Analysis of the Protist hypoxanthine-guanine phosphoribosyl transferases (HGPRTases)

By Fred R. Opperdoes


The recent publication by Sommer et al. in Molec.Biochem. Parasitol. 78 (1996) 185-193, on the Giardia lamblia HGPRTase, could not convince me that the phylogenetic tree as shown by the authors is correct. Bacteria and protists did not form separate clades and also the trypanosomes did not form a monophyletic group. The presented alignment showed sequences with highly differing lengths and a number of indels. When such an alignment is used in a phylogenetic analysis it may be expected that the branching order is affected by unequal length effects of the sequences as used in the alignment. For the above reasons I have repeated the analysis done by the authors and my results ar shown below.


I have added a number of additional bacterial species in order to increase the reliability of the resulting tree. To get the maximum number of taxa the T. brucei HGPRTase protein sequence was compared with the entire protein database (SwissProt, PIR and Genpep) using a BLAST search and all relevant protein sequences with a significant homology were included. Subsequently the sequences were aligned using Pileup of the GCG package under the standard settings for proteins. The resulting alignment seemed to be much better than that of Sommer et al. When the sequences are analyzed it becomes evident that the trypanosomatid sequences resemble more the prokaryotic sequences than the other eukaryotic sequences, including that of the other protists Toxoplasma and Plasmodium. This alignment was trimmed such that first, all sequences had the same length (pos 89-291) and second, the indels (94-99, 127-133, 164-172), where the alignment raises some doubts, were removed. All other deletions in the alignment were replaced by question marks which are inert in the phylogenetic analysis and the file was then converted to a file in Phylip format. This way the remaining minor indels do not influence the result. This file was used with the Phylip program Protdist to create a PAM distance matrix where distances are corrected for multiple mutations. The the latter file was used together with the program Neighbor to construct a neighbor-joining tree. The resulting treefile was imported in the tree-construction program TreeView and the resulting figure is shown here. Now we find three major monophyletic groups: the eukaryotes comprising metazoa and protists, the prokaryotes together with tritrichomonas and the glycosomal trypanosomatid sequences together with Giardia. Interestingly, when Giardia was removed from the alignment Tritrichomonas clustered with the Trypanosomatidae, rather than with the prokaryotes. Also T.brucei and T. cruzi are paraphyletic. This is undoubtedly the result of the long branching artefact. Indeed the evolutionary rate for the T. brucei enzyme has been twice that for the T. cruzi enzyme. Bootstrap analysis using the programs Seqboot, Protdist, Neighbor and Consense, weakly supports the monophyletic branching of T.brucei and T.cruzi (47%), but the branching points of Titrichomonas (24%), Giardia (33%) and the Tryps are not well supported by the bootstrap analysis. In conclusion.the separation of tryps, Titrichomonas, Giardia and the prokaryotes should best be presented as one single point of radiation.


Last updated: 25 August 1997.

created by :Fred Opperdoes