Overexpression of OTX2 in human neural cells links depression risk genes

Abstract

Genome wide association studies (GWAS) have implicated the OTX2 (Orthodenticle homeobox 2) gene locus in major depressive disorders (MDD) as well as genetically correlated traits. Of the genes identified by MDD GWAS, the gene for the transcription factor OTX2 stands out as it is responsible for both opening and closing of critical and sensitive brain periods. These are developmental periods where the brain is more sensitive to environmental input and are critical for normal brain development. Evidence suggests that the brain may also be more sensitive to negative environmental impact during sensitive periods. Critically, human and animal models both specifically implicate OTX2 gene expression in the response to stress and risk for depression. Based on the genetic findings, and the potential role of OTX2 as a mediator of environmental risk for depression, we identified genes regulated by OTX2 in human neural precursor cells (NPCs) using CRISPR activation (CRISPRa) to increase expression. We identified 17 significantly differentially expressed genes, including OTX2 which was increased 4-fold. In addition to OTX2, 4 genes of the 17 have been directly implicated in depression/depressive behaviours from human and animal studies (GPER1, VGF, TAFA5, P3H2). Additional differentially expressed genes are involved in processes implicated in depression (e.g. neurogenesis, neuroplasticity, response to stress). These novel findings link OTX2 expression with genes previously implicated in depression from human and animal studies, suggesting OTX2 as a master regulator of depression risk.

Introduction

The OTX2 genomic locus has been identified as significant in genome wide association studies (GWAS) of major depressive disorder (MDD) [1–4] as well as alcohol consumption and smoking initiation [5], educational attainment [6], and chronotype (an individual’s preference for early or late sleep timing) [7] – traits genetically correlated with depression [1, 5, 7]. The genomic region has also been identified in GWAS for cortical folding (cortical surface area, sulcal depth) [8], traits associated with cognitive performance and neurodevelopmental disorders [8–10].

OTX2 is noteworthy as a risk gene as it is key for both opening and closing of critical and sensitive brain periods [11]. Critical brain periods are developmental time points when environmental input is required for the proper development of the specific brain circuit and without this input, the function may be permanently compromised [12]. Sensitive brain periods are developmental time periods where the brain has increased plasticity and more responsive to environmental stimuli [11, 12]. The involvement of OTX2 in critical/sensitive brain periods was first gleaned from animal models for vision and hearing [11, 13] and suggested as key in the opening of critical brain periods for language acquisition in humans [14]. Alterations in timing during sensitive window periods have also been implicated in neurodevelopmental disorders, particularly autism spectrum disorder (ASD) and intellectual disabilities [12, 13, 15]. Sensitive windows for emotional development have been supported with evidence that adversity during these periods results in increased risk for neuropsychiatric disorders, particularly depression [16].

Evidence supports OTX2 in stress related depressive symptoms in both humans and in animal models [17, 18]. Transient down regulation of OTX2 after late postnatal but not early postnatal stress resulted in depressive-like behaviors in a mouse model [17]. Further, in treated mice, a second stressor in adult mice resulted in exaggerated depressive-like behaviours. Overexpression of OTX2 was protective to later stress in adult animals creating a partial rescue of depressive behaviours. In contrast, transient downregulation in adult mice had no lasting effect on stress susceptibility or OTX2 target gene levels. This indicates there is a sensitive window for stress in postnatal development mediated by reduced OTX2 expression that has a lasting effect on depressive behaviours in adults. This data provides a compelling model for the relationship of early life stress and OTX2 in depression risk.

Following that report, studies investigated the relationship of OTX2 and depression in children and found a correlation of methylation in the OTX2 gene with depression scores [18]. Further, the authors reported finding a history of maltreatment and methylation in the OTX2 gene significantly predicted depression in the children.

Currently there is no information on how genetic risk alters OTX2 function and consequently alters the transcriptome in human cells. The risk alleles for MDD span the OTX2 gene, the antisense gene OTX2-AS1, and the region intergenic between the two genes that are head-to-head. Risk alleles identified in current studies of MDD are not correlated with splicing or expression changes for OTX2 in brain tissues in GTEx (https://gtexportal.org/home/gene/OTX2) or PsychENCODE (http://resource.psychencode.org/#).

While the mechanism of genetic risk is unknown, the robust genetic findings for MDD from multiple GWAS studies as well as genetically correlated traits, the role of OTX2 in opening and closing critical brain periods, and critically the biological evidence from both humans and animal models, strongly support OTX2 as a risk factor for depression. In this study, we sought to identify OTX2 gene targets using CRISPR activation (CRISPRa) to overexpress OTX2 in human neural precursor cells (NPCs).

Methods

A schematic of procedures is outlined in Supplementary Figure S1.

Derivation of NPCs

The NPCs were derived from the H9 stem cell line (female). H9 stem cell colonies were plated on Matrigel-coated plates and maintained in mTSeR medium until 70% confluent. At this point, the colonies are removed from the plate and cultured in suspension as embryoid bodies (EBs) for 2 weeks using NPC media (DMEM/F12, N2, B27 & noggin) for the generation of forebrain progenitors [19, 20]. The EBs were then plated on coated plates and maintained in NPC medium plus FGF2. The neural rosettes that emerged after 3–4 days were manually selected, dissociated with accutase and plated. The NPC lines were then expanded and cryopreserved. Lines were tested for mycoplasma using the EZ-PCR™ Mycoplasma Detection Kit from Froggobio (Concord, ON, Cat. #20-700-20).

CRISPR Activation (CRISPRa) of OTX2

CRISPRa plasmids for OTX2 are obtained from Origene (GA103345, Rockville, MD). We used three activation vectors (GA103345G1, GA103345G2, GA103345G3) with 3 different guides to target OTX2 as well as one scramble negative control. Transfection used the Lonza nucleofector system (Ottawa ON, Amaxa Mouse Neural Stem Cell Nucleofector Kit, Cat# VPG-1004) with transfection efficiency monitored using expression of the tGFP gene that is cloned in the same vectors. RNA was extracted 48 hours post transfection to allow expression of OTX2. RNA was extracted using the RNeasy (Qiagen, Toronto, ON) and mRNA was polyA selected. RNA Integrity Number (RIN) was over 9.1 for all samples sequenced. OTX2 expression was checked by PCR to confirm overexpression before the RNA was sent for RNA-sequencing. The PCR conditions were as follows: primers OTX2-ex1F: ACTTCGGGTATGGACTTGCT, OTX2-ex3R: CTGTTGTTGCTGTTGTTGGC, annealing temperature 55 °C producing a 270 bp product. All experiments were done in triplicate for both the CRISPRa vectors and controls.

Transcriptome analyses

The platform for sequencing was the Illumina Novaseq X (150 bp pair ended reads). Fastq files were preprocessed with fastp version 0.23.1 [21]. Reads were aligned to the genome (GRCh38) and transcripts quantified with Salmon Version 1.10.1 [22]. Ambiguous reads were not counted. Mapping statistics are shown in Supplementary Table S1. Differentially expressed genes were called using DESeq2 [23]. Differential expression was called if the Benjamini-Hochberg false discovery rate adjusted p value was less than .05 (Table 1).

Table 1.

Differentially Expressed Genes in OTX2 CRISPRa NPCs.

Gene Symbol	log2(FC)	P-value	P-adj	Gene Name
ITGAX	5.82	6.63E-16	1.10E-11	Integrin subunit alpha X
ADHFE1-VXNa	−3.68	6.01E-09	4.96E-05	Transcript spans ADHFE1 (alcohol dehydrogenase iron containing 1) and VXN (Vexin)
PEDS1-UBE2V1	−7.56	2.02E-08	1.11E-04	Transcript spans PEDS1 (plasmanylethanolamine desaturase 1) and UBE2V1 (ubiquitin conjugating enzyme E2 V1)
OTX2a	2.22	1.43E-07	5.89E-04	Orthodenticle homeobox 2
CDK7a	−3.10	4.43E-07	1.46E-03	Cyclin dependent kinase 7
RNH1a	2.21	7.98E-07	2.20E-03	Ribonuclease/angiogenin inhibitor 1
VGFa	0.98	2.36E-06	5.58E-03	VGF nerve growth factor inducible
GPER1	3.03	3.12E-06	6.44E-03	G protein-coupled estrogen receptor 1
ETV3La	4.20	4.90E-06	9.00E-03	ETS variant transcription factor 3 like
GPR37L1	3.69	6.45E-06	1.07E-02	G protein-coupled receptor 37 like 1
TAFA5	1.11	1.33E-05	2.00E-02	TAFA chemokine like family member 5
H4C11a	0.91	1.50E-05	2.06E-02	H4 clustered histone 11
KCNC4	0.85	2.81E-05	3.32E-02	Potassium voltage-gated channel subfamily C member 4
TTNa	2.48	2.72E-05	3.32E-02	Titin
P3H2	−0.96	4.30E-05	4.27E-02	Prolyl 3-hydroxylase 2
PPL	2.80	4.08E-05	4.27E-02	Periplakin
NDUFC2-KCTD14a	−6.08	4.39E-05	4.27E-02	Transcript spans NDUFC2 (NADH ubiquinone oxidoreductase subunit C2) and KCTD14 (potassium channel tetramerization domain containing 14)

^aIndicates OTX2 ChIP-seq peak within 1.5 kb of the transcription start site.

For the differentially expressed genes, we examined evidence for binding of OTX2 at the promoter as evidence of possible direct regulation by OTX2 using OTX2 ChIP-seq peaks called in stem and neural cells (Q value < 1E-05) from the ChIP-Atlas (https://chip-atlas.org/). The neural cell tracks were from; H9 ES-derived neural progenitor cells, HRPEpiC (Human Retinal Pigment Epithelial Cells), iPS cells, hESC derived retinal cells, and medulloblastoma cell lines D283 and D341. Both D283 and D341medulloblastoma cell lines overexpress OTX2 [24].

Results

Using CRISPRa with three guides to the OTX2 gene, expression of OTX2 was increased by four-fold in NPCs compared to cells transfected with a control vector with scramble guides (Table 1). Analyses of the differentially expressed genes identified 17 genes significantly differentially expressed, four of which are directly implicated in depression in addition to OTX2 (Table 1, Fig. 1). Among the differentially expressed genes was GPER1 (G protein-coupled estrogen receptor 1). GPER1 has been previously implicated in depression with higher blood serum levels of GPER1 found in serum from patients with MDD compared to controls, and a positive correlation was found between GPER1 levels and depression scores, suggesting that serum GPER1 levels are valuable in predicting the presence of depression [25]. The gene, neuronal VGF nerve growth factor (VGF), has been linked to energy balance regulation, neurogenesis, synaptogenesis, learning and memory, and regulates depression-like behavior in animal models [26]. TAFA5 (TAFA chemokine like family member 5, also called FAM19A5) has also been implicated in depression from animal models [27]. Deletion of the gene in mice resulted in increased depressive-like behaviors and impaired hippocampus-dependent spatial memory. Overexpression of human TAFA5 selectively in the mouse hippocampus attenuated chronic stress-induced depressive-like behaviors [27]. Further, differential methylation of TAFA5 was associated with levels of depressive symptoms in a study of women and stress [28]. The fourth gene, P3H2 (prolyl 3-hydroxylase 2), was identified as having reduced methylation over time in children that had been maltreated [29]. This epigenetic marker was later shown to mediate the effect of childhood sexual/physical abuse on comorbid post-traumatic stress disorder (PTSD) and depression in adulthood [30].

Fig. 1.

Graphical representation of Log2 fold change of differentially expressed genes.

Other genes that were differentially expressed are involved in neurogenesis (ETV3L [31]) and neuroplasticity (CDK7 [32]) – processes implicated in depression and critical for antidepressant response [33, 34]. The TTN gene was reported to be differentially methylated and expressed in offspring of mice deficient for the serotonin transporter gene that were subjected to a maternal restraint stress paradigm of prenatal stress [35]. Prenatal exposure to stress in this mouse line was associated with increased depression-like behavior in the forced-swim test, particularly in female offspring [35]. An additional differentially expressed transcript spans the ADHFE1 and VXN genes. VXN was previously reported to be regulated by BDNF [36].

Except for the histone gene, H4C11, none of the differentially expressed genes were located within linkage disequilibrium (LD) regions for published GWAS for MDD. H4C11 is in the Major Histocompatibility Complex (MHC) LD region which is large with strong LD across the region. Consequently, it has been difficult to distinguish the risk genes within the MHC. Previous studies using imputation did not find evidence for association of the Human Leukocyte Antigen (HLA) alleles or C4 haplotypes (implicated in schizophrenia) with depression [37]. Thus, the risk gene/genes within the MHC locus are currently unknown. Mutations in H4C11 (HIST1H4J) result in a neurodevelopmental syndrome (Tessadori-Bicknell-van Haaften neurodevelopmental syndrome-2) characterized by intellectual disabilities, poor overall growth, profound global developmental delay with absent speech, and dysmorphic facial features [38]. Further, H4c11 (Hist1h4j) was found to be differentially expressed in female mice using subchronic variable stress in a mouse knock out model of Dnmt3a [39] suggesting a role in stress response. While the link to neurodevelopmental disorders and stress response is intriguing, currently there is insufficient evidence to support H4C11 as the risk gene for MDD within the MHC region.

Several of the differentially expressed genes, have previously been reported to be regulated by OTX2 (e.g. TTN [40], KCNC4 [41]) including OTX2 that can regulate its own expression by maintaining expression through a feedback loop mechanism [42]. We examined evidence for binding of OTX2 at the promoter of the differentially expressed genes as evidence of possible direct regulation by OTX2. We used OTX2 ChIP-seq peaks called in stem and neural cells (Q value < 1E-05) from the ChIP-Atlas (https://chip-atlas.org/). We found evidence for OTX2 binding within 1.5 kb of the transcription start site for 9 of the genes, including OTX2 (Table 1). This suggests that these genes are directly regulated by OTX2. Additional ChIP-seq peaks were in other regions of the genes, but because these may be at enhancers regulating genes distant from the gene of interest, they were not included in the tally.

Discussion

Hundreds of genetic loci have now been identified for MDD through GWAS, however for the majority of these, the molecular mechanism by which risk alleles alter gene and cell function contributing to disease risk, is unknown. The most recent unpublished genetic study reports 636 genomic regions for depression, 293 of which are novel, thus we have a wealth of new information on the genomic location of genetic risk factors contributing to depression (https://www.medrxiv.org/content/10.1101/2024.04.29.24306535v2). However, we are lagging in understanding the molecular mechanisms behind genetic risk.

Identifying functional relationships between risk genes as well as genes identified by animal models and human biomarker studies is a major step forward in beginning to understand how these genes work together to create risk. Our data connecting overexpression of OTX2 to genes previously identified in studies of methylation and depressive symptoms in humans, and genes supported in animal models of depressive behaviours, is a novel finding that links genetic risk at the OTX2 locus to these separate findings. GPER1 has been indicated as serum biomarkers for depression [25] and OTX2, TAFA5, and P3H2 have been associated with differential methylation in blood samples from individuals who have experienced chronic stress/childhood maltreatment [18, 28, 29]. Further research on these genes/proteins, or combination of these, as markers for depression risk is warranted with the goal of early detection of risk as well as altering expression and consequently risk.

We note that previous data in animal models link reduced expression of OTX2 and depressive like behaviour, whereas in this manuscript we upregulate OTX2 to identify gene targets. Also of note is that the findings for GPER1 are counterintuitive. Previously higher GPER1 levels were found in blood from MDD patients and here we found higher expression of GPER1 when OTX2 was increased. If the risk in humans is related to reduced expression of OTX2, then reduced expression of GPER1 would be predicted. Gene expression differences in animal models compared to humans, as well as opposing expression patterns in different tissues is not uncommon. Downregulation of OTX2 in NPCs as well as additional types of neural cells, will provide further information on the function of OTX2 in relation to depression models.

The OTX2 locus is also associated by GWAS with alcohol consumption and smoking initiation [5]. Further, this locus is associated with cortical folding traits (cortical surface area, sulcal depth) [8] which are related to cognition and neurodevelopmental disorders [8–10], the understanding of which is relevant to the development of psychiatric disorders.

The understanding of OTX2 function has wider scientific relevance outside of psychiatry as OTX2 is critical for eye and brain development with loss of function mutations in the OTX2 gene resulting in congenital pituitary hypoplasia, eye malformations, such as anophthalmia and microphthalmia, seizures, hearing impairments and developmental disorders [43]. Importantly, OTX2 is also an oncogene for medulloblastoma, the most common pediatric brain tumour, with overexpression of OTX2 indicated as a subtype of medulloblastoma [43, 44]. Consequently, the identification of genes regulated by OTX2 may contribute to the understanding of risk for a wide range of diseases.

Author contributions

CLB designed the experiments and analyzed the RNA-sequencing data. YF preformed the lab experiments. KGW assisted in the analyses. CLB wrote the paper with input from all authors.

Data availability

Sequencing data is available from the authors on request.

Competing interests

The authors declare no competing interests.

Ethics approval

H9 is an established human embryonic stem cell line and CRISPR modifications was approved by the Stem Cell Oversight Committee of the Canadian Institutes of Health Research. All methods were performed in accordance with the relevant guidelines and regulations.

Supplementary information

The online version contains supplementary material available at 10.1038/s41398-025-03320-8.