Enhanced genomic sequencing and in vivo footprinting

We have developed a simplified procedure for the ligation-mediated polymerase chain reaction (LMPCR) using Thermococcus litoralis DNA polymerase (Vent DNA polymerase). We show that Vent DNA polymerase produces correct, blunt-ended primer extension products with substantially higher efficiency than Thermus aquaticus (Taq) DNA polymerase or modified T7 DNA polymerase (Sequenase). This difference leads to significantly improved genomic sequencing, methylation analysis, and in vivo footprinting with LMPCR. These improvements include representation of all bands with more uniform intensity, clear visualization of previously difficult regions of sequence, and reduction in the occurrence of spurious bands. It also simplifies the use of DNase I cut DNA for LMPCR footprinting. Footprinting experiments are commonly and productively used to study protein-DNA interactions and DNA configuration in vitro. Analogous in vivo experiments done on genes in the living cell can bring a different and useful data set to the problem ofgene expression, but they require special methods for visualizing the result. Direct genomic sequencing techniques, which permit the examination of single-copy genes in large genomes, are being used increasingly for this purpose (1-4). Ligation-mediated PCR (LMPCR) is a recently introduced method that substantially increases the absolute signal and the signal-to-noise ratio obtained for genomic sequencing (2, 5, 6). It does so by coupling PCR with genomic sequencing to provide specific amplification of a sequence "ladder," while preserving the identity and relative quantitative representation of each rung in the original cleaved genomic DNA preparation. Its application has made in vivo footprinting (2) and chromosomal methylation analysis (6) more readily accessible for organisms with large genomes (e.g., mammals). While LMPCR has been used successfully by a number of investigators to obtain high quality in vivo footprint and methylation information (2, 6, 7, 8), it has had two problems that can significantly compromise data quality. These effects are minor in some regions of sequence but can be problematic in others. First, certain bands are consistently weak or missing in the genomic ladders. Second, "extra" bands occasionally appear in the genomic ladders. These bands, which aren't predicted from the sequence as independently determined from cloned DNA, are usually adjacent to expected bands and therefore convert some triplets into quartets, some doublets into triplets, and so on. We present here a solution for these problems that also permits simplification of the LMPCR procedure. These improvements stem from the use of Thermococcus litoralis DNA polymerase (Vent polymerase). This thermostable polymerase possesses no detectable terminal deoxynucleotidyltransferase activity under our conditions, and this characteristic dramatically improves LMPCR genomic sequencing. For in vivo footprinting and genomic sequencing applications, Vent polymerase yields substantially superior results, improving overall signal and, most importantly, the quality of sequence in difficult regions. We also show that starting material possessing 3'-hydroxyl ends (in this case DNase I-cut DNA), which had required modification of template ends with dideoxynucleotides in the older form of LMPCR (9), can now be used for in vivo footprinting purposes without modification. MATERIALS AND METHODS Cell Culture and DNA Preparation. L cells were grown in Dulbecco's modified Eagle's medium with 10% undialyzed calf serum (Irvine) and 2 mM glutamine. Naked and in vivo dimethyl sulfate (DMS)-treated MM14 DNA was provided by P. Mueller. DNA samples for genomic sequencing and DMS footprinting were prepared as in refs. 10 and 11. In vivo DNase I treatment was as in ref. 12, except that cells were permeabilized on ice with lysolecithin (0.25 mg/ml) for 60 sec. Addition of dideoxynucleotides prior to LMPCR where noted was as in ref. 9. LMPCR. LMPCR using T7 DNA polymerase (Sequenase version 1.0; United States Biochemical) and Thermus aquaticus (Taq) DNA polymerase (AmpliTaq; Cetus) was done as in refs. 2 and 5. LMPCR using Thermococcus litoralis DNA polymerase (Vent; New England Biolabs) was done as below. All solutions were chilled and manipulations were performed on ice except as noted. The pH values given are for room temperature. To 5 ,l (2 ,ug) ofDNA in TE (10 mM Tris-HCl, pH 7.5/1 mM EDTA) was added 25 ,l of first-strand mix [1.2x first-strand buffer (48 mM NaCI/12 mM Tris HCl, pH 8.9/6 mM MgSO4/0.012% gelatin) with 0.3 pmol of genespecific primer 1, 240 AM each dNTP, and 1 unit of Vent polymerase]. First-strand synthesis used a thermal cycle of 5 min at 95°C, 30 min at 60°C, and 10 min at 76°C. The samples were immediately iced. (It is important to minimize Vent polymerase activity during the ligation step by keeping the sample cold.) Twenty microliters of dilution solution (110 mM Tris HCI, pH 7.5/18 mM MgCl2/50 mM dithiothreitol/ 0.0125% bovine serum albumin) and 25 ,ul of ligation solution [10 mM MgCl2/20 mM dithiothreitol/3 mM ATP/0.005% bovine serum albumin with 100 pmol of unidirectional linker in 250 mM TrisHCl (pH 7.7) (thawed and added on ice) and 4.5 units of T4 DNA ligase (Promega)] were added. After incubation for 12-16 hr at 17°C, samples were iced and 9.4 IlI of precipitation solution (0.1% yeast tRNA/2.7 M sodium acetate, pH 7.0) and 220 ,ul of ethanol were added. The samples were placed at -20°C for .2 hr and then spun for 15 min at 4°C in a microcentrifuge. The pellets were washed with 75% ethanol and dried in a Speed-Vac rotary evaporator (Savant). Samples were resuspended in 70 ,ul ofwater at room Abbreviations: LMPCR, ligation-mediated polymerase chain reaction; DMS, dimethyl sulfate; MCK, muscle creatine kinase. *To whom reprint requests should be addressed. 1021 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 1022 Biochemistry: Garrity and Wold temperature and placed on ice. After addition of 30 p.l of amplification mix [3.33 x amplification buffer (133 mM NaCI/67 mM Tris, pH 8.9/17 mM MgSO4/0.03% gelatin/ 0.3% Triton X-100) with 670 p.M each dNTP, 10 pmol of gene-specific primer 2, and 10 pmol of linker primer LMPCR.1)] and 3 ILI (3 units) of Vent polymerase, samples were overlaid with 90 A.1 mineral oil and subjected to PCR using 18 cycles of 1 min at 950C, 2 min at 660C, and 3 min at 760C, with these modifications: (i) first-round denaturation was 3.5 min at 950C; (ii) 5 sec was added to the 760C step with each successive cycle (e.g., second round, 3 min 5 sec at 76°C); (iii) for cycle 18, the 76°C step was 10 min. Samples were then placed on ice and 5 ,ul of labeling mix [lx amplification buffer with 2 mM each dNTP, 2.3 pmol of gene-specific primer 3 (end-labeled as in refs. 2 and 5), and 1 unit of Vent polymerase] was added. The labeling cycle was 3.5 min at 95°C, 2 min at 69°C, 10 min at 76°C, 1 min at 95°C, 2 min at 69°C, and 10 min at 76°C. The reaction was stopped by placing the samples on ice and adding 300 p.l of stop solution (10 mM Tris HCI, pH 7.5/4 mM EDTA/260 mM sodium acetate, pH 7.0, containing tRNA at 67 pg/ml). Samples were shifted to room temperature and extracted with 400 ,.l of phenol/chloroform/isoamyl alcohol (25:24:1, vol/ vol). The aqueous phase was split into four aliquots of 94 ,ul, and 235 ,ul of ethanol was added to each. Before electrophoresis, samples were precipitated, washed, and dried as above. After resuspension in 7 ,ul of load dye (2, 5) and heating at 85°C for 5 min, samples were iced and then loaded on 6% polyacrylamide sequencing gel (2, 5). Loading one-fourth of an LMPCR mixture per lane yielded a strong signal on X-AR film (Kodak) after 3 hr with an intensifying screen at -80°C or 12 hr with no screen at -20°C. The unidirectional linker, linker primer, and muscle creatine kinase (MCK) oligonucleotides were as in ref. 2. The sequences (5' to 3') of the metallothionein I oligonucleotides were GAGTTCTCGTAAACTCCAGAGCAGC (primer 1), CAGAGCAGCGATAGGCCGTAATATC (primer 2), and AGCGATAGGCCGTAATATCGGGGAAAGC (primer 3). RESULTS AND DISCUSSION LMPCR (Fig. 1) relies on creation of a blunt end in the initial primer extension reaction to serve as a ligation substrate. Later, in the labeling reaction, precise blunt-end termination of the extension product is required. If the final labeling extensions stop short or add extra nontemplated bases, the result will be extraneous, inappropriate bands. In general, imperfect extension products may result from DNA polymerases adding a nontemplated additional base after creating a blunt end (referred to as terminal transferase activity) (13). Both polymerases commonly used for LMPCR display some terminal transferase activity. Sequenase, used in the firststrand synthesis reaction, adds an extra base to 50%o of its products. Taq, used in the PCR amplification and labeling steps, adds an extra base to -95% of its products (P. Mueller and B.J.W., unpublished data). Such activity during the first-strand synthesis creates molecules unable to participate in the blunt-end ligation. Should this activity show sequence preference, it would lead to underrepresentation or even complete loss of specific bands in the final LMPCR product. Terminal transferase activity might also explain the origin of spurious "extra" bands in an LMPCR ladder. For example, a single band in a genomic sequencing ladder would appear as a doublet ifsome products ofthe labeling reaction acquired the extra base. We hypothesized that the terminal transferase activities of Sequenase and Taq were the major source of imperfect regions in LMPCR ladders and sought a DNA polymerase that lacks appreciable terminal transferase activity. primer 1