NEW YORK (GenomeWeb) – Researchers at New England Biolabs have used Pacific Biosciences' Single Molecule Real-Time (SMRT) sequencing to examine sources of error caused by PCR, showing a low error rate can persist even with high-fidelity polymerases.
The NEB scientists plan to use the data to help them better understand the enzymes they create in the future and to give customers a more complete picture of errors that can happen in the PCR amplification step prior to sequencing.
In a report published earlier this month in PLoS One the NEB group highlighted three types of error: base substitution, DNA damage, and template switching.
"We decided to use SMRT sequencing because it enables us to directly sequence all of the individual molecules that were produced during amplification," Jennifer Ong, a senior scientist at NEB and corresponding author on the study, said in an interview.
SMRT sequencing provides multiple reads from the same DNA molecule, "So you're able to actually achieve very high accuracy," Ong added. The method was also considered by the team to be a more direct approach to determining error rates than other available technologies, such as fidelity assays based on Illumina sequencing.
The team did, however, compare the PacBio sequencing to Sanger sequencing for one of the reference polymerases it examined, as a way to compare the error rates to a traditional sequencing method, and found that the two methods yielded comparable results.
The group first began by looking at polymerase base substitution errors, Ong said, which occurs when a polymerase puts in an incorrect base during replication.
For enzymes like Taq polymerase, the majority of errors would fall into this category. Overall, the NEB researchers determined the base substitution error rates in nine different DNA polymerases using the SMRT method, reporting a substitution rate, accuracy, and fidelity relative to Taq. The firm's Q5 polymerase showed the lowest substitution rate and highest accuracy, with a fidelity 280-fold higher than Taq, while a polymerase called Deep Vent (exo-) had the highest rate and lowest accuracy in terms of base substitution.
Differentiating Family A and Family B polymerases in this set, they also found that the mutational spectrum differed, with A-type polymerases producing errors predominantly at A:T pairs and B-type dominated by G-to-A and C-to-T transitions.
Many previous reports describe these base substitution errors as number of substitutions per PCR cycle, which is challenging to compare to, Ong said. The NEB team reported the polymerase error rates as number of base substitutions per base per replication event, but also reported as "per PCR cycle" to enable comparison to other research.
"We've now developed very accurate polymerases, like Q5, which have a very low base substitution error rate," Ong said. "Because this error rate is so low, we also started looking at what other types of error occurred during PCR."
So, beyond base substitution errors caused by polymerases, Ong and her colleagues at NEB also investigated changes in DNA caused by thermal cycling.
"We found that when you take plasmid DNA and put it through thermal cycling as you would for PCR, it introduces damage to the DNA and causes mutations to appear," Ong said. Specifically, thermal cycling caused C-to-T mutations, and, using an NEB-manufactured enzyme cocktail called Pre-CR that repairs DNA damage, the researchers determined these mutations were caused by cytosine deamination.
"Because our single-molecule assay is very sensitive, we're able to detect the low level of DNA damage that gets introduced by thermal cycling," Ong noted. Interestingly, DNA damage from thermal cycling introduced a higher level of base substitution errors in PCR products than polymerase mis-incorporation by the Q5 enzyme.
Finally, another type of error the group examined was a phenomenon called template switching, which produces chimeric PCR products.
This can occur when amplifying a mixed population of templates, for example when sequencing the 16s ribosomal gene from a mixed microbial population or HLA genotyping, which both start with a mixed pool of closely related sequences.
Chimeric PCR products occur when a polymerase extends a primer but does not go to completion, and during the next round of thermal cycling that extended primer may anneal to another template and get extended. "This can cause a lot of confusion if you are trying to do haplotyping or microbial identification," Ong said, because the mixed products can indicate species or templates are present which may not actually be there.
The assay the group developed to get at the rate of template switching involved cruciform regions of lacZ flanked by inverted repeat elements. This identified three classes of template switching that produce chimeras, and the group found that some polymerases were more susceptible to these inversion events.
Replication fidelity is a primary interest at NEB, as is getting a better understanding of the enzymes it makes, Ong said. "We wanted to put this information out there on a lot of the enzymes that we sell so that customers are able to understand error rates and choose the best enzyme," she said.
But the assay is also useful to the firm as well. "Because we have developed very high-fidelity, accurate enzymes, we wanted to develop a better fidelity assay to be able to prove the accuracy of our enzymes," she said, adding, "We'll use this assay to develop more accurate enzymes in the future."