To increase the yield of DNA sequence generated by the 454 technology from small amounts of starting DNA, we investigated the efficiency of each step in the 454 library preparation process. We find that the last step, when the single-stranded library is released by NaOH, is inefficient and highly variable. When this step is replaced with heat treatment, library amounts dramatically increase. Furthermore, when sequencing templates are first isolated by NaOH treatment and subsequently by heat treatment, the sequences of both strands of individual template DNA molecules can be determined. Using this approach, we confirm that C/G base pairs observed as T/A base pairs in Neanderthal DNA sequences are due to a modification of the cytosine rather than guanine residues.
Sequencing by synthesis on the 454 platform is one of several novel sequencing techniques that allows the determination of millions of base pairs per run using a single instrument (1). Although in most applications of this technology the amount of template DNA is not limiting, a number of applications start from small quantities of DNA or RNA. For example, when bacteria that cannot be cultured (2) or when cDNA libraries from a small number of cells (3) are sequenced, template DNA amounts limit the number of DNA sequences that can be determined. The same is true for many palaeontological samples, such as Neanderthal remains (4), where the amount and quality of available bone limits the amount of DNA sequence that can be generated. Furthermore, although some sequencing methods use both strands of template molecules as starting material, others use just one strand under the assumption that the two strands are complementary to each other. This may not always be the case. For example, when a molecule is hemimethylated (5,6), when bases are modified as is often the case with ancient DNA (7), or in other circumstances, DNA strands may not be complementary to each other.
To increase DNA sequence recovery from small amounts of biological or paleontological material, we studied the template recovery at each step in the 454 library production protocol. We find that although the efficiency of recovery in most steps varies between 40% and 96%, the final step—where the single-stranded library is retrieved—is highly variable and can entail losses of over 99%. Replacing this step with more efficient means of library recovery increases the library yield 5-to 200-fold. In addition, by combining two methods of retrieval of 454 library molecules, we show that one can retrieve both strands of the same ancient DNA molecules.Materials and Methods Radioactive labeling of a PCR product
A PCR product 103 bp in length (Supplementary Figure 1, available at www.BioTechniques.com) was used as a template in two PCRs where in one only non-radioactive dATP (2.4 µM) was added, while in the other dATP (2.0 µM) and [α-32P]dATP (0.4 µM, 3000 Ci/mmol; Hartmann Analytic, Braunschweig, Germany) were used essentially as described in Reference 8,. A final reaction volume of 50 µL contained MgCl2 (5 mM; Applied Biosystems, Foster City, CA, USA), Gold buffer (1×; Applied Biosystems), BSA (0.8 mg/mL; Sigma Aldrich, St. Louis, MO, USA), L164 primer (1 µM; Metabion, Martinsried, Germany), H221 primer (1 µM; Metabion), AmpliTaq Gold DNA Polymerase (2.5 U; Applied Biosystems), dTTP (200 µM; New England Biolabs, Ipswitch, MA, USA), dCTP (200 µM; New England Biolabs), and dGTP (200 µM; New England Biolabs). After an initial incubation at 94°C for 9 min, 35 cycles at 94°C for 20 s, 53°C for 30 s, and 72°C for 30 s were performed. Both PCR products were purified using the MinElute PCR purification kit (Qiagen, Hilden, Germany); the non-radioactive PCR product was visualized in an agarose gel (Supplementary Figure 2, available at www.BioTechniques.com), and its concentration determined to be 14.8 ng/µL by absorbance (NanoDrop ND-1000 spectrometer, Thermo Scientific, Wilmington, DE, USA). This concentration was used as a proxy measure for the concentration of the radioactive PCR product.
454 library optimization
51.8 nanograms of the 103-bp radioactively labeled PCR product were used as the template for the parallel production of six 454 libraries using the 454 Life Sciences GS20 protocol (Branford, CT, USA) (see Supplementary Table 1, available at www.BioTechniques.com) but the initial nebulization step was omitted. In the final step—isolation of the single-stranded DNA library—three libraries were treated twice with 0.125 M NaOH according to the manufacturer's protocol and three were eluted in 50 µL of water for 2 min at 90°C, and then transferred to ice. Aliquots were collected from each fraction in each step, and then transferred to 20-mL scintillation vials (Carl Roth, Karlsruhe, Germany) containing 10 mL water. Their radiation content was measured (for more information on Cerenekov counting method, see las.perkinelmer.com/Content/ApplicationNotes/APP_TriCarbCerenkovCntingP32.pdf) with a scintillation counter (Packard Tri-Carb, Packard BioScience, Meriden, CT, USA) for 1 min. From the Cerenkov radiation of each aliquot, the total radiation for each fraction was calculated (for details see Supplementary Tables 2–4, available at www.BioTechniques.com).