Date Published: January 27, 2017
Publisher: Public Library of Science
Author(s): Henrik B. Hansen, Peter B. Damgaard, Ashot Margaryan, Jesper Stenderup, Niels Lynnerup, Eske Willerslev, Morten E. Allentoft, David Caramelli.
Large-scale genomic analyses of ancient human populations have become feasible partly due to refined sampling methods. The inner part of petrous bones and the cementum layer in teeth roots are currently recognized as the best substrates for such research. We present a comparative analysis of DNA preservation in these two substrates obtained from the same human skulls, across a range of different ages and preservation environments. Both substrates display significantly higher endogenous DNA content (average of 16.4% and 40.0% for teeth and petrous bones, respectively) than parietal skull bone (average of 2.2%). Despite sample-to-sample variation, petrous bone overall performs better than tooth cementum (p = 0.001). This difference, however, is driven largely by a cluster of viking skeletons from one particular locality, showing relatively poor molecular tooth preservation (<10% endogenous DNA). In the remaining skeletons there is no systematic difference between the two substrates. A crude preservation (good/bad) applied to each sample prior to DNA-extraction predicted the above/below 10% endogenous DNA threshold in 80% of the cases. Interestingly, we observe signficantly higher levels of cytosine to thymine deamination damage and lower proportions of mitochondrial/nuclear DNA in petrous bone compared to tooth cementum. Lastly, we show that petrous bones from ancient cremated individuals contain no measurable levels of authentic human DNA. Based on these findings we discuss the pros and cons of sampling the different elements.
With the introduction of next-generation sequencing (NGS), ancient DNA (aDNA) studies on humans have progressed from analyses of a few hundred basepairs of mitochondrial DNA to large-scale population genomic studies [1–3]. The feasibility of such projects ultimately rely on having access to many ancient samples with sufficient biomolecule preservation. In particular, owing to the indiscriminate nature of ‘shotgun’ sequencing, the proportion of DNA that derives from the target species—the endogenous DNA content—is a crucial factor. Because DNA degrades over time , and skeletal tissues are invaded by microbes, the endogenous DNA content is often very low in ancient samples (<1%) making genome-scale analyses impossible or, at best, very expensive . Thus, recent aDNA research on human remains has focused on identifying new suitable substrates [6–9] and optimizing DNA extraction methods [6, 10, 11], thereby increasing the success rate for identifying samples that are suitable for genome-scale analyses. Owing to high levels of endogenous DNA, the inner part of the petrous bone and the cementum layer in teeth roots are currently recognized as the optimal substrates for such research. All laboratory work was performed according to strict aDNA standards [28, 29] in dedicated clean laboratory facilities at Centre for GeoGenetics, Natural History Museum, University of Copenhagen. We believe that the results presented herein provide aDNA researchers with important new arguments, necessary for making informed decisions during the sampling process. Future research should aim at refining our understanding of DNA decay in space and time. Large-scale studies that manage to correlate aDNA preservation with parameters such as age, average temperature at burial depth, soil pH, storage time and metagenomic profile, should be highly informative—in particular when the effect varies greatly between different sites as seen in this study. Clearly the preservation environment of the Viking Age skeletons has resulted in both poor visual and molecular preservation of the teeth, but not affected the petrous bones, whereas the Bronze Age samples, despite thousands of years older, showed a much better tooth preservation. Here the objective has been of a more practical character, namely to generate empirical data that are easily translated into protocol optimizations. We will therefore end with a couple of simple take home messages that should be useful in that regard. Future refinements of the sampling processes may nuance these statements, but this is the current state of knowledge: Source: http://doi.org/10.1371/journal.pone.0170940