Cancer is caused by mutations, which can be found only in cancerous cells. To detect cancer early, we aimed to develop a method to screen human liquid or waste such as blood or feces for mutant DNA shed from cancer cells, to allow pre-symptomatic cancer screening and thus surgical or other treatment with a much higher cure rate than is now possible.

The method entailed extracting DNA of target genes from human fluids or feces, heating and cooling to re-form DNA duplexes, and enriching for heteroduplexes—DNA sequences that are mismatched due to a mutation—to allow detection of mutations by sequencing. Immobilized mismatch binding protein can perform this enrichment by binding DNA heteroduplexes while allowing other DNA to be washed away. To test the method, short, single strands of synthetic DNA, with or without mutations, were combined to form homo- or heteroduplexes that were then mixed at low ratios of hetero- to homoduplexes and enriched for heteroduplexes. A single base deletion was enriched by a factor of 29, and a mismatch was enriched by a factor of 2.

Next, we tested the method on PCR products, which are longer, double strands of DNA that are amplified from purified, human DNA. However, no enrichment was achieved. One possible cause of failure is the presence of artifactual mismatches in the PCR primers, which are short strands of synthetic DNA used to initiate amplification in PCR. For a substantial minority or even a majority of individual primer molecules, at least one synthetic DNA error may occur somewhere within each molecule. These errors act like mutations, forming mismatches that cause the duplex to be enriched. These erroneously-enriched duplexes dilute the enrichment of mutations within the targeted region, perhaps making them undetectable. To avoid this potential problem, we designed PCR primers with a special DNA sequence that attracts a DNA-cutting enzyme (Mme I) to cut off the entire primer region of the PCR product after amplification has been completed. Another possible source of artifactual mismatches is inaccurate amplification in PCR; we reduced this potential problem by using a high-fidelity PCR enzyme (Pfx) for amplification. However, both of these measures failed to lead to enrichment of PCR products.

Perhaps some component of the PCR reaction inhibits enrichment. To examine this, we performed enrichment with synthetic DNA to which was added either PCR reaction chemicals or only water. Enrichment occurred in both conditions, thus the PCR chemicals were not inhibiting enrichment.

Since we could not identify the reason PCR inhibited enrichment, we have started to pursue an alternate approach which avoids PCR before enrichment, instead using hybridization to extract target genes from human DNA, then forming duplexes and enriching for mismatches, and finally amplifying by PCR. The procedure is as follows:

1. Purify genomic DNA from tissue
2. Cut with restriction enzyme to produce DNA fragments of reasonable size
3. Purify desired sequence using immobilized DNA probes: both sense and antisense
4. Melt DNA off probes and isolate it
5. Melt purified DNA and allow sense to hybridize with antisense
6. Enrich mismatches using immobilized mismatch binding protein as before
7. Amplify enriched DNA by PCR
8. Sequence to observe degree of enrichment

We do not yet have results from this approach, but we will continue this research after the end of the grant period.