Comparative performance evaluation of bisulfite- and enzyme-based DNA conversion methods

Background: Bisulfite conversion (BC) has been the gold standard in DNA methylation profiling for decades. During this chemical process, non-methylated cytosines are converted into uracils, while methylated cytosines remain intact. Despite its popularity, BC has major drawbacks when used for sensitive applications with low-quality and -quantity DNA samples, such as the required large amount of DNA input, the caused DNA fragmentation and loss, and the resulting reduced sequence complexity. Lately, to account for BC-related disadvantages the first commercial enzymatic conversion (EC) kit was launched. While EC follows the same conversion principle as BC it uses two enzymatic steps instead of one chemical step with BC. In this study, we validated and compared the conversion performance of the most widely used BC and EC kits using a multiplex qPCR assay (qBiCo) we recently developed, which provides several indexes: conversion efficiency, converted DNA recovery and fragmentation. Results: Firstly, we implemented and standardized both DNA conversion methods. Secondly, using qBiCo, we performed a developmental validation for both conversion approaches, including testing the following parameters: repeatability, reproducibility, sensitivity and robustness. Regarding conversion efficiency, both methods performed similarly, with the limit of reproducible conversion being 5 ng and 10 ng for BC and EC, respectively. The recovery, however, is structurally overestimated for BC: 2.3 ± 0.7 and 0.7 ± 0.2 for EC. In contrast, degraded DNA input resulted in high fragmentation values after BC and low-medium values for EC (14.4 ± 1.2 and 3.3 ± 0.4, respectively). Finally, we converted 10 ng of 22 genomic DNA samples using both methods. We observed an overestimation of the BC DNA recovery (130%) and a low recovery for EC (40%). Conclusions: Our findings indicate that both DNA conversion methods have strengths and weaknesses. BC shows a high recovery, whereas EC does not cause extensive fragmentation that is characteristic to BC. EC is, therefore, more robust to the analysis of degraded DNA such as forensic-type or cell-free DNA, at least for the genomic DNA inputs tested here. We believe that the low recovery of EC could be improved by further optimizing and automating the bead-based cleanup steps. Overall, our study provides the first independent benchmarking of bisulfite- and enzyme-based conversion kits.