The Planctomycetes-Verrucomicrobia-Chlamydia (PVC) superphylum is a bacterial group whose members present features that are usually not found in bacteria but are more often associated with eukaryotes or archaea. One such feature is the presence of a complex endomembrane system inside the cell, which is sustained by proteins (such as gp4978) that display structural similarities to the eukaryotic membrane coat (MC) proteins involved in coated vesicles (e.g. clathrin, COPII Sec31 and COPI subunits alpha and beta′), as well as in the nuclear pore complex (e.g. nup85 and nup133) 1. We and others have suggested that some of these bacterial features are related by vertical descent to their archaeal or eukaryotic counterparts, leading to alternative eukaryogenesis hypotheses 2–6.
The suggested homologies were recently criticized in a ‘Think again’ article 7. Here, we will not reply to these criticisms directly, since most of them have been clarified elsewhere 6. However, we would like to address the supplementary material that accompanied this paper, as it directly criticized our paper describing the discovery of MC-like proteins in members of the PVC superphylum 1. This supplementary material ends with the statement, repeated in similar form in the main text, that ‘the conclusion of 1 that eukaryotic MC proteins have evolved from a single family of PVC-specific gp4978-like proteins contradicts the available sequence, structure and phylogenetic data’.
This conclusion was derived from the following three claims:
- (i)Our analysis was based on truncated proteins and the domain architecture of full-length gp4978 is more complex with an additional haem-binding domain at its C-terminus,
- (ii)The structure of the alpha-helical repeats in gp4978 and its bacterial homologues (HEAT repeats) are different from the alpha-helical repeats (CLTH repeats) in eukaryotic MC proteins and
- (iii)Close homologues of gp4978 are present in multiple representatives of the phylum Bacteroidetes in addition to the members of the PVC superphylum.
Of those three claims, the first two are incorrect and the last one is correct but is irrelevant to our hypothesis.
Firstly, the claim that we analysed truncated proteins is false, as can be appreciated from Table 2, Fig. 2 and Text S1 of ref. 1. As indicated in the Material and methods section 1, at the time of our analysis the genome of the Planctomycete Gemmata obscuriglobus was (and still is) unpublished. We therefore translated the genomic sequence in six frames and considered an open-reading frame as the longest fragment between a methionine and the following stop codon. Thus, the proteins we analysed are at worst too long (because the starting methionines are unknown). It can be see in Table 2 that our prediction for gp4978 is 1158 amino acids long, the corresponding protein GobsU_11075 is 1144 amino acids long. Thus, the proteins we analysed are not truncated. Concerning the architectural complexity, we hardly see how the presence of an additional 140-amino acid Ct domain (that is in fact represented in our Fig. 2) alters our conclusions about the particularity of the MC architecture. In summary, contrary to the above claim, we analysed full-length proteins and the additional Ct domain does not change our conclusions concerning the MC architecture, the main point of our paper.
The second claim that the structures of the bacterial and eukaryotic alpha-helical repeats are different is an incorrect generalization. The protocoatomer theory states that the MC architecture is composed of an Nt-beta-propeller followed by a Ct-alpha-solenoid 8. There is no doubt that all proteins we have detected in PVC members do have the MC architecture (Table 2 in ref. 1). In addition, and unlike stated in ref. 7, eukaryotic MCs do not all contain clathrin-like repeats but instead display an impressive range of structural variation on a common framework 9. It is important to state here that despite the homology between eukaryotic MCs, their sequences and structures have diverged almost beyond the point of recognition. Notwithstanding this divergence, it is now broadly accepted that the eukaryotic MCs have a common origin, a prediction we made four years before the first crystallographic support was obtained 8, 9. A structural comparison of the alpha-helical repeats using one of the most accurate tools available today, COPS (topsearch.services.came.sbg.ac.at, 10), reveals that the predicted structures of the bacterial MCs do not differ more from the eukaryotic MCs than the eukaryotic MCs do between themselves. For example, the eukaryotic MC that is most structurally similar to clathrin (PDB Code 1b89A) is COPI coatomer beta′ (3mkqC) with 124 structurally equivalent residues, and the second closest is Nup85 (3f3fD) with 113. The same clathrin structure has 118 residues that are structurally equivalent with the classical HEAT repeat protein (1b3uB), predicted to represent the bacterial MCs. Thus, structurally, the bacterial and eukaryotic MCs are at least as similar as eukaryotic MCs are to one another.
The third claim concerns the presence of MC-like proteins in another group of bacteria, the Bacteroidetes. Here, the authors are correct, but this is irrelevant to the conclusions of our paper. Furthermore, some Bacteroidetes, like some Planctomycetes, are marine organisms; it is therefore likely that members of these two groups of bacteria or their ancestors have exchanged coding genes. However, this does not modify our conclusions.
In conclusion, we did not analyse truncated proteins, our report was based on full-length bacterial proteins, and the presence of an additional Ct domain is irrelevant to our conclusions. The reported PVC proteins do have the MC architecture. The structural differences (or similarities) between bacterial and eukaryotic MCs are in the same range as the ones between eukaryotic MCs. Finally, the presence of MCs in Bacteroidetes does not modify our conclusions concerning PVC MCs.
However, we still agree that we do NOT have strong evidence establishing the homologous relationship between bacterial and eukaryotic MCs. In case of suspected remote homology, functional similarity is usually taken as an important argument supporting the likelihood of homology. The functional similarity that we have reported between the MC proteins, the main point of our studies 1, 11, were not discussed in ref. 7.
Thus, the criticisms described in ref. 7 are either incorrect or irrelevant and do not justify the statement that our conclusion contradicts available sequence, structure and phylogenetic data. Our conclusion that ‘the compartmentalized bacteria of the Planctomycetes-Verrucomicrobia-Chlamydiae superphylum have membrane coat-like proteins’ that seem to be involved in the endomembrane system of those bacteria remains valid. As suggested, PVC might have lain on the path to eukaryotic endomembrane system development. We look forward to further exciting discoveries regarding this fascinating bacterial superphylum.