Analysis of meiosis and cell cycle genes of the facultatively asexual pea aphid, Acyrthosiphon pisum(Hemiptera: Aphididae)

Authors


D. Srinivasan, Department of Ecology and Evolutionary Biology, Princeton University, 106A Guyot Hall, Washington Road, Princeton, NJ 08544, USA. Tel.: + 1 609 2586685; fax: + 1 609 2587138; e-mail: dsriniva@princeton.edu

Abstract

Phenotypic plasticity in response to environmental change is a common phenomenon, yet is poorly understood at the genetic and molecular level. Aphids exhibit a reproductive plasticity whereby seasonal changes result in asexual or sexual reproduction. To investigate the genetic basis of this reproductive plasticity, we assessed the meiosis and cell cycle gene repertoire in the genome of the pea aphid, Acyrthosiphon pisum. Aphids possess meiotic recombination genes and G1-to-S phase transition regulatory genes in gene copy numbers similar to other metazoans. However, mitotic and meiotic regulatory genes have duplicated, and several paralogues exhibit differential expression between reproductive morphs. Together, this suggests that cell cycle plasticity may be important in the evolution and mechanism of aphid reproductive plasticity.

Introduction

All organisms display phenotypic plasticity in response to changing environmental conditions (Price et al., 2003). In dramatic cases, such as polyphenisms, the same or similar genomes express alternate discrete phenotypes depending on environmental conditions (Nijhout, 2003). Phenotype plasticity is important because it allows organisms to adapt to changing or variable environments on a shorter time scale than by evolution (Moran, 1992). Despite decades of study, the genetic and molecular basis of polyphenisms remains unclear (Stearns, 1989; Nijhout, 2003).

The pea aphid, Acyrthosiphon pisum, exhibits a reproductive polyphenism that is an evolved adaptation to predictable changes in temperature and day length. In the spring, a female foundress hatches from an overwintering egg and only produces daughter clones by apomictic parthenogenesis (Blackman, 1987). These ‘asexual’ daughters are genetically identical to each other and their mother, apart from spontaneous mutation (Lushai et al., 2003), and reproduce parthenogenetically in spring/summer (long daylight, short darkness). Autumn conditions (short daylight, long darkness) induce asexuals to produce sexual males and sexual females via parthenogenesis. These sexuals create haploid gametes via meiosis, and, upon mating, the females lay overwintering diploid eggs that hatch in the spring. This polyphenism is remarkable because the same genome expresses both parthenogenesis and meiosis.

The evolution of aphid facultative asexuality may entail plasticity in the cell cycle and meiosis. We briefly describe relevant events of the cell cycle and meiosis here. The typical eukaryotic cell cycle is divided into four unique phases: G1, S, G2 and M (Nurse, 2000). During G1, the cell decides whether to enter S phase and irreversibly commit to division. During S phase, the genome and centrosome duplicate. During G2, genome integrity is confirmed before entering M phase. During M phase, the mother cell divides (via mitosis or meiosis) and the duplicated genome and centrosomes segregate equally between daughter cells. Transitions between cell cycle phases are tightly regulated by multiple conserved regulatory genes to ensure timely progression and unidirectionality.

Meiosis is a modified form of mitosis that occurs only in germ cells. The decision to enter into meiosis occurs by premeiotic S phase (Carpenter, 1994; Baltus et al., 2006; Harigaya & Yamamoto, 2007). Early in meiotic M phase (prophase), conserved proteins create double-strand breaks (DSBs) in the nuclear genome (Keeney & Neale, 2006). Processed DSB ends then initiate recombination preferentially between homologous chromosomes (Gerton & Hawley, 2005). The synaptonemal protein complex stabilizes this homologous association (synapsis) and facilitates recombination (Page & Hawley, 2004). A successful recombination event results in a crossover between homologous chromosomes that ensures homologue disjunction in the first meiotic division (MI). Sister chromatids segregate only during the second meiotic division (MII) to create haploid daughter cells (Nasmyth, 2001). No intervening DNA replication occurs between the two meiotic divisions, unlike in mitotic cell cycles (Marston & Amon, 2004).

Despite structural similarities between asexual and sexual aphid ovaries, important differences occur in meiosis and the fate of presumptive oocytes (Büning, 1985; Miura et al., 2003). Aphid sexual oocytes undergo chromosome pairing, synapsis and meiotic recombination and, upon fertilization, complete two meiotic divisions. However, asexual presumptive oocytes omit chromosome pairing, synapsis and meiotic recombination and divide only once, resulting in a discarded polar body and diploid one-cell parthenogenetic embryo that begins embryogenesis (Blackman, 1978). Thus, the reproductive polyphenism includes a plasticity in meiosis and germ cell fate that may be reflected at the genetic level.

We assessed the repertoire of meiosis and cell cycle genes in the recently sequenced pea aphid genome (S. Richards et al. unpublished data) to characterize the genetic basis of the reproductive polyphenism. Here, we identify conserved aphid meiosis and cell cycle genes, describe duplications of several key cell cycle regulatory genes and provide evidence for differential paralogue expression between reproductive morphs. Our results suggest a role for differential expression of paralogous cell cycle regulators in the aphid reproductive polyphenism.

Results and discussion

Meiosis and the cell cycle appear modified in the aphid reproductive polyphenism. Specifically, after omitting chromosome pairing and meiotic recombination, the asexual oocyte divides once and initiates embryonic mitotic divisions using centrosomes formed de novo. This modification may entail plasticity in meiosis, the meiotic cell cycle and mitotic cell cycle. We examined the genetic basis for meiotic and cell cycle plasticity, and we describe the function and characterization of genes involved in each process below.

Genes involved in meiosis are present in single copies

Many essential meiosis genes are deeply conserved from yeast to humans, and a minimal set of meiosis genes has been described (Ramesh et al., 2005). To assess the genetic basis of aphid reproductive plasticity, we performed BLAST searches of this set of deeply conserved meiosis proteins against the pea aphid genome sequence and gene predictions (S. Richards et al., unpublished data). We found single copies of almost all meiosis genes conserved among Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila melanogaster and Homo sapiens in the current pea aphid genome release (Table 1). Almost all pea aphid meiosis genes have full-length gene models and expressed sequence tag (EST) and/or experimental support suggesting these are functional genes (Table 1 and D. Srinivasan and D. Stern, unpublished data). Together, these results indicate that the meiotic recombination machinery appears intact in aphids.

Table 1.  The pea aphid genome contains single copies of genes involved in meiosis
GeneRole in meiosisNCBI protein accessionAssociated aphid ESTsInsect copy number
  • *

    Possible lineage-specific gene losses.

  • No NCBI RefSeq gene model exists for this locus.

  • Candidate Rec8 homologue identified in Nasonia vitripennis.

  • DSB, double-strand break; EST, expressed sequence tag.

Spo11DSB formationACH4818011
Mre11DSB formationXP_00194680762
Rad50DSB formationXP_00194855911
Rad54RecombinationXP_00194802321
Rad54BRecombinationXP_00194346251
Rad51RecombinationXP_00194889381
Rad51CRecombinationXP_00194461801
Dmc1*Recombination0–1
Hop2RecombinationXP_00194969231
Mnd1RecombinationXP_00194901671
Msh4Recombination controlXP_00194383410–1
Msh5Recombination controlXP_001945239110–1
Mus81Recombination controlXP_00194688511
Eme1Recombination control31
Mlh1Crossover formationXP_00194501121–2
Msh2Crossover formationXP_00194999191
Msh6Crossover formationXP_00194611861
Pms1Crossover formationXP_00194811941
Pms2Crossover formationXP_001947410101
MEI9/XPFCrossover formationXP_00194719891
Mer3Crossover formationXP_00194460401–2
Rec8*Meiotic cohesion0–1

Aphid meiosis gene copy number is consistent with that of other sequenced insects, although cases of several lineage-specific gene losses exist. For example, pea aphids possess genes that act solely in meiotic recombination and crossover control but are not found in the Drosophila or Caenorhabditis elegans genomes, such as Hop2, Mnd1, Msh4 and Msh5. Msh4-Msh5 activity promotes crossovers that exhibit positional interference with one another, unlike those promoted by Mus81-Eme1 activity (Lynn et al., 2007). Drosophila lacks Msh4-Msh5 but, similar to the pea aphid, retains the Mus81-Eme1 protein complex. The presence of both Msh4-Msh5 and Mus81-Eme1 in aphids but not in Drosophila may imply significant mechanistic differences in crossover formation between these species.

We observed several other notable meiosis gene absences in the current aphid genome. For instance, Rec8 is a meiosis-specific cohesin that promotes chromosome pairing and proper chromosome disjunction in meiosis (Revenkova & Jessberger, 2006) but is absent in the pea aphid genome. Other sequenced insects do not possess Rec8 homologues, with the possible exception of Nasonia vitripennis (GenBank accession ACR67094). However, C. elegans retains one copy (Pasierbek et al., 2001), and the water flea Daphnia pulex possesses three copies of Rec8 (Lynch et al., 2008), suggesting a loss or rapid evolution of Rec8 among insects but not among ecdysozoans. In contrast, we identified clear orthologues of the deeply conserved Rad21 mitotic cohesin in all insect genomes (Table 3) (Revenkova & Jessberger, 2006). This suggests Rad21 cohesin might function in both mitosis and meiosis in aphids or there remain unidentified meiosis-specific cohesins in the genome. The absence of Rec8 in aphids could affect homologous chromosome pairing and could help explain the observed transient and minimal chromosome pairing in asexual aphid oocytes (Blackman, 1978).

Table 3.  The pea aphid genome contains multiple copies of mitotic regulatory genes Thumbnail image of

Several other meiosis-specific proteins that act in homologous chromosome pairing, synapsis and crossover formation are absent. These include components of the synaptonemal complex (SCP1, SCP2, SCP3, SYCE1, SYCE2, C(2)M, C(3)G and ord) (Bickel et al., 1996; Revenkova & Jessberger, 2006). We were unable to identify aphid homologues, either due to inadequate sequence coverage, rapid evolution or gene loss. However, we did identify structural maintenance of chromosome (SMC) protein family members which are required for chromosome condensation, cohesion and architecture (Haering & Nasmyth, 2003). While other insects typically possess one copy of each family member, we identified two copies of SMC3 and three copies of SMC2 and SMC6, as well as single copies of SMC1, SMC4 and SMC5 (S. Richards et al. unpublished data, and data not shown). Future experiments will determine if these additional SMC paralogues have acquired additional functions or are differentially expressed.

Taken together, simple loss or expansion of the core meiosis gene set cannot explain the modified meiosis in asexual aphids. Genes involved in the decision to enter meiosis would therefore be interesting to investigate. For example, in C. elegans, a GLP-1/Notch signal maintains the mitotic state of germ cells and inhibits entry into meiosis (Hansen & Schedl, 2006). In Drosophila, several intrinsic factors such as BicaudalD and Egalitarian specify the oocyte fate and allow meiotic entry (Huynh & St Johnston, 2000). In mice, a retinoic acid signal stimulates meiotic entry in germ cells via the vertebrate-specific RNA-binding protein Stra8 (Baltus et al., 2006; Anderson et al., 2008). However, in general few such inductive signals are known in other organisms. Mechanisms underlying entry into meiosis may be unique to reproductive strategies and across taxa.

Copy numbers of G1-S phase transition genes are similar among insects

During G1, several conserved pathways determine if the cell is ready to commit to DNA replication and cell division. Principal among them, the retinoblastoma protein family (Rbf) members block and repress the activity of the E2F/Dp transcriptional regulators to prevent progression into S phase (van den Heuvel & Dyson, 2008). Two Cyclin/Cyclin-dependent kinase (Cdk) complexes relieve this repression and drive G1-S phase progression: Cyclin D with Cdk4 or Cdk6 which act in G1, and Cyclins E or A with Cdk2 which act in G1 and S phases. The p21/p27 Cdk inhibitor proteins further limit G1-S phase progression by inhibiting G1/S Cyclin/Cdk activity. Other G1/S phase regulators include Geminin and Cdt1 which, respectively, inhibit untimely and promote timely DNA replication (Luo & Kessel, 2004). Once the G1-S transition has occurred, E2F/Dp then promotes transcription of several genes required or rate limiting for DNA replication and S phase progression: ribonucleotide reductase (RNR) subunits, DNA polymerase alpha, dihydrofolate reductase (DHFR), maintenance of minichromosomes (MCM) proteins and Cyclin E (Duronio et al., 1998; Leone et al., 1998; van den Heuvel & Dyson, 2008). Together, these multiple regulatory pathways prevent inappropriate commitment to cell division.

In the pea aphid, we found G1-S phase progression genes at copy numbers similar to other insects (Table 2). Among Cdks, the pea aphid genome contains one copy each of nearly all metazoan Cdks (Cdk2, Cdk4/6, Cdk5, Cdk7, Cdk8 and Cdk10) (Fig. 1B), although only Cdk4/6 and Cdk2 drive cell cycle progression (Morgan, 1997). Most insect genomes contain only one copy of Cdk9, while the pea aphid contains two copies and Nasonia vitripennis possesses five Cdk9 paralogues. Cyclins D and E as well as E2F family members are present in pea aphids in copy numbers similar to those of other sequenced insects. While aphids and other arthropods possess only one Rbf gene, Drosophila possesses two Rbf family members. This suggests that the Drosophila model for Rbf function, whereby two Rbf family members repress E2F-mediated transcription by different mechanisms (Stevaux et al., 2002), is not a conserved pathway but instead is derived among Drosophilids.

Table 2.  The pea aphid genome contains largely single copies of G1-S phase genes
GeneNCBI protein accessionAssociated aphid ESTsInsect copy number
  • *

    Nasonia vitripennis contains five candidate Cdk9 genes.

  • Genome sequence data contains locus information for two alleles.

  • Culex quinquefasciatus contains six candidate Cdt1 genes.

  • EST, expressed sequence tag.

Cdk2XP_00194978671
Cdk4/6XP_001947810141
Cdk5XP_00194594631
Cdk7XP_001947328311
Cdk8XP_001950232151
Cdk9.1XP_001950345381*
Cdk9.2XP_0019488590
Cdk10XP_001952723181–2
Cyclin D1XP_00195235051–3
Cyclin D2XP_0019477879
Cyclin EXP_001945139201
p21XP_001947107221
E2F1XP_00194724231–2
E2F2XP_00194522813
RbfXP_00194309451–2
MCM2XP_00194488631
MCM3XP_00194369241
MCM4XP_00195016971–2
MCM5XP_001950445151
MCM6XP_001944682, XP_001945957101
MCM7XP_001952343131
MCM8XP_00194813221
MCM9XP_00194846700–1
GemininXP_00194810951–4
XP_00195124523
XP_0019522222
Cdt1XP_00194842921
XP_0019496718
RNR M1XP_001942710281–3
XP_00195065020
XP_00194942216
XP_00194937623
XP_00194467124
RNR M2XP_001944536341–2
XP_0019448970
Figure 1.

Cdk1 paralogues in the pea aphid genome. (A) Alignment of pea aphid (Acyrthosiphon pisum) Cdk1 paralogues shows deep conservation with the Drosophila, Homo sapiens, starfish Asterina pectinifera and thale cress Arabidopsis thaliana orthologues. Shaded residues indicate deep conservation while asterisks denote aphid-specific changes in conserved residues. Boxed residues 14–15 (threonine and tyrosine) are the conserved targets of Wee1 kinases and Cdc25 phosphatases. (B) Neighbour-joining tree of aphid Cdks shows clear distinctions among Cdk1, Cdk2 and other Cdks among the Schizosaccharomyces pombe, H. sapiens, Drosophila melanogaster, Aedes aegyptii and Tribolium castaneum orthologues. Scale bar indicates substitutions per site, and numbers indicate percentage of consensus support.

We next identified several downstream E2F target genes to determine if the Rbf/E2F pathway has expanded. We found that MCM family members are present in single copies in the pea aphid genome similar to other insects (Table 2). Furthermore, pea aphids and other insects retain one copy of Mcm9 while neither Drosophila nor C. elegans possess Mcm9 (Lutzmann et al., 2005). MCM proteins bind to origin of replication sites and are required for initiating DNA replication (Maiorano et al., 2006). We found single copies of other E2F target genes, such as Cyclin E, DHFR and DNA polymerase alpha (Table 2 and data not shown). In contrast, RNR subunit copy numbers have increased in pea aphids: five copies of the large subunit (one or two copies in other insects) and two copies of the small subunit (one copy in insects except Aedes aegyptii). Thus far, it is unclear if this variation in E2F target gene copy numbers is relevant to aphid meiotic and cell cycle plasticity.

In meiosis, no intervening DNA replication occurs between the two divisions. This raises the question of how an aphid oocyte can license and initiate DNA replication after division in asexuals but not in sexuals. In other organisms, Geminin is expressed from S to early M phase and temporally restrains Cdt1-dependent recruitment of the MCM complex to chromatin to only G1 and early S phases (Wohlschlegel et al., 2000). Proper replication timing therefore requires a balance between Geminin and Cdt1 activity (Saxena & Dutta, 2005). In keeping with this idea, we found that the pea aphid possesses two copies of Cdt1 and three copies of Geminin (Table 2). Other insects possess one copy of Cdt1 except for Culex quinquefasciatus (6 copies) and one copy of Geminin except for Culex (4 copies) and Aedes (2 copies). Together, these results indicate that the major G1-S control pathways largely have not expanded in aphids with the exception of Geminin and Cdt1. However, while the decision to commit to cell division remains intact, the expansion of the Geminin/Cdt1 pathway may correlate with DNA replication plasticity in aphid oocytes.

Important regulators of mitosis have expanded in aphids

Several pathways constitute a strict set of barriers to mitotic entry, centered on post-translational modifications of regulatory kinases and their downstream targets (Lindqvist et al., 2009). Principal among them is Cdk1/Cdc2, which, when activated by Cyclin A, B or B3, phosphorylates multiple proteins required for mitosis and meiosis (Ubersax et al., 2003). Wee1/Myt1 kinases and Cdc25 phosphatases negatively and positively regulate Cdk1 through phosphorylation and dephosphorylation, respectively (O'Farrell, 2001). Cks adaptor proteins can further modulate Cdk1 activity or specificity (Harper, 2001). Additionally, the Polo and Aurora kinase families control many important aspects of mitosis (Carmena & Earnshaw, 2003; Archambault & Glover, 2009).

Exit from mitosis into G1 involves several pathways. First, exit requires the proteolysis of key regulatory proteins (e.g. Cyclins B and B3) by the Anaphase Promoting Complex (APC) mediated by Cdc20/Fizzy and Cdh1/Fizzy-related (Thornton & Toczyski, 2006). Another major APC target is Securin, the Separase protease inhibitor that prevents untimely cohesin cleavage and chromosome segregation (Nasmyth, 2002). Additionally, Cdc14 phosphatase promotes mitotic exit by down-regulating Cyclin B/Cdk1 activity (Stegmeier & Amon, 2004). Together, these multiple regulatory pathways cooperate to ensure the fidelity of cell division (Nicklas, 1997).

Several aspects of G2 and M phases appear modified in asexual aphid oocytes. To determine the genetic basis for this modification, we identified and characterized several conserved G2/M and cell-division control genes (Table 3). Several cell cycle genes generally are similar in copy number to other insects. For example, we found single copies of APC subunits with the exceptions of Apc8, Apc10 and Apc11 duplications (data not shown). We identified one copy of Separase, incorrectly annotated as a Rad50 ATPase (Table 3), but not a candidate Securin. This absence of Securin may be due to assembly gaps or a lack of conservation of Securin in insects as seen for Drosophila (Jäger et al., 2001). Similarly, we observed single copies of Cyclins A, B and B3 as in other insects. In contrast, we found duplications of several key regulatory genes: Cdk1, Wee1, Cdc25, Cks, Polo and Aurora (Fig. 1, Table 3 and Figs S1–5). Below, we discuss the characterization of these key regulatory genes and their relevance to the polyphenism.

Cdk1 is the master regulator of mitosis and meiosis, and we expect its activity to be modified in asexual oocytes. We identified two aphid Cdk1 loci, which is unique among animals (Fig. 1A). Cdk1.1 and Cdk1.2 are located on different assembly scaffolds with different neighbouring sequences (EQ118846 and EQ111385, respectively), are identical at the protein level (Fig. 1A), and have several nucleotide differences in the coding region, introns and untranslated regions (data not shown). We also identified two alleles of Cdk1.1, designated alleles A and B. Cdk1.1A (XR_045879) is predicted to contain a single nucleotide insertion causing a frameshift and premature stop codon. However, this is likely a genome sequencing error since no ESTs or sequenced cDNAs contain this single base insertion (data not shown). Cdk1.1B is present on a 14 601 base pair (bp) scaffold (EQ115470) that may be a subset of the 52 614 bp Cdk1.1A scaffold (EQ118846), although it is possible that EQ115470 is a recent duplication. These Cdk1.1 alleles, however, have several coding sequence differences.

Aphid Cdk1 paralogues also show divergence in several residues conserved among metazoans and critical for function (Fig. 1A, asterisks). Metazoan Cdk1 and Cdk2 are similar in sequence often making it difficult to distinguish and correctly annotate them. Both contain the PSTAIR motif (EGVPSTAIREISLLKE) which binds to Cyclin subunits (Pavletich, 1999). However, aphid Cdk1.1 and Cdk1.2 possess unique PSTAIR sequences (EGIPATAIREISILKE) while aphid Cdk2 contains a different PSTAIR motif (EGVPSTAMREISLLKE). Clustering analysis shows that aphid Cdk2 clusters with other bona fide Cdk2 family members while the aphid Cdk1.1 and Cdk1.2 loci cluster with other Cdk1 orthologues (Fig. 1B). Unless there are compensatory changes in cyclin subunits, these different PSTAIR motifs in aphid Cdks may have consequences for Cdk activation.

The unprecedented presence of two Cdk1 loci raises the question of differential expression or function. To address this, we used primers that anneal to all three Cdk1 loci and sequenced the amplification products from cDNA derived from asexual or sexual aphids. We observed that 9 of 10 random asexual clones corresponded to Cdk1.1A, while 8 of 9 random sexual clones corresponded to Cdk1.2. No clones corresponded to Cdk1.1B. All available pea aphid ESTs derived from asexuals correspond to Cdk1.1A. Additionally, two clones from asexual cDNA correspond to a 386 bp alternative splice form of Cdk1.1A lacking 522 bases (exons 2, 3, 4 and part of exon 1, out of 6 coding exons) that would produce a protein with premature stop codons. While this assay is not quantitative, it suggests differential utilization of Cdk1 paralogues in the reproductive polyphenism.

In addition to the Cdk1 duplication, we observed a duplication of genes that regulate Cdk1 activity such as Wee1, Cdc25 and Cks. We identified three Wee1 kinase family members (Wee1.1, Wee1.2 and Wee1.3). One locus (Wee1-like) is distantly related to plant Wee1 family members by BLAST searches but clusters with neither Wee1 nor Myt1 (Fig. S1A, B). The predicted aphid Wee1 loci encode proteins of lengths similar to other Wee1 family members, and each locus is associated with at least one EST and empirical expression data (Table 3 and Fig. 2). In contrast to other arthropods, we did not identify an aphid Myt1 kinase (Fig. S2B) possibly because of either incomplete genome coverage or a lineage-specific loss. The Wee1-like locus may represent an early Wee1 duplication, an ancient horizontal gene transfer of a plant Wee1 gene or a distantly related Myt1 protein. This number of Wee1 kinases in aphids differs from other insects which typically possess one copy each of Wee1 and Myt1.

Figure 2.

Differential expression of Wee1 paralogues in sexual and asexual ovaries. Products were amplified from two different cDNA pools obtained from two different collections of sexual and asexual ovaries. No amplification product has been detected for Cdc25.3. GAPDH amplification is shown as a control for template amounts and loading.

Aphid Wee1 gene expansion is balanced by duplications of Cdc25. Cdc25 phosphatase counteracts Wee1/Myt1 inhibition of Cdk1 activity and thus promotes entry into mitosis (O'Farrell, 2001). We identified three aphid Cdc25 loci (Cdc25.1, Cdc25.2 and Cdc25.3) that are similar in length to metazoan Cdc25 proteins but are more similar to each other. These duplications are aphid-specific, and Cdc25.1 and Cdc25.2 have more recently duplicated (Fig. S2A, B). Both Cdc25.1 and Cdc25.2 loci have EST and/or experimental expression data (Table 3 and Fig. 2). However, the predicted Cdc25.3 locus (GeneID LOC100158885) is a partial gene model lacking a clear start codon and amino terminus. The true Cdc25.3 locus may consist of sequences located at the 3′ end of scaffold EQ126234 and the 5′ end of genomic scaffold EQ121782. EQ126234 contains an AUGUSTUS gene model (AUG4_SCAFFOLD15463.g2.t1) with high similarity to the amino terminus of the two aphid Cdc25 genes. We were unable to join these two gene models by PCR or amplify the conserved predicted exons of LOC100158885 from cDNA. The Cdc25.3 locus perhaps is incorrectly predicted, expressed only during a strict developmental period or in a small subset of cells, or encodes an unexpressed pseudogene.

The aphid-specific duplications of Cdc25, Wee1 and Cdk1 raise questions of their functions, expression patterns and involvement in the polyphenism. To address this, we assessed Wee1, Cdc25 and Cdk1 paralogue expression by PCR from sexual or asexual ovary cDNA. Wee1.1, Wee1.3 and Wee1-like were amplified more readily from sexual ovaries than asexual ovaries, indicating differential expression of these paralogues. However, Cdc25 and Cdk1 paralogue PCR product levels varied little between samples and reproductive morphs in this assay although subtle gene expression differences could exist.

Cks proteins are another layer of Cdk1 control. Cks proteins play several critical roles in regulating cell cycle phase transitions (Donovan & Reed, 2003). Cks proteins modulate Cdk1 substrate specificity, interact with Wee1, Cdc25 and the proteasome (Bourne et al., 1996; Patra & Dunphy, 1998; Kaiser et al., 1999), promote the destruction of the G1-S p27 Cdk inhibitor protein (Ganoth et al., 2001), and regulate transcription of several mitosis genes (Morris et al., 2003; Yu et al., 2005; Martinsson-Ahlzen et al., 2008). Importantly, mouse oocytes lacking Cks2 fail to enter anaphase of meiosis I (Spruck et al., 2003). The pea aphid genome contains three copies of Cks proteins (Table 3 and Fig. S3A): two (Cks2.1 and Cks2.2) related to the Drosophila Cks85A gene and one (Cks1) related to the Drosophila Cks30A and vertebrate Cks genes (Fig. S3B). Drosophila Cks30A, but not Cks85A, promotes progression through meiosis and early embryonic mitoses, similar to mammalian Cks2 (Swan et al., 2005). The additional aphid Cks85A paralogue could further regulate cell cycle progression through interactions with Cdc20, Cdh1, Wee1 kinases and/or Cdc25 phosphatases.

In addition to Cdk1, Polo kinases and Aurora kinases control multiple aspects of mitosis and meiosis. Most model organisms have both Aurora kinase subfamily members (Aurora A and Aurora B), and each is expressed during mitosis and meiosis and has distinct functions (Carmena & Earnshaw, 2003). Aurora A controls centrosome duplication and maturation, spindle assembly and events of the first meiotic division, while Aurora B regulates multiple aspects of chromosome behaviour in mitosis and meiosis. In the pea aphid genome, we identified one Aurora A and an aphid-specific duplication of Aurora B (Table 3, Fig. S4A, B). Similarly, Polo kinases have multiple roles in mitosis and meiosis, particularly promoting entry into and exit from mitosis and meiosis, spindle assembly and mitotic and meiotic chromosome segregation (Archambault & Glover, 2009). Another Polo family member, Polo-like kinase 4 (Plk4), is required for centrosome duplication (Bettencourt-Dias et al., 2005; Habedanck et al., 2005). Similar to other insects, we identified an aphid-specific duplication of Polo kinase (Polo kinase 1 and Polo kinase 2, Fig. S3A) and one Plk4 gene (Fig. S3B and data not shown). This is consistent with duplications of the other mitotic regulators, and additional work will help illuminate their functions and relevance to the reproductive polyphenism.

Together, our results suggest that additional control of cell division, especially Cdk1 activity, may be associated with the reproductive polyphenism. Our observation that Wee1.1, Wee1.3 and Wee1-like genes are expressed at lower levels in asexual ovarioles supports this hypothesis. However, neither Cyclin A, B nor B3 increased in copy number, suggesting that perhaps Cdk1/Cyclin targets are not different between reproductive morphs. What might this mean for Cdk1 control in the reproductive polyphenism? In Xenopus, the absence of Wee1 kinase activity promotes progression through meiosis (Nakajo et al., 2000). In contrast, the expression of Wee1 but not Cdc25c may promote meiotic arrest in mouse oocytes (Park et al., 2004). We speculate that with potentially less Wee1 activity, Cdk1 becomes active much earlier in asexual germ cells, helping to alleviate meiotic arrest and allow parthenogenesis. Additional uncharacterized modifications would be required to modify other aspects of asexual germ cell behavior.

Concluding remarks

Aphids are remarkable for their ability to express alternate discrete phenotypes given the same genome. In order for sexual organisms to evolve facultative asexuality, we assume at least three changes to occur: prevention of recombination, lack of homologue association and prevention of genome reduction (d'Erfurth et al., 2009). Evidence described here suggests the occurrence of these changes in pea aphids. Reduced recombination machinery activity may allow asexual aphids to prevent formation of crossovers necessary for the meiosis I division (D. Srinivasan and D. Stern, unpublished data). The absence of an aphid Rec8 homologue may limit homologue association in asexual aphids. Duplications and differential expression of cell cycle regulatory genes (Geminin, Cdt1, Cdk1, Wee1, Cdc25, Polo kinases, Aurora kinases, Cks proteins) may help alter the meiotic cell cycle program of asexual oocytes. Downstream targets of the expanded regulatory genes in the pea aphid should also be examined to determine if whole pathways have expanded or if plasticity is generally associated with changes in genes at the top of pathways. Further examination of expression patterns and functions in facultative asexuality will be important in understanding how aphids can exhibit plasticity in meiosis and sexual reproduction.

Experimental procedures

Identification of meiosis and cell cycle genes

We performed BLAST searches against the pea aphid genome raw traces, scaffolds (NCBI accessions EQ110773.1 to EQ133570.1), predicted RefSeq mRNAs (NCBI A. pisum mRNAs, accessions NM_001126134.1 to XM_001952843.1) and predicted RefSeq proteins (NCBI A. pisum proteins, accessions NP_001119606.1 to XP_001952878.1) at the NCBI site or Aphidbase (http://www.aphidbase.com). We used human, mouse, Drosophila, Saccharomyces cerevisiae and S. pombe homologues as query sequences for BLAST searches. Orthology was assigned when reciprocal best BLAST matches and the construction of phylogenetic trees confirmed relatedness. We generated protein alignments for phylogenetic trees using the MUSCLE algorithm in the Geneious software package (Biomatters), followed by generating maximum likelihood trees with bootstrapping with PHYML. In some instances, we also generated consensus phylogenetic trees using protein distance neighbour joining trees developed after bootstrapping the original dataset. Bayesian Inference methods implemented in MrBayes were used to construct the Wee1/Myt1 tree. Phylogenetic tree images were created in Geneious and further annotated in Macromedia Freehand.

Determination of allelism

Candidate aphid genes sharing at least 95% identity and located on different scaffolds were considered as allelic or paralogous. Four criteria were used to assign allelism: >95% nucleotide sequence identity between genomic sequences, synteny between scaffolds, size of scaffolds (e.g. alleles were often assigned to scaffolds less than 10 kilobases due to a failure to assemble more polymorphic flanking regions) and the only differences between scaffolds consisting of small insertions/deletions or single nucleotide polymorphisms. Allele names were assigned as ‘a’ or ‘b.’

Gene expression analysis

Approximately 10 sexual or asexual ovaries were dissected from third or fourth instar female nymphs in cold Dulbecco's PBS (Invitrogen, Carlsbad, CA, USA). Total RNA was isolated using the Rneasy Mini Kit (Qiagen, Valencia, CA, USA), treated with RQ1 DNase I (Promega, Madison, WI, USA) for 45 min at 37 °C to remove genomic DNA, heated to 65 °C in 2 mM EGTA to inactivate the enzyme and quantitated. One µg total RNA was used for cDNA synthesis using random primers (High Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Foster City, CA, USA). PCR was performed using cDNA template diluted 1:25 using GoTaq 2x Master Mix (Promega) (30 s denaturing at 95 °C, 30 s annealing at 56 °C and 1 min extension at 72 °C for 30 cycles). Primers used to amplify cDNA of mitosis genes by PCR are listed in the Supplemental Information.

Acknowledgements

D.G.S. was supported by a NIH NRSA Ruth L. Kirchstein Postdoctoral Fellowship (GM077928). B.F. was supported by funding from the Royal Society of Edinburgh. S. J.-P. was supported by funding from ANR Genoplante ‘Aphicible’.

Ancillary