Recharacterization of ancient DNA miscoding lesions: insights in the era of sequencing-by-synthesis

Abstract

Although ancient DNA (aDNA) miscoding lesions have been studied since the earliest days of the field, their nature remains a source of debate. A variety of conflicting hypotheses exist about which miscoding lesions constitute true aDNA damage as opposed to PCR polymerase amplification error. Furthermore, considerable disagreement and speculation exists on which specific damage events underlie observed miscoding lesions. The root of the problem is that it has previously been difficult to assemble sufficient data to test the hypotheses, and near-impossible to accurately determine the specific strand of origin of observed damage events. With the advent of emulsion-based clonal amplification (emPCR) and the sequencing-by-synthesis technology this has changed. In this paper we demonstrate how data produced on the Roche GS20 genome sequencer can determine miscoding lesion strands of origin, and subsequently be interpreted to enable characterization of the aDNA damage behind the observed phenotypes. Through comparative analyses on 390 965 bp of modern chloroplast and 131 474 bp of ancient woolly mammoth GS20 sequence data we conclusively demonstrate that in this sample at least, a permafrost preserved specimen, Type 2 (cytosine→thymine/guanine→adenine) miscoding lesions represent the overwhelming majority of damage-derived miscoding lesions. Additionally, we show that an as yet unidentified guanine→adenine analogue modification, not the conventionally argued cytosine→uracil deamination, underpins a significant proportion of Type 2 damage. How widespread these implications are for aDNA will become apparent as future studies analyse data recovered from a wider range of substrates.

The mitochondrial genome sequence of the Tasmanian tiger (Thylacinus cynocephalus)

“…We identified a substantial amount of human DNA contamination in the thylacine samples, unlike in our earlier analysis of woolly mammoth bone (Poinar et al 2006;Gilbert et al 2007b) and hair shafts (Gilbert et al 2007a(Gilbert et al , 2008Miller et al 2008). For example, from the ethanol-preserved specimen (thylacine 2 in Table 1), we identified 44,493 reads (4.3%) that originated from the human nuclear genome and 136 reads from the human mitochondrial genome.…”

“…This is most likely explained by either the probable higher thermal age (Smith et al 2003;Gilbert et al 2007a) of the thylacine samples in comparison to the mammoth sample, the effect of the preservation/storage process (tanning process or longterm storage in dilute ethanol), or a combination thereof (see Supplemental Material). On the other hand, the damage rate is less than that for a mammoth bone sample (Gilbert et al 2007b), confirming the utility of hair shafts as an excellent reservoir for ancient DNA.…”

Patterns of damage in genomic DNA sequences from a Neandertal

“…When this is done across large numbers of 454 sequence reads, the number of substitutions where any single nucleotide (e.g., C) changes to another particular nucleotide (e.g., T) should be equal to the number of substitutions where the complementary nucleotide (i.e., G) changes to the complementary nucleotide (i.e., A), unless nucleotide misincorporations occur (9). When such strandequivalent reciprocal nucleotide substitutions are analyzed in DNA sequences from Pleistocene organisms, C to T changes are more frequent than G to A changes (9,10,18). Furthermore, in contrast to DNA sequences determined from modern DNA, the rates of both G to A changes and C to T changes are elevated above the rates of the other two transitions.…”

True single-molecule DNA sequencing of a pleistocene horse bone

“…As a result, damage-free modern DNA molecules can easily outcompete homologous ancient fragments during PCR, making aDNA studies highly prone to contamination (Paabo et al 2004;Gilbert et al 2005;Willerslev and Cooper 2005). Additionally, nucleotides are misincorporated by DNA polymerases while amplifying damaged templates, particularly at sites where cytosine has been deaminated to uracil, as the latter is the chemical analog of thymine, resulting in artefactual G/C to A/T mutations (so-called type II damage) Hoss et al 1996;Hansen et al 2001;Hofreiter et al 2001;Gilbert et al 2003Gilbert et al , 2007aBinladen et al 2006;Stiller et al 2006). Cytosine deamination is therefore a common feature of all aDNA templates, with misincorporation rates that can exceed real biological mutation rates and generate spurious sequence results (Ho et al 2007).…”