Probable horizontal transfer of a gene between a protist and a cnidarian

Horizontal gene transfer has been well documented as a significant feature of genome evolution in prokaryotes [1–3]. The frequency and significance of this process in eukaryotic evolution is much less clear. Confirming that a gene has entered a species by a horizontal route can be difficult, and many previously reported cases of horizontal gene transfer from a prokaryote to a eukaryote have subsequently been invalidated [4]. Because of potential contamination, extra care must be taken when putative horizontal gene transfers are detected in sequence datasets from organisms that may harbor a significant number of parasites and commensals. In a small number of metazoan phyla, a significant fraction of the mRNA contains a leader sequence that is attached by trans-splicing [5]. These spliced leaders are distinct for each phylum and thus provide a tag that can be used to determine the origin of a trans-spliced mRNA. Thus, presence of a spliced leader on a mRNA from a trans-splicing species would confirm that the gene was not derived from a non trans-splicing contaminating organism. Hydra, a member of the basal phylum Cnidaria, is one of the few metazoan taxa in which transspliced leaders are added [5,6]. In a large EST (expressed sequence tag) project with Hydra magnipapillata (www.hydrabase.org), we have identified a cDNA homologous to the flp genes of the parabasalid protist Trichomonas vaginalis (Figure 1A,B). The flp1 and flp2 genes of T. vaginalis are of unknown function, but their expression is dramatically downregulated in response to increasing levels of iron [7]. A third flp gene (flp3) was identified by searching the unfinished T. vaginalis genome sequence with the flp1 and flp2 sequences. The flp genes are not found in any other EST dataset, nor are they found in any of the other sequenced genomes. The flp gene is conspicuously absent from the genome datasets of other protists (e.g., Plasmodium falciparum, Giardia lamblia, the 8x coverage of the Tetrahymena thermophila macronuclear genome, and the approximately 12x coverage of the Entamoeba histolytica genome). A BLAST comparison of the Hydra flp amino acid sequence with the T. vaginalis sequences yields E values of 6 × 10-22 and 5 × 10-20 for flp1 and flp2, respectively, supporting the identification of the Hydra protein as a flp homolog. The Hydra flp cDNA contains a spliced leader identical to spliced leader B (Figure 1A,C), which we previously identified in Hydra vulgaris [6]. Among the GenBank entries from a Hydra magnipapillata EST project carried out at the National Institute of Genetics in Japan, we identified two flp cDNAs that are identical in sequence to the one in Figure 1A, except for the sequence at the 5¢ end. The 5¢ sequence of these cDNAs resembles spliced leader B, but is clearly distinct from it. The spliced leader RNA genes we identified previously in H. vulgaris are located within the 5S rDNA repeats [6]. To confirm that this novel 5¢ sequence in H. magnipapillata is a spliced leader, we used PCR to amplify 5S rDNA repeats from this species. Sequence analysis of one of these repeats showed that it contained a gene with the new spliced leader followed by the expected GT splice donor dinucleotide (Figure 1C). We have termed this new sequence ‘spliced leader D’. Thus, both types of flp cDNAs contain Hydra spliced leader sequences, indicating that neither of them originated from a contaminating organism. One could, however, argue that the flp genes identified as derived from Trichomonas actually come from a metazoan. However, the Trichomonas flp genes have been isolated as cDNA clones and genomic clones and probing of Trichomonas DNA and RNA blots probed with a flp cDNA clone gave strong signals [7]. Furthermore, Trichomonas is grown in the laboratory in a defined culture medium [7]. Thus, it is unlikely that the cloned Trichomonas flp genes are a contamination from a metazoan. The anomalous phylogenetic distribution of flp suggests that this gene has been horizontally transferred [1]. The preferred approach for testing such a hypothesis is phylogenetic analysis with an appropriate outgroup species [8]. However, the limited occurrence of the flp gene makes such an analysis impossible. An alternative explanation for the distribution of flp genes is that they were ancestrally present in all eukaryotes and secondarily lost from most taxa. However, this seems unlikely as the large amount of sequence available for diverse taxa did not reveal flp genes. Rather, it seems most likely that the flp gene was horizontally transferred. The direction of transfer is at present uncertain. We favor the idea of transfer from a parabasalid protist to a cnidarian, but additional information on the distribution of flp genes in parabasalid protists and cnidarians will be required to resolve this question. It will be of obvious interest to learn more about the function of the flp genes in both Hydra and Trichomonas, in order to understand the biological significance of this gene transfer. The conserved cysteines and histidines in the Trichomonas and Hydra flp proteins may mediate metal binding, and could explain the strong response of flp gene expression to iron levels in Trichomonas [7].