Error analysis of chemically synthesized polynucleotides

Two single-stranded polynucleotide constructs, 123 and 126 nucleotides in length, were chemically synthesized using standard phosphoramidite chemistry. Clonable, double-stranded DNA fragments about 100-bp long were prepared from the polynucleotides by primer extension with a DNA polymerase and end-trimming with two restriction endonucleases, then the fragments were ligated into separate plasmids. Errors in individual insert copies were determined by dideoxy sequencing after in vivo amplification of plasmids. Five of the ten inserts sequenced contained errors, including seven single-base-pair deletions, one four-base-pair deletion and one G-->C transversion. The origins of the latter two errors are unclear, but single-base deletions are inconsistent with errors of polymerases; thus, the most common sequence errors of chemical synthesis are deletion mutations. Deletions are most likely to result from incomplete capping or de-tritylation. The observed error rate can became a significant limiting factor in applications that depend on the correctness of a polynucleotide sequence in individual insert clones.