Splice-Break: exploiting an RNA-seq splice junction algorithm to discover mitochondrial DNA deletion breakpoints and analyses of psychiatric disorders

Abstract

Deletions in the 16.6 kb mitochondrial genome have been implicated in numerous disorders that often display muscular and/or neurological symptoms due to the high-energy demands of these tissues. We describe a catalogue of 4489 putative mitochondrial DNA (mtDNA) deletions, including their frequency and relative read rate, using a combinatorial approach of mitochondria-targeted PCR, next-generation sequencing, bioinformatics, post-hoc filtering, annotation, and validation steps. Our bioinformatics pipeline uses MapSplice, an RNA-seq splice junction detection algorithm, to detect and quantify mtDNA deletion breakpoints rather than mRNA splices. Analyses of 93 samples from postmortem brain and blood found (i) the 4977 bp ‘common deletion’ was neither the most frequent deletion nor the most abundant; (ii) brain contained significantly more deletions than blood; (iii) many high frequency deletions were previously reported in MitoBreak, suggesting they are present at low levels in metabolically active tissues and are not exclusive to individuals with diagnosed mitochondrial pathologies; (iv) many individual deletions (and cumulative metrics) had significant and positive correlations with age and (v) the highest deletion burdens were observed in major depressive disorder brain, at levels greater than Kearns–Sayre Syndrome muscle. Collectively, these data suggest the Splice-Break pipeline can detect and quantify mtDNA deletions at a high level of resolution.

Ultrasensitive deletion detection links mitochondrial DNA replication, disease, and aging

“…High sequencing depth (Additional file 1 : Table S1) allowed the LostArc method to call roughly one deletion per million mtDNA circles (10 –5.9 ± 0.09 ; median ± 95% CL in the exponent). LostArc sensitivity exceeds that for sequencing after long-range PCR [ 39 ] but not for digital drop PCR (~ 10 −8 [ 35 , 40 , 41 ];). Unlike digital drop PCR, which targets deletions at specific loci, LostArc assays the full deletion spectrum across the whole mitochondrial genome.…”

“…The latter found 451 unique deletions in human brain tissue. Recently, one high-throughput study found a remarkable 4489 unique deletions, in the brain and blood, by accepting the flaws of long-range PCR enrichment and eschewing absolute frequency measurements [ 39 ]. LostArc achieves mtDNA enrichment without long-range PCR amplification by selectively degrading linear nuclear DNA with nucleases (Fig.…”

“…Roughly 470,000 unique deletion spans were recovered from skeletal muscle biopsies collected from 19 Gwt individuals and 22 Gvar patients with mitochondrial diseases (Additional file 1 : Table S1). Roughly 99% of these deletions have not been reported previously [ 35 , 39 , 47 , 70 , 72 ]. Deletion counts are converted to frequencies, the background subtracted, and then deletion patterns discovered (steps 8–10).…”

Unbiased PCR-free spatio-temporal mapping of the mtDNA mutation spectrum reveals brain region-specific responses to replication instability

“…Type I deletions are hypothesised to be formed by polymerase slippage during replication [ 50 ]. The discrepancy between human disease cases [ 51 ] and mouse models regarding direct repeats may relate to the different involvement of direct repeats in deletion formation between long-lived and short-lived mammals [ 26 ]. Type II deletions colocalize with 2D and 3D mtDNA structures [ 49 ] and occur spontaneously during replication either through strand slipping [ 48 ], Polg stalling [ 52 ], or repair of double-stranded breaks [ 53 ].…”

Accurate mapping of mitochondrial DNA deletions and duplications using deep sequencing

“…Several methods have recently been developed specifically for identification of mtDNA deletions from high-throughput short read sequencing, including MitoDel [27], Splice-Break [28], eKLIPse [29], MitoMut [30], and a PERL script provided in (Zambelli et al, 2017) [31]. These methods rely on gapped alignments to predict deletions, but fail to recognize that every such event can represent either a deletion or a duplication affecting the arc complementary to the deleted part; a consequence of the circularity of mtDNA.…”