Abstract
The forensic community currently relies upon commercially produced polymerase chain (PCR) based "kits" for developing human short tandem repeat (STR) DNA profiles from biological samples. Although robust and optimized for routine forensic casework, these kits have a limit of detection of approximately 250 pg (Collins et al., 2004), which prohibits the development of forensic DNA profiles from trace biological samples containing less than about 50 diploid cells. Such samples are termed Low Copy Number, or LCN, and previous studies aimed at manipulating the commercial kit protocols by increasing the number of PCR cycles (Gill et al., 2000) or by ratio-constant reduction of the PCR reaction volume (Gaines et al., 2002) have successfully increased the commercial kits' sensitivities by generating full DNA profiles from LCN samples. However, both strategies have problems and limitations. Increased PCR cycling introduces significant deleterious stochastic events during PCR that interfere with detecting and interpreting LCN profiles. Performing a series of duplicate PCR reactions to develop a "consensus profile" can compensate for stochastic events (sine they are random and therefore reaction-specific), but LCN samples (by definition) contain far too little DNA for multiple PCR amplifications. Ratio-constant reduction of the PCR volume strategy, known as reduced volume PCR (RV-PCR), is largely free of stochastic events and therefore does not require replicate amplifications, but its sensitivity is limited to 130 pg. or about 20 diploid cells (Gaines et al., 2002). Unfortunately, therefore, contact DNA and other LCN samples that contain the DNA from fewer than 20 diploid cells are still beyond the reach of PCR-based kits, even using the modified protocols designed to amplify LCN DNA discussed above. My research focused upon increasing the sensitivity of the RV-PCR method by evaluating the effectiveness of three additional strategies. First, I performed my RV-PCR experiments using the most recently introduced STR kit, IdentifilerTM, which had not been evaluated in prior studies and is widely reported as being more sensitive than previous PCR-based kits. Second, I tested a lower RV-PCR reaction volume (2.5gtL) than had been previously studied, hoping that increasing the concentration of DNA compared to the other reaction regents (to an even greater extent than before) would result in more effective amplification. Third, I attempted to counterbalance the inhibitory effects of ethyldiaminetetraacedic acid (EDTA) previously reported by Gaines to interfere with RV-PCR of LCN samples (Gaines et al., 2002). To do this, I added equal molar amounts of magnesium chloride, which chelates EDTA in a 1:1 molar ratio. Finally, after testing these three strategies and determining the optimal combination of reaction conditions for generating LCN profiles, I evaluated how well my new method could resolve LCN two-component mixtures and develop LCN DNA profiles from mock forensic casework samples. My results show that it is possible to generate useful (greater than 70% profile detection) and reliable DNA profiles from 30 pg of template DNA (about 5 diploid cells) using the IdentifilerTm kit and a 2.5 1iL RV-PCR reaction volume, without running multiple replicates of the PCR reaction. However, I found that the addition of equal molar magnesium chloride proved to be an ineffective counter measure to EDTA-induced PCR inhibition, demonstrating only minimal improvement in PCR performance for extracts containing less than 0.50 mM EDTA, and no benefit for those containing greater than 0.50 mM EDTA. In the mixture studies, I found that my method could detect the minor profile in a LCN two-component mixture with a limit of detection of 20:1. However, the method does not meet the current validation standards for two-component mixture analysis due to an unacceptably high level of heterozygote peak height variance, leading to the incorrect assignment of both donor DNA profiles for loci containing less than four alleles. In addition, my method was ineffective when applied to mock forensic LCN casework samples. Presumably, this is because my initial experiments were performed with dilutions of DNA extracts of known concentrations, while the casework samples contained intact cells that needed to be removed, lysed, and subjected to DNA extraction procedures prior to analysis, resulting in the loss of some or most of the LCN DNA. In conclusion, I have demonstrated that LCN samples with as little as 30 pg of DNA can be successfully and reproducibly profiled using the IdentifilerT it and an RV-PCR reaction volume of 2.5 gL. However, caution should be used when applying this technique to LCN mixtures, due to unacceptable levels of heterozygote peak height imbalances that interfere with genotype interpretation. In addition, although DNA extracts containing the DNA from as few as 5 diploid cells can be profiled using my new technique, forensic casework samples analyzed with my method will need to contain many more than 5 diploid cells for successful analysis, until the inefficiency of DNA extraction methods from LCN samples is improved. In the meantime, the application of my technique is likely to increase the number of LCN samples that can be successfully analyzed in forensic casework and thus provides a useful new tool for forensic DNA analysis.