The capacity of the honey bee to produce three phenotypically distinct organisms (two female castes; queens and sterile workers, and haploid male drones) from one genotype represents one of the most remarkable examples of developmental plasticity in any phylum. The queen–worker morphological and reproductive divide is environmentally controlled during post-embryonic development by differential feeding. Previous studies implicated metabolic flux acting via epigenetic regulation, in particular DNA methylation and microRNAs, in establishing distinct patterns of gene expression underlying caste-specific developmental trajectories. We produce the first genome-wide maps of chromatin structure in the honey bee at a key larval stage in which developmental canalization into queen or worker is virtually irreversible. We find extensive genome-wide differences in H3K4me3, H3K27ac, and H3K36me3, many of which correlate with caste-specific transcription. Furthermore, we identify H3K27ac as a key chromatin modification, with caste-specific regions of intronic H3K27ac directing the worker caste. These regions may harbor the first examples of caste-specific enhancer elements in the honey bee. Our results demonstrate a key role for chromatin modifications in the establishment and maintenance of caste-specific transcriptional programs in the honey bee. We show that at 96 h of larval growth, the queen-specific chromatin pattern is already established, whereas the worker determination is not, thus providing experimental support for the perceived timing of this critical point in developmental heterochrony in two types of honey bee females. In a broader context, our study provides novel data on environmentally regulated organismal plasticity and the molecular foundation of the evolutionary origins of eusociality.

Haemophilus influenzae uses phase variation (PV) to modulate the activity of its defence systems against phage infection. The PV of the restriction–modification (R-M) system HindI, the main defence system against phage infection and incoming chromosomal and phage DNA in H. influenzae Rd, is driven by changes of the pentanucleotide repeat tract within the coding sequence of the hsdM gene and is influenced by lack of Dam methylation. Phase-variable resistance/sensitivity to phage infection correlates with changes in lipooligosaccharide (LOS) structure and occurs by slippage of tetranucleotide repeats within the gene lic2A, coding for a step in the biosynthesis of LOS. The lack of Dam activity destabilizes the tetranuclotide (5′-CAAT) repeat tract and increases the frequency of switching from sensitivity to resistance to phage infection more than in the opposite direction. The PV of the lgtC gene does not influence resistance or sensitivity to phage infection. Insertional inactivation of lic2A, but not lgtC or lgtF, leads to resistance to phage infection and to the same structure of the LOS as observed among phase-variable phage-resistant variants. This indicates that in the H. influenzae Rd LOS only the first two sugars (Glc-Gal) extending from the third heptose are part of bacterial phage receptors.

In mammals, a family of TET enzymes producing oxidized forms of 5-methylcytosine (5mC) plays an important role in modulating DNA demethylation dynamics. In contrast, nothing is known about the function of a single TET orthologue present in invertebrates. Here, we show that the honeybee TET (AmTET) catalytic domain has dioxygenase activity and converts 5mC to 5-hydroxymethylcytosine (5hmC) in a HEK293T cell assay.In vivo, the levels of 5hmC are condition-dependent and relatively low, but in testes and ovaries 5hmC is present at approximately 7–10% of the total level of 5mC, which is comparable to that reported for certain mammalian cells types. AmTET is alternatively spliced and highly expressed throughout development and in adult tissues with the highest expression found in adult brains. Our findings reveal an additional level of flexible genomic modifications in the honeybee that may be important for the selection of multiple pathways controlling contrasting phenotypic outcomes in this species. In a broader context, our study extends the current, mammalian-centred attention to TET-driven DNA hydroxymethylation to an easily manageable organism with attractive and unique biology.

Cytosine methylation promotes deamination. In eukaryotes, CpG methylation is thought to account for CpG underrepresentation. Whether scarcity of CpGs in prokaryotic genomes is diagnostic for methylation is not clear. Here, we report that Mycoplasms tend to be CpG depleted and to harbor a family of constitutively expressed or phase variable CpG-specific DNA methyltransferases. The very CpG poor Mycoplasma penetrans and its constitutively active CpG-specific methyltransferase M.MpeI were chosen for further characterization. Genome-wide sequencing of bisulfite-converted DNA indicated that M.MpeI methylated CpG target sites both in vivo and in vitro in a locus-nonselective manner. A crystal structure of M.MpeI with DNA at 2.15-Å resolution showed that the substrate base was flipped and that its place in the DNA stack was taken by a glutamine residue. A phenylalanine residue was intercalated into the “weak” CpG step of the nonsubstrate strand, indicating mechanistic similarities in the recognition of the short CpG target sequence by prokaryotic and eukaryotic DNA methyltransferases.

TET (ten-eleven translocation) enzymes catalyze the oxidation of 5-methylcytosine bases in DNA, thus driving active and passive DNA demethylation. Here, we report that the catalytic domain of mammalian TET enzymes favor CGs embedded within basic helix-loop-helix and basic leucine zipper domain transcription factor–binding sites, with up to 250-fold preference in vitro. Crystal structures and molecular dynamics calculations show that sequence preference is caused by intrasubstrate interactions and CG flanking sequence indirectly affecting enzyme conformation. TET sequence preferences are physiologically relevant as they explain the rates of DNA demethylation in TET-rescue experiments in culture and in vivo within the zygote and germ line. Most and least favorable TET motifs represent DNA sites that are bound by methylation-sensitive immediate-early transcription factors and octamer-binding transcription factor 4 (OCT4), respectively, illuminating TET function in transcriptional responses and pluripotency support.

The honey bee genome has the capacity to produce three phenotypically distinct organisms (two diploid female castes: queen and worker, and a haploid male drone). Previous studies have implicated metabolic flux acting via epigenetic regulation in directing nutrition-driven phenotypic plasticity in the honey bee. However, the cis-acting DNA regulatory elements that establish tissue and polyphenism -specific epigenomes and gene expression programmes, remain unclear. Using a high resolution multiomic approach including assay for transposase-accessible chromatin by sequencing (ATAC-seq), RNA-seq and ChIP-seq, we produce the first genome-wide maps of the regulatory landscape across all three adult honey bee phenotypes identifying > 5000 regulatory regions in queen, 7500 in worker and 6500 in drone, with the vast majority of these sites located within intronic regions. These regions are defined by positive enrichment of H3K27ac and depletion of H3K4me3 and show a positive correlation with gene expression. Using ATAC-seq footprinting we determine queen, worker and drone -specific transcription factor occupancy and uncover novel phenotype-specific regulatory networks identifying two key nuclear receptors that have previously been implicated in caste-determination and adult behavioural maturation in honey bees; ecdysone receptor and ultraspiracle. Collectively, this study provides novel insights into key gene regulatory networks that are associated with these distinct polyphenisms in the honey bee.