Origin of magnetotaxis: Vertical inheritance or horizontal transfer?

Lin et al. (1) claim that the origin of magnetotaxis dates back to the Archean using phylogenomic methods. Although we salute the interesting results, we have serious concerns about their fundamental conclusion and interpretation of the results.

The dating of the origin of magnetotaxis was based on the inference of a magnetotactic ancestor of Proteobacteria and Nitrospirae phyla. However, whether the ancestor of Proteobacteria and Nitrospirae is magnetotactic is highly questionable. The biggest problem with their explanation is that it requires the assumption of an enormous number of independent losses of magnetosome genes in all lineages except for the direct ancestors of magnetotactic species (2). Also, the sampling of species in their phylogeny is very limited and biased: 30% of the species are outgroup species, and in Proteobacteria where most magnetotactic bacteria (MTB) were found, only 25 species from five orders were present (as shown in their figure S5) (1).

To further assess the possibility of a magnetotactic ancestor of Proteobacteria and Nitrospirae, we carefully selected 258 representative species covering 32 orders and reconstructed the species phylogeny. Assuming that the ancestor of Proteobacteria and Nitrospirae is an MTB, the loss of the entire cluster of magnetosome genes has to be assumed to occur at least 26 times (Fig. 1). To the best of our knowledge, no precedence of such a large number of gene losses has been described. Thus, our result strongly argues against the origin of MTB before the divergence of Proteobacteria and Nitrospirae (3).

Fig. 1.

The evolution of MTB in Proteobacteria and Nitrospirae. The phylogeny of Proteobacteria and Nitrospirae was inferred based on 16S rRNA sequence. GTR+G+I was selected as the best-fit substitution model using jModelTest, version 2.1.2, before phylogeny reconstruction. The maximum-likelihood tree of 16S rRNA was constructed using RAxML, version 8.2.4, with 200 times of bootstrap. The number next to the node denotes the bootstrap value. Only bootstrap values $\ge$ 50 are shown. Lineages of MTB are in blue. Losses of magnetosome genes were inferred, using the parsimonious method assuming the vertical inheritance of magnetosome genes from a magnetotactic ancestor of Proteobacteria and Nitrospirae, and are indicated by a red cross adjacent to the branch. The original tree is available at https://figshare.com/articles/16S_rRNA_tree_of_Proteobacteria_and_Nitrospirae/4897181.

An alternative explanation is horizontal gene transfer (HGT). The authors exclude the possibility of HGT based on the consistency between the phylogeny of magnetosome genes and the taxonomic phylogeny and the similar substitution rate per synonymous site (dS) between magnetosome genes and three housekeeping genes (1). However, the power of their detection of HGT is largely limited by the small number of MTB (as shown in figure 1B of ref. 1) (4). Also, dS was likely saturated due to the long evolutionary distance between analyzed species, making the comparison of dS unconvincing (5). Therefore, it is possible that magnetosome genes originated in Alphaproteobacteria and were acquired by some species from Nitrospirae and Deltaproteobacteria via ancient HGT followed by limited losses of magnetosome genes within each lineage (3, 6, 7). If so, the origin of MTB should be no earlier than the divergence of Alphaproteobacteria, which is around 2.0 giga annum (Ga) (8, 9). Also, magnetosome genes might be transferred from unknown or extinct lineages to extant MTB (2, 10). In either case, HGT can greatly simplify the evolutionary scenario of MTB as it not only avoids assuming a bewildering number of parallel gene losses but also well explains the sporadic distribution of MTB (2).

In summary, vertical inheritance alone is insufficient to interpret the evolution of MTB given the genomic data available. Hence, the origin of magnetotaxis during the Archean, which is based on the inference of a magnetotactic ancestor of Proteobacteria and Nitrospirae, is highly speculative.