Bone specimens received from DPAA are thoroughly sanded via Dremel tool to remove surface contaminants, before collecting a 1.0 gram portion which is washed with sterile water and 100% ethanol. After allowing to air-dry, these bone fragments are pulverized using a blender cup. Teeth are cleaned using a 70 mM commercial bleach solution and dried under UV light for 15 minutes. Once dry, the crown is separated from the root using a dental drill for the collection of dentin powder. Collected bone or tooth powder is dissolved in 7.5 mL of demineralization buffer (0.5 M EDTA, pH 8.0, 1% N-Lauroylsarcosine) and 200 uL proteinase K (20 mg/mL) via overnight incubation in a 56°C incubator-shaker with constant agitation. Samples are then extracted with a standard 25:24:1 Phenol/Chloroform/Isoamyl Alcohol (PCIA) method followed by concentration using a Millipore Amicon Ultra-4/30K centrifugal filtration device.

Due to the nature of high heat and fire that these skeletal elements were exposed to, a high degree of thermal alteration was to be expected. As has been well documented in the scientific community, exposure to high temperatures causes DNA degradation, often exponentially so depending on the heat samples are exposed to. Confirming these suspicions, initial testing of Tokyo Prison Fire bones revealed little to no success with traditional Sanger sequencing of mitochondrial DNA (mtDNA), autosomal STR, and Y-STR methods. This resulted in a push to process these samples through Next Generation Sequencing (NGS), also known as Massively Parallel Sequencing (MPS). NGS processing enables a deeper sequence coverage and enhanced sensitivity compared to traditional Sanger DNA processing. DNA fragments smaller than the smallest mini primer set (79 base pairs) can now be sequenced using these methods. Another benefit to NGS testing, is that the entire mitogenome can be sequenced, as opposed to just the control region targeted in traditional mitochondrial Sanger sequencing methods. Since mtDNA is passed down through the maternal lineage, it is not as unique as nuclear DNA. Therefore, being able to sequence the entire mitogenome allows for better discrimination between individuals who may have common or similar mitotypes.

Due to the large amount of damage seen in these types of samples, the samples first undergo a DNA Repair step. Uracil-Specific Excision Reagent, or USER treatment, is performed to remove uracil present due to cytosine deamination. This is followed by a MinElute purification, a silica-based purification step, in which PB buffer increases the salt concentration and allows the DNA to bind to the silica membrane. Two wash steps are performed with PE buffer to wash away any impurities, followed by eluting the DNA from the silica column by adding TE-4 to decrease the salt concentration. Samples are then quantified using a Qubit 2.0 Fluorometer with the High Sensitivity and Broad Range kits.

Following DNA Repair, samples undergo Library Preparation using the NEBNext Ultra DNA Library Preparation Kit. The first step is End Repair, which fills in abasic sites created from the USER treatment, creates blunt ends on the DNA strands, and performs A-tailing to allow for adapters to bind later on. Next is Adapter Ligation, where looped adapters are attached to the DNA fragments at the adenosine added during a-tailing. The adapters are looped for stability purposes; however, they need to be opened to allow for indexing in a later step. The middle of each looped adapter contains a uracil, so another USER treatment is performed to remove the uracil and open up the looped adapter. This is followed by an AMPure purification step, a paramagnetic bead-based cleanup method that utilizes solid phase reversible immobilization (SPRI) beads. The high salt concentration of the AMPure beads allows the DNA to bind to them. A magnetic rack pulls the DNA-bead complex out of solution, and two wash steps are performed using 80% ethanol to wash away any impurities before eluting the DNA from the beads using TE-4. Once the DNA is back in solution, the magnet pulls the beads out of solution, and the eluate containing the DNA is carried on for downstream processing.

Now samples are ready for indexing and Library PCR. Indices are added to the DNA samples to ‘barcode’ them, so when they are pooled together and loaded, the instrument can determine which sequence being synthesized and read belongs to which sample. This is done using a Combinatorial Dual Indexing strategy, meaning all samples defined within a set receive the same i5 index to barcode the samples as a set. Each sample within that set will receive a unique i7 index to barcode the samples from one another within a set. Once indices are added to the samples, they undergo limited cycle PCR to attach the indices and selectively enrich the DNA fragments with adapter molecules attached. Following PCR, another AMPure purification is performed to remove any impurities from the indexed samples. Samples are run on an Agilent 2100 Bioanalyzer using the High Sensitivity assay as a quality check to ensure that the Library Preparation was successful.

Once samples are indexed, the library product is ready for hybridization capture using Arbor Biosciences myBaits v5 kit for target enrichment of the mitochondrial DNA. During day one of hybridization capture, the library product is combined with blocking reagents, followed by a short incubation period. The blocking reagents bind to the DNA that we do not want the baits to bind to, and heat denatures the double stranded DNA (dsDNA) to single stranded DNA (ssDNA). Next, hybridization reagents and Global Mito baits are added to the library-blocker complex and incubated overnight. The baits used are specific to the type of DNA that is being enriched, in this case the biotinylated Global Mito baits are designed to cover the human mitochondrial genome and were designed based upon many different mitochondrial sequences representative of global diversity to be compatible with a range of haplogroups. The overnight incubation period allows for these baits to bind to the target DNA. During day two of hybridization capture, streptavidin coated magnetic beads are added to the DNA-bait complex and bind to the biotin of the baits. This DNA-bait-bead complex is pulled out of solution using a magnet, and any non-hybridized DNA is washed away. The complex is amplified in duplicate, which will dissociate the DNA-bait-bead complex and enrich the captured library. Following the PCR step, the complex is dissociated, and the sample is placed back on a magnet. The magnet will pull the beads out of solution, and the solution containing the DNA is processed downstream. The DNA samples then undergo another MinElute purification.

Following hybridization capture, the samples are ready to be pooled and sequenced. Since the samples have all been indexed, all samples within a set can be pooled together by volume. The pools and their associated positives are quantified on the Agilent 2100 Bioanalyzer using the 7500 and High Sensitivity assays. Once the amount of DNA in the pools is quantified, all pools are diluted, and multiple pools can then be pooled together, ensuring there is no overlap between the i5 index for each of the individual pools. Once the individual pools are pooled together, this new pool is denatured using NaOH, as ssDNA is necessary to load for sequencing. PhiX is added to the pool to create diversity within the pool, and to act as a sequencing positive control. The final pool is loaded on the Illumina MiSeq or NextSeq 550. Samples are primarily loaded on the NextSeq, as it allows for greater multiplexing, loading up to 56 libraries on one instrument. The instruments utilize sequencing by synthesis to sequence clonal clusters of ssDNA on the instrument’s flow cell. Fluorescently labeled reversible terminator bound dNTPs are added, allowing the complementary dNTP to be added and imaged, before being cleaved to allow incorporation of the next base. This allows for true base-by-base sequence of each DNA fragment.

Once the sequencing run is complete, the data is imported into the CLC Genomics Workbench v12 and run through specific AFDIL-designed workflows to analyze the data. A few key components of the workflow include removing PCR duplicates, meaning that all sequences seen are unique. This is important because it is still common to see some damage in samples that wasn’t completely removed during the USER step of library preparation due to the extremely poor quality of samples processed at AFDIL. By removing PCR duplicates, it is easier to see damage within a sample if only seeing it across a few reads, whereas if PCR duplicates were not removed, that damage may be seen across many duplicates of the same read, making it more difficult to identify as damage. Coverage thresholds set include that a position must have at least five paired reads of coverage to determine a base call, any positions with less than the 5X coverage will be excluded from analysis. To call a variant, that variant must be present in at least three reads, and in at least 10% of the total reads to be called. The sample sequence is compared to the Revised Cambridge Reference Sequence (rCRS), and the actual reported sequence is the differences from the rCRS.

AFDIL is constantly improving methods and processing, including a recent validation of a new library method, using the KAPA HyperPrep Library Kit. The steps and concepts are very similar to the NEBNext library preparation. Some major differences include utilizing Y-shaped adapters instead of the looped adapters, which eliminates the need for the second USER step. These adapters are also already indexed, and use a Unique Dual Indexing strategy, as compared to a Combinatorial Dual Indexing strategy. This means that each sample within a set will now get both a unique i7 index and a unique i5 index, which allows for greater multiplexing capabilities. This new library preparation procedure is an enhanced version of the current protocol, which is more efficient and effective.

Currently, AFDIL has processed samples from over 240 unique Tokyo Prison Fire skeletal elements across multiple commingled mass graves. These initial results indicate a consistent trend observed in DNA processing of samples from re-exhumed WWII mass graves, where bones believed to be sourced to a single individual may be spread across five or more separate burials. Despite the severe degree of thermal alteration, NGS has had a success rate of approximately 62% in producing full mitogenome DNA profiles, with 48 unique mitochondrial sequences generated. Using these DNA results, re-articulation of skeletal elements by DPAA has led to the identification of four new service members who were originally believed to have perished in the Tokyo Prison Fire, with additional tentative identifications in progress. This effort would not be possible without the analytical power afforded by the modern NGS techniques that AFDIL has adopted for forensic casework.