Insights & Perspectives Commentary

Comment on ‘‘Does constructive neutral evolution play an important role in the origin of cellular complexity?’’ DOI 10.1002/bies.201100010

W. Ford Doolittle1), Julius Lukesˇ 2), John M. Archibald1), Patrick J. Keeling3) and Michael W. Gray1)Ã

Speijer [1] has provided a critique of suppresses the effect of this mutation, be fixed by selection operating within constructive neutral evolution (CNE) so that selection pressure is relaxed and populations of species – can neverthe- and its role in the origin and evolution the mutation may be harmlessly fixed by less be sufficiently advantageous to of cellular complexity [2, 3]. Not surpris- drift. Thus, A becomes dependent on B species (fostering enhanced speciation ingly, we disagree with his assertions. for its activity by virtue of the neutral, or reduced extinction rates) to spread by Because his description of the CNE ‘‘pre-suppressive’’ effect of the A:B species or clade selection. Eukaryotic model does not precisely conform to interaction. What we submit does not sex might be one of these. Introns could our view of CNE, as we [2, 4] and happen is that the mutation occurs first, be another. No one really thinks that the Stoltzfus [3] have elaborated it, we after which the interaction with B is insertion of introns was selected for at briefly re-state the model before positively selected for because it sup- the level of individuals – that is, that all addressing Speijer’s objections. presses the deleterious effect of the introns that are currently fixed within a The underlying premise of CNE is a mutation. species were fixed because individuals pre-existing, essentially neutral inter- Speijer’s ‘‘Think again’’ article [1] that bore them were at a selective action (RNA:RNA, RNA:protein, pro- embraces several misunderstandings advantage compared to conspecifics tein:protein) between component A, about this important process. First, we that lacked them. Indeed, for individ- which has some activity, and com- note that Speijer’s critique is considerably uals within species, intron addition ponent B. The activity of A is not longer than was our Perspective in could be slightly deleterious. It might dependent on the interaction with B, Science [3], giving him space to decon- nevertheless still be true that species nor is A’s activity negatively influenced struct several points that he may consider in which many introns have become by this interaction. Thus, B could dis- components or at least entailments of our fixed do better than species with few appear from the scene without any hypothesis, but that we would not. Let us introns (speciate more frequently or effect on the ‘‘fitness’’ of A. call these misunderstandings of Type A. become extinct less often), because We imagine that a mutation occurs Type B misunderstandings reflect introns facilitate exon shuffling or the in A that compromises its activity and Speijer’s conflation of micro- and mac- elaboration of multiple gene products that normally this mutation would be roevolution, or ‘‘levels of selection’’. He through alternative splicing. There eliminated from the population by puri- fails to recognize that some features that would thus come to be more intron- fying selection. However, the pre-exist- are neutral or even disadvantageous to bearing species (and in consequence ing interaction with B fortuitously individuals – and thus not expected to more introns) in the world, thanks to species selection. We see many CNE- established features as evolving like DOI 10.1002/bies.201100039 this, influencing however subtly the future evolutionary potential – the ‘‘evolv- 1) 3) Centre for Comparative Genomics and Department of Botany, University of British ability’’ – of species, and yet their initial Evolutionary Bioinformatics, Department of Columbia, Vancouver, British Columbia, Biochemistry and Molecular Biology, Canada establishment was the consequence of Dalhousie University, Halifax, Nova Scotia, the neutral ratchet we describe. Canada 2) Type C misunderstandings are of a Biology Centre, Institute of Parasitology, *Corresponding author: diverse sort. It seems simplest to Czech Academy of Sciences, and Faculty of Michael W. Gray Sciences, University of South Bohemia, Cˇ eske´ E-mail: [email protected] attempt to correct these as we read Bude˘ jovice (Budweis), Czech Republic

Bioessays 33: 427–429,ß 2011 WILEY Periodicals, Inc. www.bioessays-journal.com 427 W. F. Doolittle et al. Insights & Perspectives .....

through Speijer’s essay, ending with a Speijer asserts, ‘‘Using ‘neutral’ theories Speijer then makes three arguable few general observations. to explain highly complex processes is claims. First he states that it is not very We do not recall that suppression by much less straight forward’’. We only likely that neutral changes will ‘‘take ‘‘remaining copies of the nonmutated partly concur. We would agree that over the complete population’’. Most gene’’ (Speijer’s Introduction) was part the case of single editing of a marsupial of the enterprise of molecular phyloge- of our model, nor indeed would we tRNA is ‘‘simple’’. Indeed it was aptly netics at the trans-species level is in fact think of that as a form of suppression. interpreted in CNE terms by the based on such neutral ‘‘takeovers’’. The Nor did we imagine that ‘‘reverse researchers who first described it, who neutral theory of molecular evolution mutation of the single first mutation also noted: ‘‘Because in other systems entails that no single pre-designated can not restore the original viable organ- many positions have become dependent neutral mutation is likely to be fixed; ism’’. It easily could do that, but if there on RNA editing subsequent to the initial at the same time it holds that some

Commentary are additional sites at which further events, the initial mutations that have neutral mutations inevitably will be dependency (through mutation) is caused the evolutionary fixation of edit- fixed. Second, Speijer claims that we possible, then a random walk through ing are probably indiscernible today’’ assert categorically that reversal of dependency space will seldom end up [7]. But with an editing system that changes must be much less likely. back at the doorstep to independence – has already become essential for per- This is untrue: we assert only that when the initial condition. Similarly, even forming one edit, the conditions are in there are more open mutational paths to those who would vigorously deny that place for more to arise: it is actually the increased complexity than decreased life shows a tendency toward complex- first cut that is the deepest. complexity, complexity will most often ity would admit that since it began Speijer then invents three criteria for increase by chance. Third, he argues simply and since there are innumerable distinguishing CNE versus selection as that many neutral changes taken ways to become more complex, com- the cause of a complex feature: first that together can be detrimental, because plexity was inevitable. That is the kind there should be a ‘‘smooth’’ increase (no complexity incurs costs. This argument of ratchet we envision. Indeed, getting rapid bursts) in complexity, second that seems to be hiding some belief in opti- ‘‘stuck with it’’ is an appropriate catch- there should not be much variation mality, a quality that Speijer surely phrase, but otherwise we feel that between lineages in the extent of com- would not expect of that other prime Speijer is making a Type A mistake here. plexity, and third that there should not exemplar of a complex system evolved In Speijer’s representation of our be good adaptationist alternatives. The through both chance and necessity, model (Step 2) we are again baffled by first two criteria seem to us to be ad hoc human institutions such as universities. the invocation of ‘‘asymmetric div- in the extreme, and would certainly not There are reasons that we have else- isions’’, which does not form part of be our criteria. Of course such evolution where described CNE as a theory about the model in any of our papers that he will proceed by fits and starts. Similarly, ‘‘cellular bureaucracy’’ [2]. ‘‘Yes, Dr. cites. Of course if there are multiple we will all admit that the rare reinte- Pangloss, there is a downside to com- copies of the mutated gene, time will gration of a reverse transcript of a ma- plexity’’, we are tempted to say. be required for segregation: here the fact ture edited mRNA can wipe out a whole In considering The creative power of that such mutations are, in our model, slew of edits at once, resetting the clock. complexity as such, Speijer discusses rendered neutral by pre-suppression That this process might vary from line- two macromolecular machines, the means that this can happen. Maybe age to lineage seems a ‘‘no-brainer’’. ribosome and mitochondrial respiratory Speijer is confused by the phrase ‘‘dupli- The third criterion ignores the extensive complexes, that we mentioned [2] as cated information’’ in Lukesˇ et al. [5]. literature on the ease with which clever possible examples of CNE. Here, we We think gRNAs arise as duplicates of biologists can invent evolutionary ‘‘Just have space to comment only on the lat- the original gene, but are transcribed So Stories’’ [8], and is highly ‘‘verifica- ter example; elsewhere we will present from the opposite strand, producing a tionist’’. The principle behind this detailed CNE scenarios for the evolution complementary RNA. We think this is a criterion seems to be ‘‘better the widely of several other complex cellular Type A mistake on Speijer’s part. believed but unprovable hypothesis we machines. In Origin and use of cellular proc- have than an equally difficult-to-prove Speijer cites the electron transport esses, we feel that Speijer makes some but less popular alternative’’. chain complex I (ETC CI) as an example Type B mistakes. Extra levels of control, Speijer’s section CNE: Conceptual of ‘‘a very complex prokaryotic fine-tuning, and ‘‘evolvability’’ are problems has its own conceptual prob- machine’’ that experienced a major indeed possible co-optations of com- lems. We would not be surprised if a increase in the number of subunits plexities first fixed by CNE, but are not population exhibited ‘‘exactly the same ‘‘during the very rapid development of the reason for fixation, at the micro- level of complexity’’. What determines the eukaryotic cell’’, with subsequent evolutionary level. Such secondary within-species heterogeneity would increases in the number of subunits ‘‘functions’’, which Speijer describes be population sizes and the potential seeming ‘‘like an afterthought’’. Speijer as ‘‘fringe benefits’’, are exaptations number and rate of occurrence of pre- suggests that the addition of the super- or species-level adaptations as we suppressible mutations, which could numerary subunits in mitochondrial CI described above, and are by no means easily be low enough that we might ‘‘has all the hallmarks of resulting from a excluded by CNE thinking. In this regard, catch very few CNE interactions in the period of intense selection’’. Rabosky and McCune [6] provide a good act of being established. And who would Bacterial CI comprises 14 subunits recent review of the relevant literature. have looked for such a thing? that constitute the evolutionary ‘‘bac-

428 Bioessays 33: 427–429,ß 2011 WILEY Periodicals, Inc...... Insights & Perspectives W. F. Doolittle et al.

terial core’’ of eukaryotic CI [9]. Of 31 subunits in the complexes they inhabit, for the evolution of cellular complexity additional subunits in bovine CI, 18 are and to be very narrowly restricted phylo- can and should be compared, CNE considered, on phylogenetic grounds, to genetically. For example, CV, the mito- should improve the rigor with which Commentary constitute a ‘‘eukaryotic core’’ present chondrial ATP synthase, has lineage- we form such explanations. in the last eukaryotic common ancestor specific novel subunits in Tetrahymena (LECA) [9, 10]. (However, a recent thermophila [12], but a different lineage- analysis of bacterial CI indicates that specific novel set in Chlamydomonas rein- References three of these eukaryote-core subunits hardtii [13]. Thus, these distinct sets have could also have been contributed by the all the hallmarks of newly evolved 1. Speijer D. 2011. Does constructive neutral evolution play an important role in the a-proteobacterial progenitor of mito- proteins that fortuitously happened to origin of cellular complexity? BioEssays 33: chondria [11].) In bovine CI, the remain- be targeted to and imported into mito- 344–9. ing subunits (13) are termed metazoan- chondria, where they could engage in 2. Gray MW, Lukesˇ J, Archibald JM, Keeling specific, as they have not been identified interactions with mitochondrial supra- PJ, et al. 2010. Cell biology. Irremediable complexity? Science 330: 920–1. outside of animals [9]. molecular assemblages like ETC com- 3. Stoltzfus A. 1999. On the possibility of con- We suppose that ETC complexes plexes. This origin model says nothing structive neutral evolution. J Mol Evol 49: 169– such as CI were built up gradually, about how important (or not) these added 81. 4. Covello PS, Gray MW. 1993. On the evolution and it is entirely conceivable that proteins may now be to eukaryotic ETC of RNA editing. Trends Genet 9: 265–8. CNE-type interactions played a role in complexes, only that their initial inter- 5. Lukesˇ J, Leander BS, Keeling PJ. 2009. the increase in complexity that accom- actions with these complexes pre-dated Cascades of convergent evolution: the corre- panied the evolution of CI, both prior any functional requirement for them. sponding evolutionary histories of euglenozo- ans and dinoflagellates. Proc Natl Acad Sci and subsequent to the emergence of the So, what is CNE and why should it be USA 106: 9963–70. eukaryotic cell. Although we can trace a considered? CNE is a ratchet-like proc- 6. Rabosky DL, McCune AR. 2009. Reinventing number of eukaryote-specific CI subu- ess capable of generating biological species selection with multiple phylogenies. Trends Ecol Evol 25: 68–74. nits to LECA, there is a ‘‘black box’’ complexity that is driven by properties 7. Bo¨ rner GV, Pa¨ a¨ bo S. 1996. Evolutionary fix- interval between the emergence of the intrinsic to macromolecules and empha- ation of RNA editing. Nature 383: 225. eukaryotic cell per se and the emergence sizes the role of neutral evolution, not 8. Gould SJ, Lewontin RC. 1978. The Spandrels of San Marco and the Panglossian Paradigm: a of the LECA that we surmise on com- positive selection. However, once in critique of the adaptationist programme. Proc parative genomic grounds. Hence, the place, CNE-generated complexities are R Soc Lond B 205:581–98. assumption of a ‘‘very rapid develop- preserved by negative or ‘‘purifying’’ 9. Brandt U. 2006. Energy converting ment of the eukaryotic cell’’ that has selection, and may later go on to acquire NADH:quinone oxidoreductase (complex I). Annu Rev Biochem 75: 69–92. ‘‘all the hallmarks of ... a period of useful functions. Thus, CNE dis- 10. Gabaldo´ nT, Rainey D, Huynen MA. 2005. intense selection’’ is problematic, to tinguishes between ‘‘origin’’ and Tracing the evolution of a large protein com- say the least. We simply do not know ‘‘present-day function’’ when consider- plex in the eukaryotes, NADH:ubiquinone oxidoreductase (Complex I). J Mol Biol 348: how many evolutionary dead ends and ing the evolution of complexity. Despite 857–70. extinct eukaryotic lineages preceded the suggestions by Speijer to the contrary, 11. Yip CY, Harbour ME, Jayawardena K, emergence of LECA, nor do we know the literature (on RNA editing, e.g. [14]) Fearnley IM, et al. 2011. Evolution of respir- how long this transition took; hence, is rife with ‘‘Just So Stories’’ that attempt atory complex I: ‘‘supernumerary’’ subunits are present in the a-proteobacterial enzyme. we can infer nothing about the process to link the origin and expansion of com- J Biol Chem 286: 5023–33. and timeline of eukaryotic CI evolution plexity with functional advantages. If 12. Nina PB, Dudkina NV, Kane LA, van Eyk JE, during this period, save for the end one doesn’t fit, try another. It is our et al. 2010. Highly divergent mitochondrial ATP synthase complexes in Tetrahymena result. assertion that most molecular biologists thermophila. PLoS Biol 8: 1–15 [e1000418]. In any event, the lineage-specific first attempt to find adaptive, positive- 13. Lapaille M, Escobar-Ramı´rez A, Degand H, accessory components of CI and other selection explanations, generally with- Baurain D, et al. 2010. Atypical subunit com- ETC complexes are of particular interest, out even considering CNE. Quite the position of the chlorophycean mitochondrial F1FO-ATP synthase and role of Asa7 protein in as it is these that can be most easily opposite of ‘‘stifling further thought’’ stability and oligomycin resistance of the rationalized as having arisen relatively on the evolution of complex biological enzyme. Mol Biol Evol 27: 1630–44. recently via a CNE pathway. These processes, CNE provides a new way to 14. Gray MW. 2001. Speculations on the origin and evolution of RNA editing. In Bass BL, ed; proteins tend to be relatively small, com- think about it. As a null hypothesis RNA Editing, Oxford: Oxford University Press. pared to the more evolutionarily ancient against which adaptationist arguments pp. 160–184.

Bioessays 33: 427–429,ß 2011 WILEY Periodicals, Inc. 429