Reply to “Lack of evidence for DNA methylation of Invader4 retroelements in Drosophila and implications for Dnmt2-mediated epigenetic regulation”

Phalke et al. reply:

In their Correspondence regarding our work published in Nature Genetics¹, Schaefer and Lyko argue that the processive DNA methyltransferase activity in every dinucleotide context at Invader4 retroelements that we observed is difficult to reconcile with other reports about the activity of DNMT2 in Drosophila. However, until now, no other studies specifically analyzed DNA methylation at Invader4 retroelements. Using high through-put bisulfite sequencing of Invader4 long terminal repeats (LTRs), Schaefer and Lyko could not detect significant DNA methylation in a wild-type w¹¹¹⁸ strain, which contradicts our published data showing early embryonic, DNMT2-dependent DNA methylation at Invader4 LTRs. Schaefer and Lyko ultimately conclude that both the primers we used and the bisulfite treatment we performed in our analysis caused this discrepancy. However, methodological differences do not explain why our experiments generated more complete conversion of cytosines in the Dnmt2-null mutant control and in older embryos as compared to all the experiments reported by Schaefer and Lyko.

Schaefer and Lyko did not exactly repeat our experiments; instead, they modified the primers we used. Whereas our primers preferentially amplified fully methylated (and potentially unconverted) clones, their primer sets were designed for unbiased amplification or for preferential amplification of clones with fully converted cytosines and with no methylation in the primer binding sequence. Also, Schaefer and Lyko did not test the Dnmt2-null mutant for DNA methylation at functional Invader4 retroelements.

In our studies, we used a newly established w¹¹¹⁸ isogenic strain developed for genome-wide construction of molecularly mapped deletions². The strain was selected from approximately 40 newly established isogenic lines on the basis of normal learning and memory behavior. In this background, we isolated variegating P{w⁺} insertions in order to identify DNMT2 targets and resolve their role in control of retrotransposon silencing in Drosophila somatic cells. This strain showed significant Dnmt2 expression in 0–3-hour-old embryos¹. We now have preliminary evidence that Dnmt2 expression is highly variable in different strains of Drosophila melanogaster (unpublished data). Consequently, the differences found between the data presented by Schaefer and Lyko and our data might be explained by strain-specific differences in the early embryonic activity of Dnmt2. Further studies are needed to resolve the possible correlation between DNA methylation at retrotransposon LTRs and early embryonic Dnmt2 expression.

Furthermore, it should be recognized that annotated Invader4 LTR elements are structurally heterogeneous and are differentially distributed within the Drosophila genome, as demonstrated by FISH analysis using different probes¹. Large clusters of Invader4 LTRs are found at the subtelomeric regions of chromosome arms 2R and 3R. The 3R subtelomeric Invader4 cluster contains six tandemly organized 984-bp repeats consisting each of a 381-bp telomere-specific repeat and of 603 bp of three rearranged crippled copies of Invader4 LTRs¹. Two of the three sequence datasets (including the only dataset from a Dnmt2-null mutant) presented by Schaefer and Lyko are exclusively derived from these defective subtelomeric elements. We previously concluded that subtelomeric-defective Invader4 elements are silenced by a DNMT2-dependent and SUV4-20–dependent pathway, but we did not provide evidence of DNA methylation at these elements¹.