The pacifastin inhibitor family web site

[Introduction] [Sequences] [Alignments] [Structures] [Phylogenetics] [References]

Created and maintained by Zoltán Gáspári
E-mail: szpari@para.chem.elte.hu

Last modified: 24.04.2003


Phylogenetics

Phylogenetic trees were constructed using a trimmed version of the alignments. Because the exact length of the inhibitors is ot known in cases where only sequence information is available, residues before the first and after the last conserved cysteines were omitted in all cases. This also reduced the number of different sequences available for analysis.

For phylogenetic tree reconstruction, the methods 'neighbour joining' (NJ) and 'parsimony' (PARS) were used as implemented in the program PAUP (version 4.0; Swofford, D. L. 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.). With both methods, a consensus of 1000 trees was generated for both the protein and nucleotide sequence alignments.

Alignment/MethodConsensus tree (nexus)Consensus tree (jpg)
Protein / NJpif_P_nj.nexpif_P_nj.jpg
Nucleotide / NJpif_N_nj.nexpif_N_nj.jpg
Protein / PARSpif_P_pars.nexpif_P_pars.jpg
Nucleotide / PARSpif_N_pars.nexpif_N_pars.jpg


One of the common feature of the trees is that the Oryza sativa inhibitors do not form a separate group but rather cluster with the insect sequences. To determine whether this is an indication of horizontal gene transfer, analysis including more plant sequences (if found) would be needed.
The trees are indicative of a number of independent gene multiplication events, as exemplified by the PLDs (pacifastin light chain domains), the Anopheles gambiae and Bombyx mori inhibitors. In the case of the locust inhibitors originating from closely related species, the gene duplication and species diversification events give rise to a more complex pattern. Independent duplication is also supported by the diversity of intermodular sequences. The only gene sequence presently available contains no introns between the modules, thus, the role of exon shuffling can not be proved. We note that there are vertebrate proteins [PatthyL] containing tandemly organized protease inhibitory domains of multiple types with no interdomain introns either, which phenomenon was explained by supposing intron loss by the authors. Gene conversion events should also be considered as a source of the observed pattern.
The differences in the tree topologies are suggested to be a result of the short length of the input sequences.