Teleost microarrays: development in a broad phylogenetic range reflecting diverse applications

Functional genomics is a relatively new area of science that can determine which genes are responsible for phenotypic change, providing insight into the molecular control of biological and physiological processes, behaviour and response to disease, toxins and environmental challenge. Gene expression profiling with microarrays is the most commonly used technology within the functional genomics field and involves fluorescent hybridization of cDNA or cRNA onto slides containing thousands to tens of thousands of gene features. Depending on their size and complexity, microarray platforms can facilitate genome-wide physiological research, supporting both hypothesis- and discovery-driven approaches to identify the response of the transcriptome to various experimental conditions. Microarrays for transcriptional analysis can be fabricated using either amplified DNA from cDNA clones or polymerase chain reaction products or through oligonucleotide synthesis. Oligonucleotides synthesized in situ on the slide are the most advanced technology and can facilitate the construction of very high-density arrays spanning the entire genome of an organism. In the post-genome sequencing era, microarrays have been developed from a wide variety of ‘model’ organisms. While fishes have long been considered ideal experimental models for physiology, developmental biology, environmental toxicology and endocrinology research, until recently, they were poorly represented in functional genomics research, in part because complete genome sequences in fish are limited to a small number of species, e.g. zebrafish Danio rerio (Hamilton), medaka Oryzias latipes (Temminck & Schlegel), pufferfishes Tetraodon nigroviridis Marion de Procé and Takifugu rubripes Temminck & Schlegel, and three-spined stickleback Gasterosteus aculeatus L. Zebrafish is one exception, having served as the ‘piscine mouse’ (Volff, 2004) for functional genomics studies in developmental biology for over a decade (Driever et al., 1994; Ingham, 1997). Given the availability of numerous mutant lines, zebrafish have also been used extensively as models for human disease (Dodd et al., 2000; Penberthy et al., 2002). There are many features of fishes that make them ideal candidates for evolutionary study, specifically for enhanced functional genomics focus. Fishes are the oldest, most diverse and species-rich class of vertebrates, having inhabited the earth for >500 million years. The common ancestor between fishes and tetrapods that gave rise to mammals, birds and reptiles existed some 360–450 million years ago (MYA) (Volff, 2004). Half of modern vertebrates are fishes, with over 25 000 species, 95% of which are in the ray-finned fish lineage (Volff, 2004; Crollius & Weissenbach, 2005). Rapid speciation and parallel evolution in some fish lineages, including sticklebacks (Gasterosteidae) and cichlids (Cichlidae), provide ideal models for studying the genetic basis of adaptation (Elgar, 2004). Because fishes inhabit virtually every available aquatic environment, there is an extreme diversity of physiological, behavioural and life-history adaptations available for study. Moreover, with water being in direct contact with many tissues and internal compartments, fishes may be more sensitive to environmental variations in temperature, oxygen, salinity and toxic chemicals, making them ideal models for environmental genomics (Cossins & Crawford, 2005). Because fishes share many developmental and physiological pathways and organ systems with mammals, studies based on model fish species are relevant to human physiology (Crollius & Weissenbach, 2005). Finally, the economic importance of fishes as food consumed by the human population cannot be discounted. The growing fish-farming industry would benefit considerably from the identification of genes involved in key production traits and disease resistance that could be used in marker-assisted selection. Microarray and quantitative trait loci technologies are two genomic approaches used to identify genes associated with phenotypic traits of interest (Rexroad et al., 2003). Wild fishes could also benefit from science-based sustainable management practices that take into consideration the evolutionary capacity of key economic species to adapt to environmental challenges. The haploid genome size of fishes is broader than that of land vertebrates, ranging from <400 Mb (megabase) in pufferfish and medaka to >100 000 Mb in some lungfish species (Elgar, 2004), while averaging 3500 Mb in mammals (Gregory, 2005). Differences in genome size have not, however, translated into large differences in gene content, with similar numbers of genes (c. 30 000) observed between T. rubripes (400 Mb genome) and humans (3000 Mb genome) (Crollius & Weissenbach, 2005). Genome duplication is recognized as one of the most important drivers of evolution (Ohno, 1970) and is postulated to have occurred twice during early vertebrate evolution before the split between tetrapods and ray-finned fishes (Meyer & Schartl, 1999) and again after the divergence of sturgeons (Acipenseridae) from the lineage that led to teleosts (Fig. 1; Taylor et al., 2001; Hoegg et al., 2004). While the process of rediploidization has led to a loss of most gene paralogues in mammals (<2% remaining; Crollius & Weissenbach, 2005), a greater number of genes still exist as paralogues in teleosts (Van de Peer et al., 2002; Irwin, 2004). Additionally, more recent tetraploidization events have also occurred in some fish lineages, with the most recent events occurring in the common carp Cyprinus carpio L. (12–15 MYA; David et al., 2003) and salmonids (25–100 MYA; Allendorf & Thorgaard, 1984). Common carp Cyprinus carpio L. are functionally tetraploid, containing two copies of most genes (Orban & Wu, 2008). In contrast, salmonids are pseudotetraploids, being in an early stage of rediploidization, with a large percentage of gene paralogues still existing within the genome. Recent genome duplication both complicates and makes more interesting genome analyses in these species. Genome sequencing can be expensive, given the large overall size of the genome, and functional genomic analyses must consider the possibility that some paralogues may, through neofunctionalization, have evolved functions different from mammalian orthologues. Furthermore, cross-hybridization between duplicated gene copies may obscure the resolution of some differentially regulated genes. The availability of species complexes containing both recent and ancient paralogues, however, offers a unique opportunity to test evolutionary hypotheses concerning the role of gene duplication in adaptation and life-history variation, and the mechanisms and selection pressures leading to neofunctionalization, subfunctionalization and degeneration of gene paralogues. Phylogenetic depiction after Nelson (1994) of teleost species for which microarrays have been developed. Microarray platforms for species highlighted in blue are described in this issue and those in red are described in previous publications: Danio rerio (Douglas, 2006), Astatotilapia burtoni (Renn et al., 2004), Sparus auratus (Sarropoulou et al., 2005), Paralichthys olivaceus (Kurobe et al., 2005), Takifugu rubripes (Hogstrand et al., 2002), Fundulus heteroclitus (Whitehead & Crawford, 2006), Austrofundulus limnaeus (Podrabsky & Somero, 2004) and Oryzias latipes (Kimura et al., 2004; Ju et al., 2007). *Species for which complete genome sequences are available contain an asterisk. MYA, millions of years ago. Note that branch lengths do not reflect true phylogenetic distances. Three species have, up to now, served as the teleost models for genomics research, D. rerio, O. latipes and T. rubripes, and while microarrays have been developed for each of these species, the zebrafish has been the most widely used for functional genomics research (Dooley & Zon, 2000), with seven whole genome commercial oligonucleotide arrays and five focused arrays available for this species (Sreenivasan et al., 2008). Various medaka microarrays (both cDNA and oligonucleotide) were recently developed not only for developmental research (Kimura et al., 2004) but also for toxicogenomics (Kim et al., 2006) and research on environmental stress (Ju et al., 2007). This issue presents 14 manuscripts describing the development and validation of arrays from 10 non-model teleost species, spanning three superorders and six orders of ray-finned fishes (Fig. 1). Multiple arrays are presented for three of the species, including rainbow trout Oncorhynchus mykiss (Walbaum) (2), Atlantic salmon Salmo salar L. (2) and fathead minnow Pimephales promelas Rafinesque (3). Five years ago, virtually all arrays based on non-model organisms were constructed using cDNA platforms, so it is highly interesting that in this issue, over half of the arrays described are based on oligonucleotide platforms, many using the state of the art high-density in situ synthesis technologies of Agilent (Agilent Technologies, Santa Clara, CA, U.S.A.). The arrays presented in this issue vary tremendously in size, complexity, degree of annotation, methodological development and motivation, but they are all available to the general research community and each provides key information that will allow researchers to assess their potential usefulness to unique questions and species (Table I). The broad phylogenetic coverage of these arrays will maximize the utility of this technology to address diverse research applications in a variety of fish genera (Fig. 1). Moreover, the inclusion of a review on heterologous array applications by Kassahn (2008) is particularly pertinent to enable researchers to assess the usefulness of arrays in alternate species. Importantly, this review outlines a number of techniques with which to deduce the reliability of cross-hybridization signals for individual probes that are important in assessing the validity of gene regulation data using a heterologous platform for comparative evolutionary studies across species. Additionally, some of the microarrays presented in this issue have been tested for heterologous application, including the salmon arrays described by von Schalburg et al. (2008) (Rise et al., 2004), European flounder Platichthys flesus (L.) by Diab et al. (2008) (Cohen et al., 2007), common carp by Williams et al. (2008) (Evans, 2005) and catfish Ictalurus sp. by Liu et al. (2008). Of particular note, Rise et al. (2004) previously demonstrated good heterologous application of the salmonid cDNA array even for distantly related subfamilies of Thymallinae (grayling) and Coregoninae (whitefish), each c. 90% similar to Salmoninae. Through a carefully constructed experiment based on a 70-mer oligonucleotide array in salmonids, von Schalburg et al. (2008) demonstrated a significant drop in signal strength at a 10% divergence level, with near background levels approached at higher levels of divergence. They concluded that the oligonucleotide array platform would probably produce higher levels of variance among probes than the cDNA arrays when used on distantly related species, which perhaps limited its applicability to genera within the family Salmoninae. Alternatively, Olohan et al. (2008) cited successful hybridizations of Coregoninae to their 60-mer salmonid oligonucleotide array, although criteria for determining success were not presented. The motivation for microarray development followed four main themes. The first two themes, toxicogenomics and environmental stress, clearly fit under the umbrella of ‘environmental genomics’, seeking to use expression profiling in fishes as indicators of the state of environmental and fish health. Toxicogenomics research application was overwhelmingly the most common, motivating, at least in part, the development of 50% of the microarrays, including those for three-spined stickleback (Geoghegan et al., 2008), European flounder (Diab et al., 2008), fathead minnow (Villeneuve et al., 2008; Klaper et al., 2008; Kane et al., 2008), salmonids (von Schalburg et al., 2008) and largemouth bass Micropterus salmoides (Lacepède) (Garcia-Reyero et al., 2008). Two application studies explored 17-beta-oestradiol response (Geoghegan et al., 2008; Garcia-Reyero et al., 2008), while others explored responses to methylmercury (Klaper et al., 2008), fadrozole (Villeneuve et al., 2008) and dibenzanthracene (Geoghegan et al., 2008). Alternately, Diab et al. (2008) demonstrated the applicability of the toxicogenomic European flounder array for research on the host immune response. Environmental stress research applications motivated, at least in part, 36% of the microarrays developed, including those for salmonids (von Schalburg et al., 2008), rainbow trout (Olohan et al., 2008), common carp (Williams et al., 2008), goby Gillichthys mirabilis Cooper (Gracey, 2008) and catfish (Liu et al., 2008). The two application studies under this theme centred on hypoxia (Olohan et al., 2008; Gracey, 2008), with others citing a diverse array of published applications. Developmental research was an additional motivating factor for the microarrays in halibut Hippoglossus hippoglossus L. (Douglas et al., 2008), fathead minnow (Kane et al., 2008) and salmonids (von Schalburg et al., 2008). Kane et al. (2008) applied their microarray to identify fry-specific transcripts in fathead minnows, whereas Douglas et al. (2008) identified genes associated with larval development in halibut. Commercial aquaculture application was an important motivator in the development of microarrays in catfish (Liu et al., 2008), rainbow trout (Salem et al., 2008), Atlantic salmon (Taggart et al., 2008; von Schalburg et al., 2008) and halibut (Douglas et al., 2008). Genomics and proteomic research for commercial application are largely motivated by the need for improving growth rates and cost-effectiveness, increasing resistance to pathogens and stressors and improving quality of broodstock (Malamed et al., 2001). Within this theme, application studies on immune response to pathogens (Liu et al., 2008), dietary lipid metabolism (Taggart et al., 2008), larval development (Douglas et al., 2008) and muscle atrophy during vitellogenesis (Salem et al., 2008) were presented in this issue. Fish biodiversity is important to humans economically, ecologically and culturally, and its maintenance is an important challenge for the next generations (Volff, 2004). As this journal issue attests, the development and application of microarray technology in fishes has expanded enormously in recent years, with representative microarrays developed in most of the teleostean superorders. What have yet to be developed are microarrays in the older evolutionary lineages, including the early ray-finned sturgeons and bichirs (Polypteridae) that predate the teleostean tetraploidization event, the lobe finfishes [coelacanths (Coelacanthiformes) and lungfishes (Ceratodontiformes)], and sharks and rays (Chondrichthyes). There has been some development of an array from one of the earliest vertebrates, the sea lamprey Petromyzon marinus L. (W. Li, pers. comm.), and a full sea lamprey genome sequence is underway by the National Advisory Council for Human Genome Research of the National Institute of Health, East Lansing, MI, U.S.A. Unlike plant genomes, animal genomes, including those of fishes, have frequent alternative splice variants of genes, the variants often having different end products and functions. Microarrays can be developed which sort out splice variant transcripts, but this of course greatly enlarges the size of the array and the work of generating it from oligos (or cDNAs). But only when this has been achieved will there be a truly effective bridge built between the transcriptome and the proteome. When that happens, molecular physiology will truly have come of age.