Inventory and phylogenomic distribution of meiotic genes in Nasonia vitripennis and among diverse arthropods


John M. Logsdon Jr, Roy J. Carver Center for Comparative Genomics and Department of Biology, University of Iowa, Iowa City, IA 52242, USA. Tel.: 1 319 335 1082; fax: 1 319 335 1069; e-mail:


The parasitoid jewel wasp Nasonia vitripennis reproduces by haplodiploidy (arrhenotokous parthenogenesis). In diploid females, meiosis occurs during oogenesis, but in haploid males spermatogenesis is ameiotic and involves a single equational division. Here we describe the phylogenomic distribution of meiotic genes in N. vitripennis and in 10 additional arthropods. Homologues for 39 meiosis-related genes (including seven meiosis-specific genes) were identified in N. vitripennis. The meiotic genes missing from N. vitripennis are also sporadically absent in other arthropods, suggesting that certain meiotic genes are dispensable for meiosis. Among an additional set of 15 genes thought to be specific for male meiosis in Drosophila, two genes (bol and crl) were identified in N. vitripennis and Apis mellifera (both for which canonical meiosis is absent in males) and in other arthropods. The distribution of meiotic genes across arthropods and the impact of gene duplications and reproductive modes on meiotic gene evolution are discussed.


Nasonia is a hymenopteran genus of parasitoid wasps in which females lay their eggs within fly pupae. There are four described species: Nasonia giraulti and Nasonia longicornis, which are endemic to North America, Nasonia oneida and the cosmopolitan Nasonia vitripennis (Campbell et al., 1993; Beukeboom & Desplan, 2003; Raychoudhury et al., 2009). Like all Hymenoptera, Nasonia have arrhenotokous haplodiploid sex determination: fertilized diploid eggs develop as females and unfertilized haploid eggs develop parthenogenetically as males. During oogenesis in N. vitripennis, haploid eggs are produced by meiosis (King & Richards, 1969). However, during spermatogenesis in haploid males, meiosis is altered: the reductional division is absent, but the equational division is maintained to produce haploid spermatids (Hogge & King, 1975; Beukeboom & Kamping, 2006). Similarly in Apis, spermatogenesis in haploid males is described as an ‘abortive meiosis I and anomalous meiosis II’ (Sharma et al., 1961).

Identifying meiotic genes in hymenopterans and other arthropods will help us to understand the regulation and mechanics of male and female gametogenesis. A recent phylogenomic inventory of >40 meiotic genes in Daphnia pulex revealed that the majority of these genes are present as multiple copies, leading to the suggestion that some copies might specifically function during parthenogenesis (Schurko et al., 2009). In this study, we present a phylogenomic inventory of meiotic genes in 11 arthropod genomes. Our primary objective is to annotate meiotic genes in the N. vitripennis genome (Werren et al., 2010), but we also survey genome projects representing Diptera (Drosophila melanogaster, Aedes aegypti, Anopheles gambiae and Culex pipiens), Coleoptera (Tribolium castaneum), Hymenoptera (Apis mellifera), Hemiptera (Acyrthosiphon pisum), Phthiraptera (Pediculus humanus), Crustacea (D. pulex) and Chelicerata (Ixodes scapularis).

First, we searched for homologues of meiosis-related genes. These genes are involved in meiotic cell cycle regulation, sister chromatid cohesion and interhomologue recombination. Second, we determined the phylogenetic distribution of 15 male-specific meiotic genes (i.e. genes that are specific to male meiosis and spermatogenesis in D. melanogaster) in arthropods to determine the evolutionary fate of these genes, most notably when male meiosis is altered (as in Nasonia and Apis). For all genes, predicted protein sequences are used in phylogenetic analyses to confirm gene identifications. Overall, this phylogenomic survey provides insight into the evolutionary patterns (e.g. gene losses and duplications) for these genes across the major arthropod groups, and also within and between parthenogenetic lineages.


I. Meiosis-related genes in Nasonia vitripennis and arthropods

We queried the genomes of N. vitripennis and 10 additional arthropods for meiosis-related genes (Fig. 1). These genes were selected because they: (1) are present in most major eukaryotic lineages (eg Ramesh et al., 2005; Malik et al., 2008); (2) are widespread in arthropods (Schurko et al., 2009); and (3) encode proteins that function during meiosis and other cellular processes (such as mitosis or DNA repair) in model eukaryotes. Eight of these genes are ‘meiosis-specific’ (boldface in Fig. 1), meaning that homologues in diverse species function and/or are expressed only during meiosis, and null mutations in model organisms are defective only in meiosis (Ramesh et al., 2005). For the purposes of this study, meiosis-related genes are divided into three broad categories: (1) cell cycle regulation genes; (2) sister chromatid cohesin genes; and (3) interhomologue recombination genes.

Figure 1.

Inventory of meiosis-related genes in Nasonia and 10 additional arthropod genomes. Gene names in bold are meiosis specific in model eukaryotes. ‘+’ indicates the presence of one gene while ‘−’ indicates the gene was not found and is likely absent. Numbers denote multiple copies. The relationships in the arthropod phylogeny are based on the literature (e.g. Giribet et al., 2001; Grimaldi & Engel, 2005; Savard et al., 2006). aGenome assembly in progress at the time this gene inventory was assembled. bSome Daphnia gene copies that might represent pseudogenes (see Schurko et al., 2009) were omitted from phylogeny. cOne of the two gene copies is meiosis specific in Drosophila.dCulex mlh1 not shown in phylogeny (short sequence could not be aligned with the other mutL homologues).

1. Cell cycle regulation genes

Cyclins, CDKs and CDC20.  Complexes of cyclin and cyclin-dependent kinase (CDK) proteins function at different levels of cell cycle regulation (reviewed by Hochegger et al., 2008). Among these complexes, cyclin B/CDK1 is the master regulatory kinase of the G2/M transition in mitosis and meiosis (Dorée & Hunt, 2002; Xiang et al., 2007). During Drosophila oogenesis, cyclin E/CDK2 is required for premeitoic S phase and regulating meiotic entry (Lilly & Spradling, 1996; Narbonne-Reveau & Lilly, 2009). The destruction of cyclins A, B and B3 and the removal of cyclin B from the spindles (which is necessary to complete meiosis) is mediated by Cks30A and the CDC20 homologues Fizzy (fzy) and meiosis-specific Cortex (cort) (Swan et al., 2005; Pesin & Orr-Weaver, 2007; Swan & Schüpbach, 2007).

We searched for genes encoding cyclin, CDK and CDC20 proteins. Homologues for the A, B, B3, D and E cyclins are present in N. vitripennis and in the 10 other arthropod genomes with multiple copies of certain genes present in some species (Fig. 1, Fig. S1). cdk1 and cdk2 are also present in Nasonia and in the other arthropods, and are clearly distinguished from the mitotic cell cycle regulators cdk4/6 and cdk10 in the phylogeny (Fig. 1, Fig. S2A). The cdc20 homologues fzy and cortex (which is specific to arthropods) were identified in Nasonia and in other arthropods, although cortex is absent in Pediculus, Acyrthosiphon, Daphnia and Ixodes (Fig. 1). The phylogeny (Fig. 2A) distinguishes fzy and cortex from the mitotic cdc20 paralogue fizzy-related (fzr) and suggests the fzy/cortex duplication occurred either early in arthropod evolution (and cortex was later lost in some lineages) or after the divergence of Ixodes, Daphnia, Acyrthosiphon and Pediculus (although neither scenario is strongly supported).

Figure 2.

Bayesian phylogenetic analyses of meiosis-related genes. (A) Phylogeny of metazoan cdc20 homologues based on an alignment of 279 amino acids. Parameter means: α= 1.82, pI = 0.04 and lnL =−10 549.39. (B) Phylogeny of eukaryotic rad21 and rec8 homologues based on an alignment of 141 amino acids. Parameter means: α= 1.63, pI = 0.04 and lnL =−10 075.75. All trees are a consensus of 900 best trees and thickened branches indicate posterior probabilities from 0.95 to 1.0. Protein identifiers for Nasonia vitripennis sequences (in bold) are in Table S1.

Polo kinases.  Polo kinases (PLKs) are a family of serine/threonine kinases involved in mitotic and meiotic cell cycle regulation (reviewed by Archambault & Glover, 2009). In Drosophila, POLO (the PLK-1 homologue) acts as the ‘trigger’ kinase for the G2/M transition (Llamazares et al., 1991) and also phosphorylates MEI-S332 to promote the removal of cohesin prior to anaphase II (Clarke et al., 2005). MATRIMONY (MTRM) inhibits POLO during early oogenesis in Drosophila, but this inhibition ends when POLO levels exceed those of MTRM (Xiang et al., 2007); POLO then phosphorylates the CDC25 phosphatase to activate cyclin B/CDK1 and meiotic entry. There are two cdc25 homologues in Drosophila: string (required for mitosis) and meiosis-specific twine (Alphey et al., 1992; Courtot et al., 1992). The functions of the other PLK family members (e.g. PLK-2, PLK-3 and PLK-4) in cell cycle progression are less well understood.

We searched for homologues of plk-1 and cdc25 (mtrm homologues could not be identified outside of Drosophila). plk-1 is present in Nasonia and in other arthropods, with multiple plk-1 copies in Acyrthosiphon and Daphnia (Fig. 1). The phylogeny distinguishes plk-1 from the other plk family members (Fig. S2B). Other plk-like sequences for Nasonia, Apis, Pediculus and Daphnia are designated ‘plk-2/plk-3’ because of their phylogenetic position basal to the vertebrate plk-2 and plk-3 clades (Fig. S2B). cdc25 is also present in Nasonia and in the other arthropod genomes, with independent gene duplications in Drosophila (string and meiosis-specific twine) and Acyrthosiphon (three copies present) (Fig. 1, Fig. S2C). However, the function of these arthropod cdc25 homologues in mitosis or meiosis is uncertain in the absence of functional studies.

2. Sister chromatid cohesin genes Cohesin is a multi-protein complex that maintains cohesion between sister chromatids until anaphase during mitosis and meiosis (reviewed by Onn et al., 2008). In general, cohesin complexes consist of two structural maintenance of chromosome (SMC) proteins (SMC1 and SMC3), one RAD21 protein (largely replaced by its paralogue REC8 during meiosis) and one stromal antigen (SA) subunit. Cohesin is removed during mitosis and meiosis by separase, a protease that cleaves RAD21 (or REC8) (Uhlmann et al., 1999). We searched arthropod genomes for homologues of rad21, rec8, SA, separase, smc1 and smc3. In addition, we also searched for timeout (tim-2) which has been implicated in cohesin loading in Caenorhabditis elegans (Chan et al., 2003).

Rad21/rec8, SA and separase.  While rad21 is present in Nasonia and in all arthropods, its meiosis-specific paralogue rec8 is only found in Nasonia, Apis, Tribolium, Bombyx, Drosophila and Daphnia (Figs 1, 2B). This suggests at least three independent rec8 losses during arthropod evolution and additional duplications in the Daphnia lineage. separase is also present in Nasonia and in the other 10 arthropod genomes (Fig. 1, Fig. S3A). Genes encoding SA proteins have undergone multiple independent duplications in eukaryotes, often resulting in meiosis-specific paralogues (e.g. meiosis-specific paralogues rec11 in the yeast Schizosaccharomyces pombe (Kitajima et al., 2003) and stag-3 in vertebrates (Pezzi et al., 2000). In Drosophila, two SA paralogues are present: SA is part of the cohesin complex while SNM is specific to male meiosis (discussed below) (Thomas et al., 2005). We identified SA homologues in Nasonia and in other arthropods, with the aforementioned gene duplications and additional duplications in Daphnia (five copies) and Aspergillus (two copies) evident in the phylogeny (Fig. S3B).

Structural maintenance of chromosome genes.  We identified smc1 and smc3 in all arthropods (Fig. 1); the phylogeny (Fig. 3) distinguishes smc1 and smc3 homologues from smc2 and smc4 (components of the condensin complex). The two smc1 homologues in Nasonia and Apis likely represent a tandem duplication; in each species, the two smc1 copies are each present on one genomic scaffold separated by ∼1 kb. In contrast, the two smc1 copies in Daphnia do not represent a tandem duplication (Schurko et al., 2009). The phylogeny (Fig. 3) shows a longer branch for XP_001603761 (Nasonia) and XP_001120789 (Apis), suggesting an accelerated rate of sequence divergence for these duplicate copies. In addition, there is expressed sequence tag (EST) evidence only for NP_001152812 (Nasonia) (Table S1) and XP_395059 (Apis). Single smc3 paralogues in Anopheles, Daphnia and Ixodes form long branches in the phylogeny at the base of the smc3 clade (Fig. 3). These divergent copies might represent remnants of an old duplication that has been lost from many eukaryotes (Schurko et al., 2009). Among the other non-cohesin smc genes, smc2 and smc4 are found in Nasonia and in all arthropods (Fig. S8).

Figure 3.

Bayesian phylogenetic analysis of metazoan smc homologues. Tree was determined from an alignment of 271 amino acids and represents a consensus of 900 best trees. Parameter means: α= 1.53 pI = 0.05 and lnL =−13 120.97. Thickened branches indicate a posterior probability of 0.95–1.0. Gene identifiers for Nasonia sequences (in bold) are in Table S1.

Timeout (tim-2). timeout (tim-2) encodes a protein that regulates the loading of cohesin onto chromosomes during DNA replication in C. elegans (Benna et al., 2000; Chan et al., 2003). TIM-2 is a paralogue of TIMELESS (TIM-1), a circadian rhythm protein in insects (Myers et al., 1995). In mammals, TIM-2 forms a complex with Tipin that is involved in the DNA damage checkpoint response (Chou & Elledge, 2006), while in yeast TIM-2 (called TOF1) and Tipin are involved in meiotic and mitotic chromosome segregation (Xu et al., 2004). Homologues of timeout (tim-2) are present in Nasonia and in all arthropods, with two copies present in Daphnia (Fig. 1). The phylogeny distinguishes tim-2 orthologues from arthropod-specific timeless (tim-1) (Fig. S3C). However, the function of tim-2 in arthropods has not been determined and the phylogeny alone cannot predict whether tim-2 orthologues are involved in meiosis as demonstrated in C. elegans and yeast.

3. Interhomologue recombination genes

DSB formation: spo11.  The initiation of meiotic recombination depends upon DSB formation by meiosis-specific SPO11 (Keeney et al., 1997), although DSB-independent synapsis occurs in some eukaryotes (e.g. Dernburg et al., 1998). Importantly, spo11 has been identified in all eukaryotes examined (Malik et al., 2007), suggesting its function is indispensable for meiosis. Consistent with this, single copies of spo11 were identified in Nasonia and in all arthropod genomes (Fig. 1, Fig. S4A).

Homologous recombination and strand exchange.  Following DSB formation, the single-stranded DNA (ssDNA) tails are resected and loaded with proteins to form nucleoprotein filaments that promote homology searches, strand exchanges and DNA repair between homologous chromosomes. We searched for genes that encode proteins involved in strand invasion (RAD51, DMC1, MND1, HOP2, RAD54, RAD54B) and in crossover resolution and DNA repair (MutL and MutS homologue proteins). We also searched for several members of the recQ family of DNA helicases that regulate levels of homologous recombination.

Rad51 and dmc1.  RAD51 and meiosis-specific DMC1, the two major eukaryotic homologues of eubacterial RecA (Bishop et al., 1992; Shinohara et al., 1992), co-localize to generate crossovers between homologous chromosomes during meiosis (reviewed by San Filippo et al., 2008). Many eukaryotes possess RAD51 paralogues; in animals, there can be as many as five RAD51-like genes (rad51B, rad51C, rad51D, xrcc2 and xrcc3). Biochemical studies suggest that these proteins form various complexes with each other and with RAD51 that may be involved in meiotic recombination (Lin et al., 2006; San Filippo et al., 2008).

We identified rad51 in Nasonia and in all arthropods; however, dmc1 is absent in eight out of 11 arthropod genomes examined (Fig. 1). The sporadic presence of dmc1 in Ixodes, Pediculus, Tribolium (and also in Rhipicephalus and Bombyx, included here) suggests that there have been at least four independent losses of dmc1 during arthropod evolution (Fig. 4A). Among additional rad51 paralogues, xrcc2, rad51c and rad51d were found in Nasonia and in the other arthropods, except for the apparent absences of xrcc2, xrcc3 and rad51c in certain taxa (Fig. 1). The phylogeny shows that Drosophila rad51 paralogues spn-B, spn-D, CG2412 and CG6318 are orthologous to xrcc3, rad51c, rad51d and xrcc2, respectively (Fig. 4A), as previously inferred (Lin et al., 2006).

Figure 4.

Bayesian phylogenetic analyses of meiosis-related genes. (A) Phylogeny of metazoan rad51 homologues based on an alignment of 217 amino acids. Parameter means: α= 1.91, pI = 0.02 and lnL =−25 494.25. (B) Phylogeny of metazoan mutS homologues based on an alignment of 322 amino acids. Parameter means: α= 1.81, pI = 0.04 and lnL =−27 991.55. All trees are a consensus of 900 best trees and thickened branches refer to posterior probabilities from 0.95 to 1.0. Protein identifiers for Nasonia vitripennis sequences (in bold) are in Table S1.

Mnd1, hop2, rad54 and rad54B.  Meiosis-specific MND1 and HOP2 form a heterodimer that is involved in the formation of the first recombination intermediates (Tsubouchi & Roeder, 2002; Petukhova et al., 2005). In humans, MND1/HOP2 stabilizes the formation of the RAD51/DMC1 nucleoprotein complex with ssDNA to promote the capture of double-stranded DNA (dsDNA) (Chi et al., 2007; Pezza et al., 2007). Single copies of mnd1 and hop2 are present in Nasonia and in other arthropods, except in Drosophila and Anopheles (Fig. 1, Fig. S4B, C). Interestingly, homologues encoding DMC1 (which interacts with MND1/HOP2) are absent in six taxa for which mnd1 and hop2 are present (Fig. 1).

Rad54 and its paralogue rad54B (also called rdh54 or tid1 in fungi) each interact with RAD51 to promote D-loop formation (Petukhova et al., 2000). Further, in humans, RAD54B stabilizes the DMC1/ssDNA nucleoprotein filament during meiosis (Sehorn et al., 2004). rad54 and rad54B are present in Nasonia and in the other arthropods, with independent losses of rad54B in Drosophila and Tribolium (Fig. 1, Fig. S4D).

Eukaryotic mutL and mutS homologues.  Eukaryotic homologues of the Escherichia coli MutL and MutS mismatch repair (MMR) proteins form heterodimers that function in chromosomal synapsis, recombination and MMR (reviewed by Kunkel & Erie, 2005). Eukaryotes encode four mutL homologues: mlh1, pms1 (mlh2 in fungi), mlh3 and pms2 (pms1 in fungi). The MLH1/PMS2 (MutL-α) and MLH1/MLH3 (MutL-γ) heterodimers function in MMR and are also involved in meiotic recombination (Baker et al., 1996; Lipkin et al., 2002; Cannavo et al., 2005). The role of MLH1/PMS1 (MutL-β) in metazoan MMR is an enigma (Räschle et al., 1999), although a meiotic phenotype is evident in pms1 (i.e. mlh2) mutants in yeast (Wang et al., 1999). In Nasonia, there are two copies of mlh1 and single copies of pms1 and pms2 (Fig. 1, Fig. S4E). While mlh1 and pms2 are also present in all other arthropods, pms1 was likely lost prior to the divergence of Coleoptera and Diptera (and possibly in Ixodes). mlh3 is only found in Ixodes, Daphnia and Pediculus, indicating that this gene was lost early in arthropod evolution.

Among mutS homologue heterodimers (reviewed by Jiricny, 2006), meiosis-specific MSH4/MSH5 recognizes Holliday junctions and stabilizes heteroduplex formation during meiotic crossing-over and recombination (Snowden et al., 2004). MSH4 has also been shown to interact with RAD51 and DMC1 in mammalian meiosis (Neyton et al., 2004). MSH2/MSH6 (MutSα) repairs short base-base mismatches and indels while MSH2/MSH3 (MutSβ) repairs longer mismatches. MSH6 also functions in repairing heteroduplex DNA during meiotic recombination in Drosophila (Radford et al., 2007) and MSH2 suppresses homeologous meiotic recombination in plants (Qin et al., 2002; Emmanuel et al., 2006). However, a role for MSH3 in meiosis has not been determined.

Homologues of msh2, msh4, msh5 and msh6 are present in all arthropods, except for meiosis-specific msh4 and msh5, which are absent in Drosophila (Fig. 1). For Nasonia, we manually annotated msh4 and msh5 because these genes have a very unusual structure: coding regions for each gene are separated by several hundred kilobases and many exons are located on different genomic scaffolds (Table S1). The mutS phylogeny shows that Nasonia msh4 and msh5 each form a close relationship with their respective orthologues in Apis (Fig. 4B). Despite the enormous physical distances between exons for these genes, ESTs have been identified for msh4 (GE426048, GE391424, GE391551, GE427889, FJ981699, FJ981700 and FJ981701) and msh5 (FJ981702) that demonstrate splicing occurs between some exons.

RecQ DNA helicases.  RecQ DNA helicase proteins function in homologous recombination regulation and in DNA repair and replication (reviewed by Wu & Hickson, 2006). In humans, RECQ2 suppresses meiotic homologous recombination by displacing RAD51 (Bugreev et al., 2007; Hu et al., 2007) and regulates DSB repair by promoting synthesis-dependent strand annealing (Adams et al., 2003; Wu & Hickson, 2003). RECQ4 and RECQ5 also suppress homologous recombination in some eukaryotes (e.g. Hartung et al., 2007; Hu et al., 2007), RECQ3 interacts with RAD51, RAD54 and RAD54B (Otterlei et al., 2006) and binds at Holliday junctions (Compton et al., 2008; Saunders et al., 2008) while RECQ1 interacts with various MMR complexes (Doherty et al., 2005).

RecQ1, recQ2, recQ4 and recQ5 are present in Nasonia and in all arthropods (except for recQ1 missing in Drosophila), while recQ3 is absent in Nasonia and in five other arthropods (Fig. 1, Fig. S5). Three copies of recQ2 are present in Nasonia and Acyrthosiphon, suggesting independent copy number expansions within these lineages.

II. Male-specific meiotic genes in Nasonia vitripennis and arthropods

While D. melanogaster is the only arthropod for which male meiosis has been extensively studied, it is distinct because recombination, synaptonemal complexes and chiasmata are absent (Hawley, 2002). While many well-characterized genes are upregulated and function only during meiosis/spermatogenesis in D. melanogaster males (e.g. Parisi et al., 2004), the phylogenetic distribution of these genes in arthropods has not been determined. Here, we search for homologues of 15 ‘male-specific meiotic genes’ (i.e. genes that are specific to male meiosis/spermatogenesis in D. melanogaster) in arthropods. We divided these genes into two categories based on their role in regulating the meiotic cell cycle and in achiasmate homologue pairing and segregation (Fig. 5A, Table S2).

Figure 5.

(A) Inventory of male-specific meiotic genes in arthropods. ‘+’ indicates the presence of a gene while ‘−’ indicates the gene was not found and is likely absent. Shaded gene names are specific to male meiosis in Drosophila melanogaster and asterisks indicate that the gene has a paralogue. The relationships in the arthropod phylogeny are based on the literature (e.g. Giribet et al., 2001; Grimaldi & Engel, 2005; Savard et al., 2006). aGenome assembly still in progress at the time this gene inventory was assembled. (B) Phylogeny of tgif paralogues based on an alignment of 129 amino acids. Parameter means: α= 0.97, pI = 0.072 and lnL =−3509.3. (C) Phylogeny of bol homologues based on an alignment of 123 amino acids. Parameter means: α= 1.64, pI = 0.08 and lnL =−2807.03. (D) Phylogeny of crl homologues based on an alignment of 137 amino acids. Parameter means: α= 0.54, pI = 0.23 and lnL =−1254.61. All trees are a consensus of 951 best trees and thickened branches refer to posterior probabilities from 0.95 to 1.0. Protein identifiers for Nasonia vitripennis sequences (in bold) are in Table S1.

1. Male-specific meiotic cell cycle genes

Meiotic arrest genes.  In Drosophila, most genes required for meiosis and spermatid differentiation are transcribed in primary spermatocytes prior to meiotic entry (Olivieri & Olivieri, 1965), although there is evidence for post-meiotic transcription of some loci (Barreau et al., 2008). Meiotic arrest genes regulate the switch in gene expression during this transition by controlling the transcription of genes that are essential for meiotic cell cycle progression and the onset of spermatid differentiation. In meiotic arrest gene mutants, spermatogenesis arrests at the G2/M transition due to the deficiency of transcripts that are required during and after meiosis. We searched for homologues of several aly- and can-class meiotic arrest genes, and we also included bol and crl, which are involved in meiotic cell cycle progression (Fig. 5A).

Aly-class meiotic arrest genes.  In Drosophila, aly-class genes (aly, comr, tomb, topi, vis and achi) regulate the transcription of genes required for both meiotic entry and spermatid differentiation (Table S2). Aly, Comr and Tomb (along with Topi and non-testis-specific Mip40 and Caf1/p55) comprise the testis-specific meiotic arrest complex (tMAC) in Drosophila (Beall et al., 2007). It has also been suggested that Vis and Achi (nearly identical TG-interacting factor (TGIF) homeodomain proteins) recruit Aly and Comr to chromatin to bind promoters in primary spermatocytes (Wang & Mann, 2003; Perezgasga et al., 2004; Jiang et al., 2007).

Among meiotic arrest genes encoding components of tMAC, aly, comr and tomb are present only in Drosophila (Fig. 5A). topi is comprised of multiple Zn-finger domains, which are also present as multiple, overlapping copies in other proteins. This complicates protein alignments and as a result, we could not distinguish topi orthologues from other related proteins. The aly phylogeny shows that the aly/mip130 gene duplication occurred within the Drosophila lineage or prior to its divergence from other dipterans (Fig. S6A). mip40 and caf1/p55, although not testis-specific, are present in Nasonia and other arthropod genomes (caf1/p55 is apparently absent in Culex) (Fig. S6B, C). Limited sequence conservation between tomb and its putative paralogue mip120 did not provide sufficient information for a phylogenetic analysis. A phylogeny of comr orthologues was not performed because a paralogue for this gene is not known.

Single copies of a TGIF-like gene were found in Nasonia and in all arthropods (Fig. 5B). The long branch for the Drosophila vis/achi duplication suggests rapid sequence divergence followed the gene duplication. By comparison, there are three TGIF paralogues in vertebrates: TGIF, TGIF2 and TGIFX (Mukherjee & Bürglin, 2007). Mammal-specific TGIFX is testis-specific (Wang & Zhang, 2004) and exhibits longer branches compared to other TGIF homologues (a similar pattern to vis and achi in Drosophila) (Fig. 5B).

Can-class meiotic arrest genes.  The five can-class genes (can, mia, nht, rye and sa) in Drosophila encode testis-specific TATA-binding protein associated factors (TAFs) (Table S2), which likely collaborate as a testis-specific protein complex in primary spermatocytes that regulates the expression of genes involved in spermatid differentiation (Lin et al., 1996; White-Cooper et al., 1998; Hiller et al., 2001, 2004). Homologues for the five testis-specific can-class meiotic arrest genes are specific to the Drosophila lineage (Fig. 5A). The phylogenies show that each can-class gene arose via duplication of taf genes in Drosophila, or prior to its divergence from other Dipterans (Table S2, Fig. S7A–E).

Boule and courtless.  The DAZ (deleted in Azoospermia) protein family (Haag, 2001) contains three members in animals: Boule, DAZL and DAZ (Xu et al., 2001). Boule is a testis-specific RNA-binding protein (Eberhart et al., 1996; Haag, 2001), except in nematodes where it is oocyte-specific (Karashima et al., 2000). In Drosophila, Boule (encoded by bol) is required for positive translational regulation of twine mRNA (Maines & Wasserman, 1999) (Table S2). We identified single copies of bol in Nasonia and in the other 10 arthropods (Fig. 5A), and the phylogeny shows that arthropod bol orthologues are distinct from the vertebrate-specific DAZL and primate-specific DAZ duplications (Fig. 5C).

The courtless (crl) gene, also called ubiquitin conjugating enzyme E2G in animals, encodes two proteins that share a common UBC domain but differ in their carboxy-terminal sequences (Orgad et al., 2000). In crl mutant males, the meiotic cycle is absent during spermatogenesis and abnormal sperm are produced. Homologues for crl were found in Nasonia and in the other 10 arthropods (Fig. 5A, D).

Achiasmate chromosome pairing genes.

Snm, mnm and tef.  The accurate paring and segregation of homologue pairs during achiasmate male meiosis in Drosophila partly depends on SNM (Stromalin in Meiosis; paralogue of the SA cohesin subunit), MNM (mod(Mdg4) in Meiosis) and Tef (Teflon) proteins, which likely comprise an autosomal homologue pairing complex (Table S2) (Tomkiel et al., 2001; Thomas et al., 2005; Arya et al., 2006). Homologues for snm (discussed above, and see Fig. S3B) and tef were only identified in Drosophila (Fig. 5A). mnm, encoded by mod(mdg4), is one of 31 alternatively spliced proteins produced by this locus (Dorn & Loewendorf, 2001; Soltani-Bejnood et al., 2007), which made us unable to distinguish mnm orthologues among the other Mod(mdg4) proteins (Fig. 5A).

Australin and borealin.  Borealin (called borealin-related, or borr in Drosophila) is a component of the multi-subunit chromosome passenger complex (CPC) (reviewed by Ruchaud et al., 2007). The CPC [comprised of borealin, aurora B kinase, inner centromere protein (INCENP) and survivin] regulates several processes during mitosis and meiosis (e.g. chiasmata resolution, centromeric cohesion and kinetochore/microtubule interactions). During meiosis in male Drosophila, Borr is replaced by its meiosis-specific paralogue Australin (encoded by aust) (Gao et al., 2008). Homologues of borr are present in Nasonia and in all arthropods, but aust is specific to Drosophila (Fig. 5A, Fig. S7F). The phylogeny also shows an independent duplication within the mosquitoes, however, the functions of these genes within the Culicidae have not been determined.


For the parasitoid wasp N. vitripennis, meiosis is essential for producing haploid eggs during oogenesis. However, spermatogenesis is ameiotic; a single equational division produces haploid sperm. Therefore, while certain meiotic genes might be dispensable during spermatogenesis, we still expect the majority of these genes to be present in the Nasonia genome to accommodate canonical female meiosis. While the presence of meiotic gene homologues in diverse arthropods does not directly assess the function during meiosis for these genes, this exhaustive phylogenomic survey provides insight into the patterns of gene loss, gain, duplication and subfunctionalization of meiotic genes during arthropod evolution.

Meiosis-related genes

We identified homologues encoding 39 meiosis-related genes (which include homologues for seven genes that are meiosis-specific in other eukaryotes) and several non-meiotic paralogues in the N. vitripennis genome and provided a comparative meiotic gene inventory in 10 additional arthropod genomes (Fig. 1, Fig. S8). For the hymenopterans Nasonia and Apis, the meiotic gene inventory was virtually identical (except for duplications of mlh1 and recQ2 in Nasonia and the absence of rad51c in Apis) and the presence of most genes was shared with the majority of arthropods. The sporadic absence of meiotic genes in different lineages suggests that certain genes are dispensable for meiosis; in such cases, other genes might be co-opted to fulfill the roles of these missing genes. Notably, 10 meiosis-related genes in our inventory have been lost in D. melanogaster, suggesting that Drosophila has a highly derived meiotic machinery that has recruited novel genes to carry out meiosis in females and males (see below).

All of the genes predicted to be involved in cell cycle regulation and sister chromatid cohesion are present in Nasonia and in most arthropods. Among these genes, Nasonia and Apis share an ancestral smc1 tandem duplication that resulted in highly divergent (i.e. long branch) smc1 paralogues in both species. The lack of ESTs only for these divergent paralogues suggests that they might represent pseudogenes; further expression and functional studies will address this question. However, meiosis-specific cortex and rec8 are absent in some lineages. cortex arose via a cdc20 duplication, but the evolutionary timing of this duplication during arthropod evolution was not resolved. rec8 is present in most animals, fungi and plants, but it has not been identified in protists (although rad21 is present in many protists) (Ramesh et al., 2005; Malik et al., 2008). This suggests that the rad21/rec8 duplication either occurred early in eukaryotic evolution (followed by rec8 loss in protists and some other lineages) or in an ancestor of animals, plants and fungi. The long branch lengths for rec8 in the phylogeny (Fig. 2B) suggest significant sequence divergence relative to its paralogue rad21, which might reflect our inability to identify rec8 homologues. Notably, the Drosophila rec8 homologue (called c(2)m) encodes for a synaptonemal complex protein (Manheim & McKim, 2003). Only functional studies can determine whether other arthropod rec8 homologues have similarly gained novel functions or retained the ancestral role in cohesin.

Homologues for meiotic genes involved in homologous recombination in model eukaryotes revealed greater variation in their distribution among arthropods. While dmc1 is the only meiosis-specific gene absent in Nasonia, it is also absent in other meiotic eukaryotes. One explanation for this is that RAD51, or other RAD51 paralogues, might substitute for DMC1 during meiosis. For example, overexpression of rad51 in yeast can partially suppress the meiotic defects in dmc1 mutants (Tsubouchi & Roeder, 2003). In Drosophila, rad51c (spn-D) and xrcc2 (CG6318) are expressed exclusively in the female germline (Staeva-Vieira et al., 2003), and rad51c mutants exhibit similar phenotypes to rad51 mutants, suggesting a function for rad51c in homologous recombination (Abdu et al., 2003). However, the sequences of these rad51 paralogues are known to be very divergent compared to dmc1 and have likely lost their RAD51-mediated strand exchange abilities (Radford & Sekelsky, 2004).

The well-characterized interactions between DMC1 and the HOP2/MND1 heterodimer leads to the prediction that if one of these genes is lost, the other two genes should also be absent (like in Drosophila and Anopheles). However, in Nasonia and five other arthropods (Fig. 1), the presence of hop2 and mnd1 is accompanied by the absence of dmc1. The only non-arthropod example of this pattern is for the microsporidian fungus Encephalitozoon (Ramesh et al., 2005), for which little is known about meiosis. While RAD51 (or a RAD51 paralogue) could plausibly interact with MND1 and HOP2 in the absence of DMC1 (Tsubouchi & Roeder, 2003), this has not been demonstrated in vivo.

For meiosis-specific msh4 and msh5 in Nasonia, ESTs demonstrate that these genes are expressed and that splicing occurs, despite the extremely large introns predicted for these genes. While large introns are unusual among arthropods, mega-introns (some >3 Mb) have been described in Drosophila (Reugels et al., 2000), some of which might be removed by recursive splicing (Burnette et al., 2005). Expression analyses of msh4 and msh5 in Nasonia and detailed examinations of transcript sequences will help to elucidate the significance of this result.

Are there genes that are specific to male meiosis?

Among the 15 genes examined that are testis-specific in D. melanogaster, only homologues for bol and crl are present in Nasonia and in other arthropods. While TGIF-like genes were also identified in these arthropods, the testis-specific achi/vis duplication is specific to Drosophila; expression and functional studies can determine whether TGIF-like genes in other arthropods are also specific to spermatogenesis.

The fact that haploid males in Nasonia and Apis lack canonical meiosis is consistent with the absence of 13 out of 15 male-specific meiotic genes in these taxa. However, these same 13 genes are also absent in the other arthropods that have diploid males and (where studied) normal meiosis in males. The specificity of most male-specific meiotic genes to Drosophila suggests that Drosophila has a derived meiotic machinery relative to other arthropods. For example, snm, mnm, aust and tef are only present in Drosophila and function in achiasmate homologue pairing, which is limited to Drosophila among the arthropods here. However, functional homologues of male-specific meiotic genes might be present in other lineages but they have yet to been identified. Testis-specific expression assays will identify transcripts that might be involved during spermatogenesis in Nasonia.

Why are two ‘male-specific’ meiotic genes (bol and crl) present in hymenopterans that do not have canonical meiosis in males? One explanation is that they might function during spermatogenesis (e.g. during cell cycle regulation and spermatid differentiation, as demonstrated in D. melanogaster) in Nasonia and Apis despite the absence of canonical meiosis. An alternative possibility is that bol and crl have additional non-meiotic roles. In Drosophila males, bol has also been implicated as a negative regulator of developmental axon pruning (Hoopfer et al., 2008) and crl is also involved in central nervous system development and courtship behaviour (Orgad et al., 2000). Gene expression profiles during spermatogenesis and oogenesis in Nasonia and in other arthropods will ultimately provide evidence for male- and female-specific genes.

Impact of gene duplications on male meiosis

Most of the 13 male-specific meiotic genes that are specific to Drosophila arose via gene duplication from ancestral genes with more general functions during mitosis and meiosis. Indeed, studies of the Drosophila sperm proteome reveal that gene duplications are a dynamic force in the creation and functional divergence of testis-specific genes (Betrán et al., 2002; Dorus et al., 2008). For Drosophila, genes with sex-biased (especially male-biased) expression patterns generally exhibit faster rates of evolution than non-sex-specific genes (Zhang et al., 2004; Ellegren & Parsch, 2007; Haerty et al., 2007) and are frequently the targets of strong positive selection (Pröschel et al., 2006). However, rapid evolutionary rates can also hinder the identification of gene homologues in different lineages. Only when genes specific to oogenesis and spermatogenesis are characterized in more diverse arthropods will we begin to understand the scope of the impact of gene duplications and the selective pressures that give rise to these genes.

Meiotic genes and parthenogenesis

Among the arthropods in our survey, arrhenotokous parthenogenesis is only found in Nasonia and Apis. However, cyclical parthenogenesis (the alternation between bouts of clonal and sexual reproduction) is characteristic of Daphnia and Acyrthosiphon. An unprecedented number of meiotic gene duplications was previously reported in D. pulex (Schurko et al., 2009), and here we show that duplicated meiotic genes are also characteristic of Acyrthosiphon (seven duplicated genes in total). The extent of the duplications in Acyrthosiphon is less than in Daphnia, but still greater when compared to the other arthropods examined. Determining whether duplicated meiotic genes are generally characteristic of the genomes of cyclical parthenogens or, rather, a manifestation of cyclical parthenogenesis will make an important contribution towards understanding the evolution and cytological manifestations of clonal reproduction.

In contrast, meiotic gene duplications are not characteristic of the genomes of arrhenotokous parthenogens; the majority of meiotic genes are present as single copies in Nasonia and Apis, and there were no unique gene losses or gains compared to obligate sexuals. Notably, the Apis genome is characterized by a high meiotic recombination rate (compared with other animals, including Nasonia) (Beye et al., 2006), but our study suggests that this feature cannot be rationalized by an expansion in the copy number of genes involved in meiotic recombination.


Although meiosis has occasionally been modified throughout eukaryotic evolution (both between the sexes and among lineages), its primary function of producing haploid gametes has been maintained. While meiosis is characterized by a core meiotic machinery, the maintenance of particular genes encoding these components tends to vary across eukaryotes. Our phylogenomic inventory reveals that meiotic gene losses and duplications have occurred frequently throughout arthropod evolution. Because all arthropods in this analysis are meiotic and sexual, our inventory supports the idea that the sporadic loss of meiotic genes does not imply an organism is ameiotic and/or asexual. In some lineages, certain meiotic genes are dispensable for meiosis, presumably because other genes can compensate for their loss. However, the presence of meiotic gene homologues does not directly assess gene functions and subsequent genetic studies are necessary to determine the roles of these genes in meiosis. Genes that are specific to male meiosis in Drosophila are largely confined to the Drosophila lineage, although it is likely that male-specific meiotic genes have similarly arisen independently in different lineages but have not yet been identified. Characterizing the transcriptomes of different taxa during spermatogenesis and oogenesis will identify such sex-specific genes and shed light on the evolution and function of male and female meiosis across metazoans.

Experimental procedures

Database mining for meiotic genes

We searched the N. vitripennis genome (Werren et al., 2010) for homologues of meiosis-related genes (Table S1, Fig. 1). To identify homologues, protein sequences from Apis and other arthropods were used as queries in BLASTP and TBLASTN searches against the Nasonia genome. Protein sequences were retrieved for genes with automated predictions. Genes without predicted models were manually annotated using known arthropod genes and proteins to determine start and stop codons and splice junctions. Similarly, we searched 10 additional arthropod genomes for these same meiosis-related genes; these genomes included representatives from Diptera (D. melanogaster, Ae. aegypti, An. gambiae and C. pipiens), Coleoptera (T. castaneum), Hymenoptera (Ap. mellifera), Hemiptera (A. pisum), Phthiraptera (P. humanus), Crustacea (D. pulex) and Chelicerata (I. scapularis) (see representative arthropod phylogeny in Fig. 1). Sequences were retrieved from the protein and whole genome shotgun databases at NCBI (, and from individual genome project databases (Tribolium, Apis and Acyrthosiphon at and; Daphnia at

For male-specific meiotic genes, we used QueryBuilder at FLYBASE ( to identify genes that are specific to male meiosis/spermatogenesis in Drosophila. From an initial set of 360 candidate genes, we examined the gene descriptions and searched the literature to exclude genes that are also involved in female meiosis and major functions unrelated to meiosis. Using these criteria, our inventory consisted of 15 male-specific meiosis/spermatogenesis genes for which we searched for homologues in other arthropods, as described above (Tables S1, S2, Fig. 5A). For some of these genes, we also searched the genomes of 11 additional Drosophila species (Drosophila ananassae, Drosophila erecta, Drosophila grimshawi, Drosophila mojavensis, Drosophila persimilis, Drosophila pseudoobscura, Drosophila sechellia, Drosophila simulans, Drosophila virilis, Drosophila willistoni and Drosophila yakuba) at NCBI.

Phylogenetic analysis

For each gene, amino acid alignments including homologues from arthropods and other eukaryotes were constructed using Clustal X (Chenna et al., 2003) and edited manually using MACCLADE 4.08 (Maddison & Maddison, 2006). Bayesian phylogenetic analyses were performed with MrBayes 3.1.2 (Ronquist & Huelsenbeck, 2003) using the WAG substitution model (Whelan & Goldman, 2001) and eight gamma-distributed plus one invariable rate heterogeneity categories. The analysis was run for 106 generations using four Markov chains (one heated and three cold) and trees were sampled every 1000 generations. Trees prior to the burnin were discarded from the final consensus tree construction, which was edited with NJplot (Perrière & Gouy, 1996).


We thank Jack Werren and Juergen Gadau for providing wasp stocks and genomic DNA, Andrew Spracklen for assistance with gene annotation and three anonymous reviewers for helpful comments on this manuscript. Sequencing of the N. vitripennis genome was performed by the Human Genome Sequencing Center at the Baylor College of Medicine (BCM-HGSC) with funding provided by the National Human Genome Research Institute (NHGRI U54 HG003273) and National Institutes of Health (NIH). We also acknowledge the genome projects for Ixodes scapularis (The J. Craig Venter Institute (JCVI) and The Broad Institute), Pediculus humanus (JCVI) and Acyrthosiphon pisum (BCM-HGSC) for making their data available. Research in JML's laboratory studying the evolution of meiotic genes is supported by grants from the NIH (GM079484) and Carver Trust.

Conflicts of interest

The authors have not declared any conflicts of interest.