Functional diversification of the nematode mbd2/3 gene between Pristionchus pacificus and Caenorhabditis elegans

Background Several members of the Methyl-Binding Domain protein family link DNA methylation with chromatin remodeling complexes in vertebrates. Amongst the four classes of MBD proteins, MBD2/3 is the most highly conserved and widespread in metazoans. We have previously reported that an mbd2/3 like gene (mbd-2) is encoded in the genomes of the nematodes Pristionchus pacificus, Caenorhabditis elegans and Caenorhabditis briggsae. RNAi knock-down of mbd-2 in the two Caenorhabditis species results in varying percentages of lethality. Results Here, we report that a general feature of nematode MBD2/3 proteins seems to be the lack of a bona fide methyl-binding domain. We isolated a null allele of mbd-2 in P. pacificus and show that Ppa-mbd-2 mutants are viable, fertile and display a fully penetrant egg laying defect. This egg laying defect is partially rescued by treatment with acetylcholine or nicotine suggesting a specific function of this protein in vulval neurons. Using Yeast-two-hybrid screens, Ppa-MBD-2 was found to associate with microtubule interacting and vesicle transfer proteins. Conclusion These results imply that MBD2/3 proteins in nematodes are more variable than their relatives in insects and vertebrates both in structure and function. Moreover, nematode MBD2/3 proteins assume functions independent of DNA methylation ranging from the indispensable to the non-essential.


Background
DNA methylation is a common regulatory mechanism implicated in transcriptional silencing in vertebrates [1]. Methylated cytosines in DNA are recognized by a set of proteins carrying a methyl-binding domain (MBD) [2,3]. These MBD proteins interpret the methylation signal by recruiting components involved in transcriptional repression or heterochromatin formation [4,5].
Outside the MBD, the five members of the family share no similarity in their primary sequence, with the exception of MBD2 and MBD3 [6,[15][16][17]. Notably, invertebrates contain a smaller number of mbd-like genes. For example, in the nematode Caenorhabditis elegans and the fruit fly Drosophila melanogaster, only one gene with similarity to the mammalian mbd2/mbd3 genes has been identified. In D. melanogaster this gene is known as mbd-2/3, whereas in C. elegans this gene is named mbd-2, given the nomenclature rules of this organism [16,17]. In contrast, a recent report of the genome of the honeybee Apis melifera describes the presence of both mbd-2/3 and mbd-1 homologues [18].
In D. melanogaster, the mbd2/3 gene is alternatively spliced to produce two isoforms. The long isoform is expressed early in development and is able to bind specifically to CpT/A methylated DNA, in contrast to vertebrate MBD proteins which bind methylated CpG [19]. The short isoform, which is expressed in late embryogenesis, lacks part of the MBD and therefore is unable to bind methylated DNA [3,20]. A null allele of the gene mbd2/3 is viable and fertile but the stability of pericentric heterochromatin is considerably affected [19].
In contrast to D. melanogaster, no obvious DNA methylation system has been found in C. elegans and other tested nematodes [16]. The unique MBD2/3 protein, Cel-MBD-2, is characterized by the complete absence of a bona fide MBD. Knock down analysis using RNA interference (RNAi) showed variable phenotypes with different degrees of penetrance, such as paralysis and death due to rupture through the vulva [16]. In contrast, RNAi of the homologous gene in C. briggsae reveals an essential function since the worms arrest early in development [16].
Previously, we initiated the analysis of genes related to DNA methylation in another nematode, the diplogasterid Pristionchus pacificus [16]. P. pacificus has been used as a satellite organism to study the evolution of developmental processes and developmental mechanisms (for review see [21]). We showed that the P. pacificus genome encodes an mbd-2 gene. However, as in C. elegans, Ppa-MBD-2 also lacks a recognizable MBD implying a function independent of DNA methylation. Remarkably, the genome of P. pacificus also encodes a homolog of the D. melanogaster DNA methyltransferase 2, dnmt-2, [16], but this protein has been shown to function in the cytoplasm of eukaryotic cells as a tRNA methyltransferase [22,23].
Here, we report the presence of mbd-2 genes in nematodes from different evolutionary clades. All encoded proteins have in common the absence of a MBD. In order to extend the functional analysis of mbd-2-like genes, we isolated a deletion mutant in P. pacificus. Ppa-mbd-2 (tu365) mutant animals are viable and fertile, but display a completely penetrant defect in the egg laying system (egg laying defective, Egl). The Egl defect is partially rescued by treatment with acetylcholine or nicotine, suggesting that Ppa-MBD-2 functions in the neuronal innervations at the vulva muscles. A molecular approach using Yeast two hybrid (Y2H) analyses reveals interaction with diverse cellular proteins but fails to show interaction of any of the tested members of the NuRD complex. On the basis of these results, we propose that the absence of the MBD and the functional diversification of the MBD2/3 proteins represent a derived character in nematodes.

Nematode genomes contain mbd-2-like genes
The only member of the MBD family found in C. elegans, C. briggsae and P. pacificus belongs to the MBD2/3 group [16]. A key determinant of the MBD-2 proteins in these nematodes is the absence of a Methyl-Binding Domain (Fig. 1A,B; SFig. 1). To extend this observation to nematodes from different clades [24], we performed an extensive search of cDNAs in public genomic databases. Complete cDNA sequences were obtained for the filarial parasite Brugia malayi (clade III) and the rhabditid Haemonchus contortus (clade V). Partial 5' cDNA sequences were obtained for the spirurid Onchocerca volvulus (clade III), the tylenchid Meloidogyne hapla (clade IVa) and the rhabditid Nippostrongylus brasiliensis (clade V). Sequence comparison of the aminoterminal regions of the predicted MBD-2 proteins identified in different databases reveals a clear absence of an MBD (Fig. 1C). In addition, the highly conserved sequence motif SIFPQ, which is characteristic of the MBD2/3 group from vertebrates to nematodes, has changed substantially in the Caenorhabditis lineage and in M. hapla (Fig. 1C). We conclude that MBD-2 proteins are present in all studied nematodes, but that they lack a canonical methyl-binding domain.
A Ppa-mbd-2 deletion mutant is viable and egg laying defective The DNA methylation system encoded by members of the DNA methyl transferase family is essential in mammals, but is dispensable in lower vertebrates and invertebrates [22,25]. In particular, no dnmt-like genes are found in the genomes of C. elegans and C. briggsae. Because MBD proteins interpret methylation signals in insects and vertebrates, we hypothesized that the absence of an active DNMT protein and the unusual structure of the MBD protein in Caenorhabditis correlates with a functional diversification. Our observations, together with the fact that no DNA methylation system has been observed in any studied nematode, indicate that MBD-2 in nematodes evolved functions independent of this modification.
To analyze the function of MBD-2 in nematodes other than Caenorhabditis, we used the satellite organism P.
pacificus. This nematode is amenable to cellular and genetic studies and the genome is almost completely sequenced [21]. We applied a reverse genetics approach in P. pacificus to obtain a deletion mutant of the Ppa-mbd-2 gene using TMP/UV mutagenesis. We screened approximately 840,000 gametes and found one allele, tu365. This allele is likely to represent a null allele because four of the six exons are removed (ca. 1.7 Kb) and exon 1 is joined to exon 6 resulting in a frame-shift ( Fig The Ppa-mbd-2(tu365) mutant is recessive, homozygous viable and displays a fully penetrant Egl phenotype (Fig.  2D). Development proceeds normally until adulthood and gravid animals can lay an average of 40 eggs before eggs start to be retained by the mother. Eventually, egg retention and hatching inside the mother generate a "bag of worms" phenotype ( Fig. 2D and Table 1). No paralyticuncoordinated or rupture phenotype is seen in Ppa-mbd-2(tu365) mutants, as is the case in Cel-mbd-2(RNAi) treated animals.
On the basis of these results, we argue that the function of MBD-2 in nematodes has changed from an essential to a dispensable gene. This argument is based on the observation that RNAi knock down of mbd-2 genes in the Caenorhabditis lineage produced lethality, whereas the P. pacificus null mutant is viable and fertile. It is important to note that mbd-2 genes are present in single copy in the genomes of C. elegans, C. briggsae and P. pacificus, as determined by Southern blot analysis (SFig. 3). Moreover, the location and structure of the mbd-2 gene differs in these nematodes. For example, in C. elegans and C. briggsae, the mbd-2 gene is located in an intron of the glycosylphosphatidylinositol anchor synthesis protein, C27A12.9, and A P. pacificus mbd-2 mutant is viable and Egg laying defective  is trans-spliced to the splice leader 1 (SL1). In P. pacificus in contrast, mbd-2 is the second gene of an operon and is trans-spliced to SL2, a key determinant of genes in operons ( Fig. 2A) [26,27]. This difference suggests the potential for changes in the transcriptional regulation of the mbd-2 genes in these species and might represent a source for the acquisition of novel functions.

Ppa-MBD-2 functions in the egg laying system
In P. pacificus, as in C. elegans, an Egl phenotype can be caused by many developmental or physiological defects. For example, misspecification of the vulva, the gonad, the sex musculature and several neurons innervating these muscles can result in an Egl phenotype. DIC-microscopy analysis of Ppa-mbd-2(tu365) mutant worms show that vulval cell fate specification is normal (Fig. 3A, B). Also, phalloidin staining, which stain actin fibers, revealed that the sex muscles are also present and correctly shaped (Fig.  3C, D). Therefore, the Ppa-mbd-2 Egl phenotype might result from defects in a neuronal function. As tu365 animals can initially lay eggs, we assumed that the neuronal system working on the egg-laying apparatus is formed properly and originally functional. Later on, neurodegeneration or failure in signal transmission might result in the Egl phenotype.
In C. elegans, the hermaphrodite specific neuron (HSN) and the ventral nerve cord neurons control the egg-laying program through the secretion of serotonin, acetylcholine and at least four neuropeptides [28]. Incubation of C. elegans with serotonin or nicotine stimulates egg laying [29,30]. In contrast, treatment with acetylcholine has opposing stimulatory and inhibitory roles in egg laying by acting through different types of receptors [31,32]. To test if the Egl phenotype of Ppa-mbd-2(tu365) can be rescued by neurotransmitters reported to work on the C. elegans egg-laying system, we incubated wild type and Ppa-mbd-2(tu365) mutant worms in M9 buffer containing different kinds of neurotransmitters or their agonists. It is important to note that wild type P. pacificus PS312 animals, in contrast to C. elegans, retain no more than two to four eggs under normal cultural conditions. However, when worms were starved for about 12 hours, they retained on average six to eight eggs (environmental Egl). In our egg-laying assay, P. pacificus PS312 animals laid an average of 3.5 to 4.5 eggs per animal independent of the presence or absence of any of the tested neurotransmitters except for serotonin (5-HT). 5-HT has a negative effect on egg laying since wild-type induced environmental Egl animals resulted in a reduction of egg laying to an average of 1.5 eggs per animal (Fig. 4). Synchronized Ppa-mbd-2(tu365) mutant animals showed no or very low egg-laying in M9 and GABA. In contrast, treatment with acetylcholine and its agonist nicotine induced a potent (six-fold) increase in egg-laying in Ppa-mbd-2(tu365) (Fig 3). 5-HT treatment also increased egg-laying, but not as efficiently as acetylcholine and nicotine (Fig. 4). These results suggest that Ppa-mbd-2 may be implicated in cholinergic neurostimulation. More generally, these initial studies suggest a neuronal wiring that differs from the corresponding process in C. elegans, although we cannot completely rule out that the wiring is the same and that only the pharmacological features of the cells are different in both species. A more detailed study of the nervous system in P. pacificus will clarify these observations further. Interestingly, in a comparative approach using different nematode species, Loer and Rivard have shown recently that P. pacificus appear to have lost a serotonin-immunoreactive HSN [33]. Therefore, these results do not exclude the possibility that the neuronal function of MBD-2 is conserved in both species. One might speculate that MBD-2 could be required in a wider set of cell types in C. elegans.

Ppa-MBD-2 interacts with diverse cellular proteins
Given that nematode MBD-2 proteins function independent of DNA methylation, they should have coevolved specific interaction partners. To test this hypothesis, we performed an unbiased Y2H screen using Ppa-MBD-2 as prey and a cDNA library of P. pacificus as bait (~1 × 10 7 clones). The protein does not show any auto-activation when present as both, bait and prey ( Table 2). We found several different in frame interactors, such as first, a SPOPlike protein which is part of E3-ubiquitin ligase [34]; second, a myosin heavy chain homolog to the C. elegans T07C4.10 predicted protein; third, a Huntingtin interacting protein; fourth, a dynein ligh chain; fifth, a protein similar to C. elegans tag-260 and two proteins with no homology in the database ( Table 2, SFig. 4). The fact that mdb-2 interacts with microtubule associated proteins such as dynein-light chain is interesting since the first gene of the operon in which Ppa-mbd-2 is placed encodes mapmodulin, a microtubule binding protein involved in vesicles transport from endosomes to trans-Golgi [35]. In this context it is important to note that genes arranged in operons have been suggested to display similar functions [26].
As some of these interactions may be physiological irrelevant, a final interpretation awaits a more detailed biochemical analysis.
Interestingly, components of the NuRD complex were not found in the Y2H screening. However, sequencing of randomly chosen clones of the Y2H cDNA library revealed the presence of many in-frame components of the NuRD complex (data not shown). Therefore, we speculate that Ppa-MBD-2 is not involved in transcriptional repression or chromatin remodeling. Taken together, Y2H screens found Ppa-MBD-2 to associate with microtubule interacting and vesicle transfer proteins

Conclusion
MBD proteins play a crucial role in DNA methylation and show an enormous multiplication and functional diversification in vertebrates. In insects, the number of genes encoding MBD proteins is smaller than in vertebrates. Our studies presented here indicate that in nematodes, only MBD2-like proteins are observed, which evolve rapidly at the sequence level. They show functions independent of DNA methylation, a phenomenon that is absent in C. elegans, C. briggsae and P. pacificus. With the limited functional data on MBD proteins from only three phyla (vertebrates, insects and nematodes) it remains currently unknown if the role of these proteins in DNA methylation represents the ancestral or derived function of the proteins. Importantly, MBD3 like proteins with a bona fide MBD are found in other lower metazoa such as in the platyhelminth Schistosoma japonicum [6]. Therefore, we speculate that the observed structure of the MBD in the Ppa-mbd-2 (tu365) mutants have normal vulva and muscle development

Behavioral assay
To characterize egg-laying response to different drugs, synchronized PS312 or mutant worms were grown in OP50 at 20°C until adult stage. Before starting the assay, PS312 worms were transferred to a NGM plate without food and incubated 12 hours at 20°C. Wild type and mutant Egl animals (that is, animals that retain in the uterus more eggs than normal) were placed individually in 100 µl of M9 buffer containing either 5 mg/ml 5-hydroxytriptamine, 3 mg/ml γ-aminobutiric acid (GABA), 5 mg/ ml acetylcholine or 0.05% nicotine (Sigma-Aldrich, USA) [29][30][31][32]40]. All assays were performed with at least 100 animals/strain/drug at 20°C and eggs layed per worm were counted after 2 hours.

Yeast-two-hybrid assay
The complete coding sequence corresponding to the gene Ppa-mbd-2 was amplified by reverse transcription PCR using the primers AG10576 ATGGCAAAAGCAGAT-GCTCTG and AG10577 CATCCAAAGTACTTCATAAC and cloned in pCR8-GW-Topo (Invitrogen, USA). Prey plasmid was done using LR-clonase into destination vector pDEST-32 BD clones (Invitrogen, USA,). Standard techniques were used to transform the plasmids into MaV203 yeast (Invitrogen, USA). Autoactivation test with this plasmid was negative. The prey pDEST-22 AD library of P. pacificus cDNA was prepared from poly(A) RNA using the Cloneminer kit following manufactures instructions (Invitrogen, USA,). Y2H screen was performed on SD plates lacking Leucine, Triptophan, Histidine plus 50 mM 3AminoThiazole. In total, 1 × 10 7 transformants were screened. Positive colonies were tested for LacZ staining, as reported [41]. In addition, positive clones were retransformed and retested.

Authors' contributions
AG designed and performed research, analyzed the data and wrote the paper. RJS designed research and wrote the paper. The authors read and approved the final manuscript.