A Systematic Review of LINE-1 Methylation Profile in Psychiatric Disorders

Long interspersed nuclear elements (LINEs) are endogenous retrotransposable elements, also known as “jumping or mobile DNA” [1]. They are DNA sequences that move from one location to another in the genome and replicate via reverse transcription using an RNA intermediate through the “copy and paste” mechanism [2]. LINEs are abundantly repeated throughout the genome, representing about 21% of its extension, and are divided into three groups: LINE-1, LINE-2, and LINE-3 [3, 4]. LINE-1 is the only active mobile DNA in humans, defined by the presence of a 5′-untranslated region (5′ UTR), three open reading frames (ORFs – ORF0, ORF1, and ORF2), and a 3′ UTR region containing a polyadenylation tail signal [2, 5‒7].

LINE-1 may influence the development of phenotypes or diseases by various mechanisms: retrotransposition, promotion of intracellular inflammation, changes in host gene expression, and impact on global hypomethylation. In the human genome, around 80–100 LINE-1 sequences are competent for retrotransposition, a process that might affect gene sequence and expression in various genomic regions [8]. Additionally, deleterious effects can occur as a response to some by-products of the retrotransposition process [9]. For instance, instability and mutations can be caused by the endonuclease activity of ORF2p, and the expression of LINE-1 can trigger the immune response, neuron degeneration, and inflammation as the result of accumulated RNA and/or DNA in the cytosol [9, 10]. Also, gene expression can be decreased by intragenic LINE-1 RNAs via the nuclear RNA-induced silencing complex [11]. The methylation levels of transposable elements, such as LINE-1, have been associated with the global DNA methylation in humans – LINE-1 individually comprises about 17% of the genome. Hypomethylated genome regions are more susceptible to cumulative oxidative damage, double-strand breaks, depurination, and other types of DNA damage. Thus, changes in the pattern of LINE-1 methylation levels may lead to genomic instability and alter biological mechanisms, such as those linked to psychiatric disorders [11, 12].

As reviewed by Suarez et al. [8], it is already known that LINE-1 is active in the human brain; moreover, it is possible that this retrotransposon could present an activity early on in the central nervous system formation (i.e., in utero development). Additionally, a higher number of copies of LINE-1 ORF2 were found in the adult hippocampus compared to other brain regions [8]. Therefore, the correct control of LINE-1 activity seems to be essential to cell metabolism and system development.

The DNA methylation process is an important regulator of the activity of LINE-1 (i.e., DNA hypermethylation leads to lower LINE-1 expression, while DNA hypomethylation leads to higher LINE-1 expression) and can be described in such cases even as a defense mechanism against retrotransposition. However, this mechanism may result in changes in the chromatin conformation due to the hypermethylation of a newly inserted transposable element, further triggering alterations in the expression of adjacent genes [11].

Furthermore, despite the existence of epigenetic defense mechanisms, in a diseased state with diminished retrotransposon repression/regulation, an upregulation of endogenous nucleic acids is observed which later promotes a host response similar to that of a viral infection or an environmental trigger. In these cases, generally, an interferon-mediated response is triggered and can possibly result in functional abnormalities and further disease phenotypes if the inflammation process becomes persistent [13].

Recent studies suggest that retrotransposons, such as LINE-1, can often be improperly regulated and hypomethylated in mental disorders, such as autism spectrum disorder (ASD) and schizophrenia (SCZ) [12, 14‒16]. However, some studies have found a LINE-1 hypermethylation in patients with SCZ and mood disorders [17, 18]. These contrasting outcomes along with the suggested role of LINE-1 in brain functioning, inflammation triggering, somatic retrotransposition, transcription interruption, insertion of mutations or DNA breaks, and genomic instability in cells demonstrate the importance of investigations toward the association of LINE-1 methylation with psychiatric disorders and a need for better understanding of its biological mechanisms. Furthermore, it is worth mentioning that environmental factors such as childhood trauma events may also affect the levels of LINE-1 methylation, further triggering changes that can be related to mental disorders [19]. However, the mechanisms involved in this process have not yet been established.

Based on these heterogeneous findings, we systematically compiled the present studies that evaluated the pattern of LINE-1 methylation across psychiatric disorders. With that, we unify the existing knowledge in the field, seeking to provide a better understanding toward the link between psychiatric disorders and LINE-1 methylation and instigate future research on this topic.

A systematic literature review was executed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Articles included in the present review were retrieved in April 2022 from the following databases: PubMed (www.ncbi.nlm.nih.gov/pubmed), MEDLINE (Medical Literature Analysis and Retrieval System Online – www.pubmed.ncbi.nlm.nih.gov), Embase (www.embase.com), and Scopus (www.scopus.com).

The following keywords were used for online searching: (LINE-1) AND (methylation) AND (i) (psychiatric disorder); (ii) (mental disorder); (iii) (mental illness); (iv) (behavior disorders); (v) (posttraumatic stress disorder); (vi) (bipolar disorders); (vii) (depression); (viii) (obsessive compulsive disorder); (ix) (anxiety); (x) (panic disorders); (xi) (phobic disorders); (xii) (autism); (xiii) (conduct disorders); (xiv) (oppositional defiant disorder); (xv) (personality disorder); (xvi) (attention deficit disorder with hyperactivity); and (xvii) (schizophrenia). Considering unique biological features of substance use disorders and eating disorders, we chose not to include these disorders in our search.

Articles were selected using the following inclusion criteria: (i) publication date ranging between January 2000 and April 2022; (ii) written in English language; (iii) LINE-1 methylation measured in human tissues, including peripheral blood and postmortem brain tissue.

We excluded studies published outside of the established year range, those written in a language other than English, conducted in animals, or review papers.

The studies retrieved from our search were assessed for eligibility by two reviewers independently (n = 58). First, the titles and abstracts were screened and then assembled in a table for subsequent exclusion of duplicates. After duplicates were excluded (n = 25), all articles remaining were read in full and further exclusions were made by following the exclusion criteria. The search process is shown in Figure 1. A total of 12 manuscripts were eligible to conduct this systematic review.

Studies included in this review used six different methods for analyzing the LINE-1 methylation pattern: pyrosequencing (n = 5); combined bisulfite restriction analysis (COBRA; n = 3); polymerase chain reaction (PCR; n = 1); methylation-sensitive restriction enzymes followed by quantitative RT-PCR (MSRE RT-PCR; n = 1); bisulfite conversion-specific one-label extension (BS-OLE; n = 1); and methylation array (n = 1, “Illumina HumanMethylation450 BeadChip array”; the resulting DNA methylation data were used to infer LINE-1 methylation levels). A summary of each test’s advantages and disadvantages can be found below.

Pyrosequencing requires bisulfite conversion of DNA and it can be used for both CpG-poor and CpG-rich regions. This method has a single-base resolution, therefore acquiring a methylation level for each CpG in the region. The main limitation is that only regions with maximum 350 bp can be analyzed [20].

COBRA requires bisulfite conversion of DNA followed by restriction digestion and quantitation step. It can be used to determine methylation levels and patterns at specific gene loci in small amounts of genomic DNA. However, the PCR amplification step can yield unwanted artifacts [21‒23].

PCR requires bisulfite conversion of DNA; the methylation is examined by two sets of primers – one specific for a methylated state and another for an unmethylated state. One limitation is that it can be challenging to find a convenient set of primers and it can also yield PCR artifacts due to over-amplification [20].

MSRE RT-PCR does not require bisulfite conversion of DNA and is a simple and rapid method. It is characterized by a high specificity, however, an important limitation of this method is that only the specific restriction sites can be analyzed, not being suitable for intermediately methylated loci [20].

BS-OLE requires bisulfite conversion of DNA and it analyzes the methylated C on a single strand DNA through a single-nucleotide primer extension. A limitation of this method is that only a single CpG site can be measured per reaction [22].

Methylation array (Illumina HumanMethylation450 BeadChip array) requires bisulfite conversion of DNA; afterward, the DNA is hybridized to two types of probes – methylated and unmethylated. This array has a coverage of over 48,000 CpG sites, being enriched for 99.3% CpG residues. It is worth noting that there is an inevitable generation of biases due to the inclusion and preselection of specific probes that cover only previously identified CpG sites. Also, the “co-methylation assumption” assumes that CpG loci close to the analyzed sites will have similar methylation or unmethylation [24].

In this review, we focused on a range of psychiatric disorders. Here we describe how these disorders were assessed by the studies selected for this review. Schizophrenia (SCZ), paranoid schizophrenia, first-episode schizophrenia (FES), first-episode psychosis (FEP) were assessed by the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) and/or International Classification of Diseases 10th Revision (ICD-10). Post-traumatic stress disorder (PTSD) was assessed with ICD-9. ASD was assessed by the DSM-IV, the Autism Diagnostic Interview Revised (ADI-R), and/or Autism Diagnostic Observation Schedule (ADOS). Major depressive disorder (MDD), bipolar disorder (BPD), and panic disorder (PD) were assessed by using the DSM-IV.

A total of 58 studies were identified in the initial online search. All duplicated articles were removed; then titles and abstracts were assessed. The remaining 33 articles were read in full, and 21 studies were further removed by following the exclusion criteria. Finally, 12 articles were eligible for this review (shown in Fig. 1).

After the selection, the included articles were grouped based on the psychiatric disorder assessed, as follows: (i) psychotic disorders (SCZ, paranoid schizophrenia, FES, and FEP); (ii) PTSD; (iii) mood disorders (MDD and BPD); (iv) anxiety disorders; and (v) ASD. The characteristics and results of all included studies are summarized in Table 1 and shown in Figure 2.

Li et al. [14] investigated the levels of LINE-1 methylation in SCZ and BPD patients at 3 different CpG sites (S1, S2, and S3) from the LINE-1 5′ UTR region. Compared to healthy controls, LINE-1 methylation levels were significantly decreased by 8% in S2 (p < 0.001) and by 5% in S3 (p = 0.002) in BPD subjects. Methylation was also decreased by 5% in S1 (p < 0.001) and by 3% in S3 (p = 0.002) in SCZ patients. Significant differences were also observed in the S1 and S2 methylation levels between these disorders, with an increased S1 methylation level (by 5%, p < 0.001) and a decreased S2 methylation level (by 12%, p < 0.001) in BPD compared to SCZ. Significant differences were found in the S3 methylation levels between male controls and male subjects with SCZ (p_covariate < 0.001) or BPD (p_covariate = 0.008) with age as a covariate.

Murata et al. [16] determined the LINE-1 methylation levels in FES, chronic SCZ, and BPD patients and their correlation with betaine (a methyl-group donor) level in peripheral blood. When compared to non-psychiatric controls, LINE-1 methylation levels were significantly decreased in FES (W = 661, p = 0.037) and chronic SCZ patients (W = 71,429, p < 0.001). It was also positively correlated with the level of betaine in plasma (R = 0.349, p = 0.012) and negatively correlated with clinical parameters such as the Global Assessment of Functioning (GAF) scale (R = −0.543, p = 0.011). Authors also found a global DNA hypomethylation for patients with BPD type I but not for those with type II.

Fachim et al. [17] determined the LINE-1 methylation levels in postmortem brain (hippocampus and prefrontal cortex) of SCZ subjects. LINE-1 mean methylation levels were elevated in the hippocampus (t = 2.786; p = 0.009) and the frontal cortex (t = 2.995; p = 0.006), when compared to healthy controls.

Misiak et al. [19] investigated the methylation of LINE-1 in FES. The FES group was divided into two categories: with and without history of childhood trauma. FES subjects with history of trauma presented a significantly lower LINE-1 methylation level when compared to FES subjects without history of trauma (p = 0.023) or healthy controls (p = 0.016). Furthermore, the scores of emotional abuses (p = 0.045) and total trauma (p = 0.037) were able to predict lower LINE-1 methylation level in FES patients.

Fachim et al. [25] examined the LINE-1 methylation levels in FEP patients, their siblings (representing a high-risk group), and healthy controls. The authors observed a statistically significant increase in LINE-1 methylation for both FEP patients and their siblings, when compared to the control group (p = 0.001).

Marques et al. [26] investigated the LINE-1 methylation profile in antipsychotic-naïve FEP patients, before and after 2 months of risperidone treatment. Authors observed lower LINE-1 methylation levels in FEP patients before treatment with risperidone when compared to healthy controls. No significant differences were observed comparing patients before and after risperidone treatment. In fact, the hypomethylation remained present when comparing FEP patients after treatment and healthy controls (p = 0.033). Of note, patients who did not respond to risperidone treatment showed lower methylation levels than those who responded (p = 0.036). Also, a positive correlation was observed between the responsiveness to treatment and the methylation levels of LINE-1 CpG1 site (p = 0.005).

Kalayasiri et al. [23] examined the LINE-1 methylation patterns in SCZ and methamphetamine-induced paranoia (MIP). Here the authors analyzed the methylation pattern from CpG dinucleotides individually: 2 methylated (mCmC), 2 unmethylated (uCuC), and the partially methylated loci (mCuC/uCmC). The percentage of mCuC was statistically different when comparing all groups, decreasing from SCZ>MIP>healthy controls (p < 0.001), whereas the percentage of uCmC presented an opposite direction, with controls having higher LINE-1 mean methylation values (p < 0.001) than the others.

Rusiecki et al. [27] examined LINE-1 methylation patterns in PTSD patients from US military soldiers. The results showed that the percent of methylated cytosines/sites was higher (p value = 0.01) for the control group at the post-deployment time point (µ: 78.5; SD: 1.7) when compared with the pre-deployment time point for the same group (µ: 77.7; SD: 1.8). Considering the post-deployment time point, authors found a negative association for LINE-1 methylation (OR: 0.82; 95% CI: 0.67–1.01) when comparing PTSD cases versus controls in the study population.

Reszka et al. [18] investigated changes in the global methylation status in women with depression diagnosed with BPD and MDD. BPD type I in women with a depressive episode showed higher percent of methylated cytosines/sites when compared to the other groups (i.e., controls, MDD, and BD-II) (p < 0.05). Additionally, the methylation level of citocines in non-psychiatric controls was positively correlated with LINE-1 methylation (r = 0.566, p < 0.00001).

Liu et al. [28] determined the levels of LINE-1 methylation in MDD patients. A significantly lower methylation level was observed in the LINE-1 5′ UTR region (p < 0.0001) in MDD patients when compared to healthy controls. Li et al. [14] and Murata et al. [16] also assessed LINE-1 methylation levels in BPD, and the results were described in the previous section.

Petersen et al. [29] investigated the leukocyte proportions and the methylation status of repetitive elements in PD patients. Authors found LINE-1 hypomethylation for PD cases in comparison to the control group (difference in mean beta value = 0.002, p < 0.001), remaining statistically significant after adjustments for sex, age, and years of study.

Tangsuwansri et al. [12] examined the DNA methylation levels of LINE-1 from ASD patients. The overall LINE-1 methylation percentage showed a statistically significant decrease (p = 0.039) only in the lymphoblastoid cell lines from ASD patients presenting a severe language impairment.

In the present study, we reviewed whether the level of LINE-1 methylation is associated with psychiatric disorders (i.e., SCZ spectrum disorders – FEP, FES; depressive and anxiety disorders; ASD; and PTSD). The studies included in this review used different methods to measure LINE-1 methylation, such as pyrosequencing, COBRA, PCR, MSRE RT-PCR, BS-OLE, and methylation array. Although the findings of this analysis are mixed between hypo- and hypermethylation, most of the studies point to an association between patients with psychiatric disorders and LINE-1 hypomethylation.

In general, for psychotic disorders, lower LINE-1 methylation level was detected in 4 out of 7 studies assessed, whereas for mood disorders, we observed both hypo- and hypermethylation (found in 3 and 1 study, respectively). ASD, PD, and PTSD patients also presented lower levels of LINE-1 methylation.

For psychotic disorders specifically, the reviewed findings show a common hypomethylation of LINE-1 in SCZ and BPD patients. However, differences in methylation levels were observed between these two disorders and also within the two types of BPD [14, 16]. This points to a singularity among disorders and even between the BPD types in terms of LINE-1 methylation, emphasizing the need for further studies to help clarify the individualities of each psychiatric disorder.

Pointing in the opposite direction, Kalayasiri et al. [23] found higher partial methylation of mCuC in paranoid SCZ and MIP patients. Additionally, the findings in the study by Fachim et al. [17] show higher LINE-1 methylation patterns in postmortem brain tissues of SCZ patients when compared to controls. Opposite results were also detected in studies performed on FEP/FES patients: Fachim et al. [25] found higher LINE-1 methylation levels in FES patients, while Marques et al. [26] and Murata et al. [16] found lower LINE-1 methylation levels in FEP patients. These contrasting outcomes demonstrate the importance of novel research on the pattern of LINE-1 methylation in SCZ spectrum disorders (FEP/FES/SCZ) to bring knowledge about the relation not only between this retrotransposon and the psychiatric disorder itself but also regarding the particularities of each tissue analyzed (postmortem brain and peripheral blood).

Moreover, FES subjects with history of childhood trauma also presented lower methylation level in comparison to both FES subjects with no history of trauma and controls [19]. The authors suggested that the experience of traumatic events during childhood was also associated with lower LINE-1 methylation level. Additionally, regardless of the patient’s self-perception of having been exposed to an adverse experience, it was possible to predict the lower levels of LINE-1 methylation based solely on the number of traumatic events experienced during childhood [19].

These results agree with the findings of Rusiecki et al. [27] that observed hypomethylation of LINE-1 in PTSD cases after military deployment, compared to the control group of post-deployment with no PTSD. This finding points to a potential association of LINE-1 methylation level with vulnerability factors related to traumatic experiences, possibly being suggestive of resilience. Consequently, the LINE-1 hypermethylation pattern found in the control group (military personnel with no PTSD) could be suggestive of possible resilience factors.

However, there is still little evidence to state whether hypomethylation at different LINE-1 CpG sites is due to trauma experience or due to the course of the disease or even if some regions of the genome are naturally more prone to undergo methylation. Future studies are needed to understand the mechanisms underlying LINE-1 methylation pattern alterations.

Although there are different studies showing LINE-1 association with both SCZ spectrum disorders and exposure to trauma, there is still little evidence of its association with anxiety and mood disorders (PD and MDD) or ASDI. In the studies assessed for this review, LINE-1 hypomethylation was observed in ASD, PD, and MDD patients [12, 28, 29], whereas Reszka et al. [18] found a LINE-1 hypermethylation level for both MDD and BPD patients versus controls. These diverging results could be related mainly to the pathophysiology that characterizes the different disorders assessed since they are considered to be multifactorial illnesses – therefore linked to one or more changes in psychological, social, and biological mechanisms.

Furthermore, mounting evidence suggests an association between global hypomethylation and genomic instability, linking LINE-1 methylation to biological mechanisms related to psychiatric disorders [11, 12]. Different studies have previously associated the methylation levels of LINE-1 with human global DNA methylation [2, 8, 11]. This was also observed by studies that compared the methylation of LINE-1 and of the transposable element Alu – another marker for global DNA methylation – against the gold standard high-performance liquid chromatography [30, 31].

Nonetheless, one of the studies assessed in this review reported that the LINE-1 methylation levels were not as efficient as expected at predicting global DNA methylation [18]. Future studies are needed to better understand these diverging results.

As reviewed by Saleh et al. [13], evidence suggests that retrotransposable elements, such as LINE-1, may interfere with gene expression as a result of genomic structural variation/instability generated by those elements and also by triggering neuroinflammation. However, there are still gaps in our knowledge of how these elements could be influencing the etiology and the pathophysiological pathways leading to disease development – not only for psychiatric but also for inflammatory and neurodegenerative disorders.

Additionally, it has also been discovered that an early dysregulated retrotransposition at neurodevelopment can possibly contribute to the phenotypes associated with neurological disabilities (e.g., ASD) and even with late-onset disorders [8].

The differences found in the LINE-1 methylation pattern of psychotic disorders among the studies previously cited [14, 16, 17, 23, 25, 26] might be influenced by some factors: (i) disease severity– suggesting that the neurobiology of the disease may be related to differences on LINE-1 methylation pattern; (ii) treatment of patients and other environmental factors – mechanisms linked to, for example, the use of antipsychotic drugs or childhood trauma may influence on DNA methylation level; (iii) different human genomic variation – specific LINE-1 methylation pattern may be associated with genomic structure; (iv) limitation of the method used to infer LINE-1 methylation levels (e.g., COBRA which is less specific compared to pyrosequencing); and (v) the different tissue samples and cellular composition used in the analysis (i.e., postmortem brain tissue samples and peripheral blood samples). Therefore, it is important to take these factors into account when studying LINE-1 methylation patterns regarding psychiatric disorders (e.g., controlling/correcting for disease severity or to antipsychotic drugs; assessing childhood trauma; or accounting for ancestry).

The association between LINE-1 methylation and different psychiatric disorders based on findings of 12 eligible studies was summarized in this review. Despite the current limited number of published articles in the literature, mounting evidence confirms the relationship between different patterns of LINE-1 methylation and psychiatric disorders.

Additionally, the studies reviewed were conducted in a variety of populations, including analysis with Asian, Latin American, North American, and European subjects, which globally portrays the different results found for LINE-1 methylation pattern and its association with different psychiatric disorders. Moreover, eleven out of the twelve studies analyzed had both female and male subjects with an age distribution ranging from 18 to 80 years old, further demonstrating that these results can be broadly used in associations with populations from different countries and with different age ranges.

In regard to LINE-1 methylation patterns, despite the majority of the studies assessed having observed LINE-1 hypomethylation in the patients’ groups, there were still some results pointing in the opposite direction (i.e., LINE-1 hypermethylation). This highlights the need for future research regarding LINE-1 methylation patterns across different psychiatric disorders to help clarify the role of LINE-1 methylation as an epigenetic biomarker for those disorders, thus, possibly, allowing for future diagnosis methods based on the pattern of LINE-1 methylation.

There is also a need to further investigate the roles of LINE-1 in the pathophysiology of each psychiatric disorder, aiming to improve our understanding of the biological mechanisms that underlie LINE-1 in this context. The understanding of how LINE-1 methylation is linked to the development of different psychiatric disorders may assist in future research focused on clinical purposes, such as new treatment approaches and methodologies.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors declare that they have no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

V.R.G.A., A.V.G.B., and D.M. were supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Brazil (CAPES) – Finance Code 001. C.M.C. was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP 2020/10599-0).

V.R.G.A.: conceptualization, methodology, writing – original draft, and investigation. D.M.: writing – review and editing, design of figures. V.K.O., A.V.G.B.: writing – review and editing. C.M.C.: conceptualization, methodology, writing – original draft, writing – review and editing, investigation, supervision, and project administration. S.I.B.: writing – review and editing, supervision, and project administration.

Carolina Muniz Carvalho and Sintia Iole Belangero contributed equally to this work.

The data that support the findings of this study are available on PubMed at http://doi.org/10.1371/journal.pone.0201071, http://doi.org/10.1016/j.jpsychires.2018.10.009, http://doi.org/10.1016/j.pnpbp.2019.109855, http://doi.org/10.2217/epi-2017-0159, http://doi.org/10.1016/j.jpsychires.2021.03.003, http://doi.org/10.2217/epi.15.68, http://doi.org/10.2217/epi-2018-0127, http://doi.org/10.2217/epi-2019-0350, http://doi.org/10.1016/j.schres.2019.02.015, http://doi.org/10.2217/epi.11.116, http://doi.org/10.1038/srep37530, http://doi.org/10.1186/s13148-020-00972-9, reference number [12, 14, 16–19, 23, 25–29]. Further inquiries can be directed to the corresponding author.