INTRODUCTION

Mitochondrial DNA (mtDNA) is thought to be strictly maternally inherited in humans and mammals.1,2,3 In 2002, paternal inheritance of mtDNA was reported in an individual with mitochondrial myopathy who was found to harbor a de novo mtDNA variant, which was paternal in origin and accounted for 90% of the individual’s muscle mtDNA.4 Subsequent studies of patients with mtDNA-related disorders failed to identify any further cases of paternal transmission.5,6,7,8,9,10 Further, a study using extreme-depth resequencing of mtDNA did not find evidence of paternal transmission in humans.11 In 2018, Luo et al. used long-range polymerase chain reaction (PCR) amplification of the entire mitochondrial genome and next-generation sequencing (NGS) to study individuals from three unrelated multigeneration families, each with unexplained multiple, high-level, heteroplasmic mtDNA variants (multiHets) suggestive of two different haplogroups. The authors proposed biparental mtDNA transmission in these families and again, they raised the hypothesis that paternal mtDNA could be transmitted to offspring.12 This report received broad international attention, both supportive and critical.13,14,15,16 Some scientists speculated that concatenated nuclear mtDNA segments (mega-NUMTs), if coamplified and sequenced together with the genuine mtDNA, can also result in the observed phenomenon.13,17 Recently, Rius et al. studied genome sequencing data of DNA extracted from blood samples of 41 pediatric patients with suspected mitochondrial disease and their parents to see whether biparental inheritance of mitochondrial DNA occurred in humans; this study showed no evidence of biparental inheritance mtDNA.18 In a later study, Wei et al. analyzed the genome sequencing data from 11,035 unrelated parent–child trios, and observed likely large, rare or unique NUMTs transmitted from the father in seven families, with heteroplasmy of 5–25%.15 However, as different methods were used, it is unknown whether any of these mega-NUMTs (Table 1 of Wei et al.)19 would also be detected by long-range PCR, the primary analysis method commonly adopted by clinical diagnostic laboratories designed to avoid interference of NUMTs by only amplifying circular mtDNA.12,20 It is critical to clarify the nature of these results, whether they truly represent biparental inheritance of genuine mtDNA or NUMTs that are also detectable by long-range PCR, as this would impact numerous fields including clinical genetics, mitochondrial medicine, molecular anthropology, population genetics, and forensic genetics.

At our clinical laboratory, samples with multiHets have been identified on a weekly basis from individuals referred for mitochondrial genome analysis. Several possible explanations for this phenomenon were excluded, including sample cross-contamination, recent blood transfusion, prior bone marrow or organ transplant, history of in vitro fertilization, or mitochondrial replacement therapy. Our cohort poses the following questions: How frequent are multiHets in a clinical laboratory? Are they relevant to the patients’ diseases? Could this represent biparental inheritance of genuine mtDNA or coamplification of mega-NUMTs?

MATERIALS AND METHODS

DNA extraction

Total genomic DNA from peripheral blood, and frozen muscle tissue was extracted using Gentra Puregene Blood Kit (Qiagen, Germantown, MD) and Gentra Puregene Tissue Kit (Qiagen, Germantown, MD), respectively. Total genomic DNA from ORAcollect buccal swabs (DNA Genotek, Ottawa, ON, Canada) were extracted on the QIAsymphony instrument using the QIAsymphony DSP DNA Midi Kit after a 2-hour incubation at 50 °C and being lysed. Total genomic DNA from urine was extracted using QIAamp Viral RNA Mini Kit (#52904) (Qiagen, Germantown, MD). All protocols were performed in accordance with the manufacturer’s instructions.

Exome sequencing and mitochondrial genome analysis

Exome sequencing and mitochondrial genome analysis was performed as previously described.20,21 The primers used for mitochondrial genome analysis were F16561 (5′-ATCACGATGGATCACAGGTCTATC-3′) and R16560 (5′-GTCTTATTTAAGGGGAACGTGTGG-3′), or F16428 (5′-GCACAAGAGTGCTACTCTCCTCG-3′) and R16427 (5′-GGGATATTGATTTCACGGAGGATGG-3′); each amplifies the entire mitochondrial genome. Each long-range PCR was performed in a 30-μl volume using 3 μl 10× Takara Buffer, 5 μl (2.5 mM) Takara dNTPs, 0.5 μl (20 μM) of each primer pair, 0.3 μl (5 units/μl) of Takara LA Hot Start Polymerase (Takara), 10 μl DNA(3 ng/μl), and 10.7 μl water. A no-DNA control was included in PCR setup for each of the fragments. Thermal cycling parameters of 1 minute at 94 °C, followed by 28 cycles of 10 seconds at 94 °C, 20 seconds at 60 °C, and 18 minutes at 68 °C, and a final extension of 10 minutes at 68 °C were used for all of the PCR amplifications. Products were visualized using 1.2% Lonza FlashGel cassette and accepted for sequencing where an appropriately sized product band was visible. Library was prepared using KAPA Hyper Prep Kit for Illumina (Kapa Biosystems) according to manufacturer’s protocol. Libraries were sequenced by 2×100 bp paired-end sequencing on the HiSeq 2500 (Illumina) platform. MtDNA sequence was assembled and analyzed relative to the revised Cambridge Reference Sequence (rCRS) and MITOMAP database (http://www.mitomap.org) as previously described.21 The average coverage (number of reads) for every nucleotide of the mitochondrial genome of a specific sample ranged from 5,000× to 25,000× for different samples, with a standard deviation of ≤10% of the average coverage. The coverage for nucleotide positions within 200 bp from the primer regions at the beginning and end of the mitochondrial genome could be lower due to removal of the primer sequences during alignment.

Determination of frequency of individuals with unexplained multiple heteroplasmic mtDNA variants

The definition of an individual with multiHets met the following criteria: (1) presence of ≥5 heteroplasmic mtDNA variants in the messenger/ribosomal/transfer RNA (mRNA/rRNA/tRNA) coding regions with heteroplasmy levels ranging from 10% to 90%; (2) no blood transfusion in the past six month; (3) no prior bone marrow transplantation (BMT) or other organ transplantation; (4) the possibility of sample cross-contamination in the procedure during blood draw or bench work is excluded by quality control, and also by repeating mitochondrial genome analysis on a new blood sample that yielded a same result for a majority of the cases. The frequency (f) of individuals with multiHets within a period of time was calculated using the formula: f = (total number of cases with multiHets)/(total number of unique cases tested for mitochondrial genome analysis)*100%

MtDNA haplogroup analysis

MtDNA haplogroup analysis was performed for each individual with all homoplasmic mtDNA variants using the online tool https://dna.jameslick.com/mthap/ by following the “Directions for Differences to rCRS (FTDNA, et al.)”. For samples with multiple heteroplasmic mtDNA variants suggesting a mixture of two types of mtDNA molecules in the same individual, the list of position differences to rCRS for each type of mtDNA molecules was used for the haplogroup classification.

PacBio Sequel sequencing and data analysis

To confirm the phase of the variants in different haplogroups, the long-range PCR products of the entire mtDNA (using primer F16561 and R16560) for the blood sample of the probands in families A, B, and C were submitted for long-read sequencing using PacBio Sequel system. Specifically, the long-range PCR products were purified and concentrated into 12-µl volume using 0.45× AMPure PB magnetic beads (Pacific Biosciences). Purified samples were quantified using Qubit 1× dsDNA HS assay kit (Thermo Fisher Scientific) and samples were normalized in equimolar ratios, totaling 1–2 µg DNA input for library preparation. The PacBio libraries were generated using the SMRTbell Barcoded Adapter Complete Prep kit–96 (Pacific Biosciences). Individual PCR products were each end-repaired and ligated to unique SMRTbell barcoded adapters in a single reaction. Ligated PacBio libraries were pooled together and purified with 0.45× AMPure PB magnetic beads. Following DNA damage repair, unsuccessful ligations were digested with two exonucleases (ExoIII and ExoVII). The Sequel Binding and Internal Control Kit 3.0 (Pacific Biosciences) was used to prepare SMARTbell libraries for sequencing. After 20 hours of sequencing, the SMRT link analysis tool (version 6.0.0.47841) automatically performed demultiplexing based on the DNA identifier associated with the corresponding barcoded adaptor codes. Binary Alignment Map (BAM) files were generated by performing CCS mapping using the GRCh38/hg38 reference (including mitochondrial genome [NC_012920.1]). We converted the PacBio CCS reads to FASTA format (consensus sequences reads longer than 15 kb for families A and C, and longer than 9.5 kb for family B, were collected) and then run Blastn (v2.6.0+) against mtDNA amplicon sequence, and human genome GRCh38/hg38 respectively. The FASTA file of the reads were also analyzed for mtDNA variants and haplogroups using the MITOMASTER tool in Mitomap (www.mitomap.org).

MtDNA content determination by real-time quantitative PCR (RT-qPCR)

Real-time quantitative PCR TaqMan method was used to quantify the copy number of mtDNA and nuclear DNA. For mtDNA, forward primer (5′- TCACAACACAAGAACACCTCTGATT-3′), reverse primer (5′- CGGTTGGTCTCTGCTAGTGT -3′), and FAM labeled TaqMan probe: 5′ FAM-CTGCCATCATGACCCTTGGCC-TAMRA 3′) was used. A DNA fragment of the TERT gene was amplified using commercially available TaqMan probes for TERT control reagents according to the manufacturer’s instructions (Thermo Fisher Scientific). Each sample was performed in triplicate for both the mtDNA and the TERT gene. Then, 2–10 ng of DNA was amplified on a QuantStudio 12 K Flex instrument (Thermo Fisher Scientific) using the TaqMan SNP genotyping Master Mix (Thermo Fisher Scientific) in the presence of a final concentration of 1× of either mtDNA probes or 1× of the nuclear DNA probes (TERT). Amplification conditions were as follows: 50 °C for 2 minutes, 95 °C for 10 minutes, and 40 cycles of 95 °C for 10 seconds, and 60 °C for 1 minute. MtDNA content was calculated using the Ct difference (delta CT) and the formula: MtDNA content = 1/2^(CtmtDNA-CtTERT).

Statistical analysis

The relationship between the level of heteroplasmy of proposed NUMTs-derived variants/haplogroups and the mtDNA content was analyzed and plotted using the statistical function and chart tool in Microsoft Excel. Student’s t-test was used to test the significance of the correlation coefficient (R), for which P < 0.05 was considered a significant correlation.

RESULTS

Of the 27,388 unrelated individuals referred for mitochondrial genome analysis, 1,957 had mitochondrial genome analysis only, while the remaining 25,431 individuals had concurrent exome sequencing. For individuals who also had exome sequencing, a majority (86%) were trio-based, in which both parents were included for segregation analysis. In total, 104 probands (approximately 1 in 263) were identified to have multiHets. Among these, 8 had mitochondrial genome analysis alone (0.41%), and 96 had both mitochondrial genome analysis and exome sequencing (0.38%). Additional testing performed across two or more generations in the four selected families is discussed below. Detailed clinical information for the families is provided in Supplemental document 1. Our results suggest that although these multiHets appear to be paternally inherited, they most likely represent coamplification of mega-NUMTs from the paternal nuclear genome.

In family A, 18 heteroplasmic and 6 homoplasmic variants were identified in the peripheral blood sample of the proband, which is consistent with the presence of two haplogroups, R0a’b and H1bg, with 75% heteroplasmy for variants in R0a’b and 25% heteroplasmy for variants in H1bg (IV-2 in Fig. 1a, IV-2bl in Fig. 1b, c, and Supplemental family A data set). Concurrent PacBio long-read sequencing confirmed the presence of full-length consensus mtDNA amplicons with either haplogroup R0a’b or haplogroup H1bg (supplemental PacBio data for family A). Additionally, a pathogenic m.7495A>G variant22 in the MT-TS1 gene was identified at 3% heteroplasmy in both Illumina and PacBio data. Testing of DNA from a buccal swab yielded the same set of variants but at different levels of heteroplasmy: the m.7495A>G variant was at 3%, variants in haplogroup R0a’b at 40%, and variants in haplogroup H1bg at 60% (IV-2buc in Fig. 1b, c and IV-2 in family A data set). The mother (III-3) was homoplasmic for the variants in haplogroup H1bg, and the father (III-2) was heteroplasmic for variants in the R0a’b haplogroup and a third haplogroup, H5b1. The paternal grandmother (II-2) was found to have the same set of variants as her son (III-2), indicating the variants from both haplogroups R0a’b and H5b1 were transmitted maternally, and there was no transmission of the mtDNA variants from the paternal grandfather (II-1) to his son (III-2). Both siblings (IV-1 and IV-3 in Fig. 1a) of the proband were homoplasmic for the variants in haplogroup H1bg, consistent with maternal transmission of mtDNA in these individuals. Follow-up testing of the proband’s urine sample identified homoplasmic variants associated with the maternal haplogroup H1bg only (IV-2ur in Fig. 1b, c and in family A data set) as well as the m.7495A>G variant at 5% heteroplasmy. The heteroplasmy level of the variants from the paternal haplogroup R0a’b was much lower in buccal than in blood in both III-2 (6% and 75%, respectively) and IV-2 (40% and 75%, respectively).

Fig. 1: Suggestive biparental mitochondrial DNA (mtDNA) inheritance pattern in family A.
figure 1

a Pedigree. The single color filled symbols represent individuals with mtDNA of homoplasmic variants belonging to a single haplogroup (II-1, III-3, IV-1, IV-3) and the multicolor filled symbols represent individuals with mtDNA of heteroplasmic variants belonging to different haplogroups (II-2, III-2, IV-2). Each color represents mtDNA of a single haplogroup. The unfilled symbols indicate individuals not tested. b Schematic of the mtDNA genotypes defined by the homoplasmic and/or heteroplasmic variants aligned from the reference mitochondrial genome. The bars of each color represent mtDNA of a single haplogroup. c Summary of the haplogroup and mtDNA variant numbers in family A. bl blood sample, buc buccal swab.

Full size image

In family B, 4 homoplasmic and 29 heteroplasmic variants were identified in the proband’s peripheral blood, which is consistent with the presence of haplogroups H2a and U3a1a (III-4 in Fig. 2a, III-4bl in Fig. 2b, c). The mother (II-2) was homoplasmic for variants in haplogroup U3a1a and the father (II-1) was heteroplasmic for variants in haplogroup H2a and a third haplogroup, K1a3a. In the proband, the heteroplasmy level of the variants within m.1–9652 (region 1 was different from that of variants within m.9653–16569 [region 2]). For the 29 variants unique to the maternal haplogroup U3a1a, 13 from region 1 were at ~26% heteroplasmy and 16 from region 2 were at ~55% heteroplasmy (Fig. 2b, c and Supplemental family B data set). Four variants shared by both haplogroups U3a1a and H2a were found to be homoplasmic. Interestingly, the father (II-1) showed the same pattern of heteroplasmy level as his daughter (III-4) for variants in haplogroup K1a3a: ~26% heteroplasmy in region 1 and ~55% heteroplasmy in region 2. Variants from haplogroup H2a were shared by both the father and the proband, and were detected at ~74% heteroplasmy in region 1 and ~45% in region 2, respectively (Fig. 2b, c). The read depth of the entire mitochondrial genome showed that the coverage of region 1 is approximately 40–50% higher than that of region 2 (Supplemental Fig. 1a), suggestive of a duplication of region 1 in the proband and the father. Since it is absent in blood of the mother (II-2) and muscle of the older brother (III-2), this apparent duplication must be from the sequence with haplogroup H2a variants in both the proband (III-4) and the father (II-1). PacBio sequencing of long-range PCR products amplified using mtDNA primer pair F16561 + R16560 on the proband’s blood DNA confirmed the presence of two types of amplicons: (1) full-length consensus mtDNA amplicons with variants from either haplogroup U3a1a or H2a; and (2) short consensus 9779 bp amplicon with rearranged mtDNA fragment belonging to haplogroup H2a. The latter includes m.16561-m.9652 and inverted m.16561-m.109 (m.109-1:16569-16561), which was amplified by single forward primer F16561 ([Supplemental Fig. 4c and supplemental PacBio data family B_Tab SP[F16561] Amp_Analyzed]). Subsequent study of PCR products amplified using another mtDNA primer pair, F16428 + R16427, observed similar result except the resultant rearranged fragment was 10,045 bp instead of 9,779 bp (Supplemental Fig. 4c and supplemental PacBio data family B_Tab SP[F16428] Amp_Analyzed), suggesting this rearranged fragment could be larger in size. Single-primer PCR using forward primer only (F16561, F16428, F15792, F14734, F13715, and F12661) generated single band with expected size (Supplemental Fig. 3a). F11628 and F10608 failed to amplify the expected PCR products of 19,645 bp and 21,685 bp, which may suggest the distal end of the rearranged fragment does not extend to m.11628. The m.9652:109 junction was confirmed by PCR with different pairs of forward primers flanking the junction (Supplemental Fig. 3b). We observed the same results in the father as in the proband (data not shown). Such large complex rearrangement, if occurred in genuine mtDNA, would be associated with a mitochondrial DNA disorder. Given that both the father and the proband do not have clinical features of a mitochondrial disorder (Supplemental document 1), and the father transmits haplogroup H2a only instead of both haplogloups H2a and K1a3a, the rearranged fragment with haplogroup H2a is more likely a nuclear component. These results strongly suggest that the paternally transmitted rearranged mtDNA fragment is mega-NUMTs.

Fig. 2: Suggestive biparental mitochondrial DNA (mtDNA) inheritance pattern in family B.
figure 2

a Pedigree. “a” and “b” in the symbol represent m.1–9652 and m.9653–16569 of the mtDNA, respectively. Each color in “a” and “b” represents mtDNA of a single haplogroup and the proportion of the color area in “a” and “b” represents the level of heteroplasmy. The unfilled symbols indicate individuals not tested (III-1, III-3, III-5, III-6, III-7, III-8, III-9). b Schematic of the mtDNA genotypes defined by the homoplasmic and/or heteroplasmic variants aligned from the reference mitochondrial genome. The bars of each color represent mtDNA of a single haplogroup. m: muscle sample. c Summary of the haplogroup and mtDNA variant numbers in family B. bl blood sample, m muscle sample.

Full size image

In family C, the peripheral blood sample of the proband (III-2) had 28 heteroplasmic variants (66% or 34% heteroplasmy) and 6 homoplasmic variants, consistent with the presence of two haplogroups, H2a and J1c3e1 (Fig. 3 and supplemental family C data set). PacBio sequencing and data analysis confirmed the presence of full-length consensus mtDNA amplicons with either haplogroup H2a or J1c3e1 (supplemental PacBio data for family C). The mother (II-2) was homoplasmic for haplogroup J1c3e1. The father (II-1) was 34% heteroplasmic for haplogroup H2a and 66% heteroplasmic for another haplogroup, T2b3e. Additional analysis of the proband’s muscle (III-2m) revealed only homoplasmic variants from maternal haplogroup J1c3e1. As muscle is known to have higher mtDNA:nDNA ratio than that of blood,23 this result suggests that the paternally transmitted haplogroup H2a does not belong to genuine mtDNA component.

Fig. 3: Suggestive biparental mitochondrial DNA (mtDNA) inheritance pattern in family C.
figure 3

a Pedigree. The single color filled symbols represent individuals with mtDNA of homoplasmic variants belong to a single haplogroup (II-2) and the multicolor filled symbols represent individuals with mtDNA of heteroplasmic variants belong to different haplogroups (II-1, III-2). Each color represents mtDNA of a single haplogroup and the proportion of the color area represents the level of heteroplasmy. The unfilled symbols indicate individuals not tested (III-1). b Schematic of the mtDNA genotypes defined by the homoplasmic and/or heteroplasmic variants aligned from the reference mitochondrial genome. The bars of each color represent mtDNA of a single haplogroup. c Summary of the haplogroup and mtDNA variant numbers in family C. bl blood sample, m muscle sample.

Full size image

In family D (Fig. 4 and Supplemental family D data set), we identified multiHets and an apparent large-scale mtDNA deletion, m.9921_13787del3867, in the peripheral blood of the proband (III-2). The variants were consistent with the presence of haplogroups N1a1a3 and T2 with heteroplasmy level of 34% and 66%, respectively. The deletion was not observed in a buccal sample of the proband and the variants in haplogroup N1a1a3 were present at a much lower level (~4%). The mother (II-5) was homoplasmic for variants in the haplogroup T2 in both her blood and buccal sample. The father (II-4) was heteroplasmic for variants in haplogroup N1a1a3 and a third haplogroup, H5. Haplogroup N1a1a3 variants were present at ~20% heteroplasmy in the father’s blood and were barely detectable in buccal (≤2%). The apparent m.9921_13787del3867 was only observed in the father’s blood but not in buccal. These findings suggest that the apparent deletion m.9921_13787del3867 is located on a paternally transmitted NUMT with variants from haplogroup N1a1a3 (referred to as NUMT1). Additionally, several variants with very low–level heteroplasmy were observed in the deletion region (m.9921_13787) from the blood of both the proband (III-2bl) and the father (II-4bl) (Fig. 4b). The presence of these very low–level heteroplasmic variants suggests that in addition to NUMT1, there are likely other copies of NUMTs with full-length mtDNA sequence shared by the proband and the father (referred to as NUMT2 and NUMT3 in Fig. 4b). The absence of very low–level heteroplasmic variant outside the deletion region could be due to NUMT2 and NUMT3 having the same variants as those in NUMT1, or preferential amplification of NUMT1 which is 3,867 bp shorter than NUMT2 and NUMT3.

Fig. 4: Suggestive biparental mitochondrial DNA (mtDNA) inheritance pattern in family D.
figure 4

a Pedigree. The single color filled symbols represent individuals with mtDNA of homoplasmic variants belonging to a single haplogroup (II-5) and the double color filled symbols represent individuals with mtDNA of heteroplasmic variants belonging to two different haplogroups (II-4,III-2). Each color represents mtDNA of a single haplogroup and the proportion of the color area represents the level of heteroplasmy. The unfilled symbols indicate individuals not tested (III-1). b Schematic of the mtDNA genotypes defined by the homoplasmic and/or heteroplasmic variants aligned from the reference mitochondrial genome. The bars of each color represent mtDNA of a single haplogroup. c Summary of the haplogroup and mtDNA variant numbers in family C. bl blood sample; buc buccal swab; RealMT real mtDNA; NUMT1, NUMT2, NUMT3 represent three different NUMT; m.9921–13787del3867 on an area with blue italic lines between two vertical lines indicates a region with the m.9921–3787del3867 deletion.

Full size image

Collectively, the results provide evidences that multiHets that are requently observed in our clinical samples represent coamplification of mega-NUMTs from the paternal nuclear genome. The heteroplasmy level of these paternally inherited variants was inversely correlated with mtDNA content of the specimen, with R varying from 0.6797 to 0.9998 and P value vary from 0.0278 to 5.73 × 10−10 (Fig. 5). The presence of these multiHets did not cosegregate with the mitochondrial disease in any of the four families, as none of the adults with multiHets were described as having features of a mitochondrial disease (Supplemental document 1). None of the deduced mega-NUMTs identified in this study (Supplemental Fig. 4b–d) have been previously reported or documented in the NUMT databases.19,24,25,26,27

Fig. 5: The heteroplasmy level of paternally transmitted variants is inversely correlated with mitochondrial DNA (mtDNA) content of the specimen.
figure 5

For the individuals with multiHets, when multiple sample types were evaluated, the heteroplasmic level of the variants of paternal origin were correlated with the mtDNA of the sample. The heteroplasmy level of the variants of paternal origin is inversely correlated with the mtDNA content. a The correlation of heteroplasmy level of the variants of paternal origin with the mtDNA content of the specimen in family A. b The correlation of heteroplasmy level of the variants of paternal origin with the mtDNA content of the specimen in family B. c The correlation of heteroplasmy level of the variants of paternal origin with the mtDNA content of the specimen in family C. d The correlation of heteroplasmy level of the variants of paternal origin with the mtDNA content of the specimen in family D. R correlation coefficient. Student’s t-test P < 0.05 is considered a significant correlation.

Full size image

DISCUSSION

In these four families and the three families described by Luo et al.12 a common feature is a male with two haplogroups transmits one haplogroup to some of his offspring, which is consistent with nuclear DNA transmission. If both haplogroups stand for genuine, circular mtDNA molecules, variants from both haplogroups rather than a single haplogroup would be expected to be transmitted from a male to his offspring. Further evidence is seen in the paternal grandmother (II-2) in family A of this study, individual III-6 in family A, and III-6 in family C from Luo et al.: a female transmitted variants from both of her haplogroups to her son. This is consistent with the cotransmission of genuine maternal mtDNA and her nuclear mega-NUMT. Large pedigrees as illustrated in Supplemental Fig. 5 are not always available to determine which genetic material is transmitted from a father to his offspring if mega-NUMTs and genuine mtDNA coexist in the father. Further investigation of extra specimen types from the individuals in these families provided additional evidence.

As segments in the nuclear genome, the copy number of NUMTs in cells from different specimen type of an individual should be consistent. In contrast, the copy number of mtDNA in cells from different specimen type differ dramatically, ranging from an average of hundreds of copies in blood to thousands of copies in muscle or urine.28 Therefore, muscle and urine have a much higher mtDNA to nuclear DNA ratio (mtDNA content) than peripheral blood or buccal. The higher the mtDNA to nuclear DNA ratio, the greater the dilution of NUMTs as template during PCR amplification. Our results show that the paternally transmitted variants, although present in the blood or buccal of some individuals, were not observed in the muscle (Fig. 3b, III-2m) or urine (Fig. 1b, IV-2ur) of these individuals and the heteroplasmy level of the paternally transmitted variants was inversely correlated with mtDNA content of the specimen (Fig. 5). This observation strongly suggests that the multiHets derived from NUMTs rather than mtDNA. The differing mtDNA content across specimen types provides an opportunity to clarify the origin of observed variants, and it is critical for patients with multiHets to determine if a pathogenic variant represents an artifact from mega-NUMTs or if it is truly present in the mtDNA. For example, a pathogenic m.7495A>G variant was identified at a low level of heteroplasmy in the blood and buccal of the proband of family A. As this variant was also identified in the urine of this patient, in which the mtDNA content is high and the likely mega-NUMT derived heteroplasmic variants were not observed, it strongly indicates this variant is a true mtDNA variant and may warrant additional invasive testing such as a muscle biopsy to determine if a diagnostic level of heteroplasmy would be observed in an affected tissue.

Additionally, complete mtDNA analysis of both parents of individuals with multiHets may also clarify the nature of possible reportable variants. Single, large-scale mtDNA deletions are generally de novo and associated with mtDNA deletion syndromes including Pearson syndrome, chronic progressive external ophthalmoplegia (CPEO), and Kearns–Sayre syndrome (KSS).29 Large-scale mtDNA duplications have only been observed in conjunction with large-scale mtDNA deletions in individuals with KSS.30 The apparent deletion in family D was not associated with a phenotype of mtDNA deletion syndrome in either the proband or the father. Further, deletions and duplications associated with KSS are always more abundant in muscle than in blood,29 whereas the duplication and variants identified in the peripheral blood of the proband and the father in family B was not observed in the muscle of an older sibling with suspected mitochondrial myopathy (Fig. 2a, III-2; Fig. 2b, III-2m), which is distinct from the common observation of accumulation of pathogenic mtDNA variants in postmitotic tissues of affected individuals. These findings provide further evidence that the deletion and duplication observed in these families are from mega-NUMTs rather than representing a diagnosis of mtDNA deletion syndromes.

In conclusion, our results indicate the previously reported “biparental inheritance of mtDNA”12 is not true paternal transmission of mtDNA, but rather the coamplification and sequencing of both rare, mega-NUMTs and genuine mtDNA using the common long-range PCR method.13,17 A hypothetical pedigree (Supplemental Fig. 5) illustrates how newly generated mega-NUMTs are transmitted from a male individual (I-1) to his descendants. These results have important clinical implications. For individuals with multiHets, follow-up evaluation of noninvasive specimens (such as buccal or urine) or postmitotic tissue (such as muscle) can reduce or eliminate the interference of mega-NUMTs. This allows for a cost-effective testing strategy that can differentiate between variants that are true mtDNA variants and variants that are technical artifacts from rare, long-range PCR amplifiable mega-NUMTs, which is critical for accurate diagnosis and genetic counseling.

URLs

MitoMap, https://www.mitomap.org/MITOMAP

Mitochondrial Pseudogene List, https://www.mitomap.org/foswiki/bin/view/MITOMAP/PseudogeneList

Human NUMTs database, https://sourceforge.net/projects/dmcrop/files/human NUMTs database/