Analysis of LINE-1 Elements in DNA from Postmortem Brains of Individuals with Schizophrenia

Abstract

Whereas some rare genetic variants convey high risk for schizophrenia (SZ), common alleles conveying even moderate risk remain elusive. Long interspersed element-1s (L1) are mobile retrotransposons comprising ~17% of the human genome. L1 retrotransposition can cause somatic mosaicism during neurodevelopment by insertional mutagenesis. We hypothesized that, compared to controls, patients diagnosed with schizophrenia (PDS) may have increased numbers of deleterious L1 insertions, perhaps occurring de novo, in brain-expressed genes of dorsolateral prefrontal cortex (DLPFC) neurons. Neuronal and non-neuronal nuclei were separated by fluorescence-activated cell sorting from postmortem DLPFC of 36 PDS and 26 age-matched controls. Genomic sequences flanking the 3'-side of L1s were amplified from neuronal DNA, and neuronal L1 libraries were sequenced. Aligned sequences were analyzed for L1 insertions using custom bioinformatics programs. Ontology and pathway analyses were done on lists of genes putatively disrupted by L1s in PDS and controls. Cellular or population allele frequencies of L1s were assessed by droplet digital PCR or Taqman genotyping. We observed a statistically significant increase in the proportion of intragenic novel L1s in DLPFC of PDS. We found over-representation of L1 insertions within the gene ontologies 'cell projection' and 'postsynaptic membrane' in the gene lists derived from PDS samples, but not from controls. Cellular allele frequencies of examined L1 insertions indicated heterozygosity in genomes of DLPFC cells. An L1 within ERI1 exoribonuclease family member 3 (ERI3) was found to associate with SZ. These results extend prior work documenting increased L1 genetic burden in the brains of PDS and also identify unique genes that may provide new insight into the pathophysiology of schizophrenia.

Figure 1

PCR confirmation of novel L1 RTP insertions in various genes. Gel images showing the ‘filled site’ (‘F’ L1HsT/FS or L1HsG/FS primer pair) and ‘empty site’ (‘E’ ES/FS primer pair) PCR amplicons. The L1 insertions were detected in both the NeuN-negative (−) and NeuN-positive (+) gDNA reactions. (a) ERI1 exoribonuclease family member 3 (ERI3). (b) Glutamate receptor, ionotropic, delta 2 (GRID2). (c) KH RNA-binding domain-containing, signal transduction-associated 2 (KHDRBS2). All L1s were confirmed in the pools in which each predicted L1 insert was initially detected. Lane ‘M’ is the NEB 100 bp marker. Lanes A, C, and E are normal control populations; lanes B, D, F, and H are SZ populations. Minus or plus signs indicate amplicons of gDNA from NeuN-negative (not sequenced, non-neuronal) or NeuN-positive (sequenced, neuronal) nuclei, respectively.

Figure 2

ddPCR allele frequency of L1s in various genes. Graphs show the absolute copy numbers of L1-containing genes (blue squares) and RPPH1 (green squares) genes, as well as the ratios (Gene-L1 : RPPH1, orange diamonds), detected in 100 ng of XmnI-digested gDNA from DLPFC various SZ samples. (a) The KHDRBS2-L1 was detected in SZ patient 063, but not in SZ patients 062 or 068. (b) The GRID2-L1 was detected in SZ patient 033, but not in SZ patients 035 or 040. (c) The SYNE1-L1 was detected in SZ patient 103, but not in SZ patients 052 or 070. In all cases, the allele frequency of the L1s was ~50%, the theoretical frequency for an L1 that is in heterozygosity in all genomes analyzed. NTC is the no template control.