I would very much like to thank the reviewers for their thoughtful and very helpful comments and criticisms. Their consideration, patience and time is greatly appreciated.
Since submitting the original draft, a paper has been published outlining the first culture of an Asgaardian organism. This has been added to the introduction during revision (Starting “To date, only one Asgardian species has been cultured,……”
Reviewer’s report 1: Damien Devos, Affiliation European Molecular Biology Laboratory (EMBL), Heidelberg)
In the manuscript “Division of labour in a matrix, rather than phagocytosis or endosymbiosis, as a route for the origin of eukaryotic cells”, the author proposes a novel model for the origin of the eukaryotic cell. The author discusses the introduction of a novel third-space that could be have played an import part in the development of the ancestral eukaryotic cell. In summary, the author proposes that biofilms have played a structural role in the development of the eukaryotes. The issue is of course of interest, of actuality and still an important unresolved question. The manuscript is well written, the ideas developed are of interest and well argued. I believe that the manuscript is of interest to a wide community of readers.
Reviewer recommendations to authors.
Main issues: One of the disturbing point is that the manuscript keep calling archae and bacteria as “Archaea” and “Eubacteria”. This is an historical mistake, as the term eubacteria was introduced to differentiate them from archaeabacteria (by Woese). But now, these terms have been defined as “Archaea” and “Bacteria”. Still mixing both is a conceptual mistake. This has to be fixed thourough the manuscript.
Author response: This was careless and has been changed throughout.
The model introduce the concept of “third-space”, a novel class of frameworks for the origins of complex cellular structures. In particular, it has three requirements L264. The third one is to be “more than a passive arena” for evolution. Please develop as this is quite limited. How so? How could it promote evolution?
Author response: New text has been added to hopefully clarify this point. Section “What are the three spaces?” Paragraph two starting “As the third space evolves from a matrix ….” to “tie together the cellular elements of the ensemble as a coordinated system.”
However, my main issue is with the point ii, L265, where the third-space is required to be ‘stable’ enough to provide the opportunity for eukaryotes to emerge. How long exactly would that be? How long do the author estimate that the presence and stability of the third-space would be required? And related to this, how long are biofilms known to be stable? Would it be realistic to think that particular biofilms could be stable for so long? If so, in which environment? Please develop this fundamental issue. Examples from biofilms or microbial mats would nicely illustrate the point.
Author response: This is difficult and I think that the reviewer has raised a very important question. I have added new text to discuss the question of stability in section “Limitations”, first paragraph starting from “The third space model is a multi-stage process and the degree of stability required need not be the same at each phase…” to paragraph 2 ending “Microbial mats provide an example of a long-lasting microbial community”. Stability, in this context, can be broken into two types, namely physical stability of the matrix ensemble, and/or its ability to efficiently reassemble if it is dispersed. Furthermore, the hypothesis is clearly for a progressive multi-stage transition, and therefore the degree of stability can ramp-up as the process continues. Initially it would be sufficient for the ensemble to efficiently reassemble, presumably driven by obligate mutualistic interactions. As the accuracy of reassembly becomes increasingly a dominant aspect of the system a shift towards physical persistence of the ensemble becomes favourable, possibly aided by the formation of an outer membrane.
Please also detail how this proposal differ from refs 95 and 96.
Author response: New text has been added in the section “Extracellular matrices as models for the third-space”, paragraph two" Starting “Jeckely (101) proposed that early eukaryotic evolution occured in a social biofilm-like context…” until “…. compaction or streamlining of structures to generate an entity more closely resembling a eukaryote”
Also, if the third-space is required to be so stable, why can we not find any intermediary forms in current biofilms? This deserves mention.
Author response: I suspect that the lack of intermediate forms is simply because they were replaced by “fitter” increasingly more eukaryotic-like cells as the process moved forward. Also, I am not sure what we would look for to find any residual intermediate forms. As the manuscript is already rather long, and this is not an easy topic to answer succinctly I would prefer not to add it to the text.
Most of the model is based on gene exchange inside the “third-space”, assimilating it to a genomic reservoir to support gene exchange L250. However, the chapter (L245–254) is quite theoretical and ill supported by examples. Could such examples be provided by example based on known interchanges in communities and biofilms? Specifically discuss bacteria/archaea interchanges.
Author response: I have added new text to expand the discussion of HGT from Bacteria to Archaea in the section "Genomic reorganization towards a eukaryotic pattern, first paragraph “The horizontal transfer of genes between prokaryotes is a fundamental property of their evolution, and has allowed for very extensive genetic mosaicism to occur in some cases, including in inter-domain mosaicism, where both archaeal and bacterial genes are prominent in the same organism (69). For example, the evolution of archaeal Halobacteria was marked by the acquisition of approximately1000 bacterial genes families (67). Interdomain mosaicism is similarly evident among Lokiarchaea species, 33% of their genes are unique to these species, with the remaining 36% of archaeal origin, 28% bacterial and 2% of eukaryotic origin (125). Bacteria may likewise carry a significant archaeal genetic content, thus in Thermotoga maritima, 24% of genes have been imported from archaeal species. Horizontal gene transfer is promoted by growth in biofilm matrices both in Archaea, as has been demonstrated, for example, in Haloferax volcanii (126), and in Bacteria ((127-129)).”
A major issue with the proposal is related to the lipid composition of eukaryotic membranes versus archaeal ones. Could the author elaborate around line 120?
Author response: I have added new text “If the host cell of the ancestral endosymbiotic partnership was archaeal, as is often proposed (33, 63, 77), then, at some stage of eukaryogenesis, it must relinquish its characteristic archaeal membrane-lipid biosynthetic pathways in favour of those of the bacterial passenger cells. The mechanisms and evolutionary rationale underpinning this transition remain unclear.”
In the allowable assumptions, I have a few responses of interest. L303, L-form bacteria are discuted and it is stated that L-form cells have been proposed to have played a significant role in early cellular evolution. However, this is a proposal about early cellular division, not about division of the eukaryotic cell. Please clarify or remove.
Author response: Yes, that is correct. I have removed the statement.
L 306, fusion of mitochondria in the eukaryotic cell is assimilated to fusion events that could have occurred between the participating cells within the third-space matrix. There is however a major difference in the fact that participating cells in the third-space matrix would be of different or differentiating types. To the best of my knowledge, fusion between different cell types have not been observed. Please response.
Author response: At fusion stage the cells in the matrix ensemble have lost cell walls and adopted an L-form lifestyle. Not much is known about fusion of L-form cells, although one report does mentions that L-form cells fuse in culture. Fusion between protoplasts of cells of different type can be accomplished. However, it should be pointed out that these fusions are in the lab not in the wild. I have added text to discuss these points in section “Genomic reorganization towards a eukaryotic pattern”, paragraph 5, starting “Cell fusion is a widespread phenomenon” to “…. and intergenic L-form cell fusions could also occur.”
. Related to this point, L 374 states that “the nascent nuclear membrane arises from the cell membranes of the fused L-form-like multi-genomes”. This is an important jump and worth much more discussion. Why would membrane surround the DNA, and how do the author envision such transition.
Author response: The DNA resides within the resident cells of the matrix, even after loss of cell wall and fusion events occur. The formation of nuclear membrane. Some of these resident cells specialize as primitive nuclei, and therefore the nuclear membrane is an adaption of their earlier cell membranes. It is not generated de novo. The discussion of this point has been expanded in section “Assembly of a primitive nucleus”.
Relatedly, how could membrane then form around the third-space?
Author response: This was not well explained in the original manuscript and was relegated to a few sentences near the end. I have now inserted a new subsection that deals with this question at more length. See first paragraph of section “Membrane encapsulation and cytosolic takeover”.
Another issue is with the assumption that eubacteria outnumbered archaea in the matrix to justify the eubacterialization of the archaeal component of the matrix. Why would the author assume this, if not to justify a posteriory a known bias in eukaryotic genomes? If this is so, this is a deduction, not an assumption.
Author response: The rationale for the assumption/deduction is now given in greater detail in Section “Genomic reorganization towards a eukaryotic pattern”, paragraph 3, starting “While transfer of genes can occur in either direction….”
It would be important to clarify how an increase in cellular complexity is required in order to “overcome problems that occur when molecules brought together from different lineages fail to interlock efficiently”. Please detail and response.
Author response: I agree that this was not very clear, and it has now been deleted.
However, the most important point to me, is related to the transformation of the cytoplasm of one prokaryote into the nuclear AND of the extracellular space into the cytoplasm. Both are dramatic modifications that are not explained at all in the current manuscript and that deserves much more details as they are so central to the proposal. L 508 and others and also figure 3 b.
Author response: The transformation of the matrix to a cytosol is now discussed in much more detail in the new section “Membrane encapsulation and cytosolic takeover.” In the original manuscript the transition from the matrix to the cytosol was referred to as “matrix replacement” and discussed too briefly near the end of the manuscript. I have changed the name of this process from “matrix replacement” to “cytosol takeover” in order to better emphasize the nature of the transition. In effect there is a slow transfer of cytosolic components from the resident cells into a membrane enclosed matrix resulting in a series of intermediate or transitionary matrices with increasing cytosolic and decreasing matrix properties. Selection for ensembles with improved matrix properties results in the increasing approach towards a cytosol-like structure enclosed in a membrane sheath. Suggestions regarding how this could occur are presented.
The formation of the nucleus occurs through the fusion of the L-form cells, a subset of which specialize by division of labour into information storage and processing centres. The text that discusses this has been edited and given a sub-heading. “ Assembly of a primitive nucleus.”
In particular, if the archaea is proposed to become the nucleus, as in other proposals, and stated L694, how typically archaeal lipids are modified to become typically eukaryotic and bacterial lipids in the nucleus?
Author response: This is, I think a misunderstanding. This section discusses other previously published models that have “third-space” characteristics and does not apply to the model presented here. These other models were presented in order to allow the reader to compare them with the current model and make decisions about their relative strengths.
L586 deserves more details.
Author response: I have revised this text from “The selection of eubacterial over archaeal-type membrane lipids might, therefore, reflect an adaption of membranes to a mesophilic lifestyle or, alternatively, result from a simple numerical predominance of eubacteria in the early matrix ensemble”; to now read “As most eukaryotes have a mesophilic lifestyle, the selection of bacterial over archaeal-type membrane lipids by the emerging eukaryote might, therefore, reflect the preferential selection of bacterial cell membrane lipids as they are more suitable for a mesophilic existence. Alternatively, it might result from a simple numerical predominance of Bacteria over Archaea in the early matrix ensemble.” (in section “Genome merger and the loss of archaeal membrane lipids”).
An important issue is that some biofilms are compatible with the existence of a bounding lipid membrane L 655. I find this argument extremely weak and not supported at all by the reference provided. Please revise.
I agree that the presence or absence of a bounding membrane in the biofilm was not the main point of the paper referenced, however the evidence given in that paper for a sheath around the colonies that takes up osmium stain as expected for a lipid membrane is good. I have revised this to be more “equivocal” about the interpretation of this structure as a membrane, and also to point out the presence of numerous extracellular vesicles in the matrix. This text is now incorporated into section. “Membrane encapsulation and cytosolic takeover”, in paragraph one, sentence starting “When examined by electron microscope tomography ….” to “…ensembles can be enclosed within a membrane-like structure.”
Minor issues: L20, the author states that “the question was addressed by enumerating the classes of potential pathways”. I however see only the two classical pathways, autogenous and fusion, followed by the novel model. There is no ‘enumerating’. I would just remove this line.
Author response: This has been changed to "The question was addressed by considering classes of potential pathways from prokaryotic to eukaryotic cells…. (In the abstract).
L 45 states that the age of the first eukaryotes is between 2.1 and 1.84 Ga, “making them much younger than Archaea or Eubacteria”. If we would agree that bacteria are ancestral, there is much less consensus on the age of the archaea. Could the author develop on the question of the age of the Archaea?
Author response: This is an important question, but as the manuscript is already very long, I preferred to change the sentence to “making them much younger than prokaryotes”, thereby avoiding the issue, rather than discuss the question at length.
Please define at first mention, eg L72, what is meant by a proto-eukaryote.
Author response: This has been modified, third paragraph of the Introduction, sentence starting “Proto-eukaryotes would be expected to have several hallmark features….”
L421, modify ‘consistent’ as this is not consistent, but this is the same thing.
Author response: This has been done.
L479 is in favor of chromosome linearization arguing that two circular chromosomes cannot segregate properly after recombination if the number of crossovers is odd. I believe that this is pure speculation and not supported by the two theoretical analyses used as refs.
Author response: This has been removed as it was not central to the hypothesis.
L481 states that mitosis is either closed or open. This is untrue as there are plenty of intermediary, as reviewed in Sazer et al., 2014 curr biol.
Author response: Yes, this is absolutely correct, I changed the text accordingly to read “Mitosis is highly variable (157). It may be closed, with the nuclear membrane retained throughout, or open, with the nuclear membrane breaking down and reforming as cell division progresses, with intermediate forms also described, such as semi-open mitosis (157-159).”
L499 states that increase in cell size follows obviously. It is unclear to me, why this should be so obvious. Please detail.
Author response: This has been revised to read “4.1. “Cell Size”. If, as proposed, eukaryotic cells developed from populations of prokaryotic cells embedded in a matrix (Fig. 3), the greater size of an ensemble versus single prokaryotic cells would lead ultimately to the greater size, on average, of eukaryotes compared to prokaryotes.”
L599, reference 160 is NOT the correct one to this statement the correct one is Devos et al., Plos biolgoy 2004.
Author response: Thanks. This has been changed as suggested.
Advantage listed a #11, L682 is not supportive as matrix membrane vesicles still needs the ability to fuse in the external environment.
Author response: This is now changed to "Membrane vesicles or folding may provide substrates for the development of complex sub-cellular membrane structures.
Figure 4, blue stream left, “ribosomes in prokaryotic cells” is stated twice
Author response: Thanks for noticing this, it has been changed.
Reviewer’s report 2: Buzz Baum, Affiliation University College London, Cell biology
The origins of the eukaryotic cell remain uncertain. Models can certainly help here. Therefore, we read this paper with interest. In this paper, “Division of labour in a matrix, rather than phagocytosis or endosymbiosis, as a route for the origin of eukaryotic cells”, the author proposes a novel mechanism by which the eukaryotic cell may have arisen from a community of prokaryotes living within a shared environment, or biofilm, which the author calls the “3rd space”. The author justifies the need for this new perspective by stating that there are “irreconcilable differences” between existing hypotheses, and on evidence that a large part of the eukaryotic genome cannot be attributed to the two proposed partners: an alpha-proteobacteria and an archaeal cell from the TACK/Asgard family - suggesting the involvement of a community of partners. In fact, the most widely cited models are very similar when compared to the 3rd space model and differ only in the path to the eukaryote cell, not in the changes in topology that underlie eukaryogenesis.
Author’s Comment: The model is about the pathways to a eukaryotic cell from a starting topology of prokaryotic primary spaces to a more complex topology. It identifies (not necessarily exhaustively) three different classes of potential pathway and focuses on one of these classes (the third space model). Each model begins with two types of prokaryotic space and ends with a eukaryotic-like structure with three secondary spaces (mitochondria, cytosol, nucleus). Other classes of topological transformation are possible, for example starting with only one primary space, or four, five or six etc. primary spaces. These are not considered here, essentially because the literature strongly supports at least two spaces, the archaeal and bacterial starting spaces. It also keeps the model manageable. The reason for exploring the third-space class of models further is that there are some versions of it that allow eukaryotic cells to evolve without requiring phagocytosis or endosymbiosis, a prerequisite feature of most (admittedly not all) other current models of eukaryotic origins.
Moreover, the author does not do enough to make the case that the eukaryotic genome is a patchwork that is likely to have arisen as the result of DNA influx from a large number of partners. The model proposes that the extracellular space in which the archaeal and bacterial partners that gave rise to the eukaryotic cell are growing when they enter into a symbiotic relationship, becomes the cytoplasm. In a process that has not been discussed in any detail this then becomes bounded by the plasma membrane.
Author’s comment. The formation of the cytosol was discussed as “matrix replacement” in the original version of the text, but I agree it was much too brief and not developed. A new section has been added to address this issue. “Membrane encapsulation and cytosolic takeover” (particularly paragraphs two and three of this section). This is discussed further below.
Finally, the author suggests that this interpretation of events explains many features of eukaryotic cell organization. While this view is novel, and is an entirely new take on the topology changes required to give rise to the eukaryotic cell, the author has not done enough to lay out the topology of the eukaryotic cell and topological changes required to generate it.
Author’s comment. The text has been edited to expand on these issues. The development of the topology of the cell is given in several places but particularly section. “Assembly of a primitive nucleus” (for the nucleus), and 4.2. “Membrane encapsulation and cytosolic takeover” (for plasma membrane and cytosol) and 4.6 “Membrane-enveloped organelles”. (perinuclear space, the endoplasm reticulum, Golgi lumen and the lysosome).
I have, in addition, edited section 1 that lays out the “rules” of the topology. This clarifies (hopefully) that spaces are functionally defined, can communicate, and what the difference is between secondary and tertiary spaces.
The annotation used here is problematic rather than helpful, in that it allows the author to suggest simple rules that make little physical sense.
Author’s comment: I kept the original annotation in section “Three distinct pathways that may lead to the tripartite eukaryotic topology”; but removed it from the rest of the manuscript as it could prove confusing when used in those more complex descriptions. I appreciate that the annotation is not a mainstream approach to biological description, but I believe that that presenting the transitions in a highly simplified and abstract form is, nevertheless, useful as it strips away the “fog” that surrounds such questions.
For example: i) The nucleus and cytoplasm are treated as distinct compartments in the model. However, they are not topologically distinct. In every cell cycle they transiently become one during mitosis. In interphase are only separated by nuclear pores, which regulate traffic, not by a membrane barrier.
Author’s comment. Regarding the dissolution of the barrier between the nucleus and cytosol in interphase, it depends on whether the organisms in question display “open” or “closed” mitosis, both forms being well documented. This question worried me as well, which is why I suggested that the model results in a closed rather than open form of mitosis (paragraph beginning “Mitosis is highly variable”). Taken together, the different forms of closed mitosis have a very wide occurrence. When mitosis is open, the reviewer is correct, the tripartite topology ‘collapses’ to a transient bipartitite state. This now inserted into the text. See section “Assembly of a primitive nucleus”, from “If nuclear division originated from fission of L-form-like multi-genomes…” to line 559 “…considered to have transiently devolved to a bipartite state.”
The biology of “open” vs “closed” mitosis has been well reviewed (Raikov, European Journal or Protistology, 1994, The diversity of forms of mitosis in protozoa: a comparative review, 30, 253–269). In metazoans, the nuclear membrane breaks down, and mitosis is therefore said to be open. When the nuclear envelope is retained, however, mitosis is said to be closed. There are further variants in which the nuclear membrane is partially retained (semi-open mitosis). The manuscript predicts that the first mitosis was probably closed. In addition to membrane status mitosis is further categorized based on the position of the spindle poles. In orthomitosis the spindle poles are at right angles to the plane of mitosis (as in animal cells), but in pleuromitosis the poles are off-angle to the plane. Closed orthomitosis occurs in various fungi, diatoms and ciliates. Pleuromitosis happens only in closed or semi-open mitosis. Closed intranuclear pleuromitosis, occurs in various fungi and foraminifera. In a further variation, the spindle may even be outside the nuclear envelope and attaches to chromosomes by “insertion points” at the nuclear membrane itself. This is called extranuclear pleuromitosis, and is found in dinoflagellates and trichomonads. Taken together, therefore, the different forms of closed mitosis have a very wide occurrence.
The connections between spaces, including the cytosol and nucleus during interphase, are covered in Section “Minimal Cellular topologies”. The secondary spaces are specifically defined as functional, and are “delineated” or “delimited”, which is not quite the same meaning as “surrounded”. There is no reason why they should not be in communication, provided the communication is “gated” in some fashion (Lines 178 to 18 starting “2. The secondary topology reduces cells to secondary spaces or domains”. The issue of nuclear pores was discussed in 4.6 Membrane-enveloped organelle".
ii) There are good arguments for considering the perinuclear space, the ER lumen, Golgi lumen and the lysosome as topologically equivalent to the outside of the cell. For example, material fuses to the plasma membrane.
Author’s comments. This is a very important point, and I agree with it to a considerable extent. Section “Membrane-enveloped organelles” has, therefore, been edited to acknowledge the connection outside the cell, and hopefully clarify it with respect to the proposed model. I am not sure, however, that the lumens of these organelles are topologically outside the cell in quite the same way that the lumen of the gut, for example, is a complete topological extension of the outer world. To mark this, I now refer to this compartment as a “topological interface”. In the proposed model, these compartments are part of the tertiary topology (as outlined in section 1, “Minimal Cellular topologies”).
iii) The cytoplasm of a eukaryotic cell is not very different to the cytoplasm of an archaeal/bacterial cell. What is the evidence to suggest it is a new structure derived from extracellular material? iv) In the model it is not clear how the plasma membrane arises and how is it maintained so that it grows and divides with the rest of the cell. In summary, the author has proposed a new topological model. Such a model must explain the topology of the eukaryotic cell in detail and explain how it arose.
Author’s comment. Points iii and iv. I agree that this was not clear in the original manuscript. It was covered briefly near the end, but in retrospect, this was inadequate. A new sub-section has been added (“Membrane encapsulation and cytosolic takeover”) that discusses both the formation of a plasma membrane and the generation of the cytosol. In the original, the transformation of the extracellular matrix to a more cytosolic structure was referred to as matrix-replacement. To better emphasize that this is the process whereby the third space becomes transformed to the cytosol, this is now called “cytosolic-takeover”. Briefly, the matrix is replaced by cytosolic components that originate from within the resident cells of the matrix ensemble. This occurs as a development and extension of excretory process that already exist within prokaryotes. The matrix undergoes a series of changes, passing from a purely extracellular polysaccharide-based substance, to a transitional state that is partly polysaccharide-based matrix and partly cytosolic (protein), increasing in cytosolic content as it approaches the final eukaryote like state. In this view, it is inevitable that the cytosol and nucleoplasm would be related to the cytoplasm of an archaeal/bacterial cell.
Reviewer recommendations to authors
The author needs to make the case for true chimeric ancestry when there are many other reasons why gene trees might point to multiple prokaryotic contributors to eukaryotes. In particular, the challenges of reconstructing relationships among such ancient sequences is fraught with challenges and artifacts (changes in base composition, long-branch attraction, and convergent selection) that can readily distort the phylogenetic trees. And this is further compounded by the HGT among bacterial lineages and possibly between bacteria and early eukaryotes.
Author’s comment. I agree we cannot be confident about very distant phylogenetic relationships, for the reasons mentioned. This is now noted in the text from Paragraph 4 of the Introduction “Recovering the detailed relationships among very ancient genomes is profoundly challenging with many opportunities for artifacts and error. Nevertheless, some general conclusions can be made, among them that eukaryotic genomes are mosaics of bacterial-derived, archaeal-derived and eukaryotic-specific genes.”
Simply bringing two cell types together seems unlikely to generate the degree of mosaicism found in the eukaryotic genome. HGT has, therefore, usually been invoked in order to explain extreme eukaryotic mosaicism. Large scale HGT did, indeed, occur between Bacteria and as outlined in section “Genomic reorganization towards a eukaryotic pattern” first paragraph starting “The horizontal transfer of genes………..” . As far as I am aware, however, there is no example of massive HGT transforming an Archaea or a Bacteria into anything other than a modified Archaea or a modified Bacteria (section “Genomic reorganization towards a eukaryotic pattern”, paragraph one to three). This is, I believe, a critical question. Conventionally, prokaryotic endosymbiosis is given as the rationale for the transition to a more complex cell type, but it remains unclear how likely, or not, such an event would be, and other possibilities should at least be considered. I have added new text to discuss this in section “Genomic reorganization towards a eukaryotic pattern”, paragraph 5: “Although massive horizontal gene transfer did occur between Bacteria and Archaea, there is no example, as far as I am aware, of massive horizontal gene transfer transforming an archaeal species or a bacterial into anything other than modified Archaea or a modified Bacteria. Something else is therefore required in addition to horizontal gene transfer. Prokaryotic endosymbiosis has been proposed as one such driver, but given its rarity in nature, other possibilities should be explored.”
Communal living is the dominant lifestyle of most prokaryotes (Flemming HC, Wuertz S. Bacteria and archaea on Earth and their abundance in biofilms. Nat Rev Microbiol. 2019). The suggestion made here is that living communally first super-charges HGT (this is supported by experiments) and second, that a transition to cell wall-free cells, protected within the matrix/proxy cytosol, allows increased cell fusion (a relic of which is the fusion/fission cycle of extant mitochondria) and this further amplifies the exchange of genetic material until structures (called here multi-genomes) emerge that are neither definitively Archaea or Bacteria. The reorganization of the multi-genomes within the proxy cytosol tends towards a eukaryote-like life form.
Second, the model is not viable without a clear notion of a cellular “space.” In what sense are the cytoplasm and nuclear lumen different spaces when there is no membrane boundary between them?
Author’s comment. Please see above, for a discussion of closed and open mitosis.
And what space or spaces do the ER lumen and perinuclear space belong in the model?
Author’s comment. They are tertiary spaces, as they derive from either in-folding of the plasma membrane, or out-folding/budding of the primitive nuclear membrane. Please see section “Membrane-enveloped organelles” for discussion of this point.
In short the author needs to base the notion of space in modern eukaryotic cell biology terms.
Third, the model needs to explain the origins of the eukaryotic cytoplasm and plasma membrane.
Author’s comment. Please see above for discussion of this point.
Is the author suggesting that the archaeal partner pumped ribosomes across its plasma membrane to fill this space?
Author’s comment There are several ways that the ribosomes could pass from the resident cells to the transitional matrix/proxy cytosol. I favour cell lysis. If a prokaryote can ‘pump out” bacteriophage by regulated autolysis I don’t see why a similar mechanism cannot be co-opted to pump out any other structure. This has been edited and is in the paragraph starting “Cell lysis, vesicle budding, or leakage across an L-form plasma membrane in section “Partitioning of protein synthesis to the cytoplasm”. Altruistic programmed cell death/cell lysis has been postulated in biofilms, and is equally applicable in a third-space model. It is one feature of the proposed model that definitively requires the presence of a cell population as programmed cell death and lysis of a population of one would halt the process dead in its tracks.
How did the plasma membrane come to be?
Author’s comment. Please see paragraph one of section “Membrane encapsulation and cytosolic takeover” and the discussion above.
In light of these major flaws, we believe that the model proposed in this paper is not appropriate for publication in its current form.
Reviewer’s report 3: Michael W. Gray, Affiliation Department of Biochemistry and Molecular Biology, Dalhousie University, Canada.
Reviewer comments to Authors.
I did not review the original submission of this manuscript, which has since been comprehensively revised to take account of the very extensive comments and criticisms of the two reviewers who did see the original. Considering the in-depth review the manuscript has already received and the resulting changes made by the author, my comments are necessarily limited. The subject of this submission is undoubtedly of considerable interest. As the author points out, none of the numerous extant models of eukaryogenesis compellingly accounts for the all of the known properties of the eukaryotic cell and its genomes (nuclear and mitochondrial). For that reason, new ideas are always welcome as a means of advancing discussion of this challenging problem. The author’s model in which a biofilm-like matrix might have been central to initiating the process of integrating separate bacterial and archaeal partners is an interesting one and, in principle, deserving of publication. The author has argued his case well and, in general, the presentation flows smoothly. I did find parts of the text heavy going, in part because my own knowledge of the biofilm literature is limited. On the other hand, the figures illustrating the main points of the model are quite helpful (but see Minor issues, below). In general, I did not find the overall model particularly persuasive in a number of its aspects. In particular, the biofilm-to-cytoplasmic membrane transition is, in my view, difficult to imagine.
Author’s comment. I agree it is difficult to imagine. The third-space model is obviously speculative, and is intended as a first pass at developing such a model, but it is based on the normal mode of life of prokaryotes (in communities) and on processes that can be observed in the wild or in laboratory conditions.
The model discussed in detail is the “strong” version, but, as outlined in section “Strong, weak and intermediate third-space models”, there are other versions, (“weak” and “intermediate”). I chose to focus on the strong version with a mixed population in order to see how far the idea could be pushed. It requires a shift in emphasis from prokaryotic endosymbiosis, or other form of cellular engulfment, which have been rarely observed, and in which only two cell types are the primary players, to an exosymbiotic process which can include more than two cell types. Everything in the model arises by small steps. Originally, I used the term “progressive” for this, but to emphasize this feature I have, in places, switched to the term “incremental” which hopefully better captures the meaning. This avoids unexplained evolutionary leaps across structural and functional chasms. The formation of the boundary membrane is a key step, and this may be hard to envisage. Additional material was, therefore, added (second paragraph of “Membrane encapsulation and cytosolic takeover”) to outline three possible ways a boundary membrane could form around a cellular ensemble in a matrix. The first paragraph of this section outlines an observed example of a bacterial colony that is, at least under some conditions, enclosed within an apparent lipidic membrane-like sheath. In addition, it points out that lipid substrates for the formation of the membrane are likely to be available in the matrix.
Once contained within a boundary membrane, the lysis of a subset of cells in the population and/or regulated secretion from the live cells of the population (both process have been observed) would deposit cytoplasmic material into the matrix. Given time, and in a population of cells linked by obligatory exosymbiosis (symbiosis and syntrophy between cells is often observed), these processes initiate the progressive replacement of the matrix by the cytoplasm. Again, replacement of the matrix by the cytosol would not occur as a step function but as an incremental process. As the ensemble transitions towards a more cell-like lifestyle, the production of the incipient cytosol becomes a more regulated and orderly process. A summary is now added in the final paragraph of section “Membrane encapsulation and cytosolic takeover”.
Also, in contrast to endosymbiotic models, the three-spaces model does not readily account for the selection of an archaeal cell as the progenitor of the nucleus and a bacterial cell as the forerunner of the mitochondrion.
Author’s comment. The strong-third space model hypothesizes that cells in the matrix-ensemble undergo genetic merging (to become the so called multi-genomes). Extensive gene transfer in both directions between Arcahea and Bacteria has been documented. The multi-genome structures contain both archaeal and bacterial genetic material. This is exactly what is seen in the eukaryotic nucleus, that is, a genome with contributions from both prokaryotic domains. It is from these structure that the nuclei develop. To emphasize the mixed nature of the nuclear precursors in the third space model I have referred to them as bacteriarcheons. The third space model assumes that systems were selected because they were “fit for purpose” (last sentence, section "Steps towards a eukaryotic cytosol and mitochondria: shared metabolism ") and not because they were specifically archaeal or bacterial. This is re-stated in section “Limitations” part 2. In this view archaeal information processing provided an advantage over the bacterial machinery. Once one part of the archaeal information processing machinery was selected, other systems that interact with it would be brought along in-step in order to maintain the biochemical and functional compatibility of the systems. Once a critical archaeal information processing system was selected, the evolution of supporting nuclear functions would, therefore, be channeled along an archaeal line. The concept of associated mechanisms being carried along by functional linkage is discussed in the final paragraph of Partitioning of protein synthesis to the cytoplasm and part 2 of "Limitations".
The selection of mitochondrial precursors is discussed in “Steps towards …. Shared metabolism”, and required an aerobic, mesophilic cell capable of generating enough ATP to provide a surplus. Clearly some member of the alpha-proteobacteria must have fit the bill.
Moreover, why the bacterial partner would selectively undergo reductive genome evolution, ceding many genes to the archaeal partner, is also not well developed. Again, endosymbiotic models better account for this distinction (it is well established, e.g., that the bacterial partner in obligate endosymbiosis generally has a much shrunken genome compared to its free-living relatives).
Author’s comment. I have now added a short discussion of this problem in text “Moreover, as the mitochondrial precursors …… mosaicism of the mitochondrial proteome”. Interestingly, exosymbiotic prokaryotic cells can also show genome reduction , so the phenomenon is not restricted to endosymbionts.
These and other concerns aside, the model presented here is still worth considering and, if possible, refining.
1.The author uses the terms “eukaryotization” and “eukaryogenesis” seemingly interchangeably throughout the manuscript. Are these terms equivalent in the author’s view? If yes, perhaps only the more common “eukaryogenesis” could be adopted to avoid possible confusion. If no, the terms should be explicitly defined in order to clarify their distinction.
Author’s comment. This has been altered to eukaryogenesis throughout
2.L68–70 (ref,. 27): Additional references linking intracytoplasmic membranes in alphaproteobacteria to mitochondrial cristae could be added: DOI:
Author’s comment. These references have been added
3. L407: Bacteria (cap.) 4. L414–415: should read “an archaeal or a bacterial species” 5. L642: coli 6. L644: vesicle-associated 7. L724: membrane-enclosed 8. L1451: delete “34.” 9. L1812: “243 of bacterial origin” in Fig. 1(A), the number is 234. Also, Fig. 1(A) retains “eubacterial” (rather than “bacterial”) and “eub/arch” (which presumably should be “bact/arch”
Author’s comments. 3–9. Each of these points have been corrected.
10. I find Fig. 1(A) confusing in that the figure legend states, “Among the bacterial clades, 41 were clearly alphaproteobacterial ...”. This implies that the 41 shown in the figure (and the 198 non-...) are actually contained in the 234/243 bacterial group to the left, rather than being separate. It would help if the figure could be revised in some way to make this point explicit.
Author’s comment. This has now stated in the legend.
11. L1815–1816: “Only 3 Clades ...” The meaning of this sentence is unclear. Does the author mean that the clades in question do not branch with either Bacteria or Archaea? In which case, why are these clades included in the bacterial group (or are they?).
Author’s comment. This paper this was based on is rather complex, and I have used only one level of their analyses. I rewrote the sentence “only 3 clades …” which now reads as “Trees were generated for eukaryotic, bacterial and archaeal gene families. These were then analyzed in terms of “configurations”, for example, those that branched cleanly between eukaryote and bacteria, were assigned as bacterial clades, etc. Only 3 clades (labelled bact/arch) have the so-called “three domain configuration”, that branched into Archaea, Bacteria, and Eukarya with no obvious bias between the three domains.”
Reviewer’s report 4: Damien Devos, Affiliation European Molecular Biology Laboratory (EMBL), Heidelberg)
Reviewer two-second version
The author has convincently addressed most of my concerns. I would now recommend the article for publication. All considered, it is an hypothesis as worthy of publication as any other. I do have however, I few minor concerns to consider: Answer to my concern with L45. I was asking about clarification of the age of the archaea in the sentence, “making them much younger than Archaea or Eubacteria”. Changing the sentence to “making them much younger than prokaryotes” is not changing anything at all. This doesn’t clarify anything about the age of archaea. I do however agree that the article is long enough to discourage a more profound discussion about the age of the archaea. Despite the author’s understanding that “eubacteria” is an inadequate historical term and as corrected in the text, fig 1 still refers to “eubacterial”, “eub/arch”,
Author comment. This has been corrected
… Ref 54 is about a prepublication deposited in bioRxiv. I am unsure about this journal’s policy concerning preprints.
Author comment. A more complete article has now appeared and is cited (reference (58)).
The new paragraph “Membrane encapsulation and cytosolic takeover” is still not satisfying to me, as it briefly provides potential lipid sources, not an explanation of how these would have formed around the third space. How would the third space be transferred internal to the lipids?The Myxococcus example is unclear, to say the least.
Author comment. The section is now re-written. I have added additional text that expands upon three possible pathways for membrane formation (paragraph 2 of Membrane encapsulation and cytosolic takeover). Th first paragraph of this section discusses extracellular lipid sources in a matrix, and a documented case that these lipids can be assembled into a sheath-like outer boundary.
Returning to the Myxococcus example, the observed existence of a membrane-like structure around the Myxococcus colonies shows that under some circumstances some bacterial communities can enclose themselves in a membrane-like structure. It is certainly true, however, that Myxococcus is not representative of typical biofilms as the colonies are motile and predatory. New text “Although Myxococcus, being motile and predatory may not be representative” was added to outline this (paragraph one 4.2. “Membrane encapsulation and cytosolic takeover”).
About our question on the assumption that eubacteria outnumbered archaea in the matrix to justify eubacterialization of the archaeal component, as a posteriory thinking, author response is not convincing and the point is not addressed in the section “Genomic reorganization towards a eukaryotic pattern”.
Author comment. The model does not require more bacterial and archaeal cell types, it can accommodate only two cell types, although I think there are advantages to a mixed population with more than two bacterial cell types. This was outlined in the previous version of the text i.e. “In its simplest version, the third-space model requires only two cell types, or a population of two cell-types, embedded in the matrix, one of which must be archaeal and the other bacterial.”
Allowing for mixed populations in the matrix, the phylogenetic makeup of the eukaryotic genome (Fig. 1) suggests that it arose from more than one bacterial genetic reservoir but does not require multiple archaeal reservoirs. The third-space model accounts for this by allowing for a population of different bacterial cells in the matrix ensemble. New text has been included in the first paragraph of section “Genomic reorganization towards a eukaryotic pattern” to include the following
"The eukaryotic genome arose by merger of archaeal and bacterial genomes. The third-space model can accommodate simple populations of only two cell types (an archaeon and an alpha-proteobacteria) but also more complex mixed populations with more than two cell types. Considering mixed populations, and focusing on the bacterial component, the relative weakness of the alpha-proteobacterial signal (the mitochondrial precursor) compared to the aggregate bacterial signal in the eukaryotic nuclear genome (Fig. 1) can be rationalized if the population contained more than one type of bacteria (an alpha-proteobacterium plus others). A mixed population model does not, however, necessarily require the involvement of more than one archaeal cell type. The minimal mixed-population third-space model suggests, therefore, more than one type of bacterial cell interacting with one archaeal cell type (presumably one of the Lokiarchaeota).