Abstract
This thesis explored the molecular basis for a novel type of allelic dropout (ADO) observed during genetic analysis of the maternally imprinted human gene, MEST (Stuffrein-Roberts 2008, Stevens et al. 2014a). Polymerase chain reaction (PCR) analysis of human genomic DNA for three single nucleotide polymorphisms (SNPs) in a region of MEST consistently gave rise to a non-Mendelian pattern where only one haplotype was detectable in each subject. The absence of observed heterozygotes appeared to be due to consistent ADO, with different alleles being lost from different subjects. This pattern of ADO was novel and warranted further examination.
Use of several different amplification and genotyping approaches indicated that the ADO occurred during PCR, not during post-amplification DNA sequencing, and that it arose from an inherent property of the genomic DNA. As an imprinted gene, the maternally inherited allele of MEST is heavily methylated. It was hypothesised that this allele was prone to ADO, with the paternal allele being consistently amplified. Several different techniques were applied to test this hypothesis. First, competitive PCR was performed on differentially methylated templates pairs, which mimicked the methylation status of genomic DNA. Second, PCR amplification was performed on genomic DNA samples, which were alternatively treated using methylation specific or methylation sensitive restriction endonucleases. Last, methylation specific PCR was performed on bisulfite treated DNA samples, enabling the amplification of paternal (non-methylated) and maternal (methylated) DNA in separate reactions. This app roach was also applied to parent-offspring trios. All data from these experimental approaches were consistent with loss of the methylated, maternal allele of the MEST promoter region during amplification under standard PCR conditions.
A role for secondary DNA structure was considered in the ADO phenomenon. Computational analyses predicted extensive potential for G-quadruplex (G4) formation across the entire amplicon, with three regions of high G4 propensity. The potential to adopt non B-DNA structure was reduced by use of a synthetic template in which predicted G4 motifs were removed by mutation, or by incorporating 7-deaza dGTP into the DNA template. These experiments demonstrated that the MEST ADO occurred due to a combination of Hoogsteen bond formation and cytosine methylation in the genomic template DNA.
Using circular dichroism spectroscopy (CD) and polyacrylamide gel electrophoresis the structural potential of the three putative G4 motifs was confirmed. G4 formation was favoured by the monovalent cation, K+ and cytosine methylation prevented G4 formation in the absence of MgCl2. Cytosine methylation also decreased the thermodynamic stability of G4, however, the re-association rates for methylated G4 were significantly increased. This effect was enhanced by MgCl2 and PCR buffer, suggesting the hypothesis that during PCR G4 structures form in genomic DNA and are maintained on the maternally imprinted, methylated allele. It was also observed that K+ concentration in the PCR buffer directly influenced genotyping outcome, reinforcing the potential for a link between G4 formation and ADO.
Two novel fluorescent techniques were developed, based on traditional radioisotopic methods, which enabled the precise mapping of G4 structure on both strands of dsDNA and genomic DNA templates. These techniques relied on either enzymatic or chemical cleavage, followed by precise sizing of fluorescent fragments on a capillary sequencer, to identify G4 footprints in DNA samples. Application of these techniques revealed extensive G4 formation across the MEST promoter region, confirming the existence of these structures in genomic PCR templates. Methylation did not appear to influence G4 topology or propensity to form under the investigated conditions. It was also demonstrated that non-structured, dsDNA templates could transition to adopt G4 topology, without strand denaturation. This indicated that Hoogsteen bond formation at the MEST G4 structures could out-compete Watson-Crick basepairs, and potentially form in vivo.
A third novel fluorescent technique was developed for the simultaneous investigation of polymerase arrest on both DNA strands of a double-stranded template. This technique used fluorescently labelled primers during PCR to monitor the extension by polymerase across both DNA strands. Fluorescent extension products were sized by capillary electrophoresis and mapped to the template sequence to identify the corresponding position of termination. This demonstrated that G4 motifs inhibit the action of Taq polymerase during PCR, regardless of methylation status. However, this effect was most pronounced with methylated templates, causing lags in amplification for multiple PCR cycles, in contrast with non-methylated templates. This was sufficient to result in a large amplification bias relative to the non-methylated template, accounting for the observed ADO of the maternal allele.
It was concluded that the combination of G4 formation and DNA methylation are both required for allelic dropout during PCR amplification of the MEST promoter region, and neither factor in isolation can explain the observed phenomenon. This research has clarified the molecular basis of this unusual “parent-of-origin” specific ADO, and led to development of several novel methods for the investigation of G4, which can be more widely applied to the analysis of non B-DNA structure. The results have significance for the accuracy of PCR tests in clinical diagnosis and more generally for the analysis of genomic DNA using PCR methods.