Introduction: Large somatic deletions of mitochondrial DNA (mtDNA) accumulate with aging in metabolically active tissues such as the brain. We have cataloged the breakpoints and frequencies of large mtDNA deletions in the human brain. Methods: We quantified 112 high-frequency mtDNA somatic deletions across four human brain regions with the Splice-Break2 pipeline. In addition, we utilized PLINK/Seq to test the association of mitochondrial genotypes with the abundance of these high-frequency mtDNA deletions. A conservative p value threshold of 5E−08 was used to find the significant loci. Results: One mtDNA SNP (T14798C) was significantly associated with mtDNA deletions in two brain regions, the dorsolateral prefrontal cortex (DLPFC) and the superior temporal gyrus. Since the DLPFC showed the most robust association between T14798C and two deletion breakpoints (7816–14807 and 5462–14807), this association was tested in the DLPFC of a replication sample and validated the first results. Incorporating the C allele at 14,798 bp increased the perfect/imperfect length of the repeat at the 3′ breakpoint of the two associated deletions. Conclusion: This is the first study to identify the association of mtDNA SNP with large mtDNA deletions in the human brain. The T14798C allele located in the MT-CYB gene is a common polymorphism that occurs in several mitochondrial haplogroups. We hypothesize that the T14798C association with two deletions occurs by extending the repeat length around the 3′ deletion breakpoints. This simple mechanism suggests that mtDNA SNPs can affect the mitochondrial genome structure, especially in brain where high levels of reactive oxygen species lead to deletion accumulation with aging.

Mitochondria are the brain’s primary energy source; these integral organelles contain a circular, 16569 bp genome of 37 genes, including 13 translated proteins, 22 tRNA, and 2 rRNAs [1]. Single-point mutations in mitochondrial DNA (mtDNA) cause multiple classical mitochondrial disorders, such as Leigh syndrome, stroke, cardiomyopathy, or neuropathy, ataxia and retinitis pigmentosa syndrome [2]. Some mitochondrial disorders reduce the lifespan, such as mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes and often involve tissues with high energy requirements such as the central nervous system, heart, and skeletal muscle [3]. There are also rare diseases and phenotypes caused by large mtDNA deletions when they are present at high amounts (e.g., Kearns-Sayre syndrome, Pearson syndrome, and progressive external ophthalmoplegia) [3]. Deletions of mtDNA in different brain regions have been associated with Parkinson disease [4] and depression like episodes in an animal model [5].

We previously demonstrated that large mtDNA deletions, such as the common deletion, occur in different brain regions at high, albeit variable levels [6, 7]. A metric, cumulative deletion read percentage correlated with age irrespective of diagnostic effects in the dorsolateral prefrontal cortex (DLPFC) and anterior cingulate cortex [8]. Furthermore, we have shown that those large mtDNA deletions are present at much lower levels in blood compared to the brain levels [7, 8]. Thus, brain tissue represents an ideal source of material to test the genetic and environmental factors that influence mtDNA deletion accumulation, as deletion breakpoints can be easily detected and compared across many aged subjects. We utilized our published Splice-Break pipeline to quantify mtDNA deletions in four brain regions, cortical and subcortical, and annotated mitochondrial single-nucleotide polymorphisms (SNPs) from whole exome sequencing (WES) data from the same subjects. After limiting the analysis to mtDNA deletions and SNPs that occurred at a high frequency, we tested their association to determine if any particular sets of mtDNA deletion breakpoints were potentially cis-regulated by mtDNA alleles.

Demographics

Three brain regions (DLPFC, superior temporal gyrus [STG], and visual cortex [V1]) were obtained from 96 postmortem human subjects from the University of California-Irvine (UCI) Pritzker Brain Bank at the UCI School of Medicine. A summary of the subjects' demographics, including sex, age, diagnosis (psychiatric disorder), brain pH, and postmortem interval (PMI), is shown in Table 1. In addition, a subgroup of the subjects also had ventral striatum tissue from the nucleus accumbens (NACC) analyzed for this study (n = 47). A second association study of DLPFC was independently conducted using primarily subjects with schizophrenia (Table 2). The individual deletions that we tested were not significantly different between diagnostic groups.

Table 1.

Demographics of subjects analyzed for the association of mtDNA deletions and mtDNA SNPs in DLPFC, STG, and V1.

 Demographics of subjects analyzed for the association of mtDNA deletions and mtDNA SNPs in DLPFC, STG, and V1.
 Demographics of subjects analyzed for the association of mtDNA deletions and mtDNA SNPs in DLPFC, STG, and V1.
Table 2.

Demographics of the sample analyzed for replication of the association of mtDNA SNP T14798C and mtDNA deletions

 Demographics of the sample analyzed for replication of the association of mtDNA SNP T14798C and mtDNA deletions
 Demographics of the sample analyzed for replication of the association of mtDNA SNP T14798C and mtDNA deletions

mtDNA Amplification, Library Preparation, and Sequencing

Genomic DNA was extracted from each of the cortical and subcortical brain regions using a phenol-chloroform method from 30 mg of frozen tissue. In the cortical tissue samples, the visible white matter was removed via microdissection prior to DNA extraction. The mtDNA was PCR-amplified using back-to-back primers that hybridize to the control region of the mitochondrial genome, and the resultant PCR product was purified by bead purification to retain both deleted and wild-type molecules. Primer sequences are as follows: F 5′-CCGCACAAGAGTGCTACTCTCCTC-3′ and R 5′-GATATTGATTTCACGGAGGATGGTG-3' (Integrated DNA Technologies). Each sample was enriched for mtDNA in a 50-μL PCR that contained the following components: 50 ng of total DNA, 1 μL of each 10 μm primer, 8 μL of 2.5 mM dNTPs, 0.5 μL of LA Taq DNA Polymerase, Hot-Start Version (Takara Bio USA, Inc., Mountain View, CA, USA), and 5 μL 10x buffer. The following cycling parameters were used: 94°C for 1 min, followed by 30 cycles of denaturation at 98°C for 10 s, and annealing/extension at 68°C for 15 min, with a final extension at 72°C for 10 min. Reactions were then kept at 4°C, for a total PCR time of 8.5 hrs. Following PCR, 5 μL (10% of the total reaction volume) was loaded into a 1% agarose gel containing 10 mg/mL ethidium bromide and the gel was run at 100V for approximately 2 hrs for visual confirmation that the PCR was successful (i.e., a band corresponding to the full-length mitochondrial genome [∼16.5 kb] was easily visible). PCR products were purified using Agencourt AMPure XP beads (Beckman Coulter, Indianapolis, IN, USA) and were quantified using the Qubit Fluorometer and dsDNA BR Assay Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. DNA shearing was performed in S220 Covaris microTUBEs with the following settings: duty factor = 5%, peak incident power = 175 W, cycles per burst = 200, time = 35. This bead-purified, sheared PCR product was used as the input for library preparation and sequencing.

Next-generation sequencing (NGS) libraries were prepared using the TruSeq kit (Illumina, San Diego, CA) for the cortical NGS libraries, which were multiplex sequenced (96 samples in one lane of a flow cell) as 150-mer paired-end reads in the RapidRun mode on an Illumina HiSeq2500 at the UCI Genomics High Throughput Facility. Library preparation for NACC samples was performed using the 384 Seq-Well kit (San Diego, CA). The NACC library was sequenced as 150-mer paired-end reads on a patterned flow cell using the Illumina NovaSeq6000 at the UCI Genomics High Throughput Facility. A minimum mitochondrial depth of 5,000× on NGS amplicon resequencing was required for samples to be evaluated for large mtDNA deletions since most of these species occur at a low read rate in homogenate tissue.

Splice-Break and Annotation of Deletion Frequency

The Splice-Break pipeline uses a long-range mitochondrial-targeted NGS of mtDNA amplicons, and customized scripts for detecting and quantifying the relative levels of mtDNA deletion breakpoints [8]. We used a customized bash script Splice-Break2 paired-end.sh (https://github.com/brookehjelm/Splice-Break2/) to perform automated filtering, normalization, and annotation of mtDNA deletion breakpoints. High-frequency deletions were characterized as those that occurred in at least 40% of samples in our initial Splice-Break study [8] that had 5′ and 3′ breakpoints at least 250bp away from the primer positions. The deletions we describe can be found in the supplemental catalog of our methods article [8], and deletion frequency from this catalog (i.e., high, mid, low, etc.) is now annotated for each deletion as part of the Splice-Break2 bioinformatics pipeline. These 112 high-frequency deletions were analyzed across four brain regions from the same subjects, and their deletion read %’s were used for association tests. A workflow diagram summarizing this process is shown in Figure 1.

Fig. 1.

Summary of laboratory and bioinformatics workflow for long-range PCR mtDNA amplification, next-generation sequencing, and mtDNA deletion quantification using the Splice-Break2 pipeline. Laboratory methods are shown in light green boxes (columns 1 and 2), and bioinformatics processes are shown in dark green boxes (column 3). * See the Splice-Break2 GitHub page (https://github.com/brookehjelm/Splice-Break2) for further details.

Fig. 1.

Summary of laboratory and bioinformatics workflow for long-range PCR mtDNA amplification, next-generation sequencing, and mtDNA deletion quantification using the Splice-Break2 pipeline. Laboratory methods are shown in light green boxes (columns 1 and 2), and bioinformatics processes are shown in dark green boxes (column 3). * See the Splice-Break2 GitHub page (https://github.com/brookehjelm/Splice-Break2) for further details.

Close modal

Whole Exome Sequencing

The DLPFC of each subject was analyzed by WES to obtain mtDNA variant calls. WES libraries were prepared with the Agilent SureSelect Exome library preparation kit, following the manufacturer’s protocols. The libraries were sequenced on an Illumina HiSeq2500 (GeneWiz, New Jersey, USA) to an average depth of 60× across 80% of the nuclear exome. The mitochondrial chromosome was sequenced at a higher depth (>1,000×) within the WES data due to the polyploid nature and high copy number of mtDNA molecules in the brain. We used WES data to call mitochondrial homoplasmic variants, in addition to nuclear variants. As a part of a separate study, we are testing nuclear variants and a variety of mitochondrial metrics (including mtDNA deletions) in the same subjects/brain regions and have set our threshold for significance as genome wide (p < 5E−08), since our a priori hypothesis includes whole genome analysis.

PLINK/Seq

The mtDNA deletion levels from Splice-Break2 and the mtDNA variants from WES (for each subject and brain region) were combined into a project in PLINK/Seq (https://zzz.bwh.harvard.edu/plinkseq/index.shtml). Custom batch processing script (bash script for Slurm is available upon request) was used to run a general linear regression model including mtDNA variant and deletion level, with pH and age as covariates. The adaptive permutation setting was enabled in PLINK/Seq.

The deletion levels (i.e., read %) of the 112 high-frequency mtDNA deletions from Splice-Break did not show group differences by diagnosis compared to controls (data not shown). The lack of any case-control differences (p value < 1.12E−04), after multiple corrections, of the levels of 112 deletions in 4 brain regions supports our rationale to investigate a cohort of postmortem brain samples from a combined cohort of both controls and psychiatric disorders.

Therefore, we tested the levels of each deletion with mtDNA SNP using a quantitative association analysis in PLINK/Seq. The supplementary tables (online suppl. Table ST1, Table ST3; for all online suppl. material, see www.karger.com/doi/10.1159/000528051) show the results of PLINK/Seq for the nominal associations of mtDNA SNPs (minor allele frequency [MAF] >0.03, p < 0.05) that also have at least one SNP:deletion pair association (p < 5E−08) occurring in ≥1 brain regions. Of these 193 nominal associations, a majority were found in SNPs with MAF <0.05. Filtering out variants with a MAF <0.05 reduced the nominal associations to 73 associations (online suppl. Table ST1). A pivot table (online suppl. Table ST2) summarizes the SNP:deletion pairs; as shown below, the genome-wide significant associations are the results of common haplogroup defining SNPs with the same mtDNA deletion. The haplogroup relationships for SNP T1189C, SNP A3480G, SNP G9055A, SNP T9698C, SNP A10550G, SNP T11299C, SNP C14167T, SNP T14798C, and SNP T16224C are depicted in online supplementary Table ST4–ST12, respectively.

We determined the percentage of the 5462–14807 and 7816–14807 deletions across four brain regions (online suppl. Fig. SF3). The 7816–14807 deletion was detected at a higher read % than the 5462–14807 deletion for all brain regions, which correlates with our previously published catalog where this deletion was more common and abundant. This deletion (7816–14807) was also significantly different between all brain regions, with the highest levels observed in the NACC followed by the DLPFC. The other deletion (5462–14807) was more similar across brain regions, but was significantly higher in a single comparison of the NACC to DLPFC. The data for deletion 5462–14807 are skewed-left due to many T individuals showing zero deletion levels, thus using a nonparametric Kruskal-Wallis test to check T and C allele differences. The combined p value for the 5462–14807 deletion was 2E−16. For the initial study, the KW p value was 1E−11, and the independent replication p values were 3.6E−06. Only one SNP:deletion pair replicated at significance levels (p < 5E−08) in two brain regions in the first cohort. PLINK/Seq identified that mtDNA SNP T14798C was significantly associated with the mtDNA deletions in the DLPFC (Table 3) and STG (online suppl. Table ST1). Specifically, carriers with the C allele had a significantly increased level of the 5462–14807 deletion in the DLPFC (p = 4.35E−21) (Fig. 2) and STG (p = 9.57E−16). Trends in the same direction for V1 (p = 3.93E−07) and NACC (p = 9.87E−04) were also observed. The NACC sample size was 47 and had a smaller number of samples compared to the cortical regions.

Table 3.

An independent set of DLPFC samples was sequenced and called through the Splice-Break2 pipeline.

 An independent set of DLPFC samples was sequenced and called through the Splice-Break2 pipeline.
 An independent set of DLPFC samples was sequenced and called through the Splice-Break2 pipeline.
Fig. 2.

The first and second studies are shown for each independently replicated deletion. a The 5462–14,807 deletion levels in DLPFC by allele frequency (C/C or T/T for SNP T14798). b The 7816–14,807 deletion levels in DLPFC by allele frequency (C/C or T/T for SNP T14798). The error bars are the least square standard error of the mean.

Fig. 2.

The first and second studies are shown for each independently replicated deletion. a The 5462–14,807 deletion levels in DLPFC by allele frequency (C/C or T/T for SNP T14798). b The 7816–14,807 deletion levels in DLPFC by allele frequency (C/C or T/T for SNP T14798). The error bars are the least square standard error of the mean.

Close modal

The same mtDNA SNP (T14798C) was also associated with a different mtDNA deletion, 7816–14807 in the DLPFC only (p = 5.76E−08) (Fig. 2), with a similar trend in the STG (p = 4.49E−03). This 7816–-14807 deletion was the second most commonly observed deletion in our previous study [8]. As observed with the 5462–14807 deletion, carriers with the C allele had an increased mtDNA deletion rate 7816–14807. For clarity, these breakpoint positions are annotated from our Splice-Break pipeline that uses MapSplice for breakpoint identification [9]; the adjusted positions of these deletions are 7814–14805 and 5464–14809 using MitoBreak standardization [10].

The conserved association of a single-mtDNA SNP with multiple deletion species is perhaps not surprising; given the two deletions, we describe as significant use the same 3′ breakpoint (annotated to 14807 by Splice-Break), which is proximal to the SNP position. The high-frequency mtDNA deletions that contained this 3′ breakpoint (14807) were further investigated for association with the T14798C allele in an independent cohort of DLPFC samples as the T14798C allele is only 9 bp from this annotated breakpoint, and we have previously observed a high level of redundant usage of both 5′ and 3′ breakpoints across the thousands of deletion breakpoints we cataloged [8]. The independent set of DLPFC samples was sequenced and called through the same Splice-Break pipeline (Table 2). For this sample of DLPFC (n = 77), the result was again significant for T14798C at the DLPFC_5462–14807 breakpoint (p = 4.53E−09, Table 3). Combining both studies (Table 3), and correcting for number of deletions tested, the DLPFC 5462–14807 and DLPFC 7816–14807 both passed FDR. The deletions that incorporated the 3′ breakpoint (14,807) which were not nominally significant in the first study also proved to be not significant in the second study. These areas that showed no base pair extension due to the T14798C SNP also did not show evidence of increased deletion (p values > 0.05, data not shown).

To test a potential mechanism for these SNP:deletion associations, we evaluated the homology of the repeat region surrounding the 3′ breakpoint (14807) compared to the repeat region of the 5′ breakpoint (varies) and determined if the repeat length was increased, decreased, or remained constant with the incorporation of the C allele at position 14,798. We define homology here as the number of bases in the 5′ and 3′ repeat sequences that are perfect matches with one another. For the 5462–14807 deletion, the cis-SNP T14798C increased the matched homology length between the 3′ and 5′ breakpoint regions from 6 to 9 bp (Table 4). An increase in homology was also observed for the 7816–14807 deletion, which increases from 6 to 7 bp (Table 4). In addition, the fold change for the mtDNA deletion levels was higher for a 3 bp homology improvement compared to 1 bp homology improvement (14.3 and 2.3, respectively), and the nonsignificant associations of other deletions with the same 3′ breakpoint (14,807) all showed zero bp homology change with T14798C (Table 4). Thus, there appears to be a relationship with fold increase and net homology gain for the associated SNP:deletion pairs. A visual of these repeats and their homology with and without the T14798C are shown in online supplementary Figure SF2.

Table 4.

The mtDNA deletion level fold change was higher for a 3 bp homology improvement compared to 1 bp homology improvement (14.3 and 2.3, respectively)

 The mtDNA deletion level fold change was higher for a 3 bp homology improvement compared to 1 bp homology improvement (14.3 and 2.3, respectively)
 The mtDNA deletion level fold change was higher for a 3 bp homology improvement compared to 1 bp homology improvement (14.3 and 2.3, respectively)

The T14798C SNP and 9 other associated SNPs are enriched in the K1 and K2 haplogroups (>99% of individuals in mitomap.org) (online suppl. Tables 1, 2). T14798C has the highest MAF (0.17) and is a top-level SNP in the K haplogroup. The remaining 9 SNPs (MT:10550, MT:11299, MT:1189, MT:14167, MT:16224, MT:3480, MT:497, MT:9055, MT:9698) have identical but slightly lower MAFs (0.10), but also showed significant association with the first deletions we describe (5462–14807), with p value ranges of 2.25E−06 to 1.19E−12.

The T14798C is also highly prevalent in the J haplogroup (53%), in the subclade J1c (99%), and in another lineage T2g (97.7%) (online suppl. Table 2). One individual with J1b haplogroup was T14798 as expected, and all J1c subclade members in this study were 14798C. The carriers of 14798C allele were compared between J (N = 11) and K (N = 15) haplogroups, and there was no difference (p = 0.45) in the mean amount of DLPFC_5462–14807 deletion levels between haplogroups that were all 14798C. Thus, the penetrance of the DLPFC_5462–14807 deletion is not altered strictly between these two haplogroups (online suppl. Fig. SF1). Analysis using ancestry multi-dimensional scaling components based on nuclear DNA exome SNPs to correct potential population stratification did not alter the main findings for the two significant deletion associations with the T14798C mtDNA allele. Further, the T14798C allele is an ethnic specific haplogroup marker for J1 and K1 (online suppl. Table ST11). Therefore, the study cannot be balanced for nonallele carriers in J1 and K1 haplogroups. Taken together, these results suggest the T14798C allele is the functional SNP, which leads to an increased probability in the formation of specific mtDNA deletions, while the other haplogroup defining SNPs that are associated with the same deletions is likely nonfunctional in this regard.

The mitochondrial, nonsynonymous SNP T14798C is associated with two large deletions of mtDNA in multiple brain regions and independent studies. These two deletions, spanning 7816–14807 and 5462–14807, were ranked the 2nd and 76th most frequently detected deletions in our Splice-Break catalog, respectively, and were detected in 93% and 47% of samples in that study [8]. To our knowledge, we are the first to report a significant association between SNPs and deletions in mtDNA. The genetic association of T14798C with these two large deletions demonstrates that the minor allele (C) leads to increased deletion formation. Other cis-acting SNPs that influence additional structural variants will likely be discovered using a larger sample, especially given the propensity for mtDNA deletion breakpoints to utilize perfect/imperfect repeat sequences for origination. Additional studies are needed to confirm that other haplogroup carriers (J1c and T2g) of the T14798C SNP would show this significant increase in deletion abundance of the 5462–14807 and 7816–14807 deletions.

The mitochondrial “common deletion,” a 4977 bp deletion that has been well studied through targeted qPCR, was shown to increase significantly with age in diverse brain regions [6, 7, 11]. Similarly, analysis of the top 30 most frequent deletions using an NGS analysis showed that 23/30 deletions, including the “common deletion,” were positively correlated with age in DLPFC. These 23 deletions with an age association retained the origin of replication of the lagging strand (OL), while the remaining 7 deletions that did not correlate had this machinery removed by the deletion [8]. This “common deletion” propensity to form is due primarily to the long, 13 bp repeat flanking the "common deletion" breakpoints. Although repeat homology and length influence mtDNA deletion formation, the variable rate across tissues, brain regions, age, and disease cases suggests environmental exposures, neurotransmitter toxicity, reactive oxygen species (ROS), and mtDNA repair capabilities are also major drivers of this structural variation.

T14798C has been reported as a functional, nonsynonymous variant in complex III (F18L) [12]. It occurs in multiple haplogroups (see results and online suppl. Fig. SF1). The SNP forms part of the ubiquinone (coenzyme Q) binding site (Qi site) of complex III, impacting the activity of complex III [13]. In addition, the T14798C variant has been shown to affect the levels of ROS produced by complex III [12]. Clomipramine, a tricyclic antidepressant, has decreased mitochondria complex III activity [14] and has been tested in cell lines derived from T14798C carriers. Glioblastoma cell lines that harbored the T14798C allele, for example, had elevated complex III activity, oxidative stress, and increased sensitivity to clomipramine levels compared to cells that did not contain the mutation [13]. Many commonly reported side effects observed in patients treated with clomipramine might be (at least partially) attributed to the inhibition of mitochondrial activity. It should be further explored if carriers of the T14798C allele are susceptible to these side effects. Furthermore, glioblastoma patients with the T14798C had a worse prognosis than noncarriers [13]. Lastly, a recent, large-scale GWAS showed this T14798C SNP was significantly associated with circulating N-formylmethionine levels in blood [15]; N-formylmethionine initiates mitochondrial protein translation. Together, these results suggest that the T14798C allele may have multiple functional effects that influence mitochondrial activity, including ubiquinone binding, protein translation, and mtDNA deletion formation.

We investigated a phenotype of somatic mtDNA deletion breakpoints in the brain and found a significant association with a cis-mtDNA SNP. While precise causative mechanisms are speculative, we consider the increase in mtDNA repeat homology as the most likely primary mechanism underlying the association, which may be further influenced (indirectly) by the functional effects of the T14798C allele and resultant increase in ROS. These results illustrate that a common single-nucleotide mtDNA variant can have cis-effects on the formation of somatic mtDNA deletions in the brain. This study supports the need for further investigations of how mtDNA deletions affect complex disease risk, drug sensitivity, and neurological function across the lifespan. Additional studies of both synonymous and nonsynonymous mtDNA SNPs (and mutations) should also consider mtDNA deletion formation as a possible functional outcome, understanding that this association may only be significant for specific species of deletion breakpoints.

We would like to thank donors’ family for donating the postmortem brain tissues. Jack D. Barchas was not available to confirm co-authorship, but the corresponding author Marquis P Vawter affirms that author, Jack D. Barchas, contributed to the paper, had the opportunity to review the final version to be published, and guarantees Jack D. Barchas’ authorship status and the accuracy of the author contribution and conflict of interest statements.

Postmortem brain tissues were obtained from the UCI-Pritzker brain bank. Written consent was obtained from the next-of-kin for each subject. The postmortem brain collection and experimental procedures were approved by UCI Institutional Review Board (IRB) (IRB# 19970074).

The authors have no conflicts of interest to declare.

This study was supported by National Institute of Mental Health (R01MH08580 to MPV); Della Martin Foundation (to SCD, postdoctoral fellowship). Postmortem brain tissue collection at the University of California-Irvine (UCI) Brain Bank was supported by the Pritzker Neuropsychiatric Disorders Research Consortium.

Sample collections were performed by Adolfo Sequeira, Alan F. Schatzberg, Jack D. Barchas, Francis S. Lee, Richard M. Myers, Stanley J. Watson, Huda Akil, and William E. Bunney. Laboratory experiments were conducted by Brandi L. Rollins, Ling Morgan, Sujan C. Das, and Brooke E. Hjelm. Data were analyzed by Brooke E. Hjelm, Christian Ramiro, Brandi L. Rollins, Audrey A. Omidsalar, Daniel S. Gerke, and Marquis P. Vawter. The manuscript was drafted and edited by Brooke E. Hjelm, Christian Ramiro, Audrey A. Omidsalar, Daniel S. Gerke, Sujan C. Das, Alan F. Schatzberg, Jack D. Barchas, Francis S. Lee, Richard M. Myers, Stanley J. Watson, Huda Akil, William E. Bunney, and Marquis P. Vawter. The study was designed and supervised by Marquis P. Vawter.

mtDNA sequencing and deletion data are publicly available through dbGaP accession code: phs002395.v1.p1. Further inquiries can be directed to the corresponding authors.

1.
Anderson
S
,
Bankier
AT
,
Barrell
BG
,
de Bruijn
MHL
,
Coulson
AR
,
Drouin
J
,
.
Sequence and organization of the human mitochondrial genome
.
Nature
.
1981
;
290
(
5806
):
457
65
.
2.
Ganetzky
RD
,
Stendel
C
,
McCormick
EM
,
Zolkipli-Cunningham
Z
,
Goldstein
AC
,
Klopstock
T
,
.
MT-ATP6 mitochondrial disease variants: phenotypic and biochemical features analysis in 218 published cases and cohort of 14 new cases
.
Hum Mutat
.
2019
;
40
(
5
):
499
515
.
3.
Arena
IG
,
Pugliese
A
,
Volta
S
,
Toscano
A
,
Musumeci
O
.
Molecular genetics overview of primary mitochondrial myopathies
.
J Clin Med
.
2022
;
11
(
3
):
632
.
4.
Bender
A
,
Krishnan
KJ
,
Morris
CM
,
Taylor
GA
,
Reeve
AK
,
Perry
RH
,
.
High levels of mitochondrial DNA deletions in substantia nigra neurons in aging and Parkinson disease
.
Nat Genet
.
2006
;
38
(
5
):
515
7
.
5.
Kasahara
T
,
Takata
A
,
Kato
TM
,
Kubota-Sakashita
M
,
Sawada
T
,
Kakita
A
,
.
Depression-like episodes in mice harboring mtDNA deletions in paraventricular thalamus
.
Mol Psychiatry
.
2016
;
21
(
1
):
39
48
.
6.
Mamdani
F
,
Rollins
B
,
Morgan
L
,
Sequeira
PA
,
Vawter
MP
.
The somatic common deletion in mitochondrial DNA is decreased in schizophrenia
.
Schizophrenia Res
.
2014
;
159
(
2–3
):
370
5
.
7.
Sequeira
A
,
Martin
MV
,
Rollins
B
,
Moon
EA
,
Bunney
WE
,
Macciardi
F
,
.
Mitochondrial mutations and polymorphisms in psychiatric disorders
.
Front Genet
.
2012
;
3
:
103
.
8.
Hjelm
BE
,
Rollins
B
,
Morgan
L
,
Sequeira
A
,
Mamdani
F
,
Pereira
F
,
.
Splice-Break: exploiting an RNA-seq splice junction algorithm to discover mitochondrial DNA deletion breakpoints and analyses of psychiatric disorders
.
Nucleic Acids Res
.
2019
;
47
(
10
):
e59
.
9.
Wang
K
,
Singh
D
,
Zeng
Z
,
Coleman
SJ
,
Huang
Y
,
Savich
GL
,
.
MapSplice: accurate mapping of RNA-seq reads for splice junction discovery
.
Nucleic Acids Res
.
2010
;
38
(
18
):
e178
.
10.
Damas
J
,
Carneiro
J
,
Amorim
A
,
Pereira
F
.
MitoBreak: the mitochondrial DNA breakpoints database
.
Nucleic Acids Res
.
2014
;
42
(
Database issue
):
D1261
8
.
11.
Rollins
BL
,
Morgan
L
,
Hjelm
BE
,
Sequeira
A
,
Schatzberg
AF
,
Barchas
JD
,
.
Mitochondrial complex I deficiency in schizophrenia and bipolar disorder and medication influence
.
Mol Neuropsychiatry
.
2018
;
3
(
3
):
157
69
.
12.
Lloyd
RE
,
Keatley
K
,
Littlewood
DTJ
,
Meunier
B
,
Holt
WV
,
An
Q
,
.
Identification and functional prediction of mitochondrial complex III and IV mutations associated with glioblastoma
.
Neuro Oncology
.
2015
;
17
(
7
):
942
52
.
13.
Keatley
K
,
Stromei-Cleroux
S
,
Wiltshire
T
,
Rajala
N
,
Burton
G
,
Holt
WV
,
.
Integrated approach reveals role of mitochondrial germ-line mutation F18L in respiratory chain, oxidative alterations, drug sensitivity, and patient prognosis in glioblastoma
.
Int J Mol Sci
.
2019
;
20
(
13
):
3364
.
14.
Daley
E
,
Wilkie
D
,
Loesch
A
,
Hargreaves
IP
,
Kendall
DA
,
Pilkington
GJ
,
.
Chlorimipramine: a novel anticancer agent with a mitochondrial target
.
Biochem Biophys Res Commun
.
2005
;
328
(
2
):
623
32
.
15.
Cai
N
,
Gomez-Duran
A
,
Yonova-Doing
E
,
Kundu
K
,
Burgess
AI
,
Golder
ZJ
,
.
Mitochondrial DNA variants modulate N-formylmethionine, proteostasis and risk of late-onset human diseases
.
Nat Med
.
2021
;
27
(
9
):
1564
75
.