The relationship between DNA methylation and genomic instability has been the subject of intensive investigation in the past few decades, especially in cancer research, but the connection between DNA methylation and genomic instability at the 1p36 region in constitutional chromosomal rearrangements remains to be elucidated. Both DNA hypomethylation and hypermethylation are associated with increased genomic instability that can lead to chromosomal rearrangements associated with malignant transformation (Brabson et al. 2021). DNA hypermethylation can lead to genomic instability by the inhibition of the recruitment of DNA repair proteins or a silencing of DNA repair genes (Toffolatti et al. 2014; Tsuboi et al. 2020; Brabson et al. 2021), and DNA hypomethylation can reduce the formation of heterochromatin leading to transposition of DNA at repetitive regions in the genome (Brabson et al. 2021). As repetitive sequences could be a source of genome instability, most of repetitive elements are usually methylated to maintain a heterochromatic state (Pappalardo and Barra 2021). It is known that alteration of the epigenetic pattern of repetitive sequences is characteristic of many complex diseases, thus it is difficult to understand if it is the cause or the consequence of the disease (Pappalardo and Barra 2021). However, there is increasing evidence showing that repetitive elements are frequently hypomethylated, which correlates with chromatin relaxation and unscheduled transcription, in various of human pathologies from cancer to psychiatric disorders (Lamprecht et al. 2010; Pappalardo and Barra 2021).

A possible explanation for the negative impact of hypomethylation of CpG–rich repeated elements on genomic stability has been proposed by De and Michor (De and Michor 2011). Significant enrichment of hypomethylated G-quadruplex sequences (G4s) in the near vicinity of somatic copy–number alteration breakpoints has been found, e.g., in many types of cancer, while information concerning this aspect in constitutional rearrangements involving 1p36 region has not been investigated. G-quadruplex structures are known to slow down the movement of DNA polymerase, increasing the likelihood of genomic rearrangements by such mechanisms as NAHR, non–homologous end joining (NHEJ), or fork stalling and template switching (FoSTeS) (Lupski 2015; Burssed et al. 2022). Their formation, however, depends on the open chromatin state and DNA accessibility, which is usually restricted by the hypermethylation of G4 sequences in normal tissues. Aberrant genome-wide DNA hypomethylation patterns resulting from carcinogenesis or a glitch in DNA methylation machinery suggested for the Hylobatidae family (Carbone et al. 2009) may therefore increase the occurrence of G4 structure formation and chromosome breaks in a tissue–specific manner (De and Michor 2011).

As far as we know, there are no published reports concerning abnormalities in methylation level in the monosomy 1p36 hotspot region. In the only article concerning methylation in a constitutional chromosomal syndrome, the haploinsufficiency of SPEN (NM_015001.3; chr1:16,174,202–16,266,951; GRCh37/hg19) was found to be associated with a distinctive X chromosome episignature in females with interstitial deletions at 1p36.21p36.13 (Radio et al. 2021). Although those reported 1p36 deletions do not overlap (12,700,001–20,400,000 bp) with our hotspot rearrangement region, patients have phenotypes similar to those seen in terminal 1p36 deletions such as intellectual disability, hypotonia, behavior abnormalities, multiple congenital anomalies, and facial dysmorphisms (Gajecka et al. 2007), and additionally obesity and increased BMI (Radio et al. 2021). Eleven individuals with interstitial deletions of 1p36.21p36.13 and truncating SPEN variants were included in the genome-wide methylation analyses, which indicated that SPEN haploinsufficiency was related to methylation changes on the X chromosome (Radio et al. 2021).

The human genome consists of different classes of repeats enriched with CpGs that are silenced into heterochromatic regions by both DNA methylation and repressive histone modifications (Brabson et al. 2021). The dynamic relationship between transcriptionally active subclasses of retroelements and DNA methylation has also been revealed, with abundant transcription of transposable elements serving as a boundary for changes in CpG methylation (Hoyt et al. 2022). The results of our analyses indicate that the 1p36 breakpoint hotspot region (chr1:4,000,000–5,500,000) meets all the prerequisite criteria for the aforementioned mechanism of a glitch in DNA methylation machinery, exhibiting high GC content and suggesting dysregulation of DNA methylation in monosomy 1p36 patients. Each class of repeat elements annotated in UCSC databases showed statistically significant higher GC content in the 1p36 breakpoint region compared to the rest of the genome, which is not surprising given its overall high GC content and high level of recombination reported for this chromosome region (D’Angelo et al. 2009). We found that the whole region has a mean GC content of 47.06%, which makes it at the 91st percentile in the human genome (40.78% mean GC content). A similar trend was observed in the 22q11.2 breakpoint region (48.47%, 94th percentile) and, to a lesser extent, the 18q21.1 region (44.21%, 80th percentile). Conversely, GC content of 9p22 region is significantly lower (39.42%, 44th percentile).

Our findings agree with the identified changes in methylation pattern in 22q11.2 deletion syndrome (Rooney et al. 2021). In genome-wide DNA methylation analyses of peripheral blood from 49 patients with 22q11.2 deletion syndrome, an evidence of a unique and highly specific episignature in typical and proximal 22q11.2 deletion syndromes was found (Rooney et al. 2021). The methylation pattern of patients with 22q11.2 deletion syndrome differed significantly from control individuals and > 1500 patients with other neurodevelopmental disorders with known episignatures (Rooney et al. 2021). Also, assessing the influence of DNA methylation on gene expression in the 22q11.2 hotspot, Starnawska et al. (2017) have suggested a relationship between changes in DNA methylation patterns at birth with development of a psychiatric disorder later in life in patients with 22q11.2 deletion syndrome. In that study, one CpG site, mapped to the STK32C gene, was associated with a later psychiatric diagnosis (Starnawska et al. 2017). In addition, differentially methylated CG dinucleotides in LRP2BP, TOP1, NOSIP, and SEMA4B were associated with intellectual disability, behavioral disorders, disorders of psychological development, and schizophrenia spectrum disorders, respectively (Starnawska et al. 2017). Pathway analysis of these genes indicated several pathways such as neurogenesis, neuron development, neuron projection development, astrocyte development, axonogenesis, and axon guidance as significantly enriched (Starnawska et al. 2017). In another study, it was found that methylation alterations in specific imprinting genes and in genes located in 6p21-p22, within the major histocompatibility complex (MHC) locus, might contribute to the development of schizophrenia spectrum disorders in 22q11.2 deletion syndrome (Carmel et al. 2021). Sixteen adult men with/without schizophrenia spectrum disorders were recruited from a 22q11.2 deletion syndrome cohort and underwent genome-wide DNA methylation analysis (Carmel et al. 2021). Differentially methylated probes and regions were enriched in two gene sets, “imprinting genes” and “chr6p21,” a region overlapping the MHC locus (Carmel et al. 2021). Most of the identified differentially methylated genes are involved in neurodevelopment and synaptic plasticity (Carmel et al. 2021).

Here, we found a complex landscape of DNA methylation at the 1p36 breakpoint hotspot region. While very few patterns can be observed in the data, the differences between patients and the parents from whom the de novo breakpoints were derived seem to be much higher than comparisons between the two parents. Moreover, these differences exhibit family–specific trends, showing either a preference for hyper- or hypomethylation changes, or both, or non-remarkable changes in DNA methylation.

It is important to remember that constitutional chromosomal and epigenetic rearrangements do not arise throughout a lifetime as in various cancers, but occur de novo in gametogenesis or early embryogenesis. It is unclear when the changes in methylation level could happen, either during germ cell development or during early embryogenesis, when global demethylation and remethylation take place. Epigenetic reprogramming in the germ cells involves the erasure of somatic methylation patterns in primordial germ cells and establishment of sex-specific germ cell methylation patterns (Zeng and Chen 2019). Whereas, reprogramming in early embryogenesis involves erasure of most methylation marks inherited from the gametes (Zeng and Chen 2019). A glitch in DNA methylation during one of those processes might predispose or lead to the chromosomal breakage. Beside embryo development and genome stability, DNA methylation is also involved in genomic imprinting, inactivation of X chromosome and repression of transposable elements (Bird 2002). In genomic imprinting only one of the two inherited alleles is expressed (monoallelic gene expression) and DNA methylation establishes imprinting marks on either paternal or maternal alleles (Elhamamsy 2017).

Another argument favoring a role of altered methylation in deletion formation is the lack of sequence features traditionally associated with genomic instability in the 1p36 breakpoint hotspot region. Previous analyses showed that motifs typically involved in genomic instability (translin sites, DNA polymerase a/b, frameshift hotspots, deletion hotspot, consensus sequences, etc.) are present in the vicinity of 1p36 breakpoint junctions but not significantly enriched (Gajecka et al. 2008). Two other important classes of genomic instability related features are direct LCRs and IP–LCRs that are known to be involved in genomic rearrangements via NAHR (Dittwald et al. 2013). We screened the whole 1p36 region (chr1:1–28,000,000) for their presence utilizing commonly used criteria (Dittwald et al. 2013). Overall, none of the LCR clusters found seem to be located in near vicinity of the 1p36 breakpoint hotspot region (chr1:4,000,000–5,500,000), which makes their participation in 1p36 genomic instability very unlikely. However, as reported by D’Angelo et al. (2009), in two terminal deletions of 1p36 that were about 25 kb apart, the breakpoint regions were refined to a genomic interval containing a series of 1p36 specific segmental duplications with 90–98% identity (D’Angelo et al. 2009), which may suggest a role in the rearrangements. Their results are consistent with previous findings indicating that genomic rearrangements directly mediated by LCRs via NAHR tend to be recurrent, sharing common size and breakpoint locations (Gu et al. 2008). Despite this, previously assessed, nonrecurrent 1p36 genomic breakpoints (Gu et al. 2008; Burssed et al. 2022), seem to be mediated by multiple mechanisms such as NHEJ (Gajecka et al. 2008) and FoSTeS (Lupski 2015).

The observed changes of DNA methylation levels in the vicinity of the breakpoint are intriguing. These results were found by the methylation array as well as pyrosequencing assays (assays 2 and 3). Assay 2 indicates a relatively lower level of methylation and assay 3 showed substantially higher level of methylation in the child as compared to the parents as well as control samples from healthy individuals. These results highlight a shift in methylation level from 15.11% (assay 2) to 70.82% (assay 3) spanning the breakpoint region. As loss of methylation results in higher genetic instability, a phenomenon frequently observed in cancer, it is tempting to speculate that the observed shift in methylation level renders DNA susceptible to breaks in this region (Esteller and Herman 2002). Our results also show that the observed changes are limited to the breakpoint region as the assays 1 and 4 located further outside the breakpoint region showed comparable methylation levels in the child, parents, and controls. It must however be noted that the methylation level shown by assays 1 and 2 in the child corresponds only to the wild-type chromosome 1, as this region is deleted on the homologous chromosome.

Study limitations include the small number of sequenced samples in this study and the assessment of somatic cells in the absence of germ cells available for testing.