DNA in ancient bone – Where is it located and how should we extract it?

Summary

Despite the widespread use of bones in ancient DNA (aDNA) studies, relatively little concrete information exists in regard to how the DNA in mineralised collagen degrades, or where it survives in the material's architecture. While, at the macrostructural level, physical exclusion of microbes and other external contaminants may be an important feature, and, at the ultrastructural level, the adsorption of DNA to hydroxyapatite and/or binding of DNA to Type I collagen may stabilise the DNA, the relative contribution of each, and what other factors may be relevant, are unclear. There is considerable variation in the quality of DNA retrieved from bones and teeth. This is in part due to various environmental factors such as temperature, proximity to free water or oxygen, pH, salt content, and exposure to radiation, all of which increase the rate of DNA decay. For example, bone specimens from sites at high latitudes usually yield better quality DNA than samples from temperate regions, which in turn yield better results than samples from tropical regions. However, this is not always the case, and rates of success of DNA recovery from apparently similar sites are often strikingly different. The question arises as to whether this may be due to post-collection preservation or just an artefact of the extraction methods used in these different studies? In an attempt to resolve these questions, we examine the efficacy of DNA extraction methods, and the quality and quantity of DNA recovered from both artificially degraded, and genuinely ancient, but well preserved, bones. In doing so we offer hypotheses relevant to the DNA degradation process itself, and to where and how the DNA is actually preserved in ancient bone.

Introduction

The long-term survival of mineralised tissues such as bone (and to some degree, teeth) is normally dependent upon rapid burial in sediments, independently of whether terrestrial or marine. Subsequently, their chemical and physical properties undergo substantial change, in a manner determined by the environment. For example, in aerated soils fungi and bacteria colonise the pore spaces of bones and begin the breakdown of mineralised tissues within a few years (Bell et al., 1996, Jans et al., 2004). In contrast, cyanobacteria are principally responsible for initial microbial attack in freshwater and marine environments (Turner-Walker and Jans, 2008, Pesquero et al., 2010), which accelerates bone degradation by increasing its porosity (Nielsen-Marsh and Hedges, 1999). Despite an increasing body of knowledge about the degradation of the bone itself, much less is known about how the DNA in the bone degrades, and indeed, even how or where it is preserved. While some have argued that DNA in the mineralised collagen of bone and teeth hypothetically undergoes a retarded rate of decomposition because of its adsorption to hydroxyapatite (e.g. Collins et al., 1995, Hagelberg et al., 1989, Lindahl, 1993), and others have argued that the mummification of individual cells, and the physical exclusion of microbes and other external contaminants from the smallest pores of skeletal tissues may play a role in DNA survival (Hummel and Herrmann, 1994), we lack a comprehensive picture of the DNA–bone relationship.

The relationship is unlikely to be simple, as significant variation exists in the quality and quantity of DNA that has been recovered from old bone – even among samples collected from environments that appear to be similar. For example, DNA recovery success rate was high in several large-scale studies of permafrost-preserved samples, including bison (Bison sp., success rate 352/442 [Shapiro et al., 2004]) and musk ox (Ovibos moschatus, success rate 207/446 [Campos et al., 2010b]). However, similar studies yielded much lower success rates, for example 27/122 saiga antelope (Saiga tatarica [Campos et al., 2010a]). Given that most of the samples were collected from places with similar environments (permafrozen soils) and were thus presumably exposed to similar diagenetic conditions (Hedges, 2002, Nielsen-Marsh et al., 2007, Smith et al., 2007), the question arises as to what caused the discrepancy? One potential answer could be differences in the types of degradation undergone during the samples’ history, both taphonomic, and post-excavation (at least one study has argued that freshly excavated bones are better for ancient DNA analyses due to acceleration of degradation in museum storage resulting from elevated temperatures and greater access to oxygen [Pruvost et al., 2007]).

The degradation of DNA in archaeological bone is not a straightforward topic, as multiple chemical processes may act to both cross-link and fragment the molecule's chemical backbone, and nucleotide bases may be either removed or altered (e.g. Lindahl, 1993, Pääbo, 1989, Hansen et al., 2006). Regardless of the underlying chemical reasons, the end result of most of these processes is the same – the lengths of amplifiable DNA molecules decrease rapidly. Although several factors affect the rate of this decay, including environmental salt content, exposure to radiation, pH, and availability of oxygen and free water, it is temperature that is believed to play the key role in the longevity of aDNA molecules (Lindahl, 1993). In brief, an exponential relationship ensures that degradation rate rapidly increases with temperature (Lindahl and Nyberg, 1972). Thus for any given age, cold preserved samples are more likely to provide usable genetic material than those of a similar age that have been buried (or stored) at warmer temperatures (Smith et al., 2001, Smith et al., 2003).

While DNA degradation is obviously a key factor in determining whether aDNA can be recovered, an alternate explanation for variable success rates may be that the observed results do not reflect on the quality of the DNA per se, but where and how it is preserved in the bone, and the efficiency of the different extraction methods used. It is clear that a comprehensive understanding of bone composition and its diagenesis is crucial for determining the location of DNA in ancient bone, and hence for selecting appropriate samples for study and the extraction techniques to apply.

Macroscopically, bone is composed of two main architectures. At the jointed ends of long bones, and in flat sheet-like bones such as the sternum and skull vault, it comprises an outer layer of compact bone that surrounds a load-bearing network of intersecting planes and buttresses called trabeculae. These are termed cortical and trabecular bone respectively (the latter is also called cancellous or spongy bone). The mid-shafts of long bones are principally hollow tubes of cortical bone (Currey, 2002). Microscopically, bone consists of a hard, apparently homogeneous intercellular material, within or upon which can be found a number of characteristic cell types including osteoblasts and the osteoprogenitor cells that give rise to them; i.e. osteocytes, osteoclasts, and bone lining cells (that are essentially inactive osteoblasts) (Fig. 1). These cells cover all available bone surfaces, the exact type of cell depending upon the physiological status of the bone tissue; i.e. resorption, formation/mineralisation or quiescence (Ortner and Turner-Walker, 2003). Osteoblasts are mononucleate immature bone cells responsible for bone formation (Fig. 1). Located on the surface of osteoid seams, they secrete osteoid, a protein mixture that subsequently mineralises with a non-stoichiometric carbonated hydroxyapatite (HAP) to become the rigid, load-bearing solid that is bone mineral. Osteoblasts also produce hormones, such as prostaglandins or alkaline phosphatase, an enzyme that has a role in the mineralisation of bone (Ortner and Turner-Walker, 2003). Osteocytes are star shaped mature bone cells that originate when osteoblasts become trapped within the matrix they produce, occupying spaces in the bone known as lacunae (Fig. 1). Osteoclasts (Fig. 1) are large, multinucleated cells located on bone surfaces in what are called Howship's lacunae (or resorption pits), and are responsible for bone resorption – the process of removing bone tissue by dissolving its mineralised matrix and breaking up the osteoid (Nijweide et al., 1986, Ortner and Turner-Walker, 2003). Compact bone is permeated by an interconnected network of pores represented by the Haversian canals and canals of Volkman, which carry blood vessels and nerves, and canaliculi and which contain the cytoplasmic processes that connect adjacent osteocytes (Fig. 1).

Structurally, the majority of bone is composed of bone matrix. This consists of both an inorganic fraction composed of cryptocrystalline carbonated hydroxyapatite (to which DNA may adsorb [Lindahl, 1993]), and an organic fraction composed principally of Type I collagen as well as various non-collagenous proteins and glycoproteins, such as glycosaminoglycans, osteocalcin, osteonectin, ostepontin, bone sialoprotein and cell attachment factor (Tuross, 2003). A simplified view of the relationships between collagen and hydroxyapatite is given in Fig. 2. Tropocollagen molecules (∼300 nm in length and ∼1.5 nm in diameter) self aggregate extra-cellularly into fibrils with mean diameters of around 50 nm (Tzaphlidou and Berillis, 2005). The fibril is stabilised by post-translational modifications and cross-links between adjacent collagen molecules. These intermolecular bonds are such that there is an offset in the alignment among the collagen molecules so that there are gaps between the end of one molecule and the beginning of the next. The collagen molecules interdigitate in such a way that there are gap zones (where there is high density of gaps) and overlap zones where the molecules are well aligned and more closely packed. The 40 nm gap zone together and the 27 nm overlap zone are responsible for the 67 nm banded appearance of collagen fibrils when seen in TEM images. The initial mineralisation of collagen takes place in the gap zone and progresses along the fibrils, small crystallites developing both within and on the surfaces of fibrils. Full mineralisation is accomplished by the replacement of water between fibrils by mineral and the bulk of the mineral load is deposited here.

While some evidence has been published that demonstrates that DNA can adsorb to hydroxyapatite and influence crystal growth (e.g. Lindahl, 1993, Okazaki et al., 2001), the nature of the collagen/DNA interactions are more obscure and have been rarely studied. However, theoretical models (Mrevlishvili and Svintradze, 2005) and in vitro experiments (Kitamura et al., 1997) strongly suggest that nuclear DNA not only binds to collagen but can act as a scaffold or matrix in the aggregation of collagen molecules into fibrils (fibrillogenesis). There is little evidence, however, that large strands of DNA are incorporated into mineralised collagen since this would distort the regular structure of the fibrils; something that has not been observed (Orgel et al., 2001). On the other hand, the possibility of short fragments of either nuclear DNA or mtDNA becoming trapped in aggregating or mineralising fibrils cannot be excluded. Furthermore, it is possible that the gap zones, which are more disordered than the overlap regions (Orgel et al., 2005) may also be sites where smaller DNA fragments may become bound to collagen molecules. These may then be encapsulated within HAP crystallites as mineralisation proceeds.

It is quite feasible that, during bone resorption and formation, large amounts of mtDNA are released into the forming osteoid matrix following the apoptosis of osteoclasts or osteoblasts. Fragments of DNA would then be available to bind to the outer surfaces of collagen fibrils in the mineralising osteoid or to the surfaces of developing HAP crystallites. Of course in aDNA studies the picture is made even more complex by the potential release of tissue decomposition products, including DNA and collagen fragments released by chemical and/or microbial degradation of un-mineralised osteoid. The two proposed mechanisms for DNA preservation in bone, i.e. binding to: (a) mineral and (b) collagen have important implications for how DNA is most efficiently extracted, considering that most protocols involve the removal and discard of the mineral phase.

In this paper we synthesise new experimental evidence and previously published data, in order to present the current knowledge on DNA degradation in bone and the efficacy of various extraction methods in retrieving DNA from both artificially degraded and truly old bone. We offer hypotheses as to the DNA degradation process itself, and where and how the DNA is actually preserved long term in archaeological and fossil bones.