Carole Deyts and Eva Candal contributed equally to this work.
Patterns & Phenotypes
An automated in situ hybridization screen in the medaka to identify unknown neural genes
Article first published online: 22 JUN 2005
Copyright © 2005 Wiley-Liss, Inc.
Special Issue: Developmental Neurobiology Special Issue
Volume 234, Issue 3, pages 698–708, November 2005
How to Cite
Deyts, C., Candal, E., Joly, J.-S. and Bourrat, F. (2005), An automated in situ hybridization screen in the medaka to identify unknown neural genes. Dev. Dyn., 234: 698–708. doi: 10.1002/dvdy.20465
- Issue published online: 18 OCT 2005
- Article first published online: 22 JUN 2005
- Manuscript Accepted: 27 MAR 2005
- Manuscript Revised: 17 MAR 2005
- Manuscript Received: 29 SEP 2004
- European Union key action. Grant Number: QLK3-CT-2001-01890
- in situ hybridization;
- unknown genes;
Despite the fact that a large body of factors that play important roles in development are known, there are still large gaps in understanding the genetic pathways that govern these processes. To find previously unknown genes that are expressed during embryonic development, we optimized and performed an automated whole-mount in situ hybridization screen on medaka embryos at the end of somitogenesis. Partial cDNA sequences were compared against public databases and identified according to similarities found to other genes and gene products. Among 321 isolated genes showing specific expression in the central nervous system in at least one of five stages of development, 55.14% represented genes whose functions are already documented (in fish or other model organisms). Additionally, 16.51% were identified as conserved unknown genes or genes with unknown function. We provide new data on eight of these genes that presented a restricted expression pattern that allowed for formulating testable hypotheses on their developmental roles, and that were homologous to mammalian molecules of unknown function. Thus, gene expression screening in medaka is an efficient tool for isolating new regulators of embryonic development, and can complement genome-sequencing projects that are producing a high number of genes without ascribed functions. Developmental Dynamics 234:698–708, 2005. © 2005 Wiley-Liss, Inc.
During the past few years, raw genome sequence data have accumulated at an ever-increasing speed, both for model organisms and for species of economic interest (Galas et al., 2003). Indeed, from the beginning of the genomic era, it became obvious that a sizeable part of the eukaryotic genomes had “escaped” detection by classical genetic methods (Dujon, 1996). Restricting ourselves to the animal kingdom, it can be estimated that even now roughly one-quarter of the predicted coding sequences encode proteins of unknown functions (Venter et al., 2001).
The fact that such a large part of the animal genome remains unstudied means that any systematic screen should yield a sizeable proportion of “unknown” genes, and this approach should remain fruitful at least until the genome is fully sequenced, annotated, and available. We show here that a whole-mount in situ hybridization (WMISH) screen for developmental regulators can, indeed, in an efficient and cost-effective way, lead to the characterization of novel genes, with the additional advantage (compared with unprocessed sequence data) that the expression pattern can, at least in some cases, allow for establishing testable hypotheses on gene function.
Our laboratory has a long-standing interest in the regulation of organ size and growth, and more specifically in the control of cell proliferation in ontogenesis. Our biological model is the medaka Oryzias latipes, a small teleost fish from Eastern Asia. We previously took advantage of the characteristics of this model (notably, its high fecundity and the transparency of its eggs and embryos) to perform a pilot expression screen aimed at discovering potential regulators of cell proliferation (Nguyên et al., 2001a). We reported that the optic tectum (OT) of the 31-stage medaka embryo is particularly well suited for such a study, due to its very peculiar mode of growth (Nguyên et al., 1999): mitotic cells are located at its margin, differentiating cells at its center, and in between lies an “arrest zone” where cell cycle-exit genes are expressed (Nguyên et al., 2001a, b). This study also demonstrated the predictive value of this approach: at least for the genes expressed in the arrest zone, there is a very strong correlation between the expression pattern and the gene function (in that case, down-regulation of the cell cycle).
We present here the continuation and expansion of this pilot study, based on the use of a WMISH automat, for which we designed a new protocol. We achieved a medium-scale (1,500 clones) screening of a cDNA libraries prepared from medaka embryonic heads, which led to the characterization (expression pattern and partial sequence) of novel genes, as a first step to understand the function of those genes in development.
Automated WMISH Screen
We performed an automated in situ screen from a cDNA library constructed from stage-31 medaka embryos' heads (Nguyên et al., 2001a). Approximately 1,500 cDNAs were amplified from arrayed bacterial clones; 748 mRNA probes were chosen according to probe's length (those larger than 500 pb), and further analyzed by WMISH.
One of the main interests of such large-scale screens is to identify genes that are still functionally uncharacterized. EST analysis was conducted for gene annotations. Of the 321 cDNA clones showing a patterned expression in the CNS, 55.14% were identified by BLAST searches as orthologues of known genes from other organisms. Additionally, 16.51% (53) of clones with regionalized expression were identified as unknown genes or genes with unknown function, while 28.35% corresponded to clones that share no significant similarity with sequences in the public databases. Expression patterns of genes within the embryo and their sequences will be available on the Medaka Expression Pattern Database (MEPD, http://www.embl.de/mepd). All clones are available upon request from the Development, Evolution and Plasticity of the Nervous System Laboratory, Gif-sur-Yvette, France (DEPSN; e-mail: email@example.com).
Unknown Genes With Restricted Expression Patterns in the Central Nervous System
We provide here eight detailed examples of clones corresponding to genes with restricted expression patterns in the embryonic brain (Figs. 1, 2) and homologous to vertebrate genes with previously unknown function (Table 1). These examples are chosen as being representative of the results of our screen, and we present them according to their expression in, or relative to, the optic tectum, which is the focus of our studies. Gene expression was studied at five developmental stages, in which the proliferating versus terminally differentiated tectal regions had been previously identified in our laboratory (Candal et al., 2004b; Nguyên et al., 1999), namely stage 19 (2 somite-stage), stage 25 (18 somite-stage), stage 28 (30 somite-stage), stage 31 (gill blood vessels formation-stage), and stage 35 (visceral blood vessels formation-stage). Our results are presented as follows: first, an example of a gene expressed outside the optic tectum, then examples of genes expressed in the proliferative and arrest zones of the optic tectum, respectively. Within the genes expressed in the tectum, the former category includes more genes than the latter, reflecting the fact that, amongst the 53 “unknown” genes of the present screen, 13.2% were found to be expressed outside the optic tectum, 74.5% in the proliferative (marginal) zone, 7.6% in the arrest (intermediate) zone, and 4.7% in the differentiation (central) zone (for a total of 86.8% of genes exhibiting a tectal expression pattern).
|Clone||Accession number||Size (aa)||Homologies by BLASTp||S||E||Functional domains|
|1123||AY675207||216||similar to HP 5930437A14 (1,102 aa)||23||6,00E-91||TSP1|
|1309||AY675211||230||unnamed protein product (990 aa)||19||5,00E-51||none|
|1053||AY675206||177||HP MGC23937 similar to CG4798 (309 aa)||34||2,00E-95||Upp, Pribosyltran|
|1182||AY675208||291||KIAA0266 (291 aa)||16||2,00E-39||Utp14|
|1231||AY675209||162||MID1 interacting G12-like protein (183 aa)||13||4,00E-31||Spot14|
|1234||AY675210||289||KIAA0355 (1070 aa)||35||1,00E-95||none|
|1737||AY675212||227||KIAA1991 (712 aa)||11||3,00E-27||RING|
|1012||AY675205||139||GAC-1 (2062 aa)||25||6,00E-67||ANK|
Clone 1123 (CDOL_001031F08_1123).
This clone was very faintly expressed at stage 19 (Fig. 1A), while its expression increased in the caudal rhombencephalon at stage 25 (Fig. 1B). Clone 1123 showed a peculiar expression pattern that dynamically changed from the caudal rhombencephalon and somites at stage 28 (Fig. 1C), to a restricted expression in the telencephalon, diencephalon, and torus semicircularis, in non-neural tissues bordering the hypothalamus (not shown), and in the proximal edge of the fins at stage 31 (Fig. 1D). Some ganglions of the autonomic nervous system were also labeled at stage 35 (Fig. 1E; see also Fig. 3A). Clone 1123 blasted only with the Rattus novergicus similar to hypothetical protein 5939437A14, and comparison with protein databases revealed that it had no human homologues. The hypothetical protein 5939437A14 contains C-terminal thrombospondin type 1 repeats (TSP1) that bind and activate TGF-beta.
Clone 1309 (CDOL_001041C05_1309).
1309-mRNA expression was absent in early embryonic stages (Fig. 1F–H) and at stage 35 (Fig. 1J), being exclusively expressed at stage 31 in the inferior hypothalamic lobes, the infundibulum, and the hypophysis (Fig. 1I). Blastx search revealed that it was similar to unnamed protein products in teleosts and mammals, and also to predicted kinases. Blastp against H. sapiens proteins revealed its similarity to an unnamed protein product without known functional domains.
Clone 1053 (CDOL_001011F04_1053).
1053-mRNA was widely expressed in the anterior brain of embryos at stages 19 and 24 (Fig. 1K,L). Its expression became rapidly down-regulated and mostly restricted to the caudal OT and fins from stage 28 onwards (Fig. 1M). Although darkening of the coroidea prevents observation of retinal cell layers, labeling appeared to be restricted to particular areas, as was confirmed in transverse sections (Fig. 3B). At stage 31 and later (Fig. 1N), 1053-mRNA was also expressed in the telencephalon and dorsal rhombencephalon, its expression becoming highly increased in the telencephalon at stage 35 (Fig. 1O). Sequence analysis revealed that it was similar to the Danio rerio hypothetical protein zgc:77421 and other hypothetical proteins in mammals. Blastp search against human proteins revealed that it is similar to the hypothetical protein MGC23937, similar to CG4798, a predicted uracil phosphoribosyl transferase (Upp). A search for functional domains allowed the identification of a Pribosyltran (phosphoribosyl transferase) domain.
Clone 1182 (CDOL_001032B09_1182).
At stage 19 (Fig. 1P), this clone was strongly expressed in the eyes, mesencephalon, and caudal pole of the embryo. Its expression became rapidly restricted to the eyes, the mesencephalon, and the primordium of the cerebellum at stage 25 (Fig. 1Q), and to the proliferative zone of the OT and fins later in development, with an increasing expression in the rhombencephalon from stage 28 onwards (Fig. 1R,S). In the retina, expression of clone 1182 was detected within the ciliary marginal (proliferative) zone (Fig. 3C) as identified by immunohistochemistry against the proliferating cell nuclear antigen (PCNA; Fig. 3D). At stage 35 (Fig. 1T), 1182-mRNA expression was also detected in the telencephalon. Blastx revealed that it was similar to a Mus musculus unnamed protein, to other hypothetical RIKEN, KIAA, or unnamed protein products in various vertebrates, and to a Rattus novergicus hypothetical protein similar to serologically defined colon cancer antigen 16. Comparison with human protein databases revealed that it was similar to the human hypothetical protein KIAA0266, which is related to the Utp14 proteins; depletion of these proteins impedes production of the 18S rRNA, indicating that they are part of the active pre-rRNA processing complex.
Clone 1231 (CDOL_001033D10_1231).
Expression of 1231-mRNA was detected in the prosencephalon and mesencephalon at stage 19 (Fig. 2A). An intense expression was observed in the olfactory epithelium and in the liver from stage 25 onwards (Fig. 2B–E). At stage 25, the 1231-mRNA was also observed in the lens and retina, OT, primordium of the cerebellum, and rhombencephalon (Fig. 2B). 1231-mRNA expression was also detected in the ears at stage 28 (Fig. 2C). From stage 31 onwards, 1231-mRNA expression was strongly detected in the telencephalon (Fig. 2D,E), diencephalon (Fig. 3E), and retina (Fig. 3E,F), and weakly detected in the mesencephalon (Fig. 3F). A strong expression was also detected in the rhombencephalon at stage 35 (Figs. 2E, 3G). Sequence analysis revealed the clone 1231 was similar to several hypothetic proteins or unnamed protein products in teleost and mammals. Blastp against human protein databases revealed that it was similar to the human MID1 interacting G12-like protein, which is related to the family of thyroid hormone-inducible hepatic proteins Spot14. Spot14-RNA is increased in rat liver by insulin, by glucose in hepatocyte culture medium, as well as by thyroid hormone.
Clone 1234 (CDOL_001033D11_1234).
1234-mRNA was ubiquitously expressed in the anterior part of the stage-19 embryo (Figs. 2F, 3H) and became restricted to the lens, retina, OT, and somites at stage 25 (Fig. 2G). At stage 28, its expression decreased in the OT (Fig. 2H). From stage 31 onwards (Fig. 2I, J), 1234-mRNA expression became strongly restricted to the tectal PZ (see also Fig. 3I) and increased in the rhombencephalon. It was also present in the distal proliferative edge of the fins. Blastp analyses revealed that this clone was similar to an unnamed protein product in teleosts and to RIKEN and KIAA hypothetical proteins in mammals. Comparison with human protein databases revealed its similarity with the human hypothetical protein KIAA0355. No functional domains were found among these proteins.
Clone 1737 (CDOL_001071E05_1737).
1737-mRNA was weakly expressed at stage 19 and 25 (Fig. 2K,L), and became differentially expressed from stage 28 onwards (Fig. 2M), when its expression was increased in the OT (see also Fig. 3J). Later in development, the expression in the tectum became restricted to the PZ; clone 1737 was also weakly expressed in the telencephalon, in the rhombencephalon, and in the distal proliferative edge of the fins (Fig. 2N,O). At stage 35, its expression increased also in the telencephalon (Fig. 2O). Clone 1737 was similar to one teleost unnamed protein product and to human hypothetical proteins. Blastp against human databases revealed that it is similar to the human KIAA1991. This protein presents a RING finger domain, a specialized type of zinc finger probably involved in mediating protein-protein interactions.
Clone 1012 (CDOL_001012A03_1012).
1012-mRNA was widely expressed at stage 19, especially in the head of the embryo, in the mid-hindbrain boundary, and also at the most caudal pole (Fig. 2P). From stage 25 onwards, its expression became mostly restricted to the retina and dorsal brain (Fig. 2Q), and at stage 28 (Fig. 2R) labeling was particularly intense in the rhombencephalon as observed in transverse paraffin sections (not shown). At stage 31 (Fig. 2S), it was predominantly detected in the telencephalon, in the arrest zone of the OT, in the rhombencephalon, and in the fin buds. A similar pattern was observed at stage 35 (Fig. 2T). The expression pattern in the arrest zone was confirmed in transverse paraffin sections (Fig. 4A), and compared with the expression pattern of Ol-KIP (Fig. 4B), a well-characterized gene involved in cell cycle exit, homologous of the p57 member of cyclin-dependent kinase inhibitors. Both clone 1012 and Ol-KIP were located just bordering the proliferating zone. Proliferation zone in the optic tectum of the medaka has been identified by immunohistochemistry against PCNA (Fig. 4B) and by the expression pattern of the neurogenic gene Ol-Delta A (Fig. 4C).
Blastx searches revealed that it is similar to several mammalian RIKEN, KIAA, or ankyrin repeat-containing hypothetical proteins. Blastp searches limited to Homo sapiens revealed that it was similar to GAC-1 (ankyrin repeat-containing cofactor-2), a protein with unknown function containing several ankyrin repeats (ANK), which mediate protein–protein interactions in very diverse families of proteins.
Expression Patterns in the Adult Central Nervous System
All the above-described clones were also detected in the young adult brain (Figs. 5, 6) where they revealed regionalized expression profiles, suggesting spatially regulated roles in the vertebrate CNS. A summary of the expression patterns found in the medaka adult brain is provided in Table 2.
Clone 1123 (CDOL_001031F08_1123).
Clone 1309 (CDOL_001041C05_1309).
1309-mRNA expression was observed in the ICL of the olfactory bulbs; in the dorsal and ventral telencephalon (Dm, Dlv, Vd, Vv, Vl) (Fig. 5F); in the habenula and preoptic area (Fig. 5G); throughout the rostral hypothalamus (HD, HV in Fig. 5I), the Hc and NDIL; in the TL, PGZ, and mpz of the tectum (Fig. 5H); in the TS and NDTL of the mesencephalic tegmentum (Fig. 5I); in the egl, CbSP, and caudal edge of the CbSg of the cerebellum (Fig. 5J); in the gc of the rhombencephalon (Fig. 5J); and in the NFS and Flm of the spinal cord.
Clone 1053 (CDOL_001011F04_1053).
Labeling was detected throughout the olfactory bulbs (OB) (inset in Fig. 5K); in the dorsal and ventral telencephalon (Dm, Dd, Dld, Dlv, Dlp, and Vv) (Fig. 5K, L); in the dorsal habenula (Hd); in the preoptic region (PMp in Fig. 5L) and the inferior suprachiasmatic nucleus of the diencephalon (SC); in the HD, HV (Fig. 5N), caudal hypothalamus (Hc) and NDIL; in the PGZ of the tectum (Fig. 5M); in the torus lateralis of the mesencephalon (NDTL); in the cerebellar granular lateral eminence (egl), in the ganglionar layer of the cerebellum (CbSP), and dorsally in the CbSg (Fig. 5O); in the NFS and other nuclei of the spinal cord (see Table 2).
Clone 1182 (CDOL_001032B09_1182).
1182-mRNA expression was observed in the dorsal and ventral telencephalon (Dm, Dld, Dlv, Vd, Vv, Vl) (Fig. 5P); in the dorsal habenula (Hd) and other nuclei in the diencephalon (A, TV, PMp, SC) (Fig. 5Q and inset); in the HD, HV, Hc, and NDIL of the hypothalamus (Fig. 5S); in the hypophysis; in the TL and PGZ of the tectum (Fig. 5R); in the NGp and NDTL of the mesencephalon (Fig. 5S); in the egl and the corpus cerebelli (CbSg and CbSP) (Fig. 5T); in the longitudinal medial fascicle (Flm), in the NFS, and in the oliva inferioris (oi) of the spinal cord.
Clone 1231 (CDOL_001033D10_1231).
Expression was detected in the ICL of the OB; in some areas of both dorsal telencephalon (Dd, Dld, Dlv, Dlp) (Fig. 6A,B), and ventral telencephalon (Vd and Vv) (Fig. 6A); in the preoptic area (PMp) (Fig. 6B); in the nucleus tuberis posterioris (TP) and in most areas of the hypothalamus (HD, HV, Hc) (Fig. 6C); in the hypophysis (Fig. 6C); in the torus longitudinalis (TL), PGZ of the tectum; in the torus semicircularis (TS), in the some mesencephalic nuclei (IR, MR, NDTL, NGp); in the CbSP and in the CbSg of the cerebellum (Fig. 6D); in the griseum central (gc) of the rhombencephalon (Fig. 6D); and in the Flm and NFS of the spinal cord.
Clone 1234 (CDOL_001033D11_1234).
Clone 1737 (CDOL_001071E05_1737).
Expression was observed in the ICL of the olfactory bulbs; in some areas of the dorsal and ventral telencephalon (Dm, Vv) (Fig. 6J); in the Hd; in some areas of the hypothalamus (HD, HV, Hc) (Fig. 6M); in the TL and PGZ of the tectum (Fig. 6K,L); in the NDTL of the mesencephalon; in the cerebellum (egl, CbSg, and CbSP) (Fig. 6N); and in some nuclei of the spinal cord (NFS and oi).
Clone 1012 (CDOL_001012A03_1012).
1012-mRNA expression was observed in the internal cellular layer (ICL) of the olfactory bulbs; in various areas of the dorsal telencephalon (Dm, Dld, Dlv, and Dlp) (Fig. 6O,P), and of the ventral telencephalon (Vv and Vs); in some preoptic nuclei (PMp in Fig. 6P, and PPp); in the periventricular dorsal and ventral hypothalamus (HD, HV, Hc), and the inferior lobes (NDIL) of the hypothalamus (Fig. 5Q); in the hypophysis (h); in the periventricular gray zone of the OT (PGZ in Fig. 6R), and the marginal proliferative zone of the OT (mpz); in the granular layer of the cerebellum (CbSg) and lobus caudalis (LC) of the corpus cerebelli (Fig. 6R); and in the nucleus fasciculus solitarius (NFS) of the spinal cord.
In recent years, ISH screening has become a widely used and efficient tool, mostly for the study of large sets of developmental regulators. Besides the direct access to the expression profile (with an excellent resolution), its power lies in the fact that it circumvents the genetic redundancy that often hinder the discovery of genes in mutagenesis screening. It should also be noted that it is applicable to a wide range of species, including those in which genetic screens are not feasible. Our model is the medaka, a teleost that has evolved as a vertebrate model for genetic studies, as it shares the main advantages of the two fish species that entered the rank of preferred experimental models: the zebrafish, because of its genetic and experimental amenability (i.e., external development, transparency of the embryo, small size, 2–3 months generation intervals, and high offspring up to 300 eggs per week); and the Japanese pufferfish, because of its compact genome (the medaka has a genome size of approximately 680–580 Mb, which is one-fourth the size of the human genome and one-half the size of the zebrafish genome).
The semi-automated protocol we present here has been carefully optimized through trials and errors, and consistently gives results that are, qualitatively, in the range of those obtained by manual screening, in the same laboratory and using the same species (Nguyên et al., 2001a). Special attention had to be paid to the duration of the proteinase K treatment, which had to be carefully adapted to each embryonic developmental stage; also, because the automat treatment is rather destructive for young (fragile) embryos, they had to be hardened by adding a post-fixation step (see Experimental Procedures section).
As reported elsewhere (see Quiring et al., 2004 and references therein), in the present screen of 1,500 clones (about 800 RNA probes) we found very little repetition, despite the fact that we used a non-normalized cDNA library. (e.g., clones 1265 and 1461 both blasted with mab-21-like protein, while clones 1321 and 1373 blasted with SOX3 protein; however, most of the clones resulted in unique matches). This indicates that embryonic cDNA libraries, and even more so those prepared from embryonic CNS, are indeed quite rich and of very low redundancy, making them excellent tools for ISH screening.
The specificity of the ISH screen presented here is twofold: firstly, we deliberately concentrated on the study of the CNS at relatively late (morphogenetic) stages, and, secondly, we supplemented the embryonic studies with ISH analyses in adult brains. Indeed, the medaka is one of the smallest vertebrates (Yamamoto, 1975) and the size of its brain is small enough to handle it the same way as an embryo (for ISH experiments). This very convenient feature is specific to the small teleosts, amongst the vertebrates used as laboratory models, and we took advantage of it to perform neuroanatomical analyses.
Our screen allowed us to identify a wealth of genes expressed in CNS development, the sequences and expression patterns of which will be uploaded in the MEPD (http://www.embl-heidelberg.de/mepd/). Of 321 mRNA probes that showed a patterned expression in the CNS, 28.35% shared no significant sequence similarity with sequences in the public databases. This category potentially includes previously unisolated medaka genes and/or 3′ untranslated regions. Of particular interest are the 16.51% “unknown” genes found in the screen. Obviously, having at one's disposal detailed expression patterns, at several developmental stages and in the adult brain, is a first step in the characterization of these genes, and allows setting forward testable hypotheses on their function. We shall briefly illustrate this point on one example:
Clone 1012 encodes a hypothetical protein with ankyrin repeats, a domain that has been recognized in more than 400 proteins, including cyclin-dependent kinase inhibitors, transcriptional regulators, cytoskeletal organizers, developmental regulators, and toxins (Sedgwick and Smerdon, 1999); it is similar to mammalian GAC-1, the function of which is at present unknown. We show here that it exhibits a clear expression in the intermediate zone (an open ring-shaped domain lying between the proliferative—marginal—and the postmitotic—central—zones) of the medaka OT at stage 31. We found genes exhibiting this expression pattern to be quite rare in our previous screen (Nguyên et al., 2001a), and, more importantly, this pattern was found to be highly predictive: all known genes in that category encode negative regulators of the cell cycle. This is the case for Ol-KIP, which corresponds to a cyclin-dependent kinase inhibitor of the p21/p27/p57 family (Nguyên et al., 2001b; Candal et al., 2004b); and also for Ol-GADD45γ (Candal et al., 2004a), a member of the GADD45 family of molecules, known to arrest the cycle at the G1 and G2 interphases (see among others Zhang et al., 2001; Vairapandi et al., 2002). Indeed, for this latter gene, we performed functional experiments that confirmed its anti-proliferative function (Candal et al., 2004a). An expression in the intermediate zone of the medaka tectum can thus be taken as a strong indication for a role of that gene in cell cycle arrest. Since the gene corresponding to clone 1012 is syn-expressed (Niehrs and Pollet, 1999) with Ol-KIP and Ol-GADD45γ, we can propose that it is a new negative regulator of cell proliferation, and therefore potentially a new tumor suppressor. Experiments aimed at testing this hypothesis can easily be designed as reported in Candal et al. (2004a).
As noted above, the present screen is novel because of its neuroanatomical aspect. Indeed, we have supplemented the “classical” embryonic WMISH with analyses of serial sections of hybridized adult brains in a large number of cases (in the range of one hundred). This allows us to draw the following conclusions: (1) almost all the genes expressed in CNS development are also expressed in the adult brain, and (2) with a pattern that is generally wider and more complex. In other words, and as illustrated in the examples given in the present report, the genes are quite representative in that respect; in the vast majority of cases, new territories of expression are turned on in the adult brain for a given gene: either new brain areas, for example (as is the case of clone 1123, which is absent in the cerebellar primordium while being expressed in the adult cerebellum) or new cell types or new zones in a given brain area (as clone 1053; that in the adult OT is expressed not only in the mpz but also in the non-proliferative PGZ). We found very few examples of genes expressed during embryonic development and turned off in the adult brain. This point is of special interest in the case of the teleost brain: as it is a structure with continuous growth (e.g., Nguyên et al., 1999; Zupanc, 1999; Wullimann and Knipp, 2000; Ekström et al., 2001), one could expect the so-called “developmental regulators” to continue to assume their function(s) in the adult CNS. Indeed, the general phenomenon of the expansion and diversification of the expression territories indicates that a sizeable part of the genome is co-opted in the adult brain to fulfill functions different than those performed in development.
In conclusion, ISH expression screening is a powerful tool to unravel new genes potentially involved in brain development and function. As it gives, by definition, direct access to the cloned cDNA, it also speeds up the experiments aimed at deciphering the function of these molecules. As shown here, it may also be helpful to propose testable hypotheses for the role of these unknown genes.
Obtainment/Preparation of Embryos
Medaka embryos and young adults of a Carbio strain (kindly provided by Jochen Wittbrodt, EMBL, Heidelberg, Germany) were used in all experiments. Embryos were collected and fixed overnight in 4% PFA (paraformaldehyde) in PBS, pH 7.4. They were then mechanically dechorionated with fine forceps, dehydrated in methanol, and stored at −20°C. Embryos were staged according to Iwamatsu (1994).
Preparation of Hybridization Probes
A directional embryonic (stage 31) medaka anterior brain cDNA library was used (Nguyên et al., 2001a). About 2,500 bacterial clones were spread on LB agar 245 x 245 mm plates (50 μg/ml ampicilline) and incubated overnight at 37°C. Then, individual bacterial colonies were picked by an automat Genetac™ (Gif/Orsay DNA Microarray Platform), grown in 96- or 384-well plates in LB (50 μg/ml ampicilline)/Hogness medium (Sambrook and Russell, 2001), and stored at −80°C.
cDNAs in 96-well plates were PCR amplified directly from bacteria using T7 and T3 primers as described in Nguyên et al. (2001). Successful amplifications were confirmed using agarose gel electrophoresis. Digoxigenin (DIG)-labeled antisense RNA probes were synthesized directly from PCR amplified templates (size between 500 and 3,000 bp). The PCR product (about 300–500 ng in 1 μl) was transcribed in 96-well plates by incubation for 2 hr at 37°C after the addition of a antisense RNA polymerase reaction mix (8U T3 RNA polymerase, Promega, Madison, WI; 10U Rnasin, Promega; 10 mM ATP, 10 mM CTP, 10 mM GTP, 6.5 mM UTP, 3.5 mM Dig-11-UTP, and 10 mM DTT, final volume 6.25 μl). After treatment with DNase RQ1 (0.5U/well, Promega), and adjustment of volume to 20 μl with DEPC water, probes were purified on Sephadex G50 columns in 96-well format. Probes were purified with Multiscreen filter plates (Millipore, Bedford, MA; MAHVN4510).
Automatized In Situ Hybridization
Protocol for WMISH was that of Quiring et al. (2004) except for the proteinase K treatment that has been adjusted to each developmental stage: samples were incubated in proteinase K solution (25 μg/ml in PBST) at room temperature for 5 min for stage-19 embryos, 10 min for stage 25, 20 min for stage 28, 30 min for stage 30, 45 min for stage 35, and 90 min for adult brains. To ensure efficient penetration of RNA probes in whole adult brains, it is preferable to use young (i.e., 2–3 months old; body length < 2 cm) adult brains. Using brains from old (i.e., 1–2 years old; body length about 4–5 cm) medakas may sometimes result in incomplete and uneven labeling. Proteinase K treatment was followed by a post-fixation step in PFA 4%/Glutaraldehyde 0.2%. Additionally, the number and duration of washing steps have been consistently increased.
Double Labeling Ol-KIP mRNA-Proliferating Cell Nuclear Antigen
In situ hybridization was performed as described above. After the developing step, embryos were fixed in Clark's solution for 10 min, dehydrated, wax-embedded, and processed for PCNA immunohistochemistry. To optimize antigen retrieval, microwave pre-treatment was performed as described in Nguyên et al. (2001b).
Histological Analysis, Data Acquisition and Storage
Whole embryos were mounted in 1% methylcellulose (Sigma, St. Louis, MO; stored at 4°C) in tap water and observed under a SV11 Zeiss dissecting microscope. Adult medaka (2 months old) were deeply anaesthetized with MS-222 and immersion-fixed for 48 hr at 4°C in PFA (4% in PBS, pH 7.4). The brains were then dissected out, dehydrated and stored in methanol at −20°C. They were then processed for WMISH as described above. They were subsequently dehydrated, wax embedded, and serially sectioned at 8 μm in transverse plane. Sections were counterstained with Nuclear Fast Red. Sections of adult brains were observed with a Leica DRM microscope. In all cases, pictures were taken with a Nikon DXM 1200 digital camera.
DNA sequencing was performed by Genome Express (Meylan, France). cDNAs were sequenced with T7, with an average length of read of 800 bp. Vector sequences were removed before searching for homologies using BLAST (NCBI, Bethesda, MD). Matches for BLASTx were considered to be significant only when the random background for matches (the expect value, E) was <10-5 with all parameters at default. Eight clones were chosen to illustrate this work. Matches against translated databases (blastx) were used to find the reading frame. Translated sequences were then re-blasted in the protein database (blastp), and conserved domains have been identified in the Conserved Domain Database (CDD; Marchler-Bauer et al., 2003; NCBI).
We thank Jochen Wittbrodt, Nicolas Pollet, and Patrick Lemaire for helpful suggestions on screen procedures; Michael Danger, Maryline Blin, and Violette Thermes for their helpful contribution during the screen; and Laurent Legendre for skillful maintenance of fish.
- 2004. Medaka as a model system for the characterisation of cell cycle regulators: a functional analysis of Ol-Gadd45gamma during early embryogenesis. Mech Dev 121: 945–958. , , , .
- 2005. Expression domains suggest cell-cycle independent roles of growth-arrest molecules in the adult brain of the medaka, Oryzias latipes. Brain Res Bull (in press). , , , .
- 1996. The yeast genome project: what did we learn? Trends Genet 12: 263–270. .
- 2001. Ventricular proliferation zones in the brain of an adult teleost fish and their relation to neuromeres and migration (secondary matrix) zones. J Comp Neurol 436: 92–110. , , .
- 2003. An historical perspective on genomic technologies. Curr Issues Mol Biol 5: 123–127. , .
- 1994. Stages of normal development in the medaka Oryzias latipes. Zool Sci 11: 825–839. .
- 2003. CDD: a curated Entrez database of conserved domain alignments. Nucleic Acids Res 31: 383–387. , , , , , , , , , , , , , , , , , , , , , , , , , , .
- 1999. Morphogenesis of the optic tectum in the medaka: a morphological and molecular study with special emphasis on cell proliferation. J Comp Neurol 413: 385–404. , , , , , , , .
- 2001a. An in situ screen for genes controlling cell proliferation in the optic tectum of the medaka (Oryzias latipes). Mech Dev 107: 55–67. , , .
- 2001b. Expression of Ol-KIP, a cyclin-dependant kinase inhibitor, in embryonic and adult medaka (Oryzias latipes) central nervous system. Dev Dyn 222: 439–449. , , , , .
- 1999. Synexpression groups in eukaryotes. Nature 402: 483–487. , .
- 2004. Large-scale expression screening by automated whole-mount in situ hybridization. Mech Dev 121: 971–976. , , , , , , .
- 2001. Molecular cloning, a laboratory manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press. , .
- 1999. The ankyrin repeat: a diversity of interactions on a common structural framework. Review. Trends Biochem Sci 24: 311–316. , .
- 2002. GADD45b and GADD45g are cdc2/cyclinB1 kinase inhibitors with a role in S and G2/M cell cycle checkpoints induced by genotoxic stress. J Cell Physiol 192: 327–338. , , , .
- 2001. The sequence of the human genome. Science 291: 1304–1351. , , , , , , , , , , et al.
- 2000. Proliferation pattern changes in the zebrafish brain from embryonic through early postembryonic stages. Anat Embryol 202: 385–400. , .
- 1975. Medaka (Killifish) biology and strains. Tokyo: Keigaku. .
- 2001. Ectopic expression of MyD118/Gadd45/CR6 (Gadd45beta/alpha/gamma) sensitizes neoplastic cells to genotoxic stress-induced apoptosis. Int J Oncol 18: 749–757. , , .
- 1999. Neurogenesis, cell death and regeneration in the adult gymnotiform brain. J Exp Biol 202: 1435–1446. .