Introduction

Although founded more than 20 years ago (1), the field of ancient DNA research continues to grow (2) and expand into new areas (3,4). Ancient DNA research shares a common problem with forensics and other approaches requiring analyses of museum and non-invasively collected specimens; the amount of endogenous DNA available in the samples is often limited. Thus, extraction techniques that retrieve as much DNA as possible from a specimen are of crucial importance. A wide range of techniques has been published to date (5,6,7,8,9), all of which aim to maximize DNA yields, while minimizing the co-extraction of PCR inhibitors (7,10). However, rather than being the product of systematic investigations, these techniques are often founded upon untested assumptions regarding the state of the various biomolecules in ancient samples, including proteins and the DNA itself. Although limited comparisons between different methods (11,12,13) have been made, no comprehensive study comparing the major protocols in use has been published to date.

We tested several published extraction methods on a number of Pleistocene cave bear bones and teeth using quantitative PCR. We also evaluated the effect of various chemicals added to the extraction buffer and investigated factors such as extraction temperature and extraction duration. We also assessed the performance of PCR amplification both with and without bovine serum albumin (BSA) (14). Finally, we measured the ability of commercially available DNA polymerases to successfully amplify ancient DNA.

Materials and Methods

Samples

We used eight cave bear bone samples and two cave bear teeth, originating from nine different caves in Austria, Switzerland, Slovenia, Croatia, Russia, and Germany (see Supplementary Table S1 available online at www.BioTechniques.com). As cave bears (Ursus spelaeus) became extinct at the end of the Pleistocene (15), all samples are of Pleistocene age and most likely older than 20,000 years. Three specimens were radiocarbon dated using accelerator mass spectroscopy, resulting in age estimates of between 27,000 and 44,000 years (3,5).

DNA Extraction Methods

Initially we tested two published extraction methods (5,8), one slightly modified method, and three commercially available DNA extraction kits, using three of the above-described samples and compared the DNA yields using quantitative PCR. These methods include the basic principles mostly used for ancient DNA purification, ultrafiltration via spin columns, and binding to silica. We also used spin columns combined with an extraction buffer used in the published silica method (5) in order to disentangle the processes of DNA solution and purification (for detailed information see the supplementary material). To assess the effect of modifying a number of factors, including changing the chemicals added to the extraction buffer, varying incubation time, and fluctuating incubation temperature, we selected one method (5) that produced high DNA yields, demanded comparatively little working time, and allowed for systematic modification. For this assessment, all components and their concentrations need to be known. We used three (or only two at the end of the test series) cave bear samples when applying this method, while optimizing individual factors. We then compared DNA yields to the best performing method in each test series to determine the relative performance of each method and applied paired Student's t-tests on each series. In cases where we found no significant difference between methods—a common finding attributable to the stochastic nature of DNA yields from ancient bone—we continued with the method that yielded the best results, provided that it did not substantially increase the workload and costs compared with other statistically indistinguishable methods.

Finally, we compared the DNA yields obtained from the optimized extraction method with several established methods. We used eight different cave bear samples in duplicate. All major principles of DNA purification (alcohol precipitation following phenol/chloroform extraction, ultrafiltration using spin columns, and binding to silica) were represented. We tested each extract's degree of inhibition with quantitative PCR using a defined amount of modern human DNA spiked with ancient extracts (16). We calculated the extent of inhibition from both the exactly measured number of human DNA template and from the efficiency of each individual quantitative PCR with and without ancient DNA extract (for details see the supplementary information). For each test we used a homogenized powder obtained by grinding the samples using a Spex 6750 freezer mill (Spex SamplePrep, Metuchen, NJ, USA) to avoid variation due to different DNA preservation in different parts of the bones. Blank controls were performed throughout all experimental steps.