.95 successfully replicated the associations in Indonesian populations, whereas other studies, including the overall meta-analysis, failed to achieve any significant associations.17, 100, 104

In contrast to these two European positive SNPs, we previously identified a functional SNP rs359895 in the promoter region of ZNF804A as significantly associated with schizophrenia in Southern Han Chinese populations. Further in vitro analysis showed that the risk SNP could affect transcription factor Sp1 binding affinity and promoter activity.96 These results confirmed ZNF804A as a susceptibility gene for schizophrenia in Asian populations. However, rs359895 is in low LD with either rs1344706 or rs7597593 in either population, implying that the association signals are likely independent.

Another risk SNP rs1366842 has also been reported in Asians by Zhang et al.105 in a small Han Chinese case–control sample. This SNP locates in the exon region of ZNF804A and results in an amino-acid change from threonine to lysine. Further meta-analyses confirmed that this SNP was significantly associated with schizophrenia (Table 3).103, 106

Table 3 Published association results for rs1366842 with schizophrenia in Asians
Full size table

In sum, although not implicated in Asian schizophrenia GWAS, these collective results suggest that ZNF804A common variants may still confer a risk of schizophrenia in this population, with distinct association signals compared with those in Europeans. Future research addressing the questions regarding which variant(s) within ZNF804A is causal for schizophrenia and what are the relevant biological mechanisms is needed.

Conclusions

In this review, we have outlined the compelling research showing that ZNF804A is a significant factor in the onset of schizophrenia. Ever since the identification of ZNF804A, the first genome-wide associated common variant for schizophrenia and bipolar disorder, additional risk genes have been discovered,10, 11, 12 yet ZNF804A is still one of the most intriguing and promising risk genes with substantial supporting evidence for schizophrenia and the broader psychosis phenotype.20, 21, 22 Previous clues of the gene’s neurobiological function have been gathered from neuroimaging and neurocognitive studies in which ZNF804A rs1344706 is associated with altered functional connectivity, relatively less impaired neuropsychological performance and reduced activation during measures of social cognition.60, 63, 65, 66, 68, 71, 73 More importantly, recent studies showed that the development and maturation of neurons in embryonic and adult stem cells appear to be sensitive to the expression of ZNF804A, with downregulation of ZNF804A in neurons causing aberrant neurite growth and decreased dendritic spine density.46 Taken together, this protein has a significant function in schizophrenia that may help with our understanding and control of this disease and other conditions where dysregulation of this gene or its pathways are involved. However, more exploration is required to reach this goal. For example, the genes regulated by ZNF804A40, 41, 42, 107 remain to be investigated in neurodevelopment and schizophrenia onset, and pathway analyses have demonstrated their involvement in several aspects of nervous system function and development associated with schizophrenia, including cell adhesion, neurite outgrowth and synapse formation.40, 41 In addition, as a transcription factor, the binding motif of ZNF804A protein on DNA sequence is unclear; thus, the genome-wide prediction of ZNF804A targets remains difficult. Further functional experiments together with bioinformatics analyses might answer this question. In addition, elucidating the biological roles of ZNF804A in animal and human studies are important next steps in understanding schizophrenia pathophysiology.

Figure 4
figure 4

Associations of ZNF804A common variants with schizophrenia in PGC2 genome-wide association studies (GWAS).10

PowerPoint slide

Full size image