Adenylate kinase

Further details of the analysis

A trypanosome ADK sequence was found in the database with a PTS1-type glycosomal targeting signal (AL474359, emblnew-acc:AC091526:115093-115737) The ORF was translated and compared with the SwissProt release 39 / TrEMBL release 17 database indexed at European Bioinformatics Institute (EBI, Hinxton UK), containing 671,000 protein sequences, using the NCBI BLASTP program and the BLOSUM 62 matrix (http://www2.ebi.ac.uk/blastall/). Click here to inspect the BLASTP output file. The three best score (E = 3e-41 to 8e-40) was obtained with the ADK from plant chloroplasts (KADC_MAIZE), followed at a distance by a large number of bacterial sequences with E scores ranging 3e-32 to 3e-26. The first eukaryotic sequence was that of Cryptosporidium with an E score of 6e-28.

Ninety-four ADK sequences with E values ranging from 2e-40 to 6e-11 were selected and aligned using the "RunDBClustalW" option in the BLASTP output. The ClustalW alignment is availabe here for inspection. After removal of one incomplete sequence and all regions with gaps a matrix with pairwise distances was prepared. Percentages of identity of the T. brucei sequence with the other ADKs ranged from 45-26%.

The alignment in Phylip format containing 94 sequences and only 90 positions was used to carry out likelihood mapping. The data set appeared to have little phylogenetic signal with 19.2% unresolved and only 72.1% fully resolved quartets. A bootstrapped neighbor-joining tree was very star-like and gave very low bootstrap support for a branching of T. brucei together with Giardia lamblia and the chloroplast ADK of maize. Since the information content of the dataset was clearly too low, a new dataset was prepared with only the T. brucei ADK and the first 21 ADKs from the BlastP output. The new bootstrapped neighbor-joining tree was much improved with respect to the previous one and now provided reasonable bootstrap support (73%) for a branching of T. brucei with Giardia lamblia and the ADK of maize chloroplasts. However, it should be noted that these three taxa represent the longest branches within the tree and thus a long-branch-attraction artefact cannot be excluded here. When the G.lamblia ADK was removed from the dataset, bootstrap support for the clustering of T. brucei and maize increased to 92%.

From this dataset all gaps were removed and the resulting Phylip file with 21 sequences and 158 positions was used for likelihood mapping. The quality of the dataset had significantly improved with only 3.5% unresolved quartets and 92.5 completely resolved quartets.

A matrix with pairwise distances was prepared. Percentages of identity of the T. brucei sequence with the other ADKs ranged from 40-35%.

This set was then used for the calculation of maximum likelihood, neighbor-joining and maximal parsimony trees. The three trees had all very similar topologies, correctly grouping the beta and gamma subdivisions of the proteobacteria. Also all three trees suggested an association (with 73-95% support) of the T. brucei ADK with that of chloroplasts.

Conclusion