Fig. 1

Overview of the procedure to detect unmodified cytosines (C), 5-methylcytosine (5mC), and 5-hydroxymethylcytosine (5hmC). Naturally, C can be converted to 5mC by DNA methyltransferases (DNMTs) and 5mC can be oxidized by ten-eleven translocation (TET) enzymes, resulting in 5hmC. There are several proposed demethylation pathways through which 5mC and 5hmC can be converted back to C. DNA samples were split in two, one half was only treated with bisulfite (BS), which converts C into thymine (T). 5mC and 5hmC are protected against this conversion, and will be read as a C on the array. The detected C signal after BS conversion is thus actually the combined 5mC and 5hmC signal. As the signals are converted to fractions, with C + 5mC + 5hmC = 1, the fraction of C in the input DNA can be determined by subtracting the C signal after BS conversion (representing the combined 5mC and 5hmC fraction in the input DNA) from 1. The other half of the DNA sample was first oxidized, which converts 5hmC into 5-formylcytosine (5fC), and then treated with BS. 5fC is not protected against the BS conversion, so it also turns into T. C detected on the array after this oxidative BS (oxBS) conversion thus represents the fraction of 5mC in the input DNA. The 5hmC fraction in the input DNA can be determined by subtracting the fraction of 5mC (detect C after oxBS) from the combined 5mC and 5hmC fraction (detected C after BS). This procedure results in three readout signals: unmodified C, 5mC, and 5hmC. Note that 5fC, and probably also 5-carboxylcytosine, are included in the unmodified C fraction.