Introduction

Genome-wide association studies (GWAS) have identified thousands of genes and genetic variants contributing to the development of complex diseases, although the biological mechanisms by which most of these genes act remains unclear. Similarly, GWAS in schizophrenia have detected numerous candidate loci (e.g.:[1,2]). One of the first genes that has achieved genome-wide level of statistical significance in schizophrenia GWAS is the Zinc finger protein 804A (ZNF804A). Despite the substantial genetic evidence of two single-nucleotide polymorphisms (SNPs) within this gene (i.e. rs7597593 and rs1344706, showing A and T alleles conferring increased risk for schizophrenia, respectively) [1,3], the function of the protein and the molecular mechanisms responsible for enhancing risk for psychosis were unknown.

Further research showed that this gene was related to neurodevelopment and plasticity, influencing the expression of genes involved in cell adhesion and important processes such as neural migration, neurite outgrowth and synapse formation [4,5]. ZNF804A is expressed in the brain and contains a C2H2-type domain associated with the zinc-finger protein family with a role in transcription. More recent evidences suggested that ZNF804A localizes to synapsis and that it plays a role in neurite formation, maintenance of dendritic spines, and activity-dependent structural plasticity [6], which are found to be altered in brains of schizophrenia patients. Additionally, the expression of ZNF804A has been observed to peak in both the rat and human brain during the prenatal period when neuronal migration ends and neuronal differentiation and maturation begins [7,8]. Empirical work and predictive bioinformatic analyses have suggested that the two GWAS-associated polymorphisms (rs7597593 and rs1344706) may modify the affinity of the gene sequence for DNA- and/or RNA-binding proteins, which might in turn alter the expression levels of the gene [9]. In this sense, for example, the A allele of rs1344706 has been associated with lower ZNF804A expression during the second semester of fetal brain development [8] and increased mRNA expression in post-mortem dorsolateral prefrontal cortex from psychiatrically normal controls [3]. Also rs7597593 was associated with post-mortem brain mRNA expression levels of ZNF804A, showing increased levels in C allele carrier female subjects [10].

In relation to the clinical phenotype of schizophrenia, this gene seems to influence the expression of several genes associated to the positive dimension of the illness [11,12], which includes psychotic symptoms such as hallucinations and delusions. Moreover, the A allele on rs1344706 was found related to elevated manic symptoms in psychotic patients [13], more severe symptoms in schizophrenia spectrum disorder patients [14], as well as poorer clinical outcome in first episode patients [15]. This allele has also been associated with altered brain macro- and micro-structure in healthy people, first episode patients, and patients with chronic schizophrenia or bipolar disorder [14,16–18]. The association of ZNF804A appears to be moderated by sex. Specifically, rs7597593 was strongly associated with schizophrenia in women, but not in men [10]. However, this sex differential associations need more attention, because the other study exploring this did not find sex moderation [3].

Convergent evidence has shown shared genetic determinants between schizophrenia and intermediate phenotypes of the disease, such as functional brain alterations, neurocognitive deficits, schizotypal personality traits or psychotic-like experiences, which are observable across the psychosis continuum (i.e. non-clinical, subclinical and clinical samples) [19,20]. In this regard, several neurocognitive and neuroimaging studies pointed out that ZNF804A variability was associated with altered functional connectivity, relatively less impaired neuropsychological performances and reduced activation during measures of social cognition (e.g.: [14,21].

Only two previous studies have explored the effect of the ZNF804A gene on psychosis proneness in non-clinical samples. The first study, observed that subjects carrying the T allele of rs1344706 were showing higher schizotypy scores than those CC, specifically in the case of the disorganization schizotypy factor [22]. A more recent study, in a large cohort of 1507 healthy young male conscripts, found that the C alleles of both rs7597593 and rs1344706 were associated to schizotypy, specifically to self-rated paranoia and ideas of reference. The rs7597593 C allele was also associated with higher perceptual aberration and positive psychotic-like experiences [23]. Both studies found an effect for this gene in their non-clinical samples, although the results reported for the rs1344706 are controversial in terms of the allele conferring risk.

Considering the contradictory findings from the above mentioned studies, the aim of the present study was to study the association between two ZNF804A polymorphisms (rs7597593 and rs1344706) and schizotypy and psychotic-like experiences in a sample of 808 non-clinical subjects. According to the hypothesis of the psychosis continuum, we hypothesize that ZNF804A variability will be associated with the positive dimension of both traits. Moreover, given the differential sex results observed in the literature with schizophrenia, we wanted to explore for the first time whether these differences between males and females were also present in psychosis proneness.

Materials and methods

Results

The initial sample contained 808 subjects, from which 706 remained after excluding those of non-European origin. From this, one did not complete the WSS, one did not complete the CAPE, and genotyping failed for 49 individuals (6.9%) for rs7597593 and for 27 individuals (3.8%) for rs1344706. Descriptive statistics for the WSS (standardized factor scores) and CAPE (raw scores), as well as genotype frequencies of the two SNPs are presented in Table 1. No differences were observed in the genotypic frequencies between males and females (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive data for the positive dimensions of schizotypy and psychotic-like experiences (mean ± standard deviation and range) and details on the genotypic frequencies for the two analysed SNPs (rs7597593 and rs1344706).

Data are given for the whole sample, as well as by sex (comparisons between males and females are given in italics).

https://doi.org/10.1371/journal.pone.0185072.t001

The association analyses between rs7597593 and positive schizotypy showed a significant association in the whole sample (F = 5.06 p = 0.007 padj<0.05, Table 2). Post-hoc pairwise comparisons revealed that TT subjects were reporting lower scores than CT (t = -3.10 p = 0.006) and further analyses grouping subjects in carriers of the C allele (i.e. CC and CT genotypes) and TT homozygotes showed that those TT were reporting lower scores compared to C carriers (t = 2.45 p = 0.007). Sex-stratified analyses showed that this effect was driven by females but not males (see Table 2). In this regard, post-hoc pairwise comparisons indicated that females with the TT genotype showed lower scores than those with the CT genotype (t = -3.25 p = 0.004, Fig 1a) and further analyses revealed that females TT were reporting lower scores compared to C carriers (t = 2.87 p = 0.002).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Association analyses between ZNF804A variation (rs7597593 and rs1344706) and the positive dimensions of schizotypy and psychotic-like experiences (mean ± standard deviation).

Data is given for the whole sample, as well as by sex (F and p-values are shown in italics, * p_adj<0.05).

https://doi.org/10.1371/journal.pone.0185072.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mean scores for the positive dimension of schizotypy (a) and psychotic-like experiences (b) in relation to rs7597593 genotype in males and females.

https://doi.org/10.1371/journal.pone.0185072.g001

The analyses between rs7597593 and positive psychotic-like experiences showed a significant association in the whole sample (F = 4.07 p = 0.02, Table 2). Post-hoc pairwise comparisons showed that TT subjects were scoring lower than CT (t = -2.82 p = 0.02) and further analyses grouping subjects in carriers of the C allele (i.e. CC and CT genotypes) and TT homozygotes showed that those TT were reporting lower scores compared to C carriers (t = 2.09 p = 0.02). Sex-stratified analyses between this SNP and psychotic-like experiences revealed that rs7597593 was associated in females, but not in males (Table 2). In this regard, post-hoc pairwise comparisons indicated that females with the TT genotype were showing lower positive psychotic-like experiences scores than CT (t = -3.14 p = 0.005) and CC (t = -2.72 p = 0.02, Fig 1b). Further analyses showed that females with the TT genotype were reporting lower scores compared to those carrying the C allele (t = 2.75 p = 0.003).

Regarding rs1344706, no association was observed with schizotypy or psychotic-like experiences, neither in the whole sample nor by sex (Table 2).

Discussion

In the present study, we aimed to provide further evidence for the implication of ZNF804A gene variation (i.e. rs7597593 and rs1344706) on psychosis proneness and explore, for the first time, whether sex was playing a role on this association as suggested by a previous study in schizophrenia. Our main result is that non-clinical females T homozygotes for rs7597593 were reporting lower positive schizotypy and psychotic-like experiences scores compared to C carriers, whereas this was not detected in males.

Given the strong association detected between genetic variants in the ZNF804A in schizophrenia GWAS [1,3,38], considerable efforts have been focused on exploring the genetic variation within this gene and its influence on schizophrenia and the broader psychosis phenotype, as well as its biological mechanisms and neuronal functions (see [9,39]). For example, the GWAS associated alleles (i.e. rs1344706 A allele and rs7597593 T allele), apart from being replicated in independent case-control studies (e.g. [40]), have been found related to higher and severe clinical symptoms [13,14], worst outcome in first episode patients [15] and brain alterations (e.g. [14,16]).

Convergent evidence from family studies have shown a phenotypic relationship between levels of schizotypy or psychotic-like experiences and schizophrenia that can be attributed to shared genetic effects, which points towards a significant overlap between the underlying genetic factors inducing both psychosis proneness and schizophrenia [41]. In this sense, despite all the new studies with ZNF804A, the variability within this gene has been understudied in relation to psychosis proneness. Only two previous studies have explored the implication of ZNF804A on psychosis-related (or attenuated) traits in non-clinical samples. In the first study, Yasuda and colleagues showed that subjects carrying the T allele of rs1344706 were reporting higher schizotypy scores (particularly for the disorganized factor) in a healthy Japanese sample [22]. In contrast, in the other study exploring this, Stefanis and colleagues observed higher schizotypy scores (specifically paranoia and ideas of reference) in rs1344706 C carriers. One possible explanation for this discordant result in terms of risk allele, could be the different population origin of both samples (i.e. Japanese and European). To date, converging data suggest that ZNF804A is undoubtedly a risk gene for the psychotic phenotype in populations of European ancestry, but this is not clear for Asian populations. An study comparing LD patterns of the genomic region covering ZNF804A between Asians and Europeans showed sharp differences supporting the genetic heterogeneity, probably because differential population histories [42].

As regards to the other polymorphism studied here (rs7597593), our results are in line with those of Stefanis and colleagues reporting that subjects with the C allele showed higher scores in schizotypy, perceptual aberration and positive psychotic-like experiences.

Regarding the differences between males and females in relation to ZNF804A genetic variability, to the best of our knowledge, this is the first study exploring this in psychosis proneness. In our sample, we detected a female-specific association between rs7597593 and both schizotypy and psychotic-like experiences. This result expand upon Stefanis and colleagues by including females, although, contrary to them, no significant results were found in the males of our sample.

In schizophrenia, the published studies exploring this sex-specific associations have shown discordant results. Zhang et al detected a strong association between the rs7597593 T allele and schizophrenia in females of European ancestry in their sex-stratified analyses, and a trend towards interaction between sex and this SNP in their schizophrenia case-control [10]. However, no significant interaction was detected in an Irish study by Riley et al (i.e. rs17508595, rs13393273, rs7597593 and rs1344706) [3] or in a recent Chinese case-control (i.e. rs1344706) [43]. In this regard, as Zhang and colleagues pointed out that both the Irish and Chinese samples were predominately male, which might have influenced the results given the analyses of female subjects could have had less statistical power to detect the association. In our study, the predominantly female sample has allowed us to detect the ZNF804A by-sex effect in psychosis proneness, although the lower number of males included is a limitation that might have influenced our results.

The results of the present study support the idea of a shared underlying aetiology for schizophrenia and related attenuated phenotypes present at the different levels of severity [44,45], although different alleles have been associated in clinical and non-clinical studies. As mentioned in Stefanis et al [23], this discordance may be an example of an allelic alteration that increase risk in one population but is protective in another, as has been observed in autism, in which genetic variation affecting the balance of glutamate receptor protein synthesis and producing either an increased or diminished synthesis results in a similar deleterious effect responsible of the phenotype. Additionally, the effect of a specific variant, such as the one studied here, may depend on the constellation of many other genetic variants with small effects, as well as other factors as environmental of personal factors. In this regard, however, it is difficult to determine the biological mechanisms underlying the associations observed, because the function of the ZNF804A gene is still not well known. Zinc finger domains are relatively small protein motifs containing multiple finger-like protrusions that make tandem contacts with their target molecule (e.g. DNA, RNA, protein, lipid substrates), having a function in gene transcription, translation, mRNA trafficking, cytoskeletal organization, protein folding and chromatin remodelling, among others. Their binding properties depend on the amino acid sequence of the finger domains, on the linker between fingers, the number of fingers, as well as on higher-order structures. Although the role of the ZNF804 protein is not clear, variation within the ZNF804A may affect its binding to the target and compromise the pathways to which they are involved, including its own mRNA expression [46,47], as seem to suggest recent neuroimaging studies [48,49]. The ZNF804A influences the expression of three genes involved directly in dopaminergic transmission (i.e. DRD2 and COMT) and cAMP signalling (i.e. PDE4B), two pathways thought to underlie many of the symptoms of psychosis [11]. Determining the genes regulated by ZNF804A may help to understand the function of this gene and how it may be involved in psychopathology. In this sense, ZNF804A was associated with positive schizotypy and psychotic-like experiences in the present study. Positive symptomatology has been related to hyperactivity of subcortical dopamine transmission, in which DRD2 and/or COMT could be involved [50,51]. Interestingly, for both DRD2 and COMT, gender-specific effects on psychosis have also been reported (e.g. [52,53]), as Zhang [10] and our study in psychosis proneness have found with ZNF804A. Also, as reviewed by Godar and Bortolato [54], sexual hormones (i.e. estrogens and androgens) influence differentially the development of schizophrenia-related symptoms hypothetically through the mediation of dopaminergic neurotransmission. It is tempting to speculate, thus, that ZNF804A could have an effect on positive symptoms through a dopaminergic-related pathway that may be influenced by sexual hormones [55]. Further functional experiments together with bioinformatics analysis might help with our understanding of these results.

The findings of the present study support the involvement of ZNF804 variability on the vulnerability for the development of psychopathology at a non-clinical level and the implication of sex in this association. However, elucidating the mechanisms by which this gene affects mental health will be relevant in the near future.

Acknowledgments

We would like to thank all the volunteers that have participated in the present study.