Reviewer's report 1
Arcady Mushegian, Stowers Institute for Medical Research (with additional contribution from Manisha Goel).
This is an interesting work, starting to trace the unusual path of evolution of malate synthase and isocitrate lyase in animal kingdom, with additional discussion of what might have been going on with these genes in bacteria.
I suggest that the authors do the following:
1. Due diligence with the databases of unfinished genomes: I did a quick tblastn against the environmental sequence genomes at NCBI and saw at least one entry that codes for the same domain tandem as the two-domain nematode protein: is it one ORF or two, from a nematode or perhaps from a bacterium? The unfinished bacterial genomes – perhaps the donor of two genes to the nematode lineage can be identified among them?
Author response: We significantly expanded the scope of searches in the revised version. Indeed, there are some very similar sequences of the ICL-MS fused nematode gene in the environmental sequence database. We find sequences that are highly similar to the nematode gene in two different configurations, with ICL and MS or not fused. Unfortunately, however, it is impossible to tell whether these sequences are from bacteria or eukaryotes, and therefore, we cannot use this information to resolve any of the issues regarding the potential donor of the ICL-MS fused gene in the nematodes.
When the authors say 'gene transfer into the nematode lineage', how do they know it is not an earlier event (search Schmidtea genome traces perhaps, also Coelenterata)? The same databases, plus ESTs, are needed to account for additional ICLs (I think I can see some in corals).
Author response: Since the submission of the first draft of this manuscript, the cnidarian Nematostella vectensis genome draft has been completed, and now we have included the ICL and MS gene sequences found in this genome into our analysis. Interestingly, the sea anemone genes appear to cluster with bacterial genes as well, albeit with different lineages of bacteria than the nematode genes. We believe that HGT of the fused gene (or operon, with subsequent fusion) in the nematode lineage is the most parsimonious solution. An earlier HGT, e.g., to the common ancestor of Metazoa, would require genes losses in addition.
Reviewer's report 2
Andrei Osterman, Burnham Institute for Medical Research
The strength of the manuscript by F. Kondrashov et al. on Evolution of glyoxylate cycle enzymes in Metazoa is in the detailed analysis of possible evolutionary scenarios that included multiple horizontal gene transfer (HGT) events from bacteria to eukaryotes beyond a symbiotic ancestor of extant mitochondria. Such a case-study is a best possible contribution to the heated debates on this exciting albeit highly controversial topic. Based on a solid comparative analysis of genomic sequences of multiple bacterial and eukaryotic species, which included the delineation of intron/exon structures and "pseudogenized" regions in genomes of Metazoa, the authors presented several plausible scenarios. While differing in details, all of them inevitably include several independent cross-kingdom HGT and gene fusion events. In that regard this paper is a highly recommended reading and thinking material. A weaker aspect of this study is a relatively low impact on our understanding of a metabolic driving force behind these amazing events. Despite a heroic attempt to build on the existing fragmental and highly controversial biochemical data, an emerging picture remains largely obscure. The above notion hardly argues against the authors of this study, but rather provides another illustration of a profound disregard of the basic metabolic biochemistry by the overwhelming majority of the experimental research community in the post-genomic era. A juxtaposition of a monumental effort (and quite a stunning progress) on elucidating minute details of signaling cascades, transcription machinery and other complex systems versus an apparent lack of any drive to finally straighten out basic questions such as: (i) presence or absence of malate synthase activity or (ii) actual function of a malate synthase homologs in placental animals, can hardly be reconciled other than by a popular misconception of the actual depth of our knowledge of basic metabolism. Contrasting this problem and putting it in a fundamental evolutionary context is another (likely unintended) impact of this article.
Overall, I firmly support the publication of the submitted article in "Biology Direct", and I believe that it is a perfect fit for the mission of this distinguished Journal.
Author response: Actually, the original motivation behind this article was to apply computational approaches in an attempt to resolve the paradox of the glyoxylate cycle in mammals and birds: several laboratories have reported that this pathway was functional but the participant enzymes could not be identified. As it happens, the paradox only deepened as we ascertained the presence of "orphan" MS in many animals and pseudogenes in mammals. So it was very much our intent, indeed, our primary goal, to attract attention to the mysterious function of the animal MS, and hopefully, to stimulate relevant biochemical experimentation. The discovery of interesting cases of HGT was, in a sense, a by-product of our research, even if it might have the greatest general impact of the observations reported here.
Reviewer's report 3
Chris Ponting, Oxford University
This is an interesting study aiming at resolving the long-standing issue of whether malate synthase and isocitrate lyase genes are functional in many animal genomes. It is argued that the malate synthase gene is functional in non-eutherian mammals, other vertebrates, echinoderms and arthropods, but that eutherians have lost this gene through pseudogenization. Meanwhile, the isocitrate lyase gene appears to be absent from all animals with the notable exception of nematode worms (where it is fused with malate synthase). It is argued that these two genes were transferred horizontally into several eukaryotic lineages.
One of the issues with which the authors had to contend was of contamination. I agree that the homologues purported to be in mosquito genome sequences appear to result instead from contamination from bacterial sources. Certainly there is evidence that the mosquito genomes are contaminated with these bacterial genes, particularly because they appear to be single exon genes present between clone gaps in unplaced sequence. Contamination might also be an explanation for the postulated horizontal gene transfer of ICL into Dictyostelium and Chlamydomonas lineages, but this wasn't (but should have been) considered by the authors.
Author response:Contamination can be a serious problem in studies such as this one. We have done our best to exclude the possibility of bacterial or other contaminations of Metazoan sequences by examining gene structure, EST sequences, and level of sequence similarity. We concluded that the ICL sequence from the mosquito Anopheles gambeae and one of the MS sequences from the cnidarian Nematostella vectensis are likely contaminants. By contrast, other Metazoan ICL and MS genes, including those of Dictyostelium and Chlamydomonas, contained introns, and most have several independent EST sequences in GenBank which effectively rules out bacterial contamination.
A main finding of this report is that the isocitrate lyase gene is absent from "completely sequenced metazoan genomes". Whilst this appears true, there are numerous isocitrate lyase ESTs from cnidarians apparent in public databases. The authors will need to determine whether these represent independent horizontal or else vertical acquisitions, or else consider whether they are contaminants of the type seen in mosquito genomes.
Author response:Indeed, these EST sequences are identical to the ICL and one of the MS genes from the genome. Thus, we have included the sequences from the recently completed genome draft of Nematostella vectensis; however, we have not included the EST sequences of other species with ESTs for which we could not obtain the cognate genomic sequences.
Table shows PAML non-synonymous and synonymous substitution rates between diverse metazoans. The vast majority of these estimates are not meaningful since saturation of substitution will have occurred, and so Table should not be kept. If the authors believe me to be in error here, they should demonstrate their point using classic tests for saturation.
Author response: Indeed many of the values reported in that table are beyond saturation. However, what is clear from this table is that the nonsynonymous divergence (which is nowhere near saturation levels) is much lower than the rate of synonymous divergence, thus demonstrating functional constraint. Therefore, we opted to keep the table after adding a note of caution to the reader regarding the interpretation of the estimates of the synonymous divergence.
I do not think the authors have presented sufficient evidence that there "was extensive HGT of bacterial MS and ICL genes into several eukaryotic lineages". They imply that these genes have been acquired independently from separate bacterial sources before being fused in nematodes. They discount the possibility that the evolutionary rate of nematode malate synthase might be particularly high without explanation and instead choose a scenario that the bacterial source is as yet unknown. I do not consider that this provides "compelling" evidence for horizontal transfer of both ICL and MS genes from bacteria, which has implications for the title of the manuscript.
Author response: Suppose the genes in the nematode, as well as Chlamydomonas, Dictyostelium and Nematostella (but not other eukaryotic species) have experienced a substantial acceleration of the evolution rate. In order for such acceleration to result in the nesting of these sequences within bacterial clades (which is what we observe) this acceleration must have been coupled with extensive convergent evolution as well. We believe that such convergent mode of evolution is highly unlikely, to say the least, and a much more parsimonious explanation for the phylogenetic tree reported here is a series of HGT events. Thus, we stand by our statement that the evidence for several cases of HGT from bacteria to specific eukaryotic lineages, including nematodes and Cnidaria, is compelling, and there was no reason to modify the title of the paper. Partly in consideration of these comments, the discussion of HGT was expanded in the revised version of the paper.
Moreover, if one considers that the horizontal acquisition of ICL by slime mold and Chlamydomonas lineages might instead be accounted for by contamination of sequences by bacterial sources, there is even less evidence for the "four additional HGTs from bacteria to eukaryotes" proposed.
Author response: As discussed above in our response to Mushegian's comments, it is highly unlikely that the source of ICL and MS sequences from Chlamydomonas, Dictyostelium and Nematostella is bacterial contamination since the genes in these species have introns and ESTs corresponding to these genes are available for all three o fthese species. We mention these observations in the revised text.