Sodium bisulfite is the undisputed king of DNA methylation research; and this king has been played over and over for years now. But a lot of players in the field are looking for something better, and the only thing that can trump a king is an ace, in this case ACE-seq. The active enzymatic demethylation of DNA leads to several intermediate cytosine modifications that may serve their own biological functions. Discrimination between these intermediates has always been a technical challenge in developing epigenomic techniques. Sodium bisulfite has become the major workhorse of DNA methylation techniques, relying on harsh chemical treatment of DNA to deaminate unmethylated cytosines to uracil. This process degrades DNA and is a challenge for low input DNA experiments. Furthermore, the standard bisulfite approaches cannot discriminate between 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC). Various approaches have been developed to discriminate between 5mC and 5hmC (TAB-seq, nanopore, single-molecule real time (SMRT), and restriction-enzyme-based approaches). But each of these methods had very different limitations, and have been producing conflicting data on 5hmC in the brain.
To address these challenges, the labs of Hao Wu and Rahul Kohli at The University of Pennsylvania developed APOBEC-coupled epigenetic sequencing (ACE-seq), an alternative to bisulfite sequencing that utilizes a non-destructive, enzymatic deamination of cytosine modifications. In their previous work, the talented team studied the AID/APOBEC family DNA deaminase enzymes, which deaminate cytosine to uracil in single-stranded DNA and are involved in innate immunity. They found that APOBEC3A (A3A) readily deaminates C and 5mC, but not the three oxidative 5mC derivative. In this work, they employed A3A to develop an assay for 5hmC that doesn’t rely on bisulfite treatment. The used the used the T4 phage as a model to develop the technique, since it has mutants available with either all cytosine methylated or hydroxymethylated. The team used Illumina high-throughput sequencing on their ACE-seq libraries, and compared to publicly available bisulfite-seq (BS-seq) data.
Here’s what they found:
- To protect all 5hmC sties from deamination, the authors tried in vitro glucosylation, they had a ∼4% protection rate of all 5hmC bases, which is at least as good as most TAB-seq studies
- Comparison of ACE-seq with bisulfite conditions in mouse embryonic stem cells (mESCs) found that 1 kb amplicons were were detectable with 1000 times lower DNA input with ACE-seq than with bisulfite, suggesting far less DNA damage
- Previous techniques have disagreed on the prevalence of 5hmC in CT dinucleotides, the authors found it to be rare: ∼5% of 5hmC as 5hmCH (H meaning “not G”) in mouse cortical neurons and absent in mESCs
- 5hmC was prevalent, but less abundant than 5mC in mouse cortical neurons (5hmC/5mC 0.3-0.5 across most genomic elements)
- Notable exceptions included nearly equal 5hmC/5mC at promoter-distal enhancers and depletion of 5hmC in imprinted regions
- Gene expression correlated with 5hmC/5mC ratio, suggesting a positive effect of 5hmC on expression
The need for about 1000 times less DNA that TAB-seq is an exciting prospect in 5hmC studies. Senior author Rahul Kohli shares, “We’re hopeful that this method offers the ability to decode epigenetic marks on DNA from small and transient populations of cells that have previously been difficult to study,” Also, the proof of principle work shown in mouse cortical neurons further supports the idea that 5hmC is important in brain, and that it has a different distribution than 5mC. ACE-seq may prove to be important for improving our understanding 5mC and 5hmC at fine cell-type resolution. So, with ACE-seq up researchers’ sleeves, the winning hand on cytosine research is within reach.
Check out the full article in Nature Biotechnology, October 2018