Mammalian oocyte is the only known cell able to activate the zygotic genome after fertilization. Its cytoplasm is also able to reprogram the nucleus of a differentiated cell in cloning experiments. Therefore, it is likely that several genes specifically expressed in the oocytes are responsible for this ability to program diploid genomes. It is the case for the so-called maternal genes such as Maternal Antigen That Embryo Required (MATER), Zygotic Arrest 1 (Zar1), Stella and nucleoplasmin 2 (Npm2) that are all required for normal embryonic development beyond the 1- or 2-cell stage [1, 2]. Moreover, new observations have shown that loss of function mutations in the two oocyte-specific genes GDF-9 and BMP-15 are responsible for severe alterations of ovarian folliculogenesis, these alterations being different depending on the hetero- or homozygous state of the individuals in the ovine species [3, 4]. So genes specifically expressed in the oocyte seem to play key roles in oogenesis, ovarian folliculogenesis, fertilization and/or early embryonic development. In an attempt to identify novel oocyte-specific genes, several groups have used both mRNA differential display [5] and in silico subtraction approaches [6]. By using the latter, we have recently identified six genes of the oogenesin family [7], these genes being present on chromosome 4 in a cluster of almost 1 Mb composed of twelve oogenesin paralogous genes. We have also identified six genes presenting similarities with NALP5/MATER [8] which are specifically expressed in the mouse oocyte, three being clusterized on chromosome 7. We have experimentally validated the specificity of expression for three oogenesin and five NALP genes [7, 8]. The fact that these two groups of genes are localized in two clusters indicates that they likely originate from duplications of an ancestral gene.

In the present work, we have used the same in silico approach to identify new oocyte-specific genes that are also localized in clusters in the mouse genome. In particular, we have identified three new loci in the mouse genome containing several genes specifically expressed in the oocyte, and two loci containing genes specifically expressed in male and female germ cells, which may thus be called "germ cell loci". We have experimentally verified the specificity of expression for eight of them by RT-PCR and five of them by in situ hybridization. Moreover, we have compared the map position of all the "oocyte-clusters" identified so far with that of the other "isolated" oocyte-specific genes. We could show that oocyte-specific clustered genes are significantly closer to telomeres than oocyte-specific isolated genes. Our results emphasize a new vision of gene regulation, where large-scale genome organization likely plays an important function.

The present work and two previous identifications of the oogenesin genes [7] and of the mater family genes [8] demonstrate that the in silico subtraction is an efficient and reliable methodology to identify new oocyte-specific genes. By localizing these genes on the mouse genome, we have discovered other "germ-cell" specific genes, organized in clusters. RT-PCR as well as in situ hybridization experiments confirmed their specificity of expression in the ovary and/or in the testis. As for GDF-9, Mater, Zar1 or Stella, these genes may play important roles in folliculogenesis, fertilization and/or early embryonic development.

Amongst the "oocyte-specific clusters" presented here, the Tcl1/1B has previously been described. Indeed, in a previous work on human TCL1 and TCL1B genes, the six murine Tcl1 and Tcl1b1-5 orthologous genes were already shown to be expressed in oocytes and two-cell embryos but not in other adult tissues except in lymphoid cell lines, where overexpression of Tcl1 in these cells leads to leukemia or lymphoma [13]. Tcl1 was also shown to be expressed in testis, as previously described [14, 15], its overexpression inducing the formation of testicular seminomas [16]. Interestingly, Tcl1^-/- female mice are sterile, as early embryos are unable to undergo normal cell cleavage up to 8-cell stage, the majority of embryos being even unable to proceed beyond the 4- to 8-cell stage [16]. Therefore, like Mater and Zar1, Tcl1 can be considered as a maternal effect gene as well. It is possible that part of the effects of this gene family is to promote cell survival and/or proliferation by activating the Akt kinase as previously observed [17, 18].

The F-box proteins, encoded by the thirteen genes localized on chromosome 9 cluster, might specifically interact with other proteins in the oocyte. Indeed, the F-box motif is present in numerous proteins known to serve as a link between a target protein and an ubiquitin-conjugating enzyme. In particular, it has been shown that the Early Mitotic Inhibitor (Emi)-1, which contains an F-box domain, is able to regulate mitosis progress by inhibiting premature anaphase promoting complex/cyclosome (APC) activation [19], the APC complex being composed of an ubiquitin ligase coupled with the SCF (Skp1-Cullin-F-box protein) complex [20]. So it is possible that the oocyte-specific F-Box proteins would participate in the stabilization of the meiotic process at the prophase I stage.

Concerning the cluster on chromosome 14, in situ hybridization showed a high level of expression of SpeerA gene in the mouse oocyte, corroborating previous results from Matzuk's group [6]. By analyzing the RT-PCR results, we found that this gene was also expressed in the testis, likely in male germ cells. The Speer family has recently been described as a new family of 14 genes. Interestingly, SPEER-1, SPEER-2, and SPEER-4D were recently shown to be expressed specifically in the mouse testis [21]. These genes have open reading frames of approximately 200-220 amino acids, and encode for proteins that exhibit a high proportion of alpha-helical secondary structure, comprising approximately 15% glutamate residues. According to their possible interaction with cytoskeletal proteins such as actin or dynein, it has been hypothesized that the SPEER may be nuclear matrix proteins involved in the reorganization of the spermatocyte/spermatid nucleus [21]. SpeerA, discovered in our study (XM_138939), shares about 34% amino acid sequence similarity with SPEER-4A gene, which could suggest that it plays a similar role in the oocyte. Of note, in contrast to most of the oocyte clusters, the four testis-specific genes neighboring the SpeerA gene were not structurally related to each other.

One of the clusters on chromosome 19 contains the well-known oocyte-specific gene Oosp-1 [12], and two structurally similar genes, these three genes clearly sharing a ZP-like domain. This suggests that these genes encode for structural proteins involved in the formation of certain components of the zona pellucida. A structural analysis of the sequences of these proteins is underway in our laboratory to further investigate the structural specificity of these proteins. The biological roles of the two genes present in the other cluster on chromosome 19, as well as the role of the testis-specific genes present on the chromosome 14 cluster are unknown.

Overall, for all of these genes, further experiments as well as structural analysis are now necessary to investigate their biological roles in oocyte and follicle maturation. Investigating such a role will require functional studies such as identification of partners and/or targeted invalidation.

As we previously discussed, the targeted invalidation of MATER gene leads to female infertility due to a blockage of early embryonic development [22]. This suggests that the oocyte-specific MATER-like genes, i.e. the Nalp9A-F genes that we have identified [8] are not able to compensate the absence of the MATER/Nalp5 product in Mater-/- mice. Similarly, invalidation of Tcl1 gene in the mouse was not clearly compensated by the five oocyte-specific Tcl1B genes. Overall, this suggests that despite an apparent redundancy in the expression of these structurally similar genes, every member of the oocyte Nalp and Tcl1 family of paralogous genes seem to have very specific biological roles in female reproduction. Invalidation of each of these genes is now necessary to study their specific biological role. Moreover, as expected, most oocyte-specific genes organized in clusters are paralogous genes, originating presumably from a common ancestral gene by successive duplication events. It is the case for the oogenesin, Nalp9, Fbxo12, Tcl1 as well as the Osp1-3 genes. It is clearly not the case for the clusters localized on 14A3 and 19D3 (Fig. 1). Nevertheless, such an organization of "oocyte- or germ-cell clusters" suggests the presence of specific genomic regulatory regions in the vicinity of these genes.

As we depicted in figure 1, the SpeerA gene is localized on a cluster on chromosome 14, in the vicinity of four unknown testis-specific genes. On chromosome 19, the oocyte-specific gene related to TWIK-Related spinal cord K+ channel is localized near the gene corresponding to AK015359 GenBank number, which is predicted in silico to be testis-specific as well. Interestingly, we have previously shown that in the oogenesin cluster on chromosome 4, the oogenesin-4 gene is also slightly expressed in the testis. Of note, Pramel1 gene, which is localized near oogenesin4, has also been described as a male germ-cell specific genes [7, 23]. Similarly, in the oocyte-specific Nalp cluster on chromosome 7, we have shown that Nalp9D is also clearly expressed in the testis, presumably in germ cells. So the notion of oocyte-specific cluster would not be systematically restricted to female germ cells, and would be extended in some cases to "germ-cells cluster" as suggested in this work.

One striking observation of our work is the difference in chromosome localization between oocyte-specific genes organized in clusters and oocyte-genes that are not. If genes were located at random on the chromosome and that the distance to the closest end is measured, then the expected range would be 0% to 50%, and one would expect to reach an average distance of 25% if enough genes are taken into consideration. In our study, isolated genes were very close to this value (24%) whereas clusters were located in average at half of this distance (12%). Gene clusters presumably originate from local gene duplications [24], which are now recognized as efficient motors of diversification in metazoans. Of note, the location of gene families near telomeres has a deep sense in terms of rapid evolution. Indeed, subtelomeric sequences are recognized as the most rapidly evolving regions in the genome, maybe due to the high level of recombination present in these regions [25]. On the other hand, clusters of genes are the entry gates for accumulation of mutations, without the risk of noxious consequences for the organism, and therefore enable the development of new functions, as clearly illustrated by the variants of the human hemoglobin beta chain, which adapts the foetus to successive oxygen environments during development [26, 27]. For mammals with a high reproductive potential, such as mice and other rodents, the possibility of rapid evolution for genes involved in ovarian function may be a great advantage. Therefore, one can hypothesize that the specific near telomere position of oocyte-specific gene clusters corresponds to an evolutionary advantage.

Interestingly, the genes that we have described here are strictly silenced in non-ovarian tissues, and one may hypothesize that the specific near telomere position of oocyte clusters would contribute towards such a silencing. In particular, it has been shown that telomeric and pericentric regions of chromosomes are mainly composed of heterochromatin in most eukaryotic genomes. Moreover, euchromatic genes relocated by chromosomal rearrangement or by transposition in the neighbourhood of heterochromatin can be silenced in drosophila [28]. In yeast, a similar phenomenon of reversible gene silencing near telomere has been called Telomere Position Effect (TPE) [29, 30]. The presence of TPE has also been observed in human cells [31]. Actually, a very specific nuclear topology would explain such a phenomenon. Indeed in yeast, the 32 telomeres cluster at the nuclear periphery in 8 to 10 groups, organized in compartments rich in histone-binding silencing factors [32]. Testing if the telomeres would be organized in very specific compartments in oocyte and in other tissues requires further investigations.

We have studied five clusters of genes specifically expressed in female, some of them being also expressed in male germ-cells. Moreover, contrarily to non-clustered oocyte-specific genes, those that are organized in clusters tend to map near chromosome ends, suggesting that this specific near-telomere position of oocyte-clusters could result in an evolutionary advantage. Understanding the biological roles as well as the mechanisms underlying such a specific expression now requires further investigation.