The main outcome of this project is the significant improvement of the gene inventory for the Australian sheep blowfly Lucilia cuprina. Amongst other applications, this new resource presents promising benefits to such areas as medical, forensic, pest control, and the understanding of genetic adaptation to insecticides.
Barring major gene expansion or contraction, and assuming that L. cuprina has the same number of genes as in D. melanogaster (~13,600) , the 7,464 unique gene clusters we found in our EST libraries would account for up to 55% of the genes present in the species. The actual percentage is much lower due to (but not limited to) the TGICL assembly parameters and the presence of 4.7% short (≤ 100 bases) sequences (Figure 2). An estimate of 24.1% gene coverage was obtained by limiting homology comparison to L. cuprina and D. melanogaster. However, fast-evolving genes and gene families that have been expanded in the blowfly lineage are under-represented in this analysis. Hence, the estimate of 24.1% could be considered the lower bound of total gene coverage. Nonetheless, this is a conservative yet reasonable estimation given that our cDNA libraries were not experimentally normalized and that only preadult developmental stages contributed to the transcript pool. The EST sequences contain a large number of recognizable protein motifs, as suggested by InterProScan results (Additional file 5), whose protein products are likely to participate in a myriad of biological and cellular processes, as also suggested by Gene Ontology analysis (Additional file 6).
Compared to D. melanogaster, L. cuprina appears to have low GC content and a different codon preference for many amino acids. Despite the fact that the comparison was based on 200 conserved gene homologs, the codon preferences for D. melanogaster are consistent with those reported by Vicario et al. . The higher effective Nc in L. cuprina (43.81) than D. melanogaster (40.89) suggests a weaker selection constraint on codon usage in L. cuprina, at least for these highly conserved genes. It is noted that the 200 sequence pairs analyzed represent only a small fraction (1.5%) of the coding sequences in the 2 species; perhaps a different pattern might emerge when less-conserved gene homologs are included. Nevertheless, these results could be useful for training gene-finding algorithms and the analysis of the full genome sequence when it becomes available.
The acquisition of > 3,280 blowfly genes allows more sophisticated experimental systems to be developed in the future. Aside from the improvement in the knowledge about the genetic composition of the species, the dataset provides a foundation for designing gene-based microarrays for expression profiling. Furthermore, the plasmid collections can also serve as a permanent source of cDNA clones for protein expression, in situ hybridization, and even for transgenic manipulation such as those described in [28–30]. The sequence knowledge of the housekeeping genes such as the ribosomal protein genes, tubulin, and actin could serve as internal controls for quantitative real-time PCR. In fact, the need for such reference genes was recently discussed in . The availability of the L. cuprina cDNA sequences would also facilitate quantification of expression profiles of many genes of interest, bypassing the time-consuming gene discovery steps. It is expected that our EST collection will be invaluable for annotating the genic regions of the L. cuprina genome, when it is eventually sequenced. Conversely, the cDNA information could itself serve as a gene database, such that short peptides generated by the high-throughput proteome sequencing, similar to those reported in the brain tissues of another blowfly, Protophormia terraenovae , could be compared, forming a transcriptomic-proteomic feed-forward loop.
We identified genes that are related to insecticide resistance in L. cuprina (Table 3). Isolation of these homologs in L. cuprina would allow their expression patterns to be accurately measured (e.g., by real-time PCR), and their roles in insecticide resistance to be evaluated. PCR assays to screen for naturally occurring DNA polymorphisms (e.g., exon-primed intron-crossing (EPIC) markers) could also be developed to monitor the temporal and spatial distribution of different alleles. While many of their D. melanogaster homologs have been implicated in insecticide detoxification [33–36], some of the genes identified are involved in other developmental processes such as ecdysone biosynthesis (disembodied and spook) [37, 38] and brain function/development (Cyp4g15) . The proportions of the new L. cuprina homologs represent only a small fraction of these 3 detoxification gene families (see [40–42]). With the advent of next-generation sequencing (NGS) technologies, large-scale genome or transcriptome sequencing has become increasingly popular. For example, transcriptomic analyses using NGS have now been reported in many non-model insect species [43–48]. Similar approaches could be extended to L. cuprina and other related blowfly species, to enable a more comprehensive assessment of novel insecticide targets.
Another important application of our newly identified ESTs was to improve the genetic map of L. cuprina. ESTs can be converted to a set of anchor loci for linkage mapping, as has been repeatedly shown in other insects [49, 50]. We adopted a conservative "reciprocal best hit with strong homology" strategy in the selection of homologous markers, in which D. melanogaster served as the primary reference. A. gambiae, which diverged from the Lucilia and Drosophila lineages about 250 MYA, acted as an outgroup to improve the confidence in orthology calling, as sequence homology needed to reach the E-value of ≤ 1e-50 threshold to be considered further. In other words, we opted for confidence rather than sensitivity in our search for orthologous markers.
The EST-derived markers constituted a substantial proportion of anchor loci in the present study and were useful for inferring chromosomal synteny (Figure 5). Linkage assignment of 41 markers allowed us to conclude that chromosomal synteny is high between the Lucilia and Drosophila lineages. Our results are typical for higher dipteran species, as suggested by previous studies [17, 51, 52]. Several chromosomal fusion/dissociation events have nonetheless been documented within the Drosophila genus. For example, the fusion of Chr 4 (Muller's element F) to an autosome was found in Drosophila willistoni . Moreover, comparison between mosquito and Drosophila reveals that patches of syntenic regions are scattered across many chromosome regions . Our mapping results suggested that gene content on each of the Muller's elements in L. cuprina can, to a large extent, be predicted from the D. melanogaster map. However, the obvious cases of synteny violation (Figure 5; Additional file 9) would mean that direct extrapolation of linkage information from D. melanogaster would require extra caution. The interspecies differences should justify future de novo construction of linkage maps for L. cuprina, with denser markers.
The 298 putative orthologs effectively form a pipeline for future comparative mapping efforts (Additional file 8). Their chromosome addresses in D. melanogaster span virtually all regions of the genome, allowing flexible control over marker density for genomic regions of interest. Several chromosomal areas in Lucilia are of significant historical importance: the Scallop/Notch [55, 56] on chromosome II and the Rop-1 [12, 57] regions on chromosome IV. In fact, several gene markers generated in the present study have already been utilized to understand the patterns of selective sweeps around the Rop-1 locus . The marker pipeline also offers a starting point for fine scale mapping of the fitness modifier locus (M), which is believed to counter the fitness disadvantage of the diazinon-resistant flies in the absence of insecticide [59–61]. Together with an appropriate genomic library, these newly acquired ESTs provide an ample supply of markers for positional cloning of the M locus.
The evolutionary origin and phylogeny relationship among blowfly species has been of great interest to many researchers, owing to its medical and forensic implications [62, 63]. With the much expanded gene repertoire, some of the L. cuprina genes identified here, especially those that show least similarity to other known sequences could be utilized to develop species diagnostic assays. The current EST sequences would greatly complement such an exploration.
While the assemblage of 29,816 ESTs into 7,464 was straight forward, the interpretation of the information contents requires regular re-adjustment, in light of the constantly expanding sequence databases in other species. In order to evaluate the coding components of the newly acquired sequences, they were sorted according to the level of homology to their counterparts in the Genbank reference protein database, producing a typical BLAST significance spectrum (Figure 1). It is anticipated that such a spectrum would change over time. As new sequences from other organisms become publicly accessible, it would simultaneously alter the structure of the existing sequence databases and hence the BLAST results. The recently released EST collections (116,737 reads) from 3 closely related taxa (Glossina, Cochliomyia and, Muscinae) clearly illustrate this notion (Figure 1).
Given that the number of non-redundant sequence clusters depends largely on the assembly settings, the "90% identity over 50 bases" requirement could be viewed as a balanced option, but might not be an optimized condition for all genes. One indication is the presence of residual sequence redundancy in the dataset, presumably due to the natural existence of splice variants, transcript isoforms, natural polymorphisms, or genuine gene families. Hence, it is worthwhile to disassemble relevant contigs that belong to the gene of interest and find the most appropriate parameters to reassemble these reads. Furthermore, we did not impose any restriction on the length of the sequences, i.e., removal of assembled contigs or reads less than a certain length (e.g., 200 bases), because such sequences could be part of the untranslated regions of many legitimate mRNA transcripts. As more similar EST sequences from closely related taxa become available, these short reads might ultimately be informative in the future. In summary, the TGICL assembly described in this paper only represents a generic, non-discriminatory clustering approach for the entire dataset, and re-assembling for the original ESTs might be necessary to produce the most accurate assembly for a given gene or a set of related genes.