Comparative analysis of the kinomes of three pathogenic trypanosomatids: Leishmania major, Trypanosoma brucei and Trypanosoma cruzi

Abstract

Background: The trypanosomatids Leishmania major, Trypanosoma brucei and Trypanosoma cruzi cause some of the most debilitating diseases of humankind: cutaneous leishmaniasis, African sleeping sickness, and Chagas disease. These protozoa possess complex life cycles that involve development in mammalian and insect hosts, and a tightly coordinated cell cycle ensures propagation of the highly polarized cells. However, the ways in which the parasites respond to their environment and coordinate intracellular processes are poorly understood. As a part of an effort to understand parasite signaling functions, we report the results of a genome-wide analysis of protein kinases (PKs) of these three trypanosomatids.

Results: Bioinformatic searches of the trypanosomatid genomes for eukaryotic PKs (ePKs) and atypical PKs (aPKs) revealed a total of 176 PKs in T. brucei, 190 in T. cruzi and 199 in L. major, most of which are orthologous across the three species. This is approximately 30% of the number in the human host and double that of the malaria parasite, Plasmodium falciparum. The representation of various groups of ePKs differs significantly as compared to humans: trypanosomatids lack receptor-linked tyrosine and tyrosine kinase-like kinases, although they do possess dual-specificity kinases. A relative expansion of the CMGC, STE and NEK groups has occurred. A large number of unique ePKs show no strong affinity to any known group. The trypanosomatids possess few ePKs with predicted transmembrane domains, suggesting that receptor ePKs are rare. Accessory Pfam domains, which are frequently present in human ePKs, are uncommon in trypanosomatid ePKs.

Conclusion: Trypanosomatids possess a large set of PKs, comprising approximately 2% of each genome, suggesting a key role for phosphorylation in parasite biology. Whilst it was possible to place most of the trypanosomatid ePKs into the seven established groups using bioinformatic analyses, it has not been possible to ascribe function based solely on sequence similarity. Hence the connection of stimuli to protein phosphorylation networks remains enigmatic. The presence of numerous PKs with significant sequence similarity to known drug targets, as well as a large number of unusual kinases that might represent novel targets, strongly argue for functional analysis of these molecules.

Figure 1

Similarity of T. cruzi ePKs to those in the 4-kinome dataset. Full-length proteins were tested by BLAST analysis against the database of catalytic domains of all human, Saccharomyces cerevisiae, Drosophila melanogaster, and Caenorhabditis elegans protein kinases. The best E-value is graphed for each kinase, which are clustered according to their classification group. Non-cat: protein kinases predicted to be non-catalytic due to lack of subdomain I or catalytic residues. Different colors were used to facilitate viewing of closely spaced spots.

Figure 2

Unrooted tree of T. brucei protein kinases. The catalytic domains of predicted functional ePKs were analyzed using MRBAYES. Bootstrap values greater than 0.95 are indicated by a dot at the node, and selected lower values are shown. No members clustered with TK or TKL kinases from human or yeast (not shown on the tree). T. brucei sequences are identified by systematic gene IDs. Selected human (Hs), S. cerevisiae (Sc) and D. melanogaster (Dm) were included to provide landmarks and these are shown in red font. Asterisks mark sequences for which the MRBAYES tree conflicted with the BLAST analysis at the group level. Kinases classified as unique through BLAST are marked with "U".

Figure 3

Comparison of L. major and human ePK classification.

Figure 4

Phylogram of CAMK ePKs. Kinase domains from the TriTryp predicted proteins classified as CAMKs by BLAST or CAMK-like from the T. brucei tree in Figure 2 were analyzed using MRBAYES. Also included are CAMKs from all families present in humans (Hs), plus several S. cerevisiae (Sc) and a P. falciparum (Pf) kinase. Nodes with a bootstrap value greater than 0.95 are marked by a dot, while values ranging from 0.7–0.95 are indicated numerically. Similar results were obtained with PAUP*. Kinase domains are indicated by systematic gene IDs, which are abbreviated in the case of L. major to Lm in lieu of LmjF. Additionally, the invariant digits in the T. cruzi systematic names were deleted (all gene names start with Tc00.1047053). Only one T. cruzi allele was included for each gene. ψ marks a T. cruzi and T. brucei gene that are predicted to be non-functional due to a lack of a recognizable subdomain 1. The CAMK-like kinases classified as unique by BLAST are marked, as are the trypanosomatid CAMKs with EF-hand accessory domains. ScCDC28, a CMGC kinase, was used as an outgroup.

Figure 5

Phylogram of STE kinases. Kinase domains from the TriTryp predicted proteins classified as STE by BLAST or STE-like from the T. brucei tree in Figure 2 were analyzed using MRBAYES. Also included are STEs from all families present in humans (Hs), plus examples from S. cerevisiae (Sc), C. elegans (Ce) and D. melanogaster (Dm). Nodes with a bootstrap value greater than 0.95 are marked by a dot, while values ranging from 0.7–0.95 are indicated numerically. Similar results were obtained with PAUP*. Gene names are shown as in Figure 4. ScTPK1, an AGC kinase, used as an outgroup.

Figure 6

Phylogram of NEK kinases. NEK sequences from human (Hs), S. cerevisiae (Sc), C. elegans (Ce), D. melanogaster (Dm), P. falciparum (Pf) and T. brucei (Tb) were analyzed using MRBAYES. TbCK2A1 was used as an outgroup. Nodes with a bootstrap value greater than 0.95 are marked by a dot, values ranging from 0.7–0.95 are indicated numerically. Similar results were obtained with PAUP*. Two NEK kinases had recognizable PH domains that did not achieve the Pfam HMM search cutoff (Tb04.24M18.60, and Tb08.10K10.710).

Figure 7

T. cruzi unique ePKs: similarity to other ePKs and Pfam kinase domain. The E-values for the relationship of individual T. cruzi ePKs as compared to the 4-kinome dataset and to the Pfam domain. Many of the T. cruzi ePKs show strong E-values against the Pfam kinase domain, despite their low similarity to the kinases in human, D. melanogaster a, C. elegans, and S. cerevisiae.

Figure 8

Domain structure of Tritryp ePKs with unusual additional domains. Examples from Table 2 are shown. ePK domains are predicted to be active.

Figure 9

Comparison of T. brucei and human atypical protein kinase classification. See Additional file 4 for systematic gene names.