Phylogenomics supports microsporidia as the earliest diverging clade of sequenced fungi

Abstract

Background: Microsporidia is one of the taxa that have experienced the most dramatic taxonomic reclassifications. Once thought to be among the earliest diverging eukaryotes, the fungal nature of this group of intracellular pathogens is now widely accepted. However, the specific position of microsporidia within the fungal tree of life is still debated. Due to the presence of accelerated evolutionary rates, phylogenetic analyses involving microsporidia are prone to methodological artifacts, such as long-branch attraction, especially when taxon sampling is limited.

Results: Here we exploit the recent availability of six complete microsporidian genomes to re-assess the long-standing question of their phylogenetic position. We show that microsporidians have a similar low level of conservation of gene neighborhood with other groups of fungi when controlling for the confounding effects of recent segmental duplications. A combined analysis of thousands of gene trees supports a topology in which microsporidia is a sister group to all other sequenced fungi. Moreover, this topology received increased support when less informative trees were discarded. This position of microsporidia was also strongly supported based on the combined analysis of 53 concatenated genes, and was robust to filters controlling for rate heterogeneity, compositional bias, long branch attraction and heterotachy.

Conclusions: Altogether, our data strongly support a scenario in which microsporidia is the earliest-diverging clade of sequenced fungi.

Figure 1

Gene order conservation in all pair-wise comparisons between six microsporidian genomes and two chytrid, three zygomycotina, two basidiomycotina, one taphrinomycotina, two saccharomycotina, and two pezizomycotina from the primary set. Results are shown for the two extreme syntenic pairs detection strategies:relaxed uncorrected, which is equivalent to that in [14] and a strict strategy correcting for segmental duplications. Exact counts are provided in additional file 1.

Figure 2

Microsporidian sister-group analysis based on the six microsporidian phylomes. Groups of bars represent the fraction of the phylomes that supports each scenario (only the three best supported are shown, see Additional file 1, Figures S2-S8 for the rest). Differently shadowed bars represent: from darker (left) to lighter (right) gray: all the trees, trees where the branch-support of the parental node of microsporidians and its sister group is higher than 0.8, trees where the alignment has a consistency score over 0.75, alignments with length larger than 500 columns, and the trees that pass all the previous filters.

Figure 3

ML tree derived from the concatenation of 53 widespread, single-copy proteins (see Additional file 1, Table S6 for the list). The alignment was trimmed as explained in the Methods section to remove non-informative positions, resulting in 25,640 positions. The tree was derived using the LG evolutionary model. All aLRT-based support measures were 1.0. Bootstrap analysis was performed based on 100 alignment replicas, and single node with support below 100 is indicated. ML tree from the same alignment was also derived using the C40 CAT model, as implemented in PhyML. Additionally, to account for potential heterotachy, we derived a ML tree with a free rates parameter covarion model recently implemented in PhyML. Finally, a bayesian tree using PhyloBAYES v3.2 was inferred. All these analyses yielded identical topologies (see additional file 1).

Figure 4

ML analysis on a partitioned dataset, according to the number of residues variability among microsporidian sequences. First partition groups 1 and 2 different residues, the second one contains columns with three different residues, the third one contains columns with four residues and the fourth group contains those columns with five or more different residues per column. The table below shows the results of the eight statistical tests implemented in CONSEL when comparing the support of each of the alternative topologies. Rows indicate the different alternative topologies considered (see Additional file 1, Figure S1): A: basal to all fungi, C: grouped with Chytrids, Z: grouped with Zygomycotina, B: grouped with Basidiomycotina, T: grouped with Taphrinomycotina, S: grouped with Saccharomycotina, P: grouped with Pezizomycotina, S+P: placed at the common ancestor of Saccharomycotina and Pezizomycotina, T+S+P: placed at the base of ascomycota, B+T+S+P: placed at the base of dikarya, B+T+S+P+Z: placed after Chytrids, A-C+Z: basal to fungi but Chytrids and Zygomycotina grouped. The columns represent the different statistical tests used: (1) Approximately Unbiased (AU) test, (2) Bootstrap probability (NP) test, (3) same as NP test, but calculated directly from the replicates (BP), (4) Bayesian posterior probability test calculated by the BIC approximation (PP), (5) the Kishino-Hasegawa (KH) test, (6) the Shimodaira-Hasegawa (SH) test, (7) the Weighted Kishino-Hasegawa (WKH) test, (8) the Weighted Shimodaira-Hasegawa (WSH) test. Dark grey represent the topology with the best likelihood, while light grey represent topologies that could not be discarded (P-value > = 0.05) by the specific test.

Figure 5

ML species tree obtained from the concatenated alignment of 53 widespread, single-copy proteins (Additional file 1, Table S6) extended with newly available three microsporidian species and one zygomycotina species (Additional file 1, Table S5). The alignment was then trimmed to remove non-informative columns and columns that contained only gaps for the nine microsporidian species considered. The ML tree was reconstructed using LG as evolutionary model and SPR as tree topology search method as recommended in PhyML. A discrete gamma-distribution with four rate categories plus invariant positions was used, estimating the gamma parameter and the fraction of invariant positions from the data. Branch supports are SH-based aLRT statistics. Nodes with support below 1.0 are marked on the tree.