Reviewer's report 1
Arcady Mushegian. Bioinformatics center. Stowers institute for medical research. Kansas City. Missouri. USA.
1. I suggest that the relationship with ERAD is addressed further, e.g. by including ERAD components into Table 1 and by adding detail to Figure 2A.
We now indicate in the table those peroxisomal proteins that show homology with components of the ERAD pathway. In order to provide more detail to figure 2.A we have included the role of ubiquitine in both the processes of ERAD and peroxisomal import. Moreover we have re-arranged the proteins and arrows so that their mechanism of action is clearer. Nevertheless the exact functioning of either ERAD or the PEX5 pathway for protein import has not completely been resolved.
2. What is the identity of actinomycete-like, cyanobacteria-like, and "unresolved" components of peroxisome in Figure 1? What is the explanation for the existence of the first two groups? What can be said about functions of "resolved" vs "unresolved" groups – any trends there?
The identity of the actinomycete-like and cyanobacteria-like proteins is indicated now in the text, they can also be seen in Figure 3 and the table. We do not have a plausible explanation for the origin of these proteins and that is now mentioned in the text. We discuss as well the observed functional dichotomy observed in the proteins with prokaryotic or eukaryotic ancestry for the resolved cases as well as the preponderance of enzymes in the unresolved cases.
3. On the ancestral reconstruction: which parsimony was used – unweighted or weighted? Are opisthocont and eukaryotic sets of PEX genes functionally coherent, or are there missing components?
We used a simple parsimony approach in deciding where certain proteins appeared in evolution: a protein is supposed to have been present at the root of the smallest partition containing all genomes that have that gene. The functional coherence of the PEX subsets is difficult to assess, since many PEX proteins have no specific function assigned. However, at least for the ancestral eukaryotic peroxisome the subset of PEX proteins recovered are all involved in peroxisomal protein import, as indicated in the text.
Reviewer's report 2
Gáspár Jékely. European Molecular Biology Laboratory. Heidelberg, Germany.
This paper makes a compelling argument for the autogenous evolutionary origin of the peroxisome. Although this was not a surprise given recent cell biological findings showing that peroxisomes grow from the endoplasmic reticulum, the autogenous origin of the organelle is now clearly backed by the systematic bioinformatic analysis of its proteome. Most interestingly Gabaldón et al. found that some components of the peroxisomal proteome (the Pex5 pathway) are evolutionary derivatives of the endoplasmic reticulum assisted decay (ERAD) pathway.
The paper is technically sound and well written, I only have a few comments.
1) I have a problem about how the authors define that a protein has eukaryotic origin. For example the Cdc48/Sec18/Pex6 family seems to have descended from archaebacterial AAA ATPases. What the tree shown in Fig. 1A shows is rather that the multiplication of this ancestral ATPase leading to several paralogs was an eukaryotic event. So the protein family clearly has prokaryotic origin, it is the formation of distinct paralogs that occurred during eukaryote evolution. This should be explained better in the text and this group should be referred to differently, like 'originated by eukaryote-specific duplication'.
We specifically want to make a distinction between horizontally transferred genes and "ancient genes" that were already present at the evolutionary split between the lineage leading to the Archaea and the one leading to the eukaryotes. Although in the case of CDC48 and Pex1 a case can indeed be made that CDC48 represents the ancestral function, given its level of sequence identity with the Archaeal sequences, and that PEX1 resulted from a gene duplication, such a clear scenario is rarely present. We have put more emphasis on the distinction between horizontally transferred genes and genes already present in ancient eukaryotes in the text, and mention the CDC48 duplication explicitly now.
2) The reconstruction of the ancestral state of the peroxisomal proteome hinges on the accepted topology of the eukaryotic tree. If Kinetoplastids are not early branching but the root lies between animals and plants, then one would probably get a different picture. This alternative reconstruction should also be presented and/or the effect of tree topology on the results should be discussed.
The consequences of using an alternative topology in the reconstruction of the ancestral proteome are now mentioned in the figure legend. They indeed lead to a larger set of ancestral Peroxisomal proteins.
3) Several of the eukaryote-specific Pex proteins are not discussed in the text. One is left wondering what could have been the evolutionary origin of these proteins. If it is not clear for most of them, this should be mentioned briefly.
We tried hard to ascertain the origin of all Pex proteins, unfortunately for the Pex proteins not discussed in the text we could not find homologies with other proteins of known function or these were too weak to be considered reliable. We now explicitly mention this fact.
Reviewer's report 3
John M. Logsdon, Jr., Department of Biological Sciences, Roy J. Carver Center for Comparative Genomics, University of Iowa, Iowa City, IA 52242 USA
This paper reports the "phylogenomic" analysis of peroxisomal proteins with an aim to distinguish between an endosymbiotic vs. endogenous origin of this organelle in eukaryotic cells. This has been a long-standing question in the evolution of eukaryotic cells and these authors have provided a compelling analysis that rejects the hypothesis that the peroxisome is of endosymbiotic origin. Instead, the data indicate an endogenous origin of peroxisomes from the endoplasmic reticulum.
The authors compiled a curated set of peroxisomal proteins from two major model systems in which global proteomic studies of the peroxisome have been done: yeast and rat. These protein sets, thus, represent a large fraction of the peroxisomal proteome. The authors then used a systematic and rigorous analysis procedure to identify all of the homologs of these proteins from among available complete genomes (prokaryotic and eukaryotic). For all peroxisomal proteins and their homologs, phylogenetic trees were reconstructed and the topologies were evaluated to determine the evolutionary history of each peroxisomal gene. The analysis methodology used is appropriately robust to the questions asked.
1) Although I wholly recommend the publication of this work in Biology Direct, it should be noted that, during the process of review, another paper reporting the same conclusions has appeared as an "Advance Access" publication at Molecular Biology and Evolution:
A. Schlüter, R. Ripp, S. Fourcade, J. L. Mandel, O. Poch, A. Pujol, "The Evolutionary Origin of Peroxisomes: An ER-Peroxisome Connection". I am satisfied that the approaches taken here are sufficiently different than those used by Schlüter et al. and thus merit separate publication. However, it would be helpful for the authors here to refer to the Schlüter et al. paper in their revision and to compare and contrast their approaches and results if at all possible. In addition, it is suggested that the authors consider changing their title so as to not so closely resemble the Schlüter et al. paper.
The Schlüter paper addresses the origin of Peroxisomal proteins without bacterial homologs. As can be seen from our analysis there is actually a conflict in the conclusion one can draw from on the one hand the presence of proteins with alpha-proteobacterial ancestry and on the other hand from the presence of proteins with ER ancestry. One can only resolve this by addressing the retargeting of proteins with alpha-proteobacterial ancestry as we have done in our analysis. We explicitly refer to the Schlüter paper and its observation of the link with the ER in the conclusion, and have changed the title of our paper.
2) Figure 3, and the verbiage associated with it (last paragraph of "Peroxisomal proteins..." section), is confusing and should either be clarified (expanded) or deleted. The figure seems too abstract to be useful. What do the dashed arrows mean?
Figure 3 depicts the moving of DNA and protein localization in evolution, which is rather central to the manuscript. We have rephrased the legend, including an explanation of the meaning of the dashed arrows.
3) Figure 4 is unnecessary to the main message of the paper and could instead be included as a supplement. In fact, it would seem that the phylogeny of this gene family would be a more relevant figure, given the verbiage in the manuscript.
We have left Figure 4 in. It includes the most relevant part of the phylogeny of the citrate synthase genes and does illustrate how the retargeting of proteins has continued in recent evolution and is even visible in the sequences.
4) The phylogenetic trees that are provided as supplementary data are supplied as a single webpage http://www.cmbi.ru.nl/~jagabald/peroxisome_trees.html with the trees given in newick format. Although providing these data in a supplemental format is perfectly acceptable, the authors should provide graphic versions of each tree. Indeed, the abbreviations used for the sequences/taxa in these trees are apparently not defined anywhere in the manuscript. Thus, a key to the taxa is at a minimum required, but even better would be a clear labeling of all of the taxon names on all of trees.
The taxonomic names of the species and the trees are being included.