Edited by: Shusaku Uchida, Kyoto University, Japan
Reviewed by: Daisuke Ibi, Meijo University, Japan; Manabu Makinodan, Nara Medical University, Japan
†These authors have contributed equally to this work
This article was submitted to Brain Disease Mechanisms, a section of the journal Frontiers in Molecular Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Accumulating evidence suggests that the epigenetic alterations induced by antipsychotics contribute to the therapeutic efficacy. However, global and site-specific epigenetic changes by antipsychotics and those shared by different classes of antipsychotics remain poorly understood. We conducted a comprehensive DNA methylation analysis of human neuroblastoma cells cultured with antipsychotics. The cells were cultured with low and high concentrations of haloperidol or risperidone for 8 days. DNA methylation assay was performed with the Illumina HumanMethylation450 BeadChip. We found that both haloperidol and risperidone tended to cause hypermethylation changes and showed similar DNA methylation changes closely related to neuronal functions. A total of 294 differentially methylated probes (DMPs), including 197 hypermethylated and 97 hypomethylated DMPs, were identified with both haloperidol and risperidone treatment. Gene ontology analysis of the hypermethylated probe-associated genes showed enrichment of genes related to the regulation of neurotransmitter receptor activity and lipoprotein lipase activity. Pathway analysis identified that among the DMP-associated genes,
Schizophrenia is a major psychiatric disorder characterized by positive and negative symptoms as well as cognitive impairments. Schizophrenia affects approximately 1% of the population and has become a worldwide public health concern with a considerable financial burden on individuals, families and society (
Epigenetics is the study of heritable and stable changes in gene expression that are not caused by changes in DNA sequences, and it involves DNA methylation and histone modifications as the major molecular mechanisms (
In addition, accumulating evidence also suggests that antipsychotics affect epigenetic status in the brain (
We previously performed comprehensive DNA methylation analyses of the antipsychotics blonanserin and perospirone using human neuroblastoma cells (
To further understand the epigenetic effects of antipsychotics, we conducted a comprehensive DNA methylation analysis of haloperidol, a typical antipsychotic, and risperidone, an atypical antipsychotic, using the same cell culture model with a HumanMethylation450 BeadChip.
Human neuroblastoma SK-N-SH cells (American Type Culture Collection, Manassas, VA, United States) were cultured as previously described (
From each sample, genomic DNA was extracted via standard phenol:chloroform methods. The DNA samples were subjected to bisulfite modification by the EZ DNA Methylation Kit D500 (Zymo Research, Irvine, CA, United States). A comprehensive DNA methylation assay was performed with the HumanMethylation450 BeadChip (Illumina, Inc., San Diego, CA, United States) according to the manufacturer’s instructions. The DNA methylation level of each probe was represented by a β value. It was calculated according to the ratio of intensities between methylated and unmethylated alleles, as in the following formula: β value = methylated intensity/(methylated intensity + unmethylated intensity + 100).
Among the 485,512 probes on the chip, we excluded unreliable probes whose detection p values ≥0.01 in at least one experimental sample. The remaining data were analyzed using ChAMP (
We cultured human neuroblastoma SK-N-SH cells with two doses of haloperidol or risperidone for 8 days. Doses were determined based on their minimum and maximum effective blood concentrations. After culture, we obtained DNA methylation profiles with the Illumina HumanMethylation450K BeadChip. After removing the unreliable probes, we obtained 411,750 probes for data analysis. Unsupervised hierarchical clustering of all β values, which are proxies of DNA methylation levels, showed that the antipsychotic groups and the two control groups, including the no treatment (NT) and DMSO group, were separated (
Cluster analysis and PCA.
We identified 3,028 and 1,320 DMPs in the high and low concentration groups for haloperidol, respectively. The average β values of DMPs in the high-dose group was significantly increased compared to those in the control group (
Average β values of DMPs in the haloperidol treatment group according to the genomic positions.
Dose | Context | # of probes | β value (haloperidol) |
β value (control) |
|
High | All DMPs | 3028 | 0.520 ± 0.189 | 0.468 ± 0.169 | 2.20E-16 |
Gene body | 1148 (38%) | 0.549 ± 0.173 | 0.486 ± 0.168 | 2.20E-16 | |
TSS1500 | 273 (9%) | 0.479 ± 0.215 | 0.469 ± 0.172 | ns | |
TSS200 | 113 (4%) | 0.429 ± 0.248 | 0.426 ± 0.184 | ns | |
5′UTR | 218 (7%) | 0.481 ± 0.192 | 0.441 ± 0.171 | 0.02244 | |
3′UTR | 105 (3%) | 0.565 ± 0.193 | 0.529 ± 0.193 | ns | |
1st exon | 54 (2%) | 0.359 ± 0.238 | 0.381 ± 0.173 | ns | |
Intergenic region | 1117 (37%) | 0.521 ± 0.176 | 0.457 ± 0.159 | 2.20E-16 | |
CpG island | 300 (10%) | 0.432 ± 0.294 | 0.444 ± 0.204 | ns | |
CpG island shore | 488 (16%) | 0.5111 ± 0.215 | 0.478 ± 0.172 | 0.009 | |
CpG island shelves | 316 (10%) | 0.556 ± 0.175 | 0.495 ± 0.175 | 1.65E-05 | |
Open sea | 1924 (64%) | 0.531 ± 0.156 | 0.465 ± 0.160 | 2.20E-16 | |
Low | All DMPs | 1320 | 0.511 ± 0.340 | 0.497 ± 0.240 | ns |
Gene body | 431 (33%) | 0.633 ± 0.302 | 0.580 ± 0.222 | 0.00395 | |
TSS1500 | 183 (14%) | 0.379 ± 0.332 | 0.406 ± 0.229 | ns | |
TSS200 | 111 (8%) | 0.277 ± 0.296 | 0.330 ± 0.196 | ns | |
5′UTR | 85 (6%) | 0.401 ± 0.341 | 0.433 ± 0.252 | ns | |
3′UTR | 48 (4%) | 0.666 ± 0.278 | 0.607 ± 0.213 | ns | |
1st exon | 45 (3%) | 0.247 ± 0.289 | 0.328 ± 0.208 | ns | |
Intergenic region | 417 (32%) | 0.537 ± 0.330 | 0.512 ± 0.229 | ns | |
CpG island | 379 (29%) | 0.346 ± 0.335 | 0.380 ± 0.228 | ns | |
CpG island shore | 316 (24%) | 0.478 ± 0.338 | 0.470 ± 0.236 | ns | |
CpG island shelves | 130 (10%) | 0.685 ± 0.269 | 0.626 ± 0.197 | 0.0441 | |
Open sea | 495 (38%) | 0.612 ± 0.303 | 0.569 ± 0.219 | 0.0124 |
We then identified 616 DMPs (386 hypermethylated and 230 hypomethylated DMPs) shared by both haloperidol dose groups (
We identified 476 and 1,025 DMPs in the high- and low-dose risperidone groups, respectively. Similar to haloperidol, DMPs in both risperidone groups showed hypermethylation and were mainly located much farther from the promoter regions, including the gene body, 3′ UTR, and intergenic region (
Average β values of DMPs in the risperidone treatment group according to the genomic positions.
Dose | Context | # of probes | β value (risperidone) |
β value (control) |
|
High | All DMPs | 476 (100%) | 0.558 ± 0.317 | 0.518 ± 0.224 | 0.0221 |
Gene body | 157 (33%) | 0.674 ± 0.266 | 0.593 ± 0.195 | 0.002389 | |
TSS1500 | 62 (13%) | 0.438 ± 0.333 | 0.429 ± 0.228 | ns | |
TSS200 | 40 (8%) | 0.387 ± 0.323 | 0.385 ± 0.221 | ns | |
5′UTR | 25 (5%) | 0.541 ± 0.324 | 0.525 ± 0.229 | ns | |
3′UTR | 15 (3%) | 0.750 ± 0.118 | 0.621 ± 0.111 | 0.004727 | |
1st exon | 16 (3%) | 0.252 ± 0.260 | 0.314 ± 0.166 | ns | |
Intergenic region | 161 (34%) | 0.549 ± 0.317 | 0.519 ± 0.225 | ns | |
CpG island | 130 (27%) | 0.430 ± 0.340 | 0.425 ± 0.225 | ns | |
CpG island shore | 116 (24%) | 0.520 ± 0.324 | 0.480 ± 0.228 | ns | |
CpG island shelves | 41 (9%) | 0.673 ± 0.273 | 0.619 ± 0.199 | ns | |
Open sea | 189 (40%) | 0.646 ± 0.269 | 0.582 ± 0.197 | 0.009143 | |
Low | All DMPs | 1025 (100%) | 0.538 ± 0.329 | 0.508 ± 0.228 | 0.016 |
Gene body | 330 (32%) | 0.673 ± 0.269 | 0.599 ± 0.195 | 5.95E-05 | |
TSS1500 | 129 (13%) | 0.387 ± 0.326 | 0.411 ± 0.221 | ns | |
TSS200 | 82 (8%) | 0.320 ± 0.320 | 0.360 ± 0.214 | ns | |
5′UTR | 68 (7%) | 0.408 ± 0.331 | 0.417 ± 0.225 | ns | |
3′UTR | 32 (3%) | 0.718 ± 0.242 | 0.623 ± 0.177 | ns | |
1st exon | 42 (4%) | 0.227 ± 0.258 | 0.312 ± 0.175 | ns | |
Intergenic region | 342 (33%) | 0.564 ± 0.318 | 0.522 ± 0.222 | 0.04743 | |
CpG island | 292 (28%) | 0.362 ± 0.330 | 0.389 ± 0.217 | ns | |
CpG island shore | 232 (23%) | 0.504 ± 0.332 | 0.479 ± 0.229 | ns | |
CpG island shelves | 107 (10%) | 0.719 ± 0.245 | 0.639 ± 0.180 | 0.007554 | |
Open sea | 394 (38%) | 0.639 ± 0.281 | 0.577 ± 0.202 | 0.0004007 |
We identified 377 DMPs shared by both doses of risperidone, which consisted of 254 hypermethylated and 123 hypomethylated DMPs (
We then examined the common DMPs shared by haloperidol and risperidone. There were 294 common DMPs, consisting of 197 hypermethylated and 97 hypomethylated DMPs (
Venn diagram and cluster analysis of common DMPs.
Common DMPs showing robust DNA methylation changes.
Probe | Gene | Genomic context | CpG island | Hal1 | Hal2 | Ris1 | Ris2 | ||||
Δβ | Δβ | Δβ | Δβ | ||||||||
cg17030628 | IL20 | TSS200 | Open sea | 0.92 | 2.48E-12 | 0.93 | 7.61E-11 | 0.94 | 1.59E-12 | 0.91 | 1.32E-11 |
cg14333542 | Intergenic region | Open sea | −0.79 | 9.61E-11 | −0.82 | 3.73E-11 | −0.82 | 7.05E-11 | −0.81 | 6.76E-11 | |
cg04124606 | Intergenic region | Open sea | −0.48 | 5.32E-10 | −0.48 | 5.96E-10 | −0.48 | 5.48E-10 | −0.48 | 5.37E-10 | |
cg19605788 | CNTD2 | Gene body | CpG island | 0.37 | 1.83E-09 | 0.36 | 2.39E-09 | 0.37 | 1.89E-09 | 0.37 | 1.89E-09 |
cg20168823 | BRUNOL4 | Gene body | CpG island | 0.36 | 2.98E-09 | 0.36 | 2.15E-09 | 0.36 | 1.96E-09 | 0.36 | 2.85E-09 |
cg17301379 | Intergenic region | CpG island | 0.40 | 1.42E-08 | 0.39 | 9.24E-09 | 0.40 | 7.84E-09 | 0.40 | 1.59E-08 | |
cg21093807 | C3orf63 | TSS1500 | CpG island | 0.27 | 7.66E-08 | 0.28 | 4.57E-07 | 0.26 | 1.09E-06 | 0.26 | 2.16E-07 |
cg08052546 | Intergenic region | CpG island shore | 0.27 | 1.03E-07 | 0.22 | 5.31E-07 | 0.24 | 1.62E-07 | 0.23 | 1.34E-07 | |
cg05545910 | TTLL1 | Gene body | Open sea | 0.28 | 3.17E-06 | 0.31 | 1.05E-07 | 0.27 | 4.24E-08 | 0.29 | 5.11E-05 |
cg10442913 | ATP1A1 | TSS1500 | CpG island | 0.22 | 7.87E-06 | 0.23 | 7.65E-06 | 0.24 | 6.21E-06 | 0.22 | 7.89E-06 |
cg07369507 | ZNF323 | TSS1500 | Open sea | −0.27 | 3.76E-05 | −0.31 | 2.55E-05 | −0.24 | 3.80E-06 | −0.29 | 1.71E-07 |
cg21581312 | LOC723972 | TSS200 | Open sea | 0.31 | 6.19E-05 | 0.41 | 2.33E-05 | 0.32 | 3.58E-06 | 0.34 | 5.52E-06 |
cg11057824 | C14orf182 | Gene body | CpG island shore | 0.23 | 0.000282 | 0.27 | 1.04E-05 | 0.25 | 2.12E-05 | 0.29 | 6.04E-06 |
GO analysis of hypermethylated DMP-associated genes altered by both haloperidol and risperidone treatment.
GO ID | GO term | Category | Corrected |
# of genes | Gene |
GO:0099601 | Regulation of neurotransmitter receptor activity | BP | 0.00345 | 5 | DLG2, GRIN2A, SHANK1, SHANK2, SLURP1 |
GO:0004465 | Lipoprotein lipase activity | MF | 0.00585 | 3 | ANGPTL4, APOA4, LMF1 |
GO:0051004 | Regulation of lipoprotein lipase activity | BP | 0.00586 | 3 | ANGPTL4, APOA4, LMF1 |
GO:0008328 | Ionotropic glutamate receptor complex | CC | 0.00586 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0098878 | Neurotransmitter receptor complex | CC | 0.00594 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0060997 | Dendritic spine morphogenesis | BP | 0.0071 | 3 | KIF1A, SHANK1, SHANK2 |
GO:0098815 | Modulation of excitatory postsynaptic potential | BP | 0.01065 | 3 | CELF4, SHANK1, SHANK2 |
GO:0099118 | Microtubule-based protein transport | BP | 0.01086 | 4 | DLG2, DYNLRB1, KIF1A, WDR34 |
GO:0098840 | Protein transport along microtubule | BP | 0.01086 | 4 | DLG2, DYNLRB1, KIF1A, WDR34 |
GO:1904115 | Axon cytoplasm | CC | 0.01098 | 3 | DLG2, KIF1A, RANBP1 |
GO:0048854 | Brain morphogenesis | BP | 0.01101 | 3 | FGF8, SHANK1, SHANK2 |
GO:1900449 | Regulation of glutamate receptor signaling pathway | BP | 0.0111 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0097106 | Postsynaptic density organization | BP | 0.0117 | 3 | DLG2, SHANK1, SHANK2 |
GO:0008066 | Glutamate receptor activity | MF | 0.01179 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0099084 | Postsynaptic specialization organization | BP | 0.01218 | 3 | DLG2, SHANK1, SHANK2 |
GO:0004970 | Ionotropic glutamate receptor activity | BP | 0.01224 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0022839 | Ion gated channel activity | BP | 0.01253 | 3 | ANO1, CLCA4, KCNT1 |
GO:0004806 | Triglyceride lipase activity | MF | 0.01327 | 3 | ANGPTL4, APOA4, LMF1 |
GO:0035235 | Ionotropic glutamate receptor signaling pathway | BP | 0.01349 | 4 | DLG2, GRIN2A, SHANK1, SHANK2 |
GO:0035176 | Social behavior | BP | 0.01574 | 3 | GNG8, SHANK1, SHANK2 |
GO terms and hub genes. A picture was drawn based on the GO analysis of hypermethylated DMP-associated genes using ClueGO. Four functional groups were identified, and the color is the same for all pathways in a group. The circles with a bold font annotation represent the leading term (the most significant term of a group of pathways). Other terms are annotated in gray, and the smallest circles indicate genes related to the pathways. If the small circle is linked to two or more colors, it indicates that the gene is enriched in different pathways. The thickness of lines between pathways and genes is based on the GO evidence code. All experimental evidence codes, such as Inferred from Experiment (EXP) and Inferred from Direct Assay (IDA), have a thick edge, and others, such as Inferred from Electronic Annotation (ICA) and Inferred from Reviewed Computational Analysis (RCA), have a thin edge. DMP, differentially methylated probe; GO, gene ontology.
We then examined the effects of haloperidol and risperidone on DNA methylation of the pharmacological targets of antipsychotics, including
We comprehensively examined DNA methylation changes induced by a typical antipsychotic, haloperidol, and a second-generation atypical antipsychotic, risperidone. To the best of our knowledge, this is the first study to examine the common effect of two distinctive antipsychotics on DNA methylation in human neuroblastoma cells. Haloperidol acts as a DRD2 antagonist to ameliorate positive symptoms. Risperidone antagonizes HTR2A as well as DRD2, thus improving both positive and negative symptoms. Our study found that haloperidol could induce DNA methylation changes at more sites than risperidone, which may be related to its stronger dopamine antagonism. In addition, our study identified the co-action sites of these two antipsychotics, both associated with regulation of neurotransmitter receptor activity and glutamate receptor activity. These results indicate that the epigenetic effects of the two antipsychotics have their own characteristics and some similarities, which are worthy of further study in the future.
The hypermethylated CpG sites in the high-dose haloperidol group were mainly located in the regions away from the promoters and the CpG island. Similarly, they were mainly located in the gene body in the low-dose group. Although risperidone induced fewer DNA methylation changes than haloperidol, risperidone also induced hypermethylation at CpG sites located away from the promoter and CpG island. These patterns of changes were closely similar to those with blonanserin (
Despite the different classes, these two antipsychotics showed very similar DNA methylation changes, especially at low doses, as evidenced by cluster analysis and PCA (
Among the genes showing robust DNA methylation changes,
Two of the other top DMP-associated genes,
Among the probes related to
From the comparison with previous studies,
In this study, we used a human neuroblastoma cell line for systematic comparison of the epigenetic effects of antipsychotics. Therefore, the DNA methylation changes identified here might be quite different from those identified in postmitotic neuronal cells. Although the SK-N-SH expresses
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below:
MB performed the experiments. YN performed the data management. JD, YN, and TK analyzed the data. JD, YN, SF, MB, and KI wrote the manuscript. KK, MB, and KI designed the experiments. All authors discussed the results and commented on the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
This study was partly supported by JSPS KAKENHI (Grant Numbers 16H06395, 16H06399, 16K21720, 18H02753, 18H05428, and 18H05430), AMED (Grant Numbers JP21dm0307001, JP21dm0307004, JP21dm0207069, JP20dm0107123, and JP21dm0207074) and Moonshot R&D (Grant Number JPMJMS2021).
This work was supported in part by the UTokyo Center for Integrative Science of Human Behavior (CiSHuB) and the International Research Center for Neurointelligence (WPI-IRCN) at The University of Tokyo Institutes for Advanced Study (UTIAS).
The Supplementary Material for this article can be found online at: