MitoCOMON: whole mitochondrial DNA sequencing by primer design and long overlapping amplicon assembly

Abstract

Background

Mitochondrial DNA sequences are used for inter- and intra-specific comparison analysis in ecological studies. Instead of using short regions as marker sequences, analyzing longer regions, such as whole mitochondrial DNA sequences, can improve the accuracy of such studies by increasing the likelihood of detecting species or specific sequences. However, current methods for sequencing whole mitochondrial DNA require primer design for each target species or long fragments of genomic DNA as a PCR template. We developed a method and accompanying tool for PCR-based long-read sequencing of whole mitochondrial DNA, named MitoCOMON, which is applicable to wide-target taxonomic clades and partially digested template DNA.

Results

PCR amplification of whole mitochondrial DNA as four fragments facilitates the successful assembly of the whole mitochondrial DNA sequence, even when a sample is a mixture of multiple species or partially degraded. The tool that we developed consists of two modules that can design a primer set for species in a target taxonomic clade and assemble the whole mitochondrial DNA sequence from amplicons which were amplified using the designed primer set. Primer sets were designed for mammal and bird species, which showed a high success rate for whole mitochondrial DNA sequencing with high sequence accuracy. Multiple whole mitochondrial DNA sequences were also assembled from samples mixed with the genomic DNA of several species without forming chimeric sequences. In addition to the accuracy, some assembled sequences also retained a long duplication at the D-loop region, suggesting that the method addresses large rearrangements. Compared with a method that amplifies the whole mitochondrial DNA as a single amplicon, our method was effective for partially degraded samples.

Conclusions

Our method and accompanying tool, named MitoCOMON, enables an easier acquisition of whole mitochondrial DNA sequences from samples with some DNA degradation without designing species-specific primers. This approach can enhance the accessibility of mitochondrial genomic data and is expected to improve the resolution of ecological analyses, including accurate species identification and individual-level discrimination.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-12010-0.

Background

The nucleotide sequence of mitochondrial DNA (mtDNA) is widely used for the analysis of inter- and intra-specific genetic variation to examine the evolution and geographic distribution of species and strains [1]. Because mtDNA shows both a conserved repertoire of genes [2] and a DNA mutation rate that is generally faster than chromosomal DNA [3, 4], mtDNA regions, such as cytochrome c oxidase subunit I (COI), cytochrome b (COB), and D-loop, may be used for inter- and intra-specific analyses [5]. Although short sequences of such marker genes are used in most studies, longer sequences can be used to detect a larger number of sequence differences at higher resolution.

An example of the advantage of the use of longer sequences may be found in phylogenomic studies of closely related individuals. For the inter-specific analysis of closely related species, whole mtDNA sequence comparison provides higher resolution of the phylogenetic relationship and resolves the incongruence among studies using short marker genes [6–8]. For intra-specific analyses, whole mtDNA sequence comparison analysis can detect a larger number of haplotypes compared with the analysis of short marker gene sequences [9, 10]. In addition to SNPs and small indels, larger genomic rearrangements, such as larger indels, duplications, and transpositions, may be detected as individual-specific characteristics by long sequence comparison analysis. Therefore, the use of longer sequences, such as whole mtDNA sequences, can successfully map the detailed distribution and evolution patterns for species and strains.

Another example is species identification. Genetic markers on mtDNA are frequently used for identifying eukaryotic species [5] by sequencing and comparing the marker sequences against databases such as RefSeq [11] and GenBank [12]. These markers are also used in metabarcoding analyses to detect multiple species present in environmental DNA samples collected from sources such as oceans, rivers, and soil, using high-throughput sequencing technologies [13, 14]. While most species identification analyses rely on short marker sequences which are typically a few hundred nucleotides in length, several studies have analyzed sequences longer than a kilobase or included multiple genetic markers, resulting in species information with higher resolution [15–18]. The application of longer sequences in species identification promises to yield more precise species profile and can significantly contribute to ecology and biodiversity research, although the use of longer sequences is currently limited by the biased repertoire of mtDNA sequences in public databases, which are skewed toward domesticated species and shorter marker sequences. Expanding the database to include longer, or ideally complete, mtDNA sequences from a wider range of species is essential for improving the accuracy of species identification.

Although the use of whole mtDNA sequences enables inter- and intra-specific analysis in more detail, it remains difficult to obtain a large number of whole mtDNA sequences when the targets are various non-model species. Currently, the most universal and unbiased method is to sequence the whole genome using a high-throughput sequencer, then assemble and extract the mtDNA sequence from the whole genome assembly. This, however, requires a large number of reads to obtain a sufficient number of mtDNA reads, resulting in a high cost. Genome skimming, which involves sequencing whole genomic DNA at low coverage to retrieve high-copy number sequences, is also used for whole mtDNA sequencing. However, this method still requires a larger number of reads compared to the following PCR-based approaches [19–21]. Another method is to amplify whole mtDNA as one or two long fragments and sequence them as long DNA sequence using a long-read sequencer, although this requires circular or long template DNA and a long amplifying polymerase with relatively low fidelity. The technique often results in low success rate for genomic DNA isolated from less fresh samples [22, 23]. Another method is to sequence many overlapping amplicons that cover the entire mtDNA [24, 25] or fragments amplified by rolling circle amplification [26–28], but these methods require a large number of primers that are distributed throughout the entire mtDNA. Moreover, it is difficult to design a primer set that is applicable to a wide range of species in a target taxonomic clade. To encourage sequencing and the use of whole mtDNA sequences for inter- and intra-specific studies, a method is needed that is applicable for a wide variety of non-model species in a target taxonomic clade, which is low-cost and does not require perfect sample condition.

Here, we present a new method to sequence whole mtDNA for a wide variety of species by designing a primer set combined with long-read sequencing. This method, named MitoCOMON (Mitochondrial DNA Complete sequencing by Merging Overlapping Nucleotides), enables the acquisition of accurate and structure-aware, whole mtDNA sequences for species in a specific taxonomic clade, combined with a tool that facilitates primer design and whole mtDNA sequence assembly.

Materials and Methods

Sample and genomic DNA Preparation

Samples of muscle tissue of Bos taurus (cattle), Sus scrofa (pig), Ovis aries (sheep), Cervus nippon (Hokkaido shika deer), Ursus (bear), Eumetopias jubatus (steller sealion), Procyon lotor (racoon), Meles anakuma (Japanese badger), Sus scrofa (wild boar), Macropus (kangaroo), Oryctolagus cuniculus (rabbit), Gallus gallus (chicken), Struthio camelus (ostrich), Anas platyrhynchos (duck), and Phasianus versicolor (green pheasant) were isolated from meat purchased from commercial meat vendor. Samples of muscle tissue of Treron sieboldii (green pigeon), Falco tinnunculus (common kestrel), Eophona personata (Japanese grosbeak), Luscinia cyanura (red-flanked bluetail), Poecile varius (Varied tit), and Agropsar philippensis (Chestnut-cheeked starling) were isolated from a deceased body found in the Toyota Forest (Aichi, Japan). Feather samples of F. tinnunculus, E. personata, and P. vairus were isolated from the same body of the bird from which the muscle tissue sample was isolated. Genomic DNA was extracted from 25 mg of muscle tissue or 1 cm of the feather quill samples using the NucleoSpin Tissue kit (Macherey-Nagel, Düren Germany).

Primer design

For the design of primer sets for whole mtDNA amplification, complete mtDNA sequences for 17,634 entries were downloaded from RefSeq [11] as of Apr 17th, 2024. Sequences of the target class, mammals, and birds (1,620 and 1,081 entries, respectively), were extracted (Tables S1 and S2) using TaxonKit 0.18.0 [29]. The sequences were searched for tRNA with tRNAscan-SE 2.0.12 [30] using the option “-M vert” to begin the sequences from the first nucleotide of tRNA-Phe and to follow the direction of its leading strand. The sequences were then aligned using MAFFT 7.526 [31] with default parameters. Throughout the alignment, the information content was calculated for each nucleotide position [32], which indicates the base diversity of a position in the alignment according to the equation:

where I was the information content and p_k was the probability of base A, T, G, or C at a position in the alignment. Using a 20 bp sliding window, the alignment was scanned and calculated for the average information content for each window and filtered for those with an average higher than 1.80. To form a highly conserved region, overlapping windows were retained and merged.

For each highly conserved region, the 20 bp window with the highest information content was selected as a candidate primer region. The consensus sequence for the selected region and the flanking 5 bp was calculated along with thermodynamic parameters (Table S3) using Primer3 2.6.1 [33] to search its partial sequence that is most suitable as a primer. Candidate primer sequences were then tested for specificity against the target taxonomic clade using PrimerProspector 1.0.1 [34]. Sequence matches between primer candidate sequences and the whole mtDNA sequence of the target taxonomic clade (mammalia or bird) were counted, and the ratio of the mtDNA entries with less than two mismatches with the primer candidate sequences was determined. The same calculation was also done for the same number of randomly selected non-target mtDNA entries (including the birds when targeting mammals and vice-versa). The primer candidate sequences with a target taxonomic clade ratio higher than 0.85 and that of the non-target taxonomic classes lower than 0.15 were selected as the final primer sequences (Tables S4 and S5). This process for primer design can be performed as a pipeline using the MitoCOMON design module.

To design the primer sets for amplifying the entire mtDNA sequence as four fragments, all combinations of the eight primer candidate sequences were tested in silico to determine whether the expected amplicons fulfilled the following conditions: 4 kb to 8 kb amplicon, and 0.5 kb to 3 kb of the overlapping sequences between neighboring amplicons. All the primer pairs fulfilling these conditions were experimentally tested by PCR using gDNA isolated from cattle for mammalian primers or chicken for bird primers. A validated set consisted of primer pairs generating an amplification of the expected length (Table 1).

Table 1.

Primer set sequences

Primer name	Sequence	Fragment
Mammal
NC_000884.1_18_37_5_0_for	AAAGCAARGCACTGAAAATG	Mammal #1
NC_000884.1_6875_6894_2_0_rev	GGTTCGAWTCCTTCCTTTCTT	Mammal #1
NC_000884.1_3697_3716_2_5_for	AAAGAGTTACTTTGATAGAGTAAATNA	Mammal #2
NC_000884.1_9836_9855_3_1_rev	TARTYTAATGAGTCGAAATCAYTT	Mammal #2
NC_000884.1_6875_6894_2_0_for	AAGAAAGGAAGGAWTCGAACC	Mammal #3
NC_000884.1_11726_11745_0_4_rev	ATTACTTTTATTTGGAGTTGCACC	Mammal #3
NC_000884.1_9836_9855_3_1_for	AARTGATTTCGACTCATTARAYTA	Mammal #4
NC_000884.1_673_692_3_4_rev	GTTTGCTGAAGATGGCGGTATATAGRC	Mammal #4
Bird
NC_036297.1_2775_2794_0_2_for	GAGGTTCAAATCCTCTCCCTAG	Bird #1
NC_036297.1_9503_9522_0_4_rev	TARRGATTGGAAGTCRATTGTAAT	Bird #1
NC_036297.1_8993_9012_3_0_for	TTYTTCTGAGCMTTCTTCCACTC	Bird #2
NC_036297.1_14852_14871_3_3_rev	TTTGGYTTACAAGACCAATGTTTTYA	Bird #2
NC_036297.1_11809_11828_0_0_for	AGYAATCCAMTGGTCTTAGG	Bird #3
NC_036297.1_231_250_0_1_rev	TACTGCTGARTACCCGTGGGG	Bird #3
NC_036297.1_14852_14871_3_3_for	TRAAAACATTGGTCTTGTAARCCAAA	Bird #4
NC_036297.1_3804_3823_0_5_rev	CTCTATGTTCACTTTATCATAGTGA	Bird #4

Fragment amplification and long-read sequencing

The amplification reactions were done using 0.1 ng of isolated mtDNA and the KAPA HiFi HS ReadyMix (Kapa Bio systems, Potters Bar, UK). The PCR protocol began with an initial denaturation step at 98°C for 3 min, followed by 30 cycles at 98°C for 30 s, 55°C for 30 s, 72°C for 3 min 30 s, and a final extension step at 72°C for 5 min. The amplified fragments were purified using Ampure XP (Beckman Coulter, Brea, CA) with the addition of the same volume of beads to each PCR product.

The amplified fragments were subjected to long-read sequencing using a MinION sequencer (Oxford Nanopore Technologies, Oxford, UK). An equimolar mixture of the four amplified fragments (4 fmol to 170 fmol in total per sample) was used for library construction with the Native Barcoding Kit 24 V14 (Oxford Nanopore Technologies, Oxford, UK). Libraries were loaded into an R10.4.1 Flow cell (Oxford Nanopore Technologies, Oxford, UK) and sequenced for 72 h (Table S6).

De novo assembly of whole mtDNA sequence(s)

Reads were basecalled and demultiplexed using Dorado 0.7.1 (Oxford Nanopore Technologies, Oxford, UK). They were initially used for the classification of each fragment by finding the primer sequences with Cutadapt 5.0 [35] using the following options: “-O 15 -e 0.2 -g ^< first primer> -g < second primer> --discard-untrimmed.” Each amplicon sequence was quality filtered using Chopper 0.9.0 [36] with the following setting: “-q 10 -l 3000.” The reads for each amplicon were clustered to obtain consensus sequences using Amplicon_sorter 2023-03-12 [37] with 300 and 1,000 reads per amplicon for single species analysis and mixture sample analysis, respectively. Consensus sequences were combined and analyzed by Minimap 2.28 [38] with “-x ava-ont” to obtain overlap information. The overlaps were filtered by removing those with an approximate per-base sequence divergence > 0.01, not at the terminal of each consensus sequence, and self-overlaps. The remained overlaps were subjected to Miniasm 0.3 [38] for de novo assembly of whole mtDNA sequence(s) with the following options: “-e 0 -s 300 -o 300 -n 0 -1 -2.” The assembled sequences were polished using Medaka 2.0.1 (Oxford Nanopore Technologies, Oxford, UK) three times with the reads after the quality filtering above. Finally, the sequence was rotated to start from the first nucleotide and to follow the leading strand of tRNA-Phe and annotated for coding sequences and RNAs using MITOS2 2.1.9 [39]. The procedure above, from read filtering to annotation, can be performed as a pipeline using the MitoCOMON assembly module.

To compare the assembled sequences with the mtDNA sequences in the database, each assembled sequence was queried with MegaBLAST [40] against the core nucleotide database at the NCBI. The nucleotide sequence of the top hit entry was downloaded and compared with the query sequence using dnadiff 1.3 [41] to calculate the number of SNPs and indels. For the assembled mtDNA sequences observed with tandem duplication at the D-loop region, the assembled sequence was compared with the top hit using Blastn [42], and the resulting alignment was visualized using gggenomes 1.0.1 [43].

Short-read sequencing and sequencing error detection

The NEBNext UltraII FS DNA PCR-free Library Prep kit (New England Biolabs, Ipswich, MA) was used to construct a library for short-read sequencing from 300 ng of the genomic DNA samples that were used for long-read sequencing. The library was diluted to 1,000 pM and subjected to sequencing by NextSeq1000 (Illumina, San Diego, CA) to obtain 150 bp paired-end reads (Table S7).

The short reads were mapped to the assembled mtDNA sequences using bwa mem 0.7.18 [44], followed by variant calling with samtools and bcftools [45, 46]. The genomic context of the position of the variant detected was validated by visualizing the mapping result using IGV 2.16.2 [47]. Self-dotplot of mtDNA sequences was produced using Gepard [48].

Genomic DNA digestion and sequencing by MitoCOMON and long-range PCR

Genomic DNA of S. scrofa (pig), O. aries (sheep), A. platyrhynchos (duck), and P. versicolor (pheasant) were digested using Covaris M220 (Covaris, MA, US) for 90, 10, or 5 s, while fixing other parameters as follows: Peak Power, 50.0; Duty Factor, 2.0; and Cycles/Burst, 200. The digested DNA samples were purified using AMPure XP beads (Beckman Coulter, CA, US).

To test the robustness of the method for samples with a shorter DNA length distribution, we conducted and compared two methods of mtDNA sequencing: MitoCOMON and long-range PCR which amplifies the whole mtDNA region using a single primer pair.

For MitoCOMON, mammal and bird primer sets were used for PCR of the digested genomic DNA of mammal and bird species, respectively. Condition of PCR, sequencing, and assembly were as described above.

For long-range PCR, the reverse complement of MiMammal and MiBird primers [49, 50] were used as primers. PCR was performed using PrimeStar GXL Polymerase (Takara Bio, Shiga, Japan) with the following program: initial denaturation at 98°C for 3 min, 30 cycles of denaturation at 98°C for 10 s, annealing at 60°C for 15 s, and extension at 68°C for 3 min, followed by a final extension at 68°C for 5 min. Amplified DNA was purified using AMPure XP beads (Beckman Coulter, CA, US). Library constructed with the Native Barcoding Kit 24 V14 (Oxford Nanopore Technologies, Oxford, UK) and sequenced using MinION sequencer (Oxford Nanopore Technologies, Oxford, UK) with R10.4.1 Flow cell (Oxford Nanopore Technologies, Oxford, UK) and sequenced for 72 h. The long reads were filtered to retain those with primer sequences at both ends, then filtered to identify those with primer sequences inside the read to remove chimeric reads using Cutadapt 5.0 [35]. To obtain amplicons with the predicted length, the reads were filtered to retain reads longer than 15,000 bp and shorter than 17,000 bp using Seqkit v2.4.0 [51]. The amplicon reads were subjected to amplicon_sorter 2023-03-12 [37] to obtain the consensus sequence, followed by error correction using Medaka 2.0.1 (Oxford Nanopore Technologies, Oxford, UK).

Results

Overview of MitoCOMON

MitoCOMON consists of two modules: primer set design and PCR fragment assembly (Fig. 1). The design module outputs primer sequence candidates from an input of whole mtDNA sequences from the target species. First, by calculating the information content and consensus sequence of each nucleotide position in the alignment, the entire mtDNA sequence may be divided and classified into highly conserved and less conserved regions. Next, for each highly conserved region, the consensus sequence of the most conserved 20 bp region is extracted as a primer candidate sequence. The primer candidate sequences are then filtered by calculating basic thermodynamic parameters and their specificity against the target species. Based on the primer candidate sequences, a primer set may be designed by considering the primer pairs fulfilling the conditions of the number of fragments and the overlap length between the neighboring fragments. The overlap regions were designed to contain at least one less conserved region to find the appropriate neighboring fragment pairs in the course of sequence assembly even when the sample included multiple different species. In this work, we designed the primer sets that cover the whole mtDNA with four amplicons, but the design can be adjusted to generate a larger number of shorter amplicons if needed. Multiple candidate pairs can be verified for their amplification efficiency by experimental validation before selecting the primer pairs to include in the final set. Subsequently, mtDNA samples are amplified using the designed primer set, and sequencing of the resulting amplicons is done using a long-read sequencer. A MinION sequencer (Oxford Nanopore).

Fig. 1.

An overview of MitoCOMON. See the main text for details. The figure of Long-read sequencer is from TogoTV (© 2016 DBCLS TogoTV, CC-BY-4.0 https://creativecommons.org/licenses/by/4.0/deed.ja)

Technologies, Oxford, UK) is recommended because it has flow cells that provide varying read amounts, which can be selected by the sample size of a study.

The assembly module is used to assemble the long reads of the sequenced amplicons obtained using the designed primer sets. As each long amplicon is pooled and sequenced, the module first clusters the reads for each amplicon. For each cluster, a consensus sequence is calculated and polished to obtain the sequence of each amplicon for each species. Then, the amplicon sequences are assembled by comparing the overlapped terminal sequences between the neighboring amplicons to obtain a circular assembly graph and a whole mtDNA sequence. As the overlapping regions are designed to include less conserved region(s), of which the sequence should be different between the different species, separate circular assembly graphs for each species may be generated when the source sample is a mixture of multiple species.

In the following sections, we demonstrate an application of MitoCOMON for designing the primer sets of mammals and birds to amplify four fragments of whole mtDNA as a proof of concept. The assembly module was applied to samples with both single and multiple species.

Primer design of mammals and birds for amplifying four fragments of mtDNA

We applied MitoCOMON to mammal and bird species by designing primer sets and assembling the mtDNA of various species. We downloaded all of the mtDNA sequences (17,634 entries) from RefSeq, containing 1,620 mammal species and 1,081 bird species (Tables S1 and S2). After extracting and aligning the mtDNA of mammals and birds, 39 and 56 highly conserved regions were detected, respectively (Fig. 2). As expected, the highly conserved regions were distributed unequally throughout the mtDNA. They were densely distributed around the rRNA, which is known to be highly conserved and sparsely distributed in the D-loop region. The most conserved sequences in the highly conserved regions were extracted and filtered using thermodynamic parameters, which resulted in 28 and 34 sequences, respectively (Tables S4 and S5).

The sequences were filtered by calculating the specificity against the mtDNA of the target species (Fig. 3). Retaining the query sequences with high specificity for the target species and lower specificity for the non-target species resulted in 13 and 20 sequences for mammals and birds, respectively. From the qualified primer candidate sequences, primer sets were designed to cover the entire mtDNA as four fragments with an overlap sequence longer than 500 bp between neighboring fragments. After experimentally validating the amplification efficiency of the primer pairs, the final primer sets were designed (Table 1).

Fig. 2.

Distribution of information content for the alignment of mtDNA. A Mammals and B birds. The dashed line is the information content of 1.80

Assembly of whole mtDNA sequences with high accuracy and retaining structure variation

Using the designed primer sets, we amplified, sequenced, and assembled whole mtDNA sequences for 11 and 10 species of mammals and birds, respectively (Table 2). Four fragments were separately amplified, and an equimolar amount of the purified amplicons was pooled and sequenced using a MinION sequencer (Oxford Nanopore Technologies, Oxford, UK). By applying the assembly module of the MitoCOMON pipeline, we successfully obtained the entire mtDNA of 10/11 and 9/10 species of mammals and birds, respectively. Two species failed to amplify and assemble one of the four amplicons, nevertheless the primer sequence showed a complete match with the mtDNA sequence of the same species in the database.

Table 2.

Assembly of complete mtDNA sequences

Species name	Common name	Length bp	Blastn top hit ID	Blastn top hit species	Identity %	SNPsa	Indelsa
Bos taurus	Cattle	16,340	MF663794.1	Bos taurus	99.93	11	1
Sus scrofa	Pig	16,775	JN601075.1	Sus scrofa	99.81	0	32
Ovis aries	Sheep	16,543	KF938317.1	Ovis aries	99.93	11	0
Cervus nippon	Hokkaido shika deer	16,543	NC_006973.1	Cervus nippon yesoensis	99.94	10	0
Ursus	Bear	16,760	AB863014.1	Ursus thibetanus japonicus	99.19	88	28
Eumetopias jubatus	Steller sealion	16,678	NC_004030.2	Eumetopias jubatus	99.22	34	96
Procyon lotor	Racoon	16,619	AB462046.1	Procyon lotor	99.80	3	30
Meles anakuma	Japanese Badger	16,471	NC_009677.1	Meles anakuma	99.66	47	9
Sus scrofa	Wild boar	16,730	OK329967.1	Sus scrofa	99.73	23	22
Macropus b	Kangaroo	N/Ac	N/A	N/A	N/A	N/A	N/A
Oryctolagus cuniculus	Rabbit	17,524	PP357264.1	Oryctolagus cuniculus	99.89	5	8
Gallus gallus	Chicken	16,783	CP115610.1	Gallus gallus	99.99	1	1
Treron sieboldii	Green pigeon	17,495	NC_062674.1	Treron sphenurus d	96.39	373	71
Falco tinnunculus	Common kestrel	17,657	NC_011307.1	Falco tinnunculus	99.16	64	80
Struthio camelus	Ostrich	18,302	Y12025.1	Struthio camelus	99.70	16	4
Anas platyrhynchos	Duck	16,603	OR269156.1	Anas platyrhynchos	99.98	3	1
Phasianus versicolor	Green pheasant	16,688	NC_010778.1	Phasianus versicolor	99.71	44	4
Eophona personata	Japanese grosbeak	16,775	KX812499.1	Eophona personata	99.19	126	10
Luscinia cyanura	Red-flanked bluetail	18,633	NC_026067.1	Luscinia cyanura	99.82	21	2
Poecile varius	Varied tit	16,773	LC541463.1	Poecile varius	99.87	20	1
Agropsar philippensis b	Chestnut-cheeked starling	N/A	N/A	N/A	N/A	N/A	N/A

^aNumber of SNPs and indels were calculated by comparing the assembled sequence and the sequence of Blastn top hit using dnadiff.

^bAssembly failed because of the failure of amplification of at least one fragment.

^cN/A, not applicable.

^dMitochondrial DNA of Treron sieboldii was not available in the Genbank.

For three species of birds (F. tinnunculus, E. personata, and P. vairus), we also obtained feather samples in addition to the muscle tissue samples and tested our method from genomic DNA isolation to fragment sequence assembly, which also resulted in the successful assembly of whole mtDNA sequences (Table S8).

To confirm the quality of the assembled sequence, it was analyzed using Blastn [42] against the GenBank database to roughly assign a species. All assembled sequences yielded a top hit with the expected species, except for Treron sieboldii (Green pigeon), whose mtDNA was not registered in the database. Although assembled sequences showed > 99% sequence identity with their top hit, except for Treron sieboldii, the number of SNPs and indels detected was different between species. Species of domesticated animals, such as cattle and pigs, showed a low number of SNPs and indels, whereas species of wild animals, such as bears and the common kestrel, exhibited a large number of SNPs and indels. The results indicate a large bias in the coverage of intra-specific genetic diversity of the species in the database.

The accuracy of the assembled sequences was validated as the MinION sequencer is relatively error-prone compared with short-read sequencers. We sequenced the whole genome using the NextSeq 1000 (Illumina, San Diego, CA) for the same genomic DNA samples without PCR and mapped the reads to the assembled sequences to detect errors. Each mtDNA sequence was detected with up to 9 errors per sample (Fig. 4A).

Fig. 4.

Errors in the assembled mtDNA sequences identified by mapping short-read sequencing. A Distribution and classification of detected errors. B An example of the self-dotplot of the Ursus mtDNA sequence enlarged at the complex multi-nucleotide repeat region. Red lines indicate the errors classified as “Inside multi-nucleotide repeat” group. See Figure S2 for self-dotplot of other samples

To determine the cause of the errors, the sequence of the region surrounding the error was inspected in detail. We found that the errors could be classified into four groups: “Inside mononucleotide repeat,” “Inside multi-nucleotide repeat,” “Possible heteroplasmy,” and “Possible sequencing error” (Fig. 4A). “Inside mononucleotide repeat” and “Inside multi-nucleotide repeat” errors are caused by a length change of a mononucleotide repeat longer than seven nucleotides, possibly introduced by replication slippage during PCR [52] and/or by base calling errors during the MinION reads [53]. In many mammal mtDNA and some bird mtDNA samples, complex multi-nucleotide repeats present in the D-loop region were hot spot of these errors (Fig. 4B, Fig. S2). “Possible heteroplasmy” indicates that the error involved a single nucleotide, but the corresponding nucleotide in the short reads consisted of two different bases, both with a high fraction. This was observed in the short reads of Poecile varius isolated from a feather sample and indicated that the nucleotide at the corresponding position was 36% and 64% of the adenine and guanine reads, respectively. This suggested that the assembled sequence shows only one type of base (adenine), but the mtDNA population in the sample likely involved a high ratio of molecules with another base at the corresponding position. The last “Possible sequencing error” represents a group of errors that were not grouped in the prior three groups, which could involve errors caused by sequencing itself. There were 30 errors detected in the assembled sequences, yielding an estimated error rate of 0.008% (30/373,877).

Fig. 3.

Specificity of primer candidate regions against target and non-target species. A Mammal and B birds. The plot corresponding to the regions with a ratio of target species higher than 0.85 detected with zero or one mutation and with a ratio of non-target species lower than 0.15 detected with zero or one mutation are colored red

Two mtDNA sequences for the bird that were assembled, Struthio camelus (Ostrich) and Luscinia cyanura (Red-flanked bluetail), showed a longer length compared with the average 16 kb of the bird mtDNA. We compared the assembled sequences with those registered in the RefSeq database and found a duplication of the 1.7 kb region that included the nd6 and D-loop regions (Fig. 5). The result was supported by the mapping depth of the Illumina reads to the unamplified reference sequence, which showed increased depth of the duplicated region (Fig. S1).

Assembly of whole mtDNAs from mixed samples

As we designed the overlapping sequences between amplicons to include less conserved regions, it is theoretically possible that separate mtDNA sequences were assembled from amplicons derived from a mixture of mtDNA of different species. To test this, we prepared a mixture of genomic DNA from two, three, or five species of mammals or birds and applied this method.

From a mixture of the same amount of two, three, or five species, we successfully constructed the whole mtDNA sequence of the mixed species separately with comparable amounts of sequencing error (Table 3). For the mixture of two species, we also prepared a mixture that introduced a 10-fold difference in the ratio of genomic DNA weight, which resulted in the successful construction of separate whole mtDNA sequences. The results suggest the robustness of our method against the weight bias of the mixed species.

Table 3.

Assembly of complete mtDNA sequence from mixed gDNA samples

Name	Contig #	Length bp	Blastn top hit ID	Blastn top hit species	Common name	Identity %	SNPs	Indels
Cattle + Pig (1:1)	1	16,768	JN601075.1	Sus scrofa	Pig	99.83	0	29
	2	16,339	MF663794.1	Bos taurus	Cattle	99.93	11	1
Cattle + Pig (1:10)	1	16,761	JN601075.1	Sus scrofa	Pig	99.80	1	32
	2	16,339	MF663794.1	Bos taurus	Cattle	99.93	11	1
Cattle + Pig (10:1)	1	16,761	JN601075.1	Sus scrofa	Pig	99.81	0	32
	2	16,339	MF663794.1	Bos taurus	Cattle	99.93	11	1
Duck + Green pheasant (1:1)	1	16,576	OR269156.1	Anas platyrhynchos	Duck	99.81	3	28
	2	16,662	NC_010778.1	Phasianus versicolor	Green pheasant	99.55	44	30
Duck + Green pheasant (1:10)	1	16,603	OR269156.1	Anas platyrhynchos	Duck	99.98	3	1
	2	16,688	NC_010778.1	Phasianus versicolor	Green pheasant	99.71	44	4
Duck + Green pheasant (10:1)	1	16,603	OR269156.1	Anas platyrhynchos	Duck	99.98	3	1
	2	16,688	NC_010778.1	Phasianus versicolor	Green pheasant	99.71	44	4
Three mammal mixa	1	16,768	JN601075.1	Sus scrofa	Pig	99.85	0	25
	2	16,543	KF938317.1	Ovis aries	Sheep	99.93	11	0
	3	16,339	MF663794.1	Bos taurus	Cattle	99.93	11	1
Three bird mixb	1	16,602	OR269156.1	Anas platyrhynchos	Duck	99.97	3	2
	2	16,774	LC541463.1	Poecile varius	Varied tit	99.88	20	0
	3	16,688	NC_010778.1	Phasianus versicolor	Green pheasant	99.71	44	4
Five mammal mixc	1	16,774	JN601075.1	Sus scrofa	Pig	99.88	1	19
	2	16,543	KF938317.1	Ovis aries	Sheep	99.93	11	0
	3	16,750	AB863014.1	Ursus thibetanus japonicus	Bear	99.19	97	20
	4	16,543	NC_006973.1	Cervus nippon yesoensis	Hokkaido shika deer	99.94	10	0
	5	16,339	MF663794.1	Bos taurus	Cattle	99.93	11	1
Five bird mixd	1	16,603	OR269156.1	Anas platyrhynchos	Duck	99.98	3	1
	2	16,774	KX812499.1	Eophona personata	Japanese grosbeak	99.19	126	9
	3	16,774	LC541463.1	Poecile varius	Varied tit	99.88	20	0
	4	16,688	NC_010778.1	Phasianus versicolor	Green pheasant	99.71	44	4
	5	17,495	NC_062674.1	Treron sieboldii	Green pigeon	96.39	373	71

^aMix of gDNA of S. scrofa (pig), O. aries (sheep), and B. taurus (cattle).

^bMix of gDNA of A. platyrhynchos (duck), P. varius (varied tit), and P. versicolor (green pheasant).

^cMix of gDNA of S. scrofa (pig), O. aries (sheep), B. taurus (cattle), Ursus (bear), and C. nippon (deer).

^dMix of gDNA of A. platyrhynchos (duck), P. varius (varied tit), P. versicolor (green pheasant), E. personata (Japanese grosbeak), and T. sieboldii (green pigeon)

Robustness of MitoCOMON for samples with a shorter length distribution

One of the purposes of developing MitoCOMON was to assemble mtDNA sequences from partially fragmented DNA samples. To determine whether MitoCOMON could be applicable for samples with a shorter DNA length distribution, we applied MitoCOMON to the genomic DNA of two mammal and two bird species at four levels of digestion (90 s, 10 s, 5 s, and 0 s; longer treatment period corresponds to a higher level of digestion) (Fig S3). For comparison, long-range PCR, which amplifies the entire mtDNA as a single fragment using a single pair of primers, was also evaluated for the same sample. For some of the highly digested DNA templates, assembly by MitoCOMON was successful, whereas long-range PCR failed (Table 4). For example, whole mtDNA was successfully reconstructed from the 10 s samples using MitoCOMON, suggesting that the method is applicable when a detectable amount of 4 to 8 kb DNA was present in the electropherogram, even when the overall length peak was much shorter (Fig. S3). Of note, for the sample in which both methods were successful, the length of the assembled sequences was always longer for the MitoCOMON product, as long PCR cannot assemble the region (about 230 bp) between the primers. The difference in the length observed for the MitoCOMON assembly of S. scrofa from the samples with different digestion level was due to a copy number change in the complex repeat sequences in the D-loop. Taken together, the results suggest that MitoCOMON can be applied to partially digested samples compared with the long-range PCR method.

Table 4.

Results of whole mtDNA assembly by MitoCOMON and Long PCR.

	90 s		10 s		5 s		0 s
	MitoCOMON	Long-range	MitoCOMON	Long-range	MitoCOMON	Long-range	MitoCOMON	Long-range
Sus scrofa	16,671a	Fb	16,772	F	16,769	16,557	16,775	16,549
Ovis aries	F	F	16,543	16,325	16,543	16,325	16,543	16,325
Anas platyrhynchos	F	F	16,603	F	16,603	16,364	16,603	F
Phasianus versicolor	F	F	16,688	F	16,688	F	16,688	16,462

^aNucleotide length of the assembly when the whole mtDNA assembly was successful.

^bF, Whole mtDNA assembly failed.

Discussion

We developed a method, named MitoCOMON, that can design a primer set to amplify the whole mtDNA of a targeted taxonomic clade as four overlapping amplicons. The amplicons were pooled and sequenced by long-read sequencers and assembled as whole mtDNA sequence(s) by simply matching the overlapping sequences at both termini. We demonstrated that our method reconstructed the whole mtDNA sequence(s) from genomic DNA of a single species or a mixture of multiple species, and even when the template DNA was partially digested.

Compared with methods that amplify and sequence whole mtDNA as a single fragment, our method offers two major advantages. First, the length of the amplicons is up to 8 kb, so they can be amplified using a polymerase with high-fidelity, which is widely used for the amplification of short marker genes. Because amplicons obtained by the long-range PCR method are longer than 15 kb, which exceeds the limit of the amplification length of the typical high-fidelity enzyme, a polymerase that is suitable for amplification of long fragments, yet with relatively low fidelity, is usually used. We demonstrated that the combination of high-fidelity polymerase and a MinION sequencer can produce highly accurate whole mtDNA sequences (Fig. 4). Second, amplicons can be amplified from partially digested DNA samples. Some studies conduct mtDNA sequencing of a single species by amplifying the entire mtDNA as one or two fragments, whose success rate was approximately 50% [22, 23]. Our analysis revealed more successful cases for digested DNA compared with the long-range PCR method, which amplifies the entire mtDNA as a single fragment (Table 4). Therefore, MitoCOMON is a useful method to obtain accurate mtDNA sequences from samples under a wider range of conditions.

We demonstrated that sequencing mtDNA using a long-read sequencer can recover the entire mtDNA sequence with structural variation, such as long duplication (Fig. 5). The duplication of the long region in the D-loop is known to occur in several bird species [54]. Because the duplication length was larger than the length of short reads, the assembly of short reads would have missed such duplications if not carefully analyzed with their mapped read depth. Long-read sequencing for mtDNA analysis can reveal structural variants of mtDNAs that have been missed, which are important for understanding the diversification of mtDNA throughout evolution.

Fig. 5.

Comparison of whole mtDNA sequences with duplication. A Struthio camelus (Ostrich) and B Luscinia cyanura (Red-flanked bluetail). Protein coding genes, red; rRNA, green; tRNA, light blue

A limitation of primer design using MitoCOMON is that the selection of the target taxonomic group requires careful consideration. While we successfully designed primer sets for mammals and birds, it was not possible to design a primer set targeting Actinopterygii (ray-finned fish) due to the high sequence diversity among species in this taxonomic group. Conversely, designing a primer set that strictly excludes certain taxonomic groups can also be challenging when those groups are closely related to the target taxonomic group (e.g. reptiles when targeting birds). In such cases, with reconsidering the target species, it is necessary to target narrower taxonomic groups to enable primer set design.

Although the amount of mtDNA information is increasing in public databases, it is still important to increase the amount of mtDNA sequence information from a wider variety of species and strains for the usage as a reference database for a more precise species identification analysis. In addition, it is important to increase not only the short marker sequences but also the entire mtDNA sequences, so that the application of long-read sequencing for species identification produces useful results. For example, for species identification of eDNA, several groups recently amplified and sequenced a few kb-long regions, including a control region and 12S rRNA instead of typical shorter marker sequences. This enabled a more accurate species identification, but resulted in fewer number of assigned species compared to the analyses using shorter markers, in part because of the lack of long mtDNA sequence information in the database [15, 16]. Not limited to the long amplicon analysis of eDNA, more whole mtDNA information will be also useful for species identification of full or partial mtDNA sequences obtained from shotgun metagenomic analyses, non-targeted eDNA sequencing, trace DNA sequencing, and ancient DNA sequencing. Because longer amplicon sequence includes a higher number of sequence variations useful for accurate species identification, increasing the sequence data of entire mtDNA are crucial for application of long-read sequencing to various DNA analysis to obtain higher resolution of information of ecology, biodiversity, and evolution.

Conclusions

We have developed a novel method and accompanying tool, MitoCOMON, which enables the efficient acquisition of complete mitochondrial DNA sequences for species in a target taxonomic clade from samples exhibiting partial DNA degradation. This approach can enhance the accessibility of mitochondrial genomic data and is expected to improve the resolution of ecological analyses, including accurate species identification and individual-level discrimination.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MitoCOMON

Mitochondrial DNA Complete sequencing by Merging Overlapping Nucleotides

mtDNA

Mitochondrial DNA

Author contributions

YF, MK, and HT conceived the ideas and designed methodology; MK prepared samples; YF analyzed the data and led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Funding

Not applicable.

Data availability

The code of MitoCOMON is available at GitHub (https://github.com/ToyotaCRDL/MitoCOMON). The sequence data generated in this study have been deposited in the DNA Data Bank of Japan (DDBJ) under the BioProject accession number PRJDB35418. Under this BioProject, the raw sequencing data (both long- and short-reads) are available with accession numbers DRR699304-DRR699386, and the assembled whole mtDNA sequence are available with accession numbers LC880087-LC880105.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.