Efficient Cloning of SAGE Tags by Blunt-End Ligation of Polished Concatemers

The serial analysis of gene expression (SAGE) has become an important technique for high-throughput transcriptome analysis. Since its first description in 1995 (1), SAGE has provided both quantitative and qualitative information concerning global gene expression in a variety of biological systems. Thereby, it has deepened our knowledge concerning developmental processes as well as pathogenic mechanisms (2–4). Furthermore, SAGE was shown to be a useful tool for genome annotation (5). The common application of SAGE, however, was delayed and is still hampered by a technically demanding protocol. Recently, improvements have been made toward a more widespread use of this method (6–9). Several of the methodical advances concerned the cloning of the concatemers, as this was shown to be a crucial yet inefficient step. One protocol described the removal of biotinylated linkers by streptavidin-coated magnetic beads before cloning (10). The introduction of a heating step before separation of the concatemers by gel electrophoresis resulted in a better correlation between the migration distance in the gel and the concatemer length (11). This allowed the cloning of concatemers with, on average, significantly greater length. None of these protocols, however, has addressed the major problem underlying inefficient cloning of any given concatemer, which is the problem of “corrupted concatemer termini”. Corrupted concatemer termini do not contain the nucleotide termini expected from normal endonuclease cleavage; consequently, concatenation has an inherent tendency to stop at these termini, as they prevent further elongation. Corrupted termini can arise from exonuclease activity present during/after restriction endonuclease cleavage, or from star activity of the restriction endonuclease used. These corrupted termini result in poor cloning efficiency of the concatemers, as these are subsequently cloned into a vector cleaved with a restriction endonuclease that yields cohesive ends compatible only with intact termini. One approach to overcome this problem involves polishing the concatemer termini, thus allowing bluntend ligation. Although blunt-end ligation is generally less efficient than cohesive end ligation, it can result in a higher efficiency under conditions of enriched corrupted cohesive termini. To test this hypothesis, we used the exonuclease activity of bacteriophage T4 DNA polymerase to convert the 3′-protruding concatemer termini to bluntend DNA, as the commonly used fill-in reaction by Klenow is prohibited by the presence of protruding 3′ ends (12). Concatemers were purified from biotinylated linkers by streptavidin-coated magnetic beads (Dynal, Hamburg, Germany) (10), kept at 65°C for 5 min to break concatemer aggregates (11) and separated in a 1% agarose gel (universal agarose; Peqlab, Erlangen, Germany). Fractions of 300–500 bp (fraction I) and 500–1000 bp (fraction II) in length were cut out and eluted from the gel matrix with the E.Z.N.A. Gel Extraction Kit (Peqlab). To compare blunt-end ligation with the established protocol of cloning concatemers into the SphI restriction site, each fraction was divided into two aliquots of 15 μL. To one aliquot, 1.2 U T4 DNA polymerase (New England Biolabs, Frankfurt, Germany) were added in the presence of 100 μM each dNTP, and the reaction was run for 15 min at 12°C in 1× reaction buffer in a total volume of 20 μL. The reaction was stopped by heat-denaturing the enzyme at 75°C for 20 min. This yielded a polished concatemer sample ready for blunt-end ligation. The other aliquot was kept at 4°C during this procedure (unpolished concatemer sample). To clone concatemers, a modified pBluescript® vector (Stratagene, Amsterdam, The Netherlands) was used that contained an additional SphI recognition site between the BamHI and the EcoRI recognition site. Five micrograms of vector were cleaved either with 50 U SphI (New England Biolabs) or with 50 U Eco32I (MBI Fermentas, St. Leon-Rot, Germany) for 90 min at 37°C in a total volume of 100 μL. Eco32I represents an isoschizomer of EcoRV and was recently reported to improve blunt-end cloning by a decreased intrinsic single nucleotide deletion activity when compared to EcoRV (13). This should allow for more efficient blue/white selection. After cleavage by restriction endonuclease, vectors were dephosphorylated using 10 U calf intestinal phosphatase (New England Biolabs), purified by agarose gel electrophoresis, and extracted using the same gel extraction kit as above (Peqlab). The entire unpolished samples were ligated into 100 ng SphI-cleaved pBluescript; the entire T4 DNA polymerase-polished samples were directly ligated after the heat denaturation step into 100 ng Eco32I-cleaved pBluescript by adding 15 U T4 DNA ligase (MBI Fermentas), 5 μL 10 × ligase buffer and water to a total volume of 50 μL. Ligations were carried out at 15°C for 19 h. Thereafter, T4 DNA ligase was removed by StrataClean resin (Stratagene), and the ligation products were precipitated with ethanol in the presence of 70 μg glycogen (Peqlab) and resuspended in 10 μL water. Then, 1.5 μL each ligation product was added to 50 μL ElectroTen-Blue electroporation competent cells (≥ 1 × 1010 transformants/μg; Stratagene). After electroporation at 1.8 kV (E. coli pulser; Bio-Rad Laboratories GmbH, München, Germany) in 0.2-cm-wide cuvettes (Peqlab), transformation mixtures were spread on 145-mm-wide LB agar plates containing 50 μg/mL ampicillin (Sigma, Taufkirchen, Germany) and the X-gal/IPTG system (Peqlab) for blue/white selection. Transformed cells were counted the next day. A substantial increase of transformants was observed for T4 DNA polymerase-treated concatemers when compared to the corresponding SphI ligation (Table 1). A 3.4-fold increase for fraction I and a 2fold increase for fraction II were observed. On all plates, less than 1% blue colonies were detected. To check for insert length, the inserts of 40 randomly picked white clones from Eco32I ligations were amplified by colony-PCR and separated in a 1% agarose gel. Thirty-seven of 40 selected clones (92.5%) from fraction I and 36 of 40 selected clones (90%) from fracBenchmarks