Expression profiling of rainbow trout testis development identifies evolutionary conserved genes involved in spermatogenesis

The formation of male gametes, spermatozoa, occurs in the testes by means of a highly coordinated differentiation process called spermatogenesis that includes three phases. Firstly, primary germ cells or spermatogonia actively divide and after a fixed number of mitoses, differentiate to become spermatocytes. During meiosis, spermatocytes undergo DNA recombinations and then two successive divisions to produce haploid spermatids. Finally, during spermiogenesis, spermatids differentiate into mature, motile spermatozoa. In addition to a complex germ cell intrinsic genetic program, the entire spermatogenetic process also strongly relies on numerous endocrine and paracrine signalling pathways from testicular somatic cells [1–5]. In mammals, this cellular communication network that triggers efficient spermatogenesis notably involves the production of androgens by Leydig cells in response to the luteinizing hormone (LH) and the sustain of germ cell survival and differentiation by the so-called "nurse" Sertoli cells, which are themselves responsive to both androgens and the follicle-stimulating hormone (FSH). Whereas these overall features of spermatogenesis appear to have been conserved in vertebrates, a number of reproductive strategies have emerged throughout evolution, leading to marked differences in terms of structural organisation of the testes or regarding the frequency, timing and rate of spermatogenesis [6]. These spatio-temporal differences in the development of the germinal compartment are believed to involve original regulatory events.

In seasonally breeding rainbow trout, Oncorhynchus mykiss, spermatogenesis occurs during a marked annual reproductive cycle during which the testis undergoes a dramatic increase in size (from 0.05% up to 8% of the body weight), while germ cells develop partially synchronously, to finally allow the massive production of sperm within a short period of time once a year [7, 8]. Many efforts have been made in commercial fish farms to control reproductive cycles in order to obtain all-year-round spawning fish or to prevent the slowing of growth linked to sexual maturation. Successful results have been obtained to some extent through manipulating the photoperiod or the hormonal status of fish. However, the mechanisms underlying such effects are still poorly understood and these applications in broodstock management remain limited to a few situations [9, 10]. Further research efforts concerning the cellular pathways involved during spermatogenesis recrudescence are thus necessary if we are to understand and control testis maturation in teleost fish species. Moreover, the seasonally-driven testis development in trout provides a new model for deciphering the hormonal and paracrine regulatory processes that govern early germ cell proliferation and differentiation until spermatozoa secretion.

A microarray experiment investigating testicular gene expression during the reproductive cycle in trout is reported here. The use of entire gonad samples at various stages together with isolated germ cells allowed us to follow expression patterns during testis development and more importantly, to predict cellular origins of differentially-expressed genes. We also carried out a cross-species comparison and identified a complement of evolutionary conserved genes displaying very similar expression during spermatogenesis in trout and mice. Finally, by using non-testicular trout tissue expression data we found potential testis-specific genes that may have unique functions during spermatogenesis. Together with the confident annotations we provide for the genes represented on the trout microarray, this dataset thus constitutes a necessary and solid basis to further investigate the control of spermatogenesis in fish.

We have already applied preliminary large scale expression profiling (i.e., using 1K macroarrays) with the aim of studying the underlying molecular mechanisms of testis development at the beginning of the trout reproductive cycle [20]. In the present study we have confirmed and considerably extended previous findings, firstly by increasing the number of genes covered by the microarray, but also by refining the samples analysed, the gene annotation and the data-mining procedures. The use of testes at various developmental stages highlighted several gene activation or repression patterns throughout the reproductive cycle. In addition, using isolated germ cell populations we were further able to distinguish between somatic cell expression clusters and those corresponding to proliferating and differentiating germ cells. The identification of somatic genes was also reinforced by the use of spawning testes. Indeed, due to a dilution-effect resulting from the accumulation of germ cells, somatic gene expression decreased in maturing testes. On the contrary, in stage VIII gonads, where no more transcriptionnally active germ cells reside apart from a few spermatogonial stem cells, somatic genes are not diluted anymore. Moreover, this stage highlights somatic genes that undergo specific regulation as testes mature: low expression in stage VIII predicts their down regulation while high expression in stage VIII predicts their up regulation. We have thus reported the accurate spatio-temporal expression patterns for hundreds of testicular transcripts associated with the mitotic, meiotic and post-meiotic phases of germ cell development as well as with the somatic compartment of maturing testes in trout.

The establishment and description of transcriptomes underlying spermatogenesis have been mainly limited to a few mammalian species, especially in mice, rats and humans [21]. The broadening of such approaches to other vertebrates is important if we are to uncover the different regulatory pathways triggering tubular development, spermatogonial stem cell self-renewal, rapid proliferation and further differentiation of latter germ cells. In this respect, the cyclical and progressive features of trout spermatogenesis make it an original and interesting model for studying hormonal cues and paracrine communications sustaining germ cell development. In comparison to mammals where Sertoli cells sustain several generations of germ cells at the same time, the cystic organization and synchronous development of germ cell clones indeed make the trout testes a convenient way of studying cellular communications involved in the specific steps of spermatogenesis. The present dataset is likely to be of great utility in studying and identifying such networks. An important step forward will then be to understand how these networks are regulated and what factors (such as sexual steroids or other hormones) are responsible for the different expression profiles we have identified.

To gain a functional insight into the genes undergoing differential expression during the spermatogenetic process we undertook a sophisticated annotation procedure of trout clone cDNAs. This step is of particular importance when studying species such as trout for which the genome is not available and only little information exists in public databases (e.g. only 861 RefSeq entries in NCBI release 34 and 1920 entries in UniProtKB release 15.0 for Oncorhynchus mykiss). The starting point of our procedure was the use of expressed sequence tags (ESTs) instead of EST contigs that could lead to irrelevant annotations resulting from assembling errors. The high sequence coverage of trout clones, i.e. an average of 2.18 ESTs per clone, allowed us to map most of the corresponding transcripts on the genome of a closely related organism. For instance, up to 90% of the trout clones were linked to a potential fish orthologous gene in at least one of the 4 fish genomes. When this first mapping step was achieved, the use of mammalian orthologs and associated GO terms as provided by the Ensembl database allowed us to improve the somehow poor annotation of fish genomes. We believe that this annotation pipeline is more accurate than directly mapping trout sequences onto phylogenetically distant species and that such a strategy should be used in non-model species for which few genomic resources exist. As a matter of fact, the different expression patterns we conferred to the somatic compartment of the testis, the proliferation phase of germ cells or their subsequent differentiation, were consistently associated with annotations that are in agreement with what we know about these different testicular cell types and their biology. This further validated the biological significance and our interpretation of identified expression clusters. Obviously, some of the inferred orthologs and ontologies may still be erroneous, due to either partial, supernumerary duplications in fish genomes [22, 23] or simply incorrect orthological relationships and annotations provided by the Ensembl database. Furthermore, even for accurately annotated and true orthologous genes, the biological processes they are involved in may differ from one organism to another. However, the use of statistical tests to investigate the content of expression clusters bypasses these issues by highlighting only relevant, either enriched or depleted, GO terms.

The search for mammalian trout gene orthologs was also carried out with the aim of comparing underlying expression programs in testis development and spermatogenesis in fish and mice. Genes that exhibit similar expression patterns in different species are very likely to play key roles in a given process. Their expression conservation indeed suggests that evolutionary forces act strongly on promoter regulatory regions and corresponding transcriptional regulators so that the encoded proteins achieve their functions both in a specific cell type and at a precise developmental stage. By comparing trout and mouse expression data we were able to identify genes not only expressed in the testes of both species but that display similar developmental expression patterns. These included genes more likely to be expressed in the germline and that tend to accumulate during the course of the spermatogenetic process, but also somatic genes that exhibit decreasing expression from puberty onwards as a consequence of their relative dilution. Here again the annotation was of great importance to compare expression profiles of accurate pairs of genes. From this point of view, it is important to note that the subgroup of correlated genes contained a number of genes with one-to-many or many-to-many orthologous relationships. Whereas this category of orthologs may lead to false correlated genes is still a possibility. We also believe that for such ambiguous sequence-based orthologs, expression conservation is a relevant way of identifying true orthologs at a functional level. The complement of conserved genes we have reported here is therefore likely to contain key factors fulfilling important functions during spermatogenesis that have been maintained throughout evolution and deserve special attention. On the other hand, the absence of correlation observed for a number of genes may underscore original pathways that may account for specific regulations in each species. It could also reflect species-specific features of spermatogenesis such as morphological organization and spatio-temporal development of the testis, and may thus represent molecular events that are difficult to detect or to observe in other species. However, it remains hard to distinguish between genes exhibiting true diverging expression profiles from discrepancies due to either wrong orthological relationships or artefacts due to the different technologies used to produce the two datasets.

Whereas several aspects of spermatogenesis such as events associated with meiosis and spermiogenesis are well conserved across species, testis organization and biology differ on certain points between amniote (reptiles, birds, mammals) and anamniote (fishes and amphibians) vertebrates [8]. Importantly, the number of spermatogonial generations, i.e. the number of mitoses spermatogonia undergo before meiosis, and the signalling regulating their differentiation is more specific to each species. In teleost fish testes, the molecular mechanisms triggering testis maturation are still poorly understood. A few growth factors that are produced by Sertoli cells and regulate germ cell proliferation and fate have however been identified. These include the TGF family members activin [24], anti-Müllerian hormone (AMH) [25] and gonadal soma-derived factor (GSDF) [26], as well as an ortholog of the platelet-derived endothelial cell growth factor (PD-ECGF) [27]. Importantly, Amh expression was found to be inhibited by 11-ketotestosterone, a teleost fish-specific androgen, and this steroid-mediated suppression of AMH action was sufficient to allow full completion of spermatogenesis in the eel [25]. This led AMH to be considered as a spermatogenesis-preventing substance and suggested that the initiation of meiosis in fish may occur through the suppression of an inhibitory activity rather than direct activation. In this regard, the cluster of somatic genes exhibiting apparent down regulation during the reproductive cycle, which includes the Amh and Inhba genes themselves, is of peculiar interest (cluster B). In an attempt to discover further factors that may sustain testicular maturation and could regulate spermatogonial cell proliferation and the onset of meiosis, we focused on those genes encoding extracelullar and transcription factors within the somatic expression clusters. In addition to the aforementioned genes Amh and Inhba, several well-known genes with important implications in testis development or spermatogenesis could be found: Dmrt1 and Sox9a are orthologs of mouse genes expressed in Sertoli cells and encode key transcription factors for testis differentiation [28, 29]; Etv5 encodes a transcription factor, ERM, of which the expression in adult Sertoli cells controls spermatogonial stem cell self-renewal in mice [30]; Inha encodes a TGF family member that inhibits spermatogonia proliferation [31] and FSH production by the pituitary gland through a negative regulatory loop. Furthermore, we investigated the expression of previously uncharacterized factors in trout testes. Ghr2 encodes an isoform of the growth hormone (GH) receptor and was found to be expressed in Leydig cells. This localization is consistent with our finding that GH regulates steroidogenesis around spawning [12, 32]. This factor could also take part in the local production of Insulin-like Growth Factor 1 (IGF1), which in turn might regulate the proliferation and/or differentiation of spermatogonia as previously proposed [33–35]. In addition, we have demonstrated that Igfbp7, which encodes an IGF-binding protein, is expressed in Leydig cells. This factor could therefore control the availability of other secreted factors and thus modulate germ cell proliferation. IGFBP7 is indeed a strong growth suppressor that can bind IGF1 and IGF2 as well as activin [36, 37]. Furthermore, IGFBP7 was shown to inhibit activin-enhanced FSH-stimulated expression of Cyp19a1 and to suppress estradiol production in rat granula cells [38]. This suggests that IGFBP7 could similarly control the local production of steroids in the testis.

The large number of differentially-expressed genes we have reported here, i.e. almost 46% of the genes analysed, is most likely to reflect the important changes in cellular composition throughout spermatogenesis and the wide variety of underlying molecular and cellular functions in this process. In particular, germ cells undergo an extraordinary cellular differentiation process and fulfil unique functions in the organism. Their expression program therefore significantly differs from that of somatic cells and notably involves the expression of many testis specific gene products. To address this issue in trout, we used expression data from 5 non-testicular tissues to identify transcripts only detected in the testes or the gonads. Not surprisingly, most of these genes were expressed within the germline, especially in the meiotic and post-meiotic germ cells. However, a few of them were also found in somatic expression clusters, notably including Dmrt1 and Lhgcr, and may be involved in original pathways leading to successful sustaining of germ cell development. We investigated the expression of such a gene, Tgfbr3, and found it was expressed in Leydig cells. Tgfbr3 encodes the Transforming growth factor (TGF)-beta receptor type III (TGFBR3 or betaglycan), which orchestrates the TGF-beta superfamily signalling by binding both BMP and TGF-beta members [39]. Notably, TGFBR3 was shown to function as an inhibin co-receptor with activin receptor type II (ActRII) and to mediate functional antagonism of activin signalling by conferring inhibin sensitivity [40]. Betaglycan might thus play an important role within the testis in regulating paracrine actions of activin and inhibin produced in response to gonadotropes. We also demonstrated that Tcf23, a gene detected only in gonads from both sexes, was expressed by peritubular cells. Tcf23 encodes the basic helix-loop-helix factor TCF23 (also called OUT), a transcriptional repressor involved in myoblast differentiation in mammals [41], for which expression in testicular contractile cells is thus relevant. Tissue-specific expression is often correlated with important physiological roles for a gene in that tissue and the testis and gonad specific genes we identified may have important implications during spermatogenesis or testis development. Besides, these testis-specific genes are also of great interest to identify common cis DNA regulatory elements responsible for their restricted expression in the male gonad.

In this study we investigated the underlying expression program in trout testis development and spermatogenesis using cDNA microarrays. In addition to genes preferentially expressed in germ cells at each differentiation stage, our expression profiling analysis has revealed several distinct somatic gene expression clusters that may sustain germ cell survival and testis maturation throughout the male reproductive cycle. Furthermore, to point out and provide the community with the most promising and important genes, we identified potential trout testis-specific genes as well as evolutionary-conserved genes involved in spermatogenesis in vertebrates. In addition to the accurate annotation of trout genes represented on the microarray, our dataset represents a valuable and solid base for further investigating the molecular mechanisms involved in testis differentiation and the onset of spermatogenesis in fish.