An empirical method for the estimation of nonspecific PCR primer derived genomic reactivity.

A total of 36 PCR primers were subjected to thermal cycling, in a reaction composed of a single primer only, in order to assess their individual tendencies toward "nonspecific" amplification of human genomic DNA background. Nonspecific amplification of genomic DNA was estimated by Southern hybridization of the amplification products with an Alu-specific DNA probe. The yield of amplified Alu-hybridizing material was estimated by scanning densitometry. In this way a quantitative estimate of "mispriming" was obtained for the primers tested. Human Alu and total primate GenBank databases were scanned in order to obtain an estimate of the prevalence of homologous primer binding sites for each primer. Database searches were conducted at both 70% primer binding site homology and 100% homology at the 3'-terminal third of the primer. The observed levels of amplified, Alu-hybridizing material correlated best, but only marginally, with homology-based estimates of potential cross-reactivity at the 3' terminus of the primer (R = 0.193). A set of p53 tumor suppressor gene primers, exhibiting low and high extremes of background amplification, were tested for sensitivity in the presence and absence of added genomic DNA. The primer that gave the highest level of background amplification was the one whose performance was most severely affected by added genomic DNA. Empirical assessment of the tendencies of individual primers to amplify irrelevant DNA at low levels of the intended target may permit a useful "noise-specific" adjunct to primer design by computational methods.