Dyneins across eukaryotes: a comparative genomic analysis

Abstract

Dyneins are large minus-end-directed microtubule motors. Each dynein contains at least one dynein heavy chain (DHC) and a variable number of intermediate chains (IC), light intermediate chains (LIC) and light chains (LC). Here, we used genome sequence data from 24 diverse eukaryotes to assess the distribution of DHCs, ICs, LICs and LCs across Eukaryota. Phylogenetic inference identified nine DHC families (two cytoplasmic and seven axonemal) and six IC families (one cytoplasmic). We confirm that dyneins have been lost from higher plants and show that this is most likely because of a single loss of cytoplasmic dynein 1 from the ancestor of Rhodophyta and Viridiplantae, followed by lineage-specific losses of other families. Independent losses in Entamoeba mean that at least three extant eukaryotic lineages are entirely devoid of dyneins. Cytoplasmic dynein 2 is associated with intraflagellar transport (IFT), but in two chromalveolate organisms, we find an IFT footprint without the retrograde motor. The distribution of one family of outer-arm dyneins accounts for 2-headed or 3-headed outer-arm ultrastructures observed in different organisms. One diatom species builds motile axonemes without any inner-arm dyneins (IAD), and the unexpected conservation of IAD I1 in non-flagellate algae and LC8 (DYNLL1/2) in all lineages reveals a surprising fluidity to dynein function.

Figure 1. Use of HMMs to identify DHCs encoded in the genomes of 24 diverse eukaryotes.

A) Performance of the Pfam dynein heavy HMM, PF03028.5, against group-specific HMMs. The y-axis shows score for group-specific HMM giving highest-scoring match. All predicted polypeptides in the 24 genomes with a score >0 on either axis are shown. B) Histogram of the distribution of matches to group-specific HMMs used to define the dynein heavy dataset (score > threshold). On the basis of the distribution, a liberal threshold was chosen to encompass the vast majority of dynein-like sequences without the inclusion of excessive numbers of false positives and extremely divergent sequences that accumulate in the low-score tail.

Figure 2. A Bayesian phylogeny for the DHC sequences from 24 diverse eukaryotes.

Prefixes: Caeel, Caenorhabditis elegans; Chlre, C. reinhardtii; Crypa, C. parvum; Dicdi, D. discoideum; Drome, D. melanogaster; Giala, G. lamblia; Homsa, H. sapiens; Leima, L. major; Ostlu, O. lucimarinus; Phatr, P. tricornutum; Physo, P. sojae; Plafa, P. falciparum; Sacce, S. cerevisiae; Schpo, S. pombe; Takru, T. rubripes; Tetth, T. thermophila; Thaps, T. pseudonana; Toxgo, T. gondii; Trybr, T. brucei. (No DHC: A. thaliana, C. merolae, E. histolytica, O. sativa and P. trichocarpa). For display, the tree has been rooted by bisecting the longest internal branch,although the true position of the root is unknown. Topology support for selected nodes is indicated (Bayesian partial bootstraps/ML/NJ/MP). Bootstrap values give a conservative estimate of the confidence that a particular group of sequences are monophyletic (94). Generally, groups with >90% bootstrap support were considered to be well supported and those with >70% bootstrap support to have some support. Italicized grey values give clade support, excluding the similarly highlighted sequences. Bootstrap values for all nodes from all four inference methods are given in File S3. Cyto 1, cytoplasmic dynein 1; Cyto 2, cytoplasmic dynein 2.

Figure 3. Bayesian phylogenies for: A) dynein IC and B) Tctex1/Tctex2 family LC sequences from 24 diverse eukaryotes.

Prefixes as in legend to Figure 2. For display, trees have been rooted by bisecting the longest internal branch. Topology support for selected nodes is indicated (Bayesian partial bootstraps/ML/NJ/MP). Bootstrap values for all nodes under all four methods are given in Files S4 and S5.

Figure 4. The distribution of dynein and IFT components across 24 diverse eukaryotes.

A) Cladogram showing the likely evolutionary relationships of the organisms analysed and the inferred DHC repertoire in ancestral organisms (changes to the repertoire are shown). B) Presence (dot) or absence (circle) of identifiable orthologues of five DHC classes (DHC); six IC groups (DIC); nine LC groups (DLC); one LIC (DLIC); and 10 components of the IFT system (IFT). The names LC11, LC14, LC16, LC18 and LC19 refer to the LCs identified as ‘M_r= 11 000’, ‘M_r = 14 000’, etc., isolated from Chlamydomonas flagella (47, 74, 95). Orthologues were identified by RBB analysis (File S7) and, where necessary, iterative-HMM searches followed by phylogenetic inference (Figure 3). Grey dots indicate sequences failing the iterative-HMM cutoff but giving reciprocating BLAST-hits. Organisms building flagella/cilia are shown in bold. Notes: (a) Caenorhabditis elegans cilia are immotile; (b) Plasmodium falciparum builds flagella by an IFT-independent mechanism; (c) Briggs et al. (56) suggest that there may be cryptic orthologues of IFT57 and IFT72 encoded in the T. gondii genome that are not found in the predicted protein datasets used here.

Figure 5. Relationships between OADβ family sequences (Figure 2) inferred by Bayesian, ML, NJ, MP or BLASTp means.

Prefixes as in legend to Figure 2, with the addition of Trigr: Tripneustes gratilla. Trees are rooted using Chlre_DYHG and Homsa_DYH8 (OADα family members). Different fonts have been used to represent particular groups of sequences. Bars, substitutions per site; *except for BLASTp tree (see Materials and Methods for distance metric definition).