Functional CpG Methylation System in a Social Insect

See allHide authors and affiliations

Science  27 Oct 2006:
Vol. 314, Issue 5799, pp. 645-647
DOI: 10.1126/science.1135213


DNA methylation systems are well characterized in vertebrates, but methylation in Drosophila melanogaster and other invertebrates remains controversial. Using the recently sequenced honey bee genome, we present a bioinformatic, molecular, and biochemical characterization of a functional DNA methylation system in an insect. We report on catalytically active orthologs of the vertebrate DNA methyltransferases Dnmt1 and Dnmt3a and b, two isoforms that contain a methyl-DNA binding domain, genomic 5-methyl-deoxycytosine, and CpG-methylated genes. The honey bee provides an opportunity to study the roles of methylation in social contexts.

Among the many important functions of CpG DNA methylation, sex-specific regulation of gene expression (imprinting) in vertebrates stands out because it provides insight into intragenomic conflict (1, 2). Provided social insects have CpG methylation, they would be ideal models to further explore the kin-conflict theory of imprinting, because insect societies are composed of many different types of relatives and they interact with each other in many evolutionarily important contexts (2, 3).

However, although widely conserved from yeast and fungi to plants to vertebrates, DNA methylation in insects is enigmatic. Evidence of CpG-methylated sequences exists for several insect species (46), but no bona fide invertebrate deoxycytosine methyltransferases (DNMTs) have been described. Conversely, the model insect Drosophila melanogaster shows limited DNA methylation, predominantly in asymmetric CpT and CpA dinucleotides (7), and this is attributed to the only DNMT family member encoded in its genome, dDNMT2, a tRNAAsp methyltransferase (8, 9). The Drosophila situation renders the fragmentary findings from the other insect species uninterpretable.

Here, we report that a social insect, the honey bee, Apis mellifera, has a fully functional CpG methylation system. We present biochemical, molecular, and genomic analyses, made possible by the recently sequenced honey bee genome (10).

In addition to Dnmt2, we identified one ortholog for de novo methylation (AmDnmt3) and two orthologs for maintenance methylation (AmDnmt1a and AmDnmt1b) (Fig. 1A and fig. S1A). AmDnmt1a and AmDnmt1b encode 70% identical ∼1400–amino acid proteins, which have 55% identity and 70% similarity to human DNMT1 over almost their entire length. Likewise, AmDNMT3 shows strong sequence similarity to hDNMT3A and hDNMT3B, with 33 and 32% similarity over the whole gene and 61 and 66% identity in the catalytic domains, respectively.

Fig. 1.

DNA methylation in the honey bee. (A) Expression of AmDNMT1 and AmDNMT3 genes. Reverse transcription polymerase chain reaction (RT-PCR) of total RNA from multiple tissues and stages of development (24-hour-old embryo, adult worker ovary, queen larva, drone genitals and sperm, and adult worker brain; lanes 1 to 5, respectively) indicates expression of AmDnmt1a (top), AmDnmt1b (middle), and AmDnmt3 (bottom). Data are from a representative experiment repeated twice with similar results. For each experiment, total RNA was extracted from 50 eggs (24 hours old), one adult queen ovary, one queen larva, four adult drone genitals, and 10 brains from 8-day-old worker bees. M = 100–base pair ladder. (B) AmDNMT1A and AmDNMT3 proteins are catalytically active DNA methyltransferases in vivo. DNMT assays were performed on honey bee protein extracts prepared from either 40 embryos (lane 3), or 20 mg of tissue from 2-day-old, 3-day-old, or 5-day-old larvae (lanes 4 to 6, respectively). Xenopus egg extract (lane 1) was used as a positive control for DNMT activity, and bovine serum albumin (BSA) (lane 2) was used as a negative control. Experiments were carried out four times, each in duplicate, resulting in a highly reproducible pattern of activity. Results from a representative DNMT activity assay are shown. The two bars represent duplicates from one experiment. dpm, disintegrations per minute. (C) AmDNMT1 and AmDNMT3 proteins are catalytically active DNA methyltransferases in vitro. In vitro transcribed and translated full-length AmDNMT1A, AmDNMT3 catalytic domain (amino acid 401 to the carboxyl terminus) (AmDnmt3c), and empty vector (control) were analyzed for DNA methyltransferase activity using an unmethylated doublestranded DNA template. The average of three independent experiments (±SD) is shown. (D) Honey bee genome contains 5-methyl deoxycytosine. Genomic DNA from adult honey bees (a, b, d, and e), larvae (c), or rat liver (f), was digested to nucleotide monophosphates and analyzed by reverse phase high-performance liquid chromatography for dCM. Digested bee DNA was spiked (a) with exogenous dCM and analyzed in parallel with the digested DNA (b), clearly identifying a peak in adult honey bee coincident with dCM enrichment (indicated by arrows). DNA from 3- to 4-day-old larvae (c) and adults (d) both contain dCM. The dCM peak from the larger sample (73.8 μg) of adult bee DNA (e) co-elutes with that from a lesser amount (17.8 μg) of rat liver DNA (f). The x axis is time (min) and the y axis is absorbance at 260 nm.

The predicted AmDNMT-encoding genes are expressed in tissue-specific and developmentally regulated patterns (Fig. 1A). Protein extracts from bees (Fig. 1B) demonstrate the presence of catalytically active DNMT enzymes. In vitro biochemical analyses of the recombinant AmDNMT proteins confirmed that AmDNMT1A and AmDNMT3 are catalytically active DNMTs (Fig. 1C) with similar characteristics to their vertebrate orthologs (9).

The honey bee CpG methylation system is functional in vivo. We isolated 5-methyl deoxycytosine (dCM) (11) from bee genomic DNA (Fig. 1D) and using the methods of (12) have so far identified six genes methylated in vivo (figs. S2 to S7). These analyses also revealed that non-CpG methylation is either extremely rare or nonexistent in honey bee. In each instance, the methylation was found exclusively in transcribed regions and predominantly in predicted exons with low G+C content and few CpGs overall. One methylated gene (GB16767) encodes an ortholog of mSin3A-associated protein 130 kD (SAP-130); Sin3 complexes regulate gene expression from yeast through humans (13), providing the potential for potent downstream genome-wide effects in honey bees.

Further analysis of the honey bee genome identified a gene homologous to the family of methyl-DNA binding domain (MBD) proteins (fig. S1B), some of which are effectors of DNA methylation (14). At least two expressed splicing variants exist, both of which contain the most highly conserved amino acids in the MBD (fig. S1B). AmMBD-l preferentially, but not exclusively, interacted with a methylated DNA probe in vitro (fig. S8, B and C), similar to MBDs in other species (14, 15). The protein sequence, size, and in vitro binding characteristics indicate that AmMBD-l is most similar to the vertebrate MBD3 subfamily. MBD3 proteins across species vary in their methyl-DNA binding specificities in vitro (14) and function in vivo as integral components of the Mi-2 chromatinremodeling complexes in both vertebrates and Drosophila (15, 16). Alternatively, the translation of DNA methylation marks in the honey bee may be achieved through other mechanisms (17).

On the basis of our findings, it is now possible to reflect on the earlier insect work (46) and predict that vertebrate-like systems of methylation are widespread in insects. If so, then Drosophila is of interest, not as a general model of insect methylation, but for unexplored evolutionary aspects of genome regulation, including the lack of a canonical telomeric (TTAGG) repeat (18), the pairing of homologous chromosomes throughout interphase (19), and the lack of symmetrical or CpG methylation (7).

Our findings also mean that the honey bee has the mechanisms that underlie imprinting, so tests of the kin-conflict theory in social insects are likely to be fruitful. Toward this end, we note several differences in honey bee methylation relative to vertebrate systems. First, methylation in vertebrates represses expression of repetitive DNAs and retrotransposons to maintain genome integrity, but in honey bees (and other insects) intermediate and high repetitive DNA elements and transposons are not methylated (6). This suggests that the role in mobile element repression has been lost or evolved after insects and vertebrates diverged (6, 9). Second, overall levels of methylation appear to be lower in the honey bee than in vertebrates, arguing against DNA methylation as a global mediator of bee heterochromatin. Third, honey bees possess two paralogs for methylation maintenance, making them the first animal discovered to express multiple somatic DNMT1 proteins. Fourth, all detected methylation was limited predominantly to the coding regions of genes. It remains to be seen how any of these differences relate to the functions of methylation in social contexts.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article