In silico ribozyme evolution in a metabolically coupled RNA population

Reviewer 1: Gáspár Jékely

In this paper Könny? and colleagues extend their Metabolically Coupled Replicator
System (MCRS) to simulate the early evolution of metabolically coupled ribozymes on
a mineral surface. The main novelty in their approach is the explicit calculation
of RNA secondary structures and minimal free energies for the evolving sequences and
the use of the calculated parameters for the model simulations. This makes the model
more realistic than the earlier versions. I suggest that the authors state more explicitly
what are the new insights gained from using the more realistic secondary structure
and free energy calculations. For example the authors observe that: “The shorter (faster
replicable) mutants of the parasitic mutants are heavily selected against due to the
disadvantage of aggressive parasites”. Was this already shown in the less explicit
versions of the MCRS model? An extra paragraph in the Conclusions section could summarize
the novel results.

Two extra paragraphs have been added to Conclusions section summarizing the novelties
of the results.

It would be interesting to see results with larger metabolic neighbourhood (h?=?7×7)
and higher replicator diffusion. One would expect that parasitic sequences can diverge
more since they will not as easily demolish metabolism in their own vicinities, due
to larger diffusivity.

The results of the simulations with h?=?7×7 and D?=?4 (Figure3 ) show that increasing metabolic neighbourhood size, even if it decreases replicator
frequency, does not affect the stable persistence of the system. Its stationary characteristics
(the equilibrium distributions of replicator frequencies, enzymatic activities and
sequence chain lengths) have also been largely insensitive to the actual value of
the dispersal parameter D (Figure4 ). This is in good accordance with our previous results [19] obtained with toy model
simulations scanning a wide range of the (D, h, r) parameter space. The earlier study
revealed a positive effect of increasing dispersal (D) which prevents the aggregation
of conspecific replicators, and a negative effect of increasing h, though approximating
the mean-field situation that is known to go extinct. The deleterious effect of larger
h can be compensated by increasing D to some extent, but the compensatory effect is
limited: at too large metabolic neighbourhood sizes the system collapses anyway (cf.
[19]: Figures2 and4 ). The parameter setting suggested by the Reviewer (large h, large D) is, unfortunately,
practically not feasible in the explicit model, because even on a high-capacity grid
computer it would take months to run a single simulation. The huge difference between
the CPU time demands of the toy models and the present one is due to the “handling
time” of sequence folding. However, the results of the toy models are likely to carry
over to the explicit case in this respect, too, since in all other respects we experienced
qualitative matches.

Importantly, the authors should provide their code as an Additional file or deposit
it to a public repository for others to reproduce or extend the model calculations.

As the code is the result of a long process of development, and it will be further
developed for later studies, we do not find it convenient to publish it at this stage.
However, we are willing to send the code to the reviewer or to anyone for further
studies or for reproducing our results, on an individual basis.

Minor comments:

pg 19–20 The discussion about the dynamics of parasites is repetitive in this section.
I suggest to delete or shorten this part: “Mono-active replicators easily mutate to
parasitic ones … their own vicinities and starve to death”.

We have shortened this part.

Typos:

page 11: “we assume that different RNS molecules” change to RNA

Corrected.

page 17: “i.e., larger metabolic neighbourhood) that fosters parasite invasion, which
is evident on Figure 4A”. – change to Figure 3A

Corrected.

Reviewer 2: Anthony Poole

This is a very elegant study which has been explained in very accessible language.
I found the results very insightful, and need say little other than that, for those
interested in the RNA world, this is a paper well worth reading.

We thank the Reviewer for their positive judgment of the study.

For my money, the most exciting result is that this model suggests that catalytic
promiscuity in early ribozymes may have been extremely short-lived. This bears thinking
about, particularly given the view, popular in protein science circles that early
enzymes were promiscuous (both in their substrate specificity and enzymatic reactions).
It is also noteworthy that group selection appears as a feature of the model. This
is broadly consistent with the cooperative networks that Lehman and colleagues observed
for fragmented ribozymes (Nature 491:72–77). I would be interested to see a brief
discussion of that work and how it relates to the authors’ findings.

Lehman and co-workers showed that mixtures of RNA fragments that self-assemble into
self-replicating ribozymes can form catalytic cycles and more complex networks. We
agree that some aspects of our model show some similarity to some of Lehman’s, even
though the two models (Lehman’s and ours) are essentially different in their basic
assumptions. Ours assumes cooperation in the evolutionary sense, so that a complete
metabolic neighbourhood (a cooperating “team” of potentially competing replicators)
is capable of replicating a focal sequence which thus will have two identical copies
locally (apart from mutations). The model of Lehman, on the other hand, assumes collective
autocatalysis: in which the complex networks arise due to the fact that the members
of the set catalyze each other’s formation, rather than replication. Yet, it is true
that both models are prone to being parasitized and ultimately exterminated by parasitic
replicators in a mean-field framework, both requiring some form of spatial structure
as a potential defense: “Longer term evolutionary optimization would have required
spatial heterogenity or compartmentalization to provide lasting immunity against parasitic
species or short autocatalytic cycles.” (Vaidya et al. Nature 491:77 (2012)).

Just a minor quibble about this statement in the Background: “recent organisms still
carry reliable clues suggesting that RNA had played a central role both in the metabolism
and in the genetics of very early forms of life [3,4]”. The papers cited here are
both excellent, but address the more chemical aspects of the origin of RNA itself
and of RNA catalysis. By contrast, the recent paper by Hoeppner et al. (PLoS Comp
Biol 8: e1002752. http://dx.doi.org/10.1371/journal.pcbi.1002752) used a comparative genomic approach to look at whether there are ‘clues’ of the
RNA world in modern organisms, so is perhaps more appropriate, given the sentence.

Thanks for drawing our attention to the paper – we have cited it in the corresponding
part of the text.

Minor comments/typos (can be deleted from the review once addressed):

Page 7, the use of the term “catching” – perhaps “binding” is more appropriate here.
Enzymologists routinely talk about substrate binding and product release.

We rephrased the sentence in a more clear way.

Page 11, “hydroxil” should be hydroxyl

Corrected.

P11, RNS should be RNA

Corrected.

P26, “thedesing” should be the design

Corrected.

Figure 2 X-axes: Ferquency

Corrected.

Reviewer 3: Armen Mulkidjanian

Könny? and co-workers have attempted to fill the gap between purely mathematical,
“toy” modeling of very early evolution and the physico-chemical realm within which
such evolution may have proceeded. Specifically, the model of Könny? and co-workers
explicitly accounts for the primary and secondary structure of replicators. Hopefully,
the authors would continue their efforts to model the physics and chemistry of the
early evolution. Therefore, the comments below contain certain recommendations which
could be realized either upon revising of the given manuscript or in the future work
of the authors.

Major comment:

1) My major concern is the plausibility of the metabolic part of the model. The authors
assumed that “the different RNA species cooperate to produce monomers for their own
replication, and possibly also to supply other “common goods” for the replicator community”.
Thereby, only three catalytic activities were assumed to be sufficient to perform
all these functions in the model of the authors. Obviously, the assumption of only
three catalytic activities is an oversimplification, which is quite understandable
in the given context. In a wider context, however, the source of monomers and “common
goods” for the first replicators is one of the open questions in the origin of life
research. Apparently much more than three catalytic activities should have been simultaneously
needed to produce different monomers and, in addition, the “common goods”. Furthermore,
there is no evidence of ribozymes capable of synthesizing nucleotides from scratch;
the chemistry of the RNA catalysis, as revealed so far, is not very encouraging in
this respect. A possible solution for this conundrum is that monomers and other common
goods (e.g. amino acids and sugars) could be initially produced in abiotic reactions
[1–5]. Only later, step by step, the first replicating entities may have learned to
synthesize nucleotides. Syntheses of nucleobases and amino acids were recently demonstrated
in the solutions that contained formamide or urea and in the presence of UV-light,
see [6,7] for reviews. Environments with high levels of amides/urea could be envisioned
on the primordial Earth [6,8,9]. The hypothesis of abiotic origin of monomers got
a major boost after Sutherland, Powner and their co-workers succeeded in synthesizing
nucleotides from scratch in geologically plausible, one-pot settings [10,11]. In addition,
these authors have shown that natural nucleotides were particularly UV-stable, so
that their relative fraction selectively increased under UV illumination [10], in
support of earlier theoretical predictions [12]. In such a case, the very first replicators
would require only catalytic abilities of assembling abiotically formed monomers –
a task that should have been feasible for ribozymes. Hence, scenarios of “abiotic
syntheses” match the simplified approach of Könny? and co-workers in that they imply
only few catalytic activities. In a framework of such scenarios, however, an essential
source of monomers for replication would be the decay of other replicators. Furthermore,
the abilities to accelerate this decay by cleaving neighboring sequences (which would
correspond to a reversion of the assembly reaction) as well as to use the fragments
for building the own “body” would be extremely advantageous. A scenario of “abiotic
syntheses”, of course, would require a separate model that would differ from the model
in the given manuscript. Still, in the view of anticipated importance of replicator
decay/cleavage, an analysis of the present model in relation to the decay processes,
namely a consideration of the model outcome as a function of decay parameters (currently
absent from the manuscript) might be of use for readers.

We absolutely agree with the Reviewer in that even a very simple realistic metabolism
requires a lot more different types of catalysts than postulated in our model. However,
we are also sure that the only way such a metabolically competent ribozyme set could
have evolved is through the retroevolutionary mechanism explained in some detail elsewhere
[20–21], which must have started from abiotiocally produced monomers at its very beginning,
exactly as the Reviewer suggests. We have inserted a paragraph into the Background
section pointing this out explicitly and citing the relevant literature.

Minor comments:

2) The manuscript would benefit from a graphical presentation of the focal cell with
its neighborhood. Without such a figure, the sentence “For a replication event to
occur the focal replicator s must be complemented by all three different enzymatically active molecules in its
metabolic neighborhood (MET(h,s), the set of h sites concentric on the site of the focal replicator s)” is not quite clear.

We added a figure that explains metabolic and replicator neighbourhood configurations.

3) It is not clear how the model accounts for “the time consumption of “releasing”
the product and “catching” the next substrate for catalysis”. The whole section on
sub-additive effects is rather incomprehensible and showed be re-written in a more
clear way.

We rephreased it in a more clear way.

4) The statement that “replication is possible only in the unfolded state” should
be defined as one of assumptions of the model. Generally, it is possible to imagine
that unfolding could proceed concurrently with replication. That is how replication
occurs in our cells.

Indeed, we cannot exclude the possibility of the simultaneous unfolding and replication
of RNA molecules on the basis of first principles, but note that our postulate of
the temporal separation of the two processes is in fact a worst-case assumption: the
chance of catalytically active, low-energy folds to be replicated is small. Therefore
if this assumption has any effect on the results, then it is negative, but we actually
think that changing it may not alter the results in the qualitative sense.

References

[1] A.I. Oparin, The Origin of Life, Moskowskiy rabochiy, Moscow, 1924.

[2] J.B.S. Haldane, The Origin of Life, in: C.A. Watts (Ed.) The Rationalist Annual
1929, pp. 3–10.

[3] N.H. Horowitz, On the Evolution of Biochemical Syntheses, Proc. Natl. Acad. Sci.
USA, 31 (1945) 153–157.

[4] S.L. Miller, L.E. Orgel, The Origins of Life, Prentice Hall Englewood Cliffs,
NJ, 1973.

[5] A. Lazcano, S.L. Miller, On the origin of metabolic pathways, J. Mol. Evol., 49
(1999) 424–431.

[6] S.A. Benner, H.J. Kim, M.A. Carrigan, Asphalt, water, and the prebiotic synthesis
of ribose, ribonucleosides, and RNA, Acc Chem Res, 45 (2012) 2025–2034.

[7] R. Saladino, C. Crestini, S. Pino, G. Costanzo, E. Di Mauro, Formamide and the
origin of life, Phys Life Rev, 9 (2012) 84–104.

[8] A.Y. Mulkidjanian, A.Y. Bychkov, D.V. Dibrova, M.Y. Galperin, E.V. Koonin, Origin
of first cells at terrestrial, anoxic geothermal fields, Proc Natl Acad Sci U S A,
109 (2012) E821-830.

[9] D.V. Dibrova, M.Y. Chudetsky, M.Y. Galperin, E.V. Koonin, A.Y. Mulkidjanian, The
role of energy in the emergence of biology from chemistry, Orig Life Evol Biosph,
42 (2012) 459–468.

[10] M.W. Powner, B. Gerland, J.D. Sutherland, Synthesis of activated pyrimidine ribonucleotides
in prebiotically plausible conditions, Nature, 459 (2009) 239–242.

[11] M.W. Powner, J.D. Sutherland, J.W. Szostak, Chemoselective multicomponent one-pot
assembly of purine precursors in water, J Am Chem Soc, 132 (2010) 16677–16688.

[12] A.Y. Mulkidjanian, D.A. Cherepanov, M.Y. Galperin, Survival of the fittest before
the beginning of life: selection of the first oligonucleotide-like polymers by UV
light, BMC Evol Biol, 3 (2003) 12.