Reproducible DNA fingerprinting with the random amplified polymorphic DNA (RAPD) method.

Much interest has recently arisen in methods for DNA fingerprinting based on the polymerase chain reaction (PCR). Among these, the Random Amplified Polymorphic DNA (RAPD) method, developed by Williams etal. (1), is currently receiving particular attention (2) because of its extreme simplicity and requirement for minimal amounts of genomic DNA. The basic strategy involves the PCR amplification of random fragments of genomic DNA with single or multiple (3) primers of arbitrary sequence. Polymorphism between individuals (or strains) is detected as differences between the pattern of DNA fragments amplified from the different DNAs using a given primer(s). Although RAPD PCR is an extremely powerful tool for such tasks as gene mapping, population and pedigree analysis, phylogenetic studies and bacterial strain identification, the reproducibility of RAPD fingerprints can be quite problematic (4-6). Since the pattern of fragments amplified is, in large part, a function of the sites on the template to which productive annealing of the oligonucleotide primer can occur, differences between DNA preparations that affect primer annealing could be one major source of irreproducibility of RAPD patterns. Here we show directly that ethanol precipitable contaminants in DNA are a major cause of irreproducibility. In preparing DNA for RAPD analysis using standard extraction techniques we found a perfect correlation between the procedure used to collect ethanol precipitated DNA and the reproducibility of the RAPD pattern. We always obtained highly reproducible patterns from DNAs that were wound on a glass rod, while patterns from DNAs collected by centrifugation displayed different degrees of variability. Since centrifugation of an ethanol precipitate of a previously wound DNA sample did not introduce variability, we concluded that winding the DNA must free it of the material that causes the variability exhibited by the centrifuged samples. The results in the figure demonstrate that the RAPD pattern of centrifuged DNA differs from that of wound DNA (lanes Al and Bl, respectively), that the pattern produced by wound DNA can be produced from centrifuged DNA if the latter is reprecipitated from ethanol but collected by winding on a glass rod (lanes A2), and that an altered pattern is obtained using wound DNA to which its supernatant material has been added back (lanes B2). Although RNase A digestion modified the RAPD pattern produced from centrifuged DNA, this treatment did not result in the same RAPD pattern produced with wound DNA (results not shown). Therefore, contaminating RNA may only be partly responsible for the variability observed with centrifuged samples. On the other hand, DNase treatment modified the RAPD pattern produced from wound DNA (results not shown), suggesting that either the presence of very short DNA fragments or shortened templates, or both, may lead to RAPD variability. These findings support the hypothesis that the ethanol precipitable contaminants include very low molecular weight DNA and/or RNA which, in some way, alter the formation of productive template/primer complexes. Whatever their nature, our results unequivocally show that ethanol precipitable contaminants in DNA extracted by standard techniques are a major, if not sole, source of variability in RAPD