Using a rather restrictive definition of amino acid similarity (see legend of Figure 13.1), we scored aminoacids identical and similar to RAG1 for every protein aligned (Table 13.1). Surprisingly, a relatively high score of about 20 amino acids similar and identical to RAG1 was obtained for all homeodomains, despite their diversity. In fact, as many as 8 out of 60 residues (stars in Figure 13.1) are conserved in all proteins examined here and these are identical or similar to those of the RAG1 DNA-binding domain. The other residues similar to RAG1 vary according to the homeodomain family. Among all homeodomains scanned, the median Hox proteins present the best score, up to 30 (Figure 13.1 and Table 13.1). In particular, they are unique in presenting a (V/L)CL(T/S) motif quite similar to the homologous VCLT motif of RAG1. This motif is located at the loop between helix2 and helix3 of the homeodomain.

transib-1 is2lvyek3nd2lk22—has^

transib-3 ¡sslsysd^g^lkss—[jatjg transib-5 lasgvd-smssmskl^-ltffa-fsff urchin rag1 ^ai^gsijhyj-k^dcasnsargajj homo ragi ¡232qh2lsl-52ra32h!3ïrelk fly antp ¡jk^gr-qty- jtlejjek^j

Figure 13.2 Similarity between the NH2 arm of the Transib transposase,

RAG1 and Hox proteins. Same grey tones as in Figure 13.1.

This location and the presence of the highly reactive cysteine residue are indicative of a putative protein-protein interaction motif.

As expected from the similarity with RAG1, metazoan homeodo-mains are also similar with Transib transposases (Figure 13.2). The similarity-domain-1 defined by Kapitonov and Jurka (2005) corresponds to the NH2 part of the RAG1 DNA-binding domain and to the NH2 arm of metazoan homeodomains, which is known to make contact with DNA.

Not all transposases belonging to DNA type II transposons present a helix-loop-helix DNA-binding domain. Not all metazoan transcription factors possess a homeodomain. Thus we find that the similarities observed here are more likely to result from common descent, that is, represent true homologies, rather than convergences. This means that RAG1 and metazoan homeobox genes are issued from transposons of the same family, not necessarily from the same and single lateral transfer event. Their common ancestor, as ancestor of two different transposons, might be far more distant than the common ancestor of all metazoans.

We know that certain homeobox genes are present outside the metazoans, in particular in fungi and plants. When phylogenetic analyses include fungi and plant homeobox genes, animal homeobox genes appear polyphyletic (Bharathan et al. 1997). Some clades are metazoan-specific, such as the 'ANTP super-class' including Hox, paraHox, NK, engrailed, BarH and related genes, others pre-date the divergences between plants, animals and fungi, such as a clade comprising Knotted from plants, Cup genes from fungi and exd/Pbx from animals. A very ancient homeobox gene might have been present in an ancestor eukaryote, giving rise to the present-day diversity of homeobox genes in the three lineages by multiple duplications followed by diversification. This 'classical' scenario requires that a number of the different gene families thus generated have been lost in each lineage (Bharathan et al. 1997).

Alternatively, if homeobox genes are issued from transposons, a scenario involving multiple transfers of related but not identical transposons at different times reconciles the gene tree with the species tree more parsimoniously. In particular, although a number of homeobox genes belonging to different families, including the ANTP-class, have been isolated from sponges, no Hox or paraHox genes have been detected (Richelle-Maurer et al. 2006). We thus suggest that an ANTP-super' transposon has invaded the common ancestor of all metazoans, and that in a second event, after the divergence between Porifera and Eumetazoa, the same or a related transpo-son has invaded the eumetazoan ancestor, generating the 'Hox-extended' family, comprising the Hox, paraHox, Mox and Evx/eve genes. After transfer, these transposons would have been 'domesticated' during evolution (Volff 2006). The role of transposable elements as a source of genetic evolutionary novelties is now better acknowledged (Biemont and Vieira 2006). Domestication must have involved a reduction of transposase activity, but it does not need to be rapid. Maintenance of some transposase activity may account for the high numbers of duplications and rearrangements of the Hox-extended family observed in the Cnidaria (Chourrout et al. 2006).

possible transposition events in the hox complex in the drosophilidae lineage

We can expect that during the course of transposon domestication, the transposase/recombinase activity of metazoan homeodomains has been progressively reduced or lost. Nevertheless, we wondered whether some transposition events could have occurred. We focused on Hox genes because, as they are not present in sponges, they might be the more recent and less derived homeobox genes. However, we surmised that transposition of Hox genes would have been lethal in organisms that develop progressively by posterior addition (Hughes and Jacobs 2005), because temporal collinearity would require the integrity of the Hox complex (Duboule 1994, Deutsch and Le Guyader 1998, Monteiro and Ferrier 2006).

Progressive development has been lost in certain bilaterian taxa, among which long-germ band insects. In these animals the Hox complex has a disturbed structure, as shown in the silk moth

Bombyx mori (Lepidoptera) (Yasukochi et al. 2004) and in the Drosophi-lidae (Negre et al. 2003). In Drosophila pseudoobscura, the Deformed Hox gene is in the same orientation in the ANT complex as the other Hox genes, whereas it is inverted in D. melanogaster (Randazzo et al. 1993). Despite this rearrangement, expression and function of Dfd and those of the neighbouring genes do not differ between the two Drosophila species. We thus inferred that the inversion must involve a segment larger than the mere Dfd transcript, including all relevant cis-regulatory sequences. Figure 13.3 shows the inversion by plotting the sequences of the Dfd region from these two species against each other.

We have drawn the structure of the Hox cluster on a phylogenetic tree of 12 Drosophila species whose complete genome sequence is currently available (Figure 13.4). It reveals that the Dfd gene has been inverted twice independently: once during the evolution leading to the melanogaster subgroup, once during the evolution to the willistoni group.


0 0

Post a comment