Human Molecular Genetics

Expression analysis of Dres genes

The sequence of a significant number of human genes can be found in publicly available EST databases. One of the major challenges of biomedical research for the next decade will be to identify systematic approaches to help understand their function (10). An obvious and productive shortcut to this goal is to start characterizing human genes sharing sequence similarities with those for which a functional characterization has already been performed in other species. This was the idea behind the DRES project: to study human genes showing significant sequence identity to Drosophila mutant genes (19-21). We wanted to take advantage of the functional studies (e.g. biochemical analyses and characterization of mutant phenotypes) previously performed in another organism to study the function of an interesting subset of human genes.

Sequence similarity often corresponds to the structural relationship of the predicted protein products, suggesting that the biochemical functions performed are also similar. However, due to the major changes that have occurred during evolution, leading to striking anatomical and physiological differences between man and fly, the physiological roles of related genes/proteins can be completely different in the two species, even if their biochemical function has been conserved. Nature often uses the same `core groups' of interacting genes or `pathways' in completely different organs, tissues or cellular compartments. It is, therefore, critical to understand not only how a gene/protein works but also where (i.e. in which tissue/organ) and when (i.e. at what developmental stage) it exerts its function. One way to achieve this is to analyze gene expression patterns at different stages during development. This study demonstrates both the feasibility and value of using RNA in situ hybridization for the systematic analysis of gene expression. Obviously, optimization of the speed of each step involved in the process and the use of automation would significantly increase the number of genes tested. Exploring the complexity of the spatial and temporal distribution of expression of Dres genes has provided us with very useful information on several aspects of gene function, which are discussed below.

Expression pattern conservation

For some Dres genes, we found a clear conservation of expression pattern in related structures between mouse and Drosophila. For example, the Drosophila gene Ras opposite (32,33) is highly expressed in the nervous system, where its product acts as a modulator of neurotransmitter release. Similarly, Dres65, highly homologous to Ras opposite, is exclusively expressed in neural tissue in the mouse (Fig. 2C-E). Along these lines, Dres91 and Dres92 are significantly homologous to the Drosophila CDP-diacylglycerol synthase (CDS) gene, involved in light-induced photoreceptor degeneration (34). The Dres92 transcript has been found in the anlage of thymus at E12.5 (Fig. 2B) and in adult retina, while Dres91 is also expressed in the developing neural retina (Fig. 2F and G), indicating that these murine genes might be the real functional CDS homologs and may represent candidate genes for human retinopathies (35). Another striking example of functional conservation found was Dres30, a murine homolog of the Drosophila Serrate gene that is involved in the dorso-ventral patterning of the wing (36-38). Dres30, subsequently renamed Jagged2, is expressed from E9.0 in the surface ectoderm of branchial arches and in the apical ectodermal ridge of limb buds, confirming that similar molecular mechanisms are involved in limb development in insects and vertebrates (31). Noticeably, mutations in Jagged2 were recently found to be responsible for abnormal digit formation in the mouse mutant syndactylism (sm), displaying hyperplasia of the apical ectodermal ridge (39).

It is indeed remarkable that some Dres genes, which are defined by a highly significant sequence homology with Drosophila, often have a similar pattern of expression and exert related biological functions in the development of the metazoan body plan. On the other hand, we found several examples of Dres genes with expression patterns that are strikingly different between mouse and Drosophila. For example, the Drosophila retinal degeneration (rdg)A, rdgB and rdgC genes are strongly expressed in the retina and are involved in the degeneration of photoreceptors (40-42). Expression patterns observed for their respective murine homologs, Dres8, Dres9 and Dres10, suggest different roles for these genes in mammalian development. Dres9, being predominantly expressed in the neural retina, has an expression pattern remarkably similar to rdgB (43). Instead, Dres8 and Dres10 have patterns of expression unequivocally different from their relatives rdgA and rdgC. The Dres8 transcript is detected in a variety of tissues, including the nervous system, gut and vibrissae follicles and Dres10 is restrictively expressed in the peripheral nervous system, within sensory neurons of neural crest origin (44). The fact that the expression patterns of Dres8 and Dres10 are not conserved suggests that their functions have been recruited in other tissues for a novel specialized use.

Expression analysis and genetic programs

A systematic, high resolution expression analysis of developmentally regulated genes can also contribute to the definition of cell types and stages and to the definition of the genetic program underlying the development of complex structures, such as the eye. Among the collection of Dres genes analyzed in this study, several were found to be expressed in the neural retina. However, the spatial distribution of gene expression observed was diverse, in some cases showing layer-specific expression patterns and the presence of boundaries of gene expression. Figure 3 shows the different spatial distribution of six Dres cDNAs in the neural retina at E12.5. Dres45, Dres65 and Dres129 are expressed along the major gradients of differentiation that proceed from the inner to the outer layers and from the center toward the periphery of the retina. In particular, while Dres65 and Dres129 transcripts are exclusively detected in the differentiated ganglionic cell layer, Dres45 is also expressed in some cells of the outer proliferative neuroblast layer. These cells expressing Dres45 seem to be aligned in radial columns, in accordance with the radial pattern of cellular differentiation of the retina. In contrast, Dres16 is uniformly expressed in the entire neural retina and Dres9 is mainly expressed in the outer proliferative neuroblast layer. The most striking pattern of expression was observed for Dres93, which shows a domain of expression spatially restricted to the infero-nasal portion of the retina. These few examples of differential spatial distribution of gene expression in the neural retina demonstrate how broad is the spectrum of genes involved in the development of a single structure and how fine is the tuning of gene expression that needs to be unraveled.

Expression analysis and disease gene identification

Several genes responsible for human disease have already been identified through the screening of EST databases (11,45,46). In these cases, disease gene identification is usually based on the so-called positional candidate gene approach, which relies on the association of positional and functional data (47), i.e. the co-localization of a disease locus and a gene with a suggestive pathogenetic role in the disease phenotype. On the same basis, DRES can be considered potential candidate genes for human diseases. Some DRES genes have already been found to be involved in human disorders, such as the RIEG gene (DRES112), a novel homeobox gene similar to the Drosophila orthodenticle and aristaless genes, found to be mutated in Rieger syndrome (48).

Even if there are examples in which the disease phenotype reflects only a subset of the tissues which express the gene, such as the CBP gene mutated in Rubinstein-Taybi syndrome (49), the knowledge of gene expression patterns contributes significantly to the positional candidate approach by pointing to a dysfunction of a particular organ or tissue. For example, a promising `positional candidate' disease gene identified in this study is Dres119, a fourth human homolog of the Drosophila eyes absent gene. In the mouse we found Dres119 to be expressed in craniofacial mesenchyme, external auditory meatus, dermomyotome, kidney and limbs. Its human counterpart, DRES119, maps to chromosome 6q23, making it a suitable candidate for oculodentodigital syndrome (ODD), an autosomal dominant disorder previously mapped to the same region and affecting development of the face, eyes, limbs and dentition (50). Another candidate disease gene is Dres139, the homolog of the Drosophila kekkon gene, which may be involved in Cornelia de Lange syndrome, characterized by distinctive facial features such as micrognathia, hirsutism, synophrys and bushy eyebrows, low anterior hairline and small teeth, in association with growth and mental retardation, upper limb and digital defects (i.e. short tapering fingers) and delayed skeletal maturation. The transcript distribution for Dres139 is highly restricted to developing cranial and thoracic bones, vibrissae and sinus hair follicles, tooth buds and in the tip of the digits (Fig. 1C-G). In addition, Dres139 maps to mouse chromosome 3, within the human syntenic region for 3q26-qter, where the locus for de Lange syndrome is located (MIM 122470).

A possible involvement of certain DRES genes in particular disorders can also be hypothesized solely on the basis of their expression, especially if expressed in specific organs, such as Dres61, Dres92 and Dres93. The Dres61 transcript, an Eag-like potassium channel, could be detected exclusively in the primordium of the thyroid (Fig. 2A), suggesting an important role in the development of this organ and possibly in the pathogenesis of thyroid dysgenesis. Dres92, homologous to the Drosophila gene CDS, is selectively expressed in the thymic rudiment at E12.5 (Fig. 2B), while a novel homolog of empty spiracles, Dres93, has a restricted domain of expression in the developing neural retina (Fig. 3C).

Expression analysis and mouse models

The data obtained from systematic high resolution expression studies will certainly add important biological information to the creation and use of the preferred model for mammalian development, the laboratory mouse. Recently, there has been dramatic progress in the efficiency of methods addressing gene function in mouse. Positional cloning of mouse mutant genes, induction of new mutants by chemical mutagenesis, targeted mutation by homologous recombination, analysis of polygenic traits and comparative gene mapping are some of the strategies being used to study the phenotypic effects of mutation of specific murine genes (51). A study of the spatio-temporal transcript distribution is the first requirement in order to fully comprehend which cell types, developmental stages and cellular processes can be altered by the mutation, as well as the compensatory effects that may be due to expression of related genes.

Furthermore, systematic RNA in situ hybridization may provide very useful information in order to choose which mouse genes would be more interesting targets for gene inactivation strategies. Recently, systematic targeted mutagenesis approaches have made possible the creation and storage of several thousand ES cell lines each carrying a knock-out of a different gene (52). Many of these genes are unknown. Therefore, knowing the expression patterns of these genes would be of great value before deciding on the generation of null mutant mice.

Expression database

To store, group and search the data obtained in this initial expression study and especially in light of the next phase of large scale data collection, we will deposit the complete expression profiles as edited panels of captured in situ images in the existing DRES database (20), accessible through a World Wide Web interface (http://www.tigem.it/dros.html ). To effectively integrate the high resolution expression data with the extensive structural documentation on genes available in public repositories, we are also planning to submit the expression profiles to the comprehensive Mouse Gene Expression Information Resource (MGEIR) through the mouse GXD Gene Expression Database, both of which are currently under development (53-55).

The comparative approach initiated with the DRES genes, in structural (sequence homology) and functional (expression analysis) terms, represents an effective strategy to obtain useful information on gene function. Moreover, the results have important implications regarding both mouse developmental biology and the evolutionary conservation of genes that orchestrate the development of the metazoan body plan. It is conceivable that all metazoans have a set of core biochemical processes and that duplicated mammalian pathways have adapted to specialized biological functions. If high resolution expression studies are applied on a large scale to ESTs and the information is comprehensively collected in databases, it will then be possible to scrutinize their sequence and/or expression patterns for particular motifs. These data will provide an essential directory to compare sequence similarities with temporal and spatial patterns of gene expression. Ultimately, this information may be of considerable help in predicting clinical consequences of gene dysfunction.

Bioinformatics

EST entries, including homology data, were retrieved from dbEST (8). Computational analysis of EST sequences was performed using the Blast basic local alignment search tool (56), either at TIGEMNet (http://www.tigem.it ) or at the National Center for Biotechnology Information (NCBI), National Library of Medicine (http://www.ncbi.nlm.nih.gov/Recipon/bs_seq.html ).

Murine cDNA clone retrieval and characterization

EST cDNA clones were obtained from IMAGE Consor-tium Clone Distributors (http://www-bio.llnl.gov/bbrp/image/idist_add.html ). Automated fluorescent DNA sequencing was performed using Perkin Elmer 377 Prism machines with both the dye terminator and dye primer cycle sequencing chemistries on double strand plasmid templates.

In situ hybridization

Expression was detected in tissue sections of E10.5, E12.5 and E17.5 mouse embryos using radioactive in situ hybridization according to a published protocol (57). Embryos, embedded in paraffin, were fixed in 4% paraformaldehyde in phosphate-buffered saline. Sets of serial sections cut in three different planes (sagittal, coronal and transverse) covering the entire embryos were hybridized with [³⁵S]UTP-labeled antisense or sense riboprobes. The probes were transcribed from the 3[prime]-untranslated region of the murine ESTs. Sense probes did not show any specific hybridization. Slides were counterstained in Hoechst 33258 dye to stain the cell nuclei. The resulting red color represented the hybridization signal.

The whole mount in situ hybridization experiments on E8.0, E9.5 and E10.5 mouse embryos were performed as previously described using digoxygenin-labeled RNA antisense and sense probes (58). Panels for Figures 1-4 were obtained by capturing and processing images using Nikon SMZ-U dissecting and Ziess Axiophot microscopes and Adobe Photoshop software.