Ancient DNA Methods Improve Forensic DNA Profiling of Korean War and World War II Unknowns

Abstract

The integration of massively parallel sequencing (MPS) technology into forensic casework has been of particular benefit to the identification of unknown military service members. However, highly degraded or chemically treated skeletal remains often fail to provide usable DNA profiles, even with sensitive mitochondrial (mt) DNA capture and MPS methods. In parallel, the ancient DNA field has developed workflows specifically for degraded DNA, resulting in the successful recovery of nuclear DNA and mtDNA from skeletal remains as well as sediment over 100,000 years old. In this study we use a set of disinterred skeletal remains from the Korean War and World War II to test if ancient DNA extraction and library preparation methods improve forensic DNA profiling. We identified an ancient DNA extraction protocol that resulted in the recovery of significantly more human mtDNA fragments than protocols previously used in casework. In addition, utilizing single-stranded rather than double-stranded library preparation resulted in increased attainment of reportable mtDNA profiles. This study emphasizes that the combination of ancient DNA extraction and library preparation methods evaluated here increases the success rate of DNA profiling, and likelihood of identifying historical remains.

Keywords: ancient DNA; degraded DNA; forensic DNA profiling; human identification; massively parallel sequencing (MPS); mitochondrial DNA.

Figure 1

These historical photographs show some of the important events that occurred after service members were killed in action during the Korean War. (A) In this photograph, taken on 19 August 1950, a battlefield cemetery in Daegu (formerly Taegu), Korea, contains dozens of temporary graves where the bodies of soldiers killed in action were laid to rest. This image was reprinted with permission of the U.S. Army Quartermaster Museum in Fort Lee, Virginia. (B) This photograph of the Kokura embalming laboratory was originally printed in the American Graves Registration Service Group 8204th Army Unit APO 3 Brochure, April 1955. Examination tables were equipped with exhaust systems for formaldehyde treatment. (C) These remains from a Korean War service member were covered with a powdered hardening compound prior to shipment to the United States. This photograph was reprinted courtesy of the DPAA. (D) In Operation GLORY, the first of the United Nations war dead from North Korea were received at Moji Port near Kokura, Japan (ca. fall of 1954). This photograph was originally printed in the American Graves Registration Service Group 8204th Army Unit APO 3 Brochure, April 1955.

Figure 2

Relative informative sequences recovered per extraction protocol across 15 historical skeletal remains with the ancient DNA extraction fractions (A) separate and (B) merged. The total informative sequences per library was calculated by multiplying the number of DNA molecules in each library as determined by qPCR by the proportion of sequences longer than 35 bp that mapped to the human reference genome. The relative informative sequences were then calculated by taking the ratio of total informative sequences for each extraction/sample combination to the total number of informative sequences from the best performing extraction protocol per sample. The box plots follow the standard Tukey representation to show the distribution of the relative informative sequences per extraction type (n = 15). The significance of different relative amounts recovered per extraction was evaluated with a Wilcoxon test with a correction for multiple testing. The ancient DNA protocols (aDNA37: 37 °C digestion; aDNA56: 56 °C digestion; Dabney merged: merging of total informative sequences per extraction/sample from aDNA37 and aDNA56) are colored in blue; forensic protocols (PM: PCIA with Min Elute; PUM: PCIA with Min Elute and USER treatment; PFM: PCIA with Min Elute and FFPE treatment) in green; and the combined forensic ancient DNA protocol (A_D: AFDIL digestion with Dabney purification) in yellow.

Figure 3

The (A) percent mapped (B) average fragment size (bp) and (C,D) observed 5′ C to T substitution frequencies of DNA recovered from historical and ancient DNA samples. Each bar represents a single sample for fifteen historical samples (purple) and seven Late Pleistocene modern human skeletal samples from ~37,000–46,000 years ago (dark green). In (A–C) the x-axis is the same for each and the black line represents the average for the ancient skeletal (dark green) samples and the dashed lines show two standard deviations above and below this average. In (C) the red line represents the 10% threshold typically used to determine if a sample contains ancient DNA. In (D) each point represents the 5′ C to T substitution frequency at a position from the 5′ end of DNA fragments recovered from a single sample. Note that only historical samples are shown in (D).

Figure 4

Relative informative nuclear DNA sequences recovered per library preparation protocol across 15 historical skeletal remains. The relative informative sequences per library was calculated as described in Supporting Information Text S2. The box plots follow the standard Tukey representation to show the distribution of the relative informative sequences per extraction type (n = 15). The significance of different relative amounts recovered per library prep type was evaluated with a Wilcoxon test with a correction for multiple testing.

Figure 5

(A) The distribution of the percentage of sequences at least 30 bp long that mapped to the human mtDNA reference genome after capture using the MPI workflow. (B) The relative number of mapped mtDNA sequences across library preparation methods for both the CLC and MPI workflows. The box plots follow the standard Tukey representation to show the distribution of the relative informative sequences per extraction type (n = 15).

Figure 6

Average coverage of different qualities of samples with MPI and SRSLY library preparations after enrichment for human mtDNA using the CLC workflow. Samples are colored based on the quality as determined from the initial casework evaluation before changing the extraction and library preparation methods (Table 1). The left plots underwent MPI library preparation and the right plots SRSLY library preparation.