Interspecific profiling of gene expression informed by comparative genomic hybridization: A review and a novel approach in African cichlid fishes

Modern genomic approaches have facilitated great progress in our understanding of the molecular and genetic underpinnings of ecological and evolutionary processes. Analysis of gene expression through heterologous hybridization in particular has enabled genome-scale studies in many ecologically and evolutionarily interesting species. However, these studies have been hampered by the difficulty of comparing-on a common array platform-gene-expression profiles across species due to sequence divergence altering the dynamics of hybridization. All too often, comparisons of expression profiles across species were limited to contrasting lists of gene or even of just functional categories. Here we review these issues and propose a novel solution. Exploiting the diverse cichlid lineages of East Africa as our model-system, we then present results from an experimental case study that compares the neural gene-expression profiles of males and females of two species that differ in mating system. Using a single microarray platform that contains genes from one species, Astatotilapia burtoni, we conducted a total of 16 direct comparisons for neural gene-expression level between individual males and females from a pair of sister species, the polygynous Enantiopus melanogenys and the monogamous Xenotilapia flavipinnis. Next, we conducted a meta-analysis with previously published data from two different intra-specific expression studies to determine whether sex-specific neural gene expression is more closely associated with behavioral phenotype than it is with gonadal sex. Our results indicate that the gene expression profiles are species-specific to a large extent, as relatively few genes show conserved expression patterns associated with either sex. Finally, we describe how competitive genomic DNA hybridizations between the two focal species allow us to assess the degree to which divergence of sequences biases the results. We propose a masking technique that correlates interspecific expression ratios obtained with cDNA with hybridization ratios obtained with genomic DNA for the same set of species and determines threshold sequence divergence to reduce false positives. Our approach should be applicable to a wide range of interesting questions related to the evolution and ecology of gene expression.