Neurobiological roots of psychopathy

Psychopathy is an extreme form of antisocial behavior, with about 1% prevalence in the general population, and 10–30% among incarcerated criminal offenders. Although the heritability of severe antisocial behavior is up to 50%, the genetic background is unclear. The underlying molecular mechanisms have remained unknown but several previous studies suggest that abnormal glucose metabolism and opioidergic neurotransmission contribute to violent offending and psychopathy. Here we show using iPSC-derived cortical neurons and astrocytes from six incarcerated extremely antisocial and violent offenders, three nonpsychopathic individuals with substance abuse, and six healthy controls that there are robust alterations in the expression of several genes and immune response-related molecular pathways which were specific for psychopathy. In neurons, psychopathy was associated with marked upregulation of RPL10P9 and ZNF132, and downregulation of CDH5 and OPRD1. In astrocytes, RPL10P9 and MT-RNR2 were upregulated. Expression of aforementioned genes explained 30–92% of the variance of psychopathic symptoms. The gene expression findings were confirmed with qPCR. These genes may be relevant to the lack of empathy and emotional callousness seen in psychopathy, since several studies have linked these genes to autism and social interaction.


Introduction
In developed countries, a relatively small group of antisocial recidivistic offenders commits the majority of all violent crimes. The prevalence of antisocial personality disorder (ASPD) is 1-3% in the general population and 40-70% in prison populations, and the corresponding figures for its most severe manifestation, psychopathy, are about 1% in the general population and 10-30% among incarcerated offenders [1][2][3][4][5]. ASPD is characterized by aggression, hostility, callousness, manipulativeness, deceitfulness, and impulsivity, and psychopathy is an extreme manifestation of ASPD. Severe antisocial and criminal behavior has a substantial genetic component [6]. This far only one study has reported contributing genes reaching genome-wide significance for ASPD [7], although two studies have found association between singlenucleotide polymorphisms and broad spectrum of antisocial behavior [8,9]. LINC00951, the gene associated with imprisoned offenders with ASPD, codes for long intergenic noncoding RNA, which is expressed especially in the frontal cortex and cerebellum. Its function is not known [7], but it has been linked to autoimmune disease [10]. The gene linked to adult antisocial behavior, ABCB1, is also highly expressed in the brain, and implicated in substance abuse [9]. No underlying molecular pathways of severe antisocial and criminal behavior are known, but there is preliminary evidence on dysregulation of the endogenous opioid system and brain opioid receptors [11][12][13] in antisocial individuals. Also abnormal glucose metabolism leading to hypoglycemia has been observed as the strongest predictor for violent crimes [14]. A recent study has also found association between immune-related gene sets and antisocial behavior [9]. We aimed to study the neurobiological background of psychopathy by using induced pluripotent stem cell (iPSC)derived cortical neurons and astrocytes, and included also nonpsychopathic substance abusers in addition to healthy individuals as control groups in order to distinguish the putative role of the coexisting substance dependence.

Material and methods
We generated and fully characterized iPSC lines from six antisocial violent offenders, three nonviolent substance abusers and six control subjects without antisocial traits or substance abuse disorders. Due to the explanatory nature of the study, no power analysis based on predefined effect size was done. The clinical and sociodemographic characteristics of the study subjects are shown in Table 1. We chose to differentiate the cells into cortical neurons expressing markers of glutamatergic and GABAergic neurons and to astrocytes. Methods for iPSC production and their characterization, derivation of neurons and astrocytes and their analyses are reported in detail in Supplementary Material.

Description of subjects
Six male offenders were identified by the history of their criminal convictions from the Finnish National Crime Register and recruited through the penal system and classified as extremely violent offenders as described in Tiihonen et al. [5]. Three individuals having substance dependence without violent behavior were recruited from the local substance abuse rehabilitation center, and six healthy controls were recruited from the staff of Niuvanniemi Hospital. The participants were interviewed with Structured Clinical Interview for DSM-IV-Disorders to exclude individuals with a psychosis diagnosis, and to assess whether or not the subject fulfilled criteria for ASPD. Also, any history of substance abuse (alcohol, heroin, buprenorphine, amphetamine, cannabis, other) was obtained through a questionnaire. The history of criminal convictions was obtained from the National Crime Register [5]. Psychopathy ratings with the Hare Psychopathy Checklist revised (PCL-R) [1] were done by accredited rater OV using official crime register data and forensic mental examination reports (violent offenders), and clinical interview (individuals with substance dependence, healthy controls). Informed consent was obtained from all subjects. This study was approved by the Ethics Committee for Pediatrics, Adolescent Medicine and Psychiatry, Hospital District of Helsinki and Uusimaa, and the Criminal Sanctions Agency of Finland.

Generation of hiPSCs and their characterization
The hiPSC lines were derived from individuals' skin fibroblasts (Supplementary Table 1

hiPSC differentiation to astrocytes
We have adapted a previously published protocol for the differentiation of hiPSC-derived astrocytes [16]. Briefly, due to the same origin of neurons and astrocytes, hiPSCs were differentiated into neuroepithelial cells by using the same procedure as for neuron differentiation for the first 10 days. The neural progenitors were detached to ultralow attachment dishes and expanded in astrocyte sphere medium, i.e., DMEM/F12 medium supplemented with 1% N 2 supplement, 2 mM Glutamax, 50 IU/ml penicillin, and 50 μg/ml streptomycin (all from Gibco), 5000 KY/ml Heparin (LEO), 10 ng/ml bFGF and 10 ng/ml EGF (both from R&D Systems). Half of the medium was renewed every other day and the spheres were manually cut once a week. According to our experience, this method generates a homogenous population of astrocyte progenitor cells within 4 months of differentiation. The astrocytes were further maturated on Matrigel-coated plates by treatment with 10 ng/ml CNTF and 10 ng/ml BMP4 (both from PeproTech) for 1 week in a density of 600,000 cells/6-well plate or 80,000 cell/24well plate. Methods concerning immunocytochemistry, RNA isolation, gene expression profile, qRT-PCR, and quantitative proteomic analysis are described in detail in Supplementary Material.

Results
Supplementary Tables 2 and 3 show all differentially expressed genes in neurons and astrocytes, respectively, up to nominal significance (p < 0.05) between violent offenders and control subjects. Since cultivation of neurons failed from cells of one healthy control, there were 14 individuals in the analyses concerning neurons, and 15 individuals in the analyses concerning astrocytes. Differentially expressed genes surviving correction for multiple comparisons in cortical neurons are shown in Table 2. Of these genes, in neurons, ribosomal RPL10P9 pseudogene showed over tenfold upregulation in violent offenders as compared with healthy controls and nonviolent individuals with substance abuse. Also zinc finger protein 132 (ZNF132) gene was markedly upregulated, and cadherin 5 (CDH5) gene markedly downregulated among the neurons derived from cells of violent offenders. Figure 1 displays the qPCR replications of these results (except for RPL10P9 due to no suitable primers being available), and correlations between gene expression levels and psychopathy score (PCL-R). Pearson's correlations between gene expression and PCL-R score were 0.67 (p = 0.013, N = 13; see Fig. 2) for RPL10P9, 0.96 (p = 0.000, N = 13) for ZNF132, −0.65 (p = 0.015, N = 13) for CDH5, and −0.55 (p = 0.05, N = 13) for opioid receptor delta 1(OPRD1). CDH13 gene encoding a cadherin that regulates axon growth during neural differentiation, has been previously linked to extremely violent behavior [5], but it did not achieve statistically significance (p = 0.24) in this study, possibly because it is most prominently expressed by oligodendrocytes in the brain. Although the result for RPL10P9 could not be verified with qPCR due to missing suitable primers, the same result for significant correlation with PCL-R score and upregulation of the gene in the gene array was also discovered in the astrocytes differentiated from the hiPSCs lines [Pearson's correlation 0.66 (p = 0.007, N = 14)] (Fig. 2). Altogether, the data indicate the robustness of this finding and underline the importance of RPL10P9 in the pathophysiology of psychopathy. In the astrocytes, also mitochondria encoded 16S RNA (MT-RNR) 2 showed a four-to sixfold upregulation of RNA expression in the violent offenders (Fig. 2).
Interestingly, opioid receptor delta 1 (OPRD1) gene was upregulated in nonviolent offenders as compared with normal controls, but the expression for OPRD1 was even lower in the violent offenders as compared with the two other groups, although these results reached only borderline statistical significance (Fig. 1c, Suppl. Table 2). As the OPRD1 gene codes for an opioid receptor delta protein, a protein involved in mediating the effects of opioids often used for substance abuse purposes, and as many violent antisocial offenders often suffer from substance abuse disorders, it was striking that the expression level of this protein was low in the violent offenders. Thus, we did further qPCR analysis for OPRD1 in order to see whether differences between the groups arise in this more accurate analysis. The qPCR analysis pointed toward the same trend of lower expression of OPRD1 in the violent offender group, although this result again reached only borderline statistical significance. Taking all together, ZNF132, RPL10P9, CDH5, and OPRD1 genes explained 30-92% of variance of the psychopathy symptomatology, as measured by the PCL-score. Moreover, in fibroblasts, no differences between the studied groups were detected (Suppl. Fig. 4). These findings point out that ZNF132 is mainly overexpressed in neurons and most likely affect the transcriptional regulation of other genes. Results from pathway analyses from neurons are displayed in Supplementary  Table 4, and show enrichment in several immune responserelated pathways. No statistically significant enrichments were observed in astrocytes.
Data from proteomic analysis are shown in Fig. 3. Here the largest effect sizes were observed for opioid-binding protein/cell-adhesion molecule (OPCML) in the proteomic analysis and for PSMD3, PEG10, and PCDH19 in phosphoproteome analysis. Of the proteins with significantly higher levels in the proteome analysis, OPCML was the most elevated protein in the violent offenders as compared with controls (6.6-fold change, p = 9.5 × 10 −3 ). However, also nonviolent substance abusers showed higher OPCML values than controls, indicating that this finding may not be necessarily specific for psychopathy but could be associated with substance dependence.

Discussion
To our knowledge, this is the first study to reveal significant alterations in gene expression related to psychopathy. Our results showed that expression levels of RPL109, ZNF132, CDH5, and OPRD1 genes in neurons explained 30-92% of the severity of psychopathy, and RPL109 expression was significantly associated with degree of psychopathy also in astrocytes. It is remarkable that all the aforementioned genes except OPRD1 have been previously linked to autism [17][18][19][20][21][22], and might thus contribute to the emotional callousness and lack of empathy observed in psychopathic violent offenders. The strongest association was observed for ZNF132, a member of zinc finger Kröppel family associated with several developmental and malignant disorders [23]. It has been also reported that autism gene CHD8 modifies the expression of ZNF132 [18]. The exact function of ZNF132 is unknown but it may be involved in transcriptional regulation. Interestingly, the highest expression levels of ZNF132 mRNA were seen in cortical neurons of violent subjects, while in hiPSCs, no difference between violent and nonviolent subjects was observed. ZNF132 is expressed highly in the cerebellum [24], and a recent study has found that cerebellum can regulate social behavior by controlling dopamine release [25], suggesting that this may contribute to mental disorders, such as autism and schizophrenia. Our results imply that cerebellum may also have a role in severe antisocial behavior. We observed enrichment in several immune responserelated pathways. This is an interesting finding since a recent study on adult antisocial behavior found enrichment in 7 gene sets, most of which being immune related [9]. This suggests that altered immune response contributes to the pathophysiology of antisocial behavior. In proteomic analysis, the most robust finding was upregulation of OPCML. It has been shown to have an accessory role in opioid receptor function, and the gene encoding the protein is highly conserved in mammals. In rats, the accessory role to activate opioid receptors has been shown to be specific for the mu receptor ligands. Differences in OPCML gene expression have also been detected in patients with schizophrenia, although protein level measurements from post mortem brains have not differed between patients and healthy controls [26].
In phosphoprotein analysis, several proteins were upregulated. Of these, PEG10 is a paternally imprinted gene that uses a rare mechanism for encoding for two different protein products by using the −1 ribosomal frameshift translation, which is well known from retroviruses and retrotransposons, but is extremely rare in humans [27]. In adult mice, the protein is expressed only in the brain and testes, and blocks TGF-B signaling. A paternally imprinted gene such as this one could explain why psychopathy is inherited from father to son. In this study, three of six offenders had a biological father convicted into prison, while none of the mothers had been imprisoned. PCDH19 is a protocadherin, which has been linked to epilepsy [28], autism [29] and behavioral problems, aggression, and photosensitivity. PCDH19 is thought to be a calciumdependent cell-adhesion protein that is primarily expressed in the brain, and has been shown to cause a decrease in the amount of neurosteroids, including adrenocorticotropic hormone, in females. ACSL4 has been associated with X-chromosome linked mental retardation [30] as well as insulin secretion [31]. On the other hand, PSMD3 and MYO1e were substantially downregulated compared with controls. Of these, PSMD3 is an enzyme, an aberration of which contributes to pathogenesis of neurodevelopmental  and neurodegenerative disorders [32,33] and insulin resistance [34]. This finding suggests that downregulation of PSMD3 contributes to abnormal glucose metabolism which results into impulsive violent behavior among severely antisocial individuals as has been reported in several studies [14,35]. MYO1e has been associated with autism in a single study [36].
In conclusion, expression of ZNF132 in neurons and RPL10P9 in both neurons and astrocytes is markedly abnormal among habitually violent offenders and these findings are strongly associated with the degree of psychopathic symptoms. The changes in protein levels observed here point to alteration in insulin sensitivity and glucose metabolism, and previous literature has shown that abnormal glucose metabolism is the only predictor for violent crimes which can surpass the accuracy of PCL-R [35]. We also observed changes in the opioid system, which has been shown to support prosocial functions, such as empathy, among humans and nonhuman primates [12,13,37,38]. Our results showing a decrease in the expression of opioid delta receptor gene are in line with these previous findings. A recent theory suggests that a deficient endogenous opioid system contributes to antisocial personality, proposing that antisocial individuals attempt to stimulate their dysfunctional opioid system by the rewarding effect of substance abuse, and impulsive, sensationseeking, aggressive, and promiscuous behavior [11]. Our data suggest that dysfunction of the opioid system contributes to the phenotype of psychopathy, supporting the recently presented idea that partial opioid receptor agonists, such as (+)-naloxone might be the first effective treatment for psychopathy [11].

Data availability
All data needed to evaluate the conclusions in the manuscript are provided in the manuscript or the supplementary material.
Author contributions JT and JK conceived the study. ŠL planned and supervised iPSC lines characterizations, differentiation the neurons and sample preparation for RNA and protein sequencing. IO and OV performed skin biopsies and rating of symptoms, KAP derived iPSC lines with YG and characterized them. IH grew and differentiated neurons and PLJV characterized them. MK differentiated the neurons, and prepared RNA and protein samples for sequencing, run qRT-PCR.
JT wrote the first draft of the manuscript with help of JK, ŠL prepared the figures and tables, and all authors contributed to the final version.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.