Sexual reproduction involves two main events: meiosis and fertilization, and creates new genotypes by shuffling allelic combinations. Although the predominance of sexual reproduction in eukaryotes supports this innovation as a successful reproduction strategy, asexuality has evolved independently multiple times from sexual ancestors in almost all major taxa [1–4], such as in stick insects [3] and Ranunculus plants [4]. How asexuality has evolved in sexual organisms is unclear. In aphids, asexuality was acquired once about 250 million years ago by a common sexual ancestor [5]. Most aphid species alternate between sexual reproduction and asexual parthenogenetic reproduction according to seasonal variations. In spring and summer, aphids reproduce asexually by parthenogenesis and produce clonal parthenogenetic female progeny by viviparity. The autumnal shortening of the photoperiod induces the concentration of juvenile hormone (JH) to decrease in the aphid haemolymph., and particular form of parthenogenetic female called the sexuparae are produced. Sexuparae females produce sexual females and males that subsequently mate to produce overwintering eggs. Although parthenogenetic viviparous females and sexual oviparous females exhibit major differences in morphology and behavior, they share the same genome. This phenomenon, called reproductive polyphenism, is an example of aphid phenotypic plasticity [6].

The cellular and cytogenetic bases of reproductive polyphenism have been described for several aphid species [7, 8]. In aphids, three generations are represented within one viviparous female: the mature embryos developing inside the maternal abdomen carry the first developmental stage of the third generation. This phenomenon is known as the 'telescoping of generations' [9]. The embryonic developments of asexual and sexual females are similar until the formation of undifferentiated germaria, which takes place at the end of germ band retraction (stage 18) [8]. The 32 germ cells within the germaria enter into early meiotic prophase (zygotene) and a partial association between bivalents occurs [7]. After this stage, the embryonic development and maturation of germ cells diverge between sexual and asexual females. Ancestral sexual development occurs under short photoperiod: the basal cells of germaria progress through meiosis; the sexual oocyte displays chromosome pairing, synapsis and meiotic recombinations. Meiosis is blocked in first meiotic prophase until the egg is fertilized and laid. Under long photoperiod, sexual oogenesis is inhibited [7]. The asexual oocyte ovulates in the posterior part of the germaria and forms a new follicle where it skips the first meiotic division and undergoes a single maturation division resulting in a discarded polar body and a diploid clonal oocyte. Synchronous mitotic divisions follow immediately without fertilization to initiate early embryogenesis, which continues within the parthenogenetic female abdomen and progresses through a series of 20 developmental stages [8].

We identified genes regulated in sexual and asexual aphid oogenesis by combining transcriptomic analyses and in situ hybridizations. In aphids, sexual and asexual reproduction can be analyzed in individuals possessing identical genomes. Sequencing of the LSR1 pea aphid genome [10] and the recent development of new genomic resources offer an unique opportunity to understand the molecular bases of sexual and asexual reproduction. Our results demonstrate that sexual and asexual reproduction pathways require unique and shared genetic programs.

By coupling quantitative transcriptomic and qualitative WISH approaches, we identified 33 transcripts associated with the onset of divergent transcripts expression leading to the establishment of the sexual or asexual phenotype in aphids. The involvement of these candidate genes in determining the developmental direction of oogenesis is supported by: i) a strong signature of oogenesis among the regulated transcripts (7/33); and ii) the oocyte- and/or germarial-specific localizations of 16 of the transcripts. These data illustrate both shared and unique regulatory patterns of transcript expression for the two modes of reproduction.

First, we identified transcript expression profiles shared by sexual and asexual oogenesis. These transcripts are expressed in sexual and asexual germaria and/or oocytes but display differences in their expression levels and/or in their localization. In various organisms, gld2, bicC, pop2 and orb are all involved in regulating the poly(A) tail length of maternal mRNAs. This process is a common mechanism of translational control essential for progression of meiosis and axis patterning during oocyte maturation [12–14]. The regulation of poly(A) tail length of maternal mRNAs results from a balance between concomitant deadenylation (translation repression) and polyadenylation (translation activation) [15]. In Drosophila melanogaster, gld2, bicC and orb have been described as part of the cytoplasmic polyadenylation complex that polyadenylates mRNAs and activates their translation [13, 16]. Gld2-dependent mRNA polyadenylation and translational activation have been shown to be essential for oogenic meiosis [13]. On the other hand, mRNA deadenylation by the major deadenylation complex in Drosophila, CCR4-NOT, which includes the deadenylases CCR4, POP2, and four NOT proteins [17], promotes negative regulation of target mRNAs. Deadenylation of specific maternal mRNAs could be involved in the precise temporal and spatial mRNA localization necessary for axial patterning during oogenesis [12]. Expression of Api-gld2, Api-bicC, Api-pop2 and Api-orb in sexual and asexual germaria and/or in oocytes suggests that the balance between cytoplasmic polyadenylation and deadenylation in oocytes acts in the maturation of haploid as well as diploid oocytes in the pea aphid as it has been shown for other sexual organisms. Moreover, expression of the oocyte-specific linkers histoneH1 and uhrf1 in asexual and sexual oocytes suggests that epigenetic regulation is involved in pea aphid oogenesis. The linker histone H1 interacts with DNA to set-up and maintain the organization of the chromatin between the nucleosomes. Histone H1 proteins form a complex family of related proteins that have distinct species, tissue and developmental specificities and are involved in the epigenetic control of gene expression. The replacement of a somatic linker H1 by an oocyte specific one (H1oo) during oogenesis has been described in a wide range of species (reviewed in [18, 19]) and may promote a loosening in chromatin structure that appears to correlate with the initiation of meiosis [19]. The presence of Api-histoneH1 in germaria and oocytes suggests that this oocyte-specific variant of H1 might be involved in both sexual and asexual oogenesis in aphid. uhrf1 encodes an ubiquitin-like protein that is involved in two epigenetic silencing pathways through its regulation of DNA and histone methylation [20, 21]. Although, uhrf1 has not been implicated in any oogenesis process to date, our results suggest a role for this protein in asexual and sexual oogenesis in aphids. Altogether, our results suggest that epigenetic regulatory mechanisms and the balance between cytoplasmic mRNA polyadenylation and deadenylation may have roles in regulating haploid and diploid oocyte maturation. Moreover, as orb transcript is the earliest to be differentially expressed during the developmental time course, we propose that Orb might play a major role in determining oocyte fate.

Second, we identified one gene, lsd1, that is specifically expressed in pea aphid sexual germaria and oocytes. The lsd1 transcript was not significantly detected either in asexual embryos according to microarray data (Additional file 3) or in asexual ovaries according to WISH. lsd1 encodes a protein involved in the regulation of lipid droplet storage in eukaryotic cells [22]. Lipid droplets, like mature yolk bodies, are the major energy storage for oocytes and developing embryos in Drosophila[23]. Our results suggest that lsd1 is required only for sexual oogenesis during which it may be involved in yolk accumulation. The absence of lsd1 transcripts in asexual embryos is consistent with the absence of yolk in parthenogenetic oocytes [8]. Thus, identification of lsd1 expression as a specific feature of sexual oogenesis validates the accuracy of our approach. Despite the existence of genes known to be involved in meiotic recombination in the pea aphid genome [24], we identified no such genes as specific markers of sexual oocytes. However, germline expression of meiotic genes such as spo11 was shown to be similar in sexual and asexual aphids. Expression of this gene seems to be modulated between asexual and sexual aphids by alternative splicing [25].

Finally, we identified three transcripts that were specific to asexual oogenesis. No significant transcription signal has been observed for these genes either in sexual embryos using microarrays (Additional file 3) or in sexual ovaries using WISH. That the polypeptide encoded by ACYPI39770 does not bear any similarity to a protein or protein domain of any other organism, suggests that ACYPI39770 is either an orphan gene or a gene with an unusual accelerated evolution rate. In Drosophila, cyclin J and lodestar are involved in sexual oogenesis and early embryogenesis [26, 27]. In aphids, their expression profiles suggest that they may haveroles in asexual oogenesis only. The absence of lodestar and cyclin J transcripts from aphid sexual oocytes and embryos was particularly unexpected. Three lodestar paralogs were identified within the pea aphid genome (ACYPI000753, ACYPI000590 and ACYPI006971). Although the three copies were represented on the microarray by copy-specific oligonucleotides, only ACYPI000753 was differentially expressed in sexual and asexual embryos. Localization of ACYPI000753 transcripts by in situ hybridization showed that its expression was specific to asexual oocytes. However, although the sequence of the riboprobe used to locate ACYPI000753 transcripts was absent from the ACYPI006971 sequence, it shared 87% nucleotide identity with the unregulated paralog ACYPI000590. Therefore, cross hybridization between the ACYPI000753 riboprobe and ACYPI000590 transcripts cannot be excluded. Nevertheless, we detected no ACYPI000753 expression in sexual oogenesis. This unexpected result might reflect the functional specialization of at least one of the lodestar paralogs in asexual oogenesis.

Among the 16 transcripts found to be specifically localized in germaria and/or oocytes , lsd1 was the only one for which the expression level was found to be higher in sexual ovaries than in asexual ovaries according to the microarray results. This disproportion is consistent with the striking over-representation of transcripts that were upregulated in asexual embryos (29/33), as evidenced by the microarray analysis. Sexual oogenesis and asexual oogenesis display major inherent differences. Within embryos, the sexual oocyte nucleus remains blocked in meiosis prophase 1 whereas the asexual oocyte undergoes a modified but complete meiotic division immediately after ovulation [7]. This modified meiosis involves specific changes within the oocyte; self-organized asters are formed, that then recruit centriole precursors and pericentriolar material [28]. This process may be accompanied by the activation of microtubule stabilization factors or by the inactivation of microtubule destabilization factors specific to asexual oogenesis [28]. The predominance of genes that are upregulated in asexual embryos may be partially explained by the production of molecular components that are specifically required for asexual oogenesis but not for sexual oogenesis.

In this study, we identified a number of putative embryonic target genes of the signaling cascade initiated by photoperiodic cues and transduced by JH. These genes are likely to be among those that direct sexual or asexual differentiation of the developing embryos. Our analysis of their expression patterns represents the first time the molecular bases of aphid sexual and asexual oogenesis have been addressed. Altogether, our results showed that aphid asexual oogenesis requires the same genes as sexual oogenesis but that these two distinctive differentiation processes display specific gene expression programs.

AG, SJP and DT: INRA, UMR 1099 BiO3P, Biologie des Organismes et des Populations appliquée à la Protection des Plantes, 35653 Le Rheu cedex, France

SS and TH: Okazaki Institute for Integrative Bioscience, National Institute for Basic Biology, National Institutes of Natural Sciences, Higashiyama, Myodaiji, Okazaki, 444-8787, Japan