Albert de Roos: A design hypothesis for the evolution of the nucleus
by Guest Author[Albert de Roos is a cell biologist from Amsterdam, who has previously graced Telic Thoughts with this guest post about applying engineering principles to evolution. We've invited him to write about his latest article, published in the journal Artificial Life. Not because we agree with everything he has to say (nor vice versa), but because we find it to be an interesting approach, which may jolt researchers into thinking about evolution in new terms. Don't forget to check out Albert de Roos' blog]
A design hypothesis for the evolution of the nucleus
By Albert de Roos, Ph.D. Cell Biology
Recently, I published an article about the origin of the nucleus. Basically, I pose that the nucleus arose in evolution when a nucleus-like cell generated an extra plasma membrane around itself. Or in other words, when we take the current nucleus, we are looking at the direct descendent of a free living ancestor cell. Genetic material that is wrapped in a double membrane with large simple pores in them that keeps macromolecules such as DNA, RNA and proteins inside, while nutrients and waste is free to diffuse in and out.
This article is not 'just another theory' about the origin of the nucleus, but it is derived from an engineering framework named 'design by contract'. This concept is used in the development of software where components of systems communicate according to defined interfaces or contracts. As long as you don't change existing interfaces, you can extend the system. You can directly apply that to evolution: you can add new functionality, as long as you keep existing interfaces intact. The conserved mechanisms for translation, transcription, splicing etc. can be considered to reflect these constant interfaces in this engineering view. Conservation is thus regarded as an inevitable consequence of extension since any evolutionary process that would require extensive rework in critical systems would never survive because of the direct fitness costs.
Looking at the origin of the eukaryotic cell, the only scenario that would fulfil a conservation of existing interfaces for transcription and translation is when the nucleus evolved first in evolution and a plasma membrane was wrapped around it. Everything could remain intact and the new cytoplasm could work initially as a controlled microenvironment buffering external influences and eventually giving the possibility for multicellularity. In line with a model of evolution that sees evolution as an expanding set of functionalities, the cell wrapped itself with another layer of functionality. This is common practice in software engineering when legacy systems are wrapped in new technology and many nuclear systems can thus be regarded as biological legacy systems.
Two main assumptions are held in the view where the nucleus represents the ancestor cell. One is the functioning of the nucleus as an independent cell or unit, and the other is that the resulting evolutionary scenario is feasible. It is very easy to seen the nucleus as an independent unit because it contains all the DNA, transcription and translation machinery. It also can divide itself into two, for instance in closed mitosis. The syncytial blastoderm in Drosophila show that the nuclei can independently divide and later form a multicellular embryo. Also in plants, the nucleus and its surrounding cytoskeleton seems to be the basic building block.
The evolutionary scenario is also relatively easy. Starting with DNA surrounded by lipid vesicles and DNA binding proteins, a double membrane may spontaneously self-assemble around the DNA, a process which happens at every division in telophase. The generation of a surrounding plasma membrane may have occurred through the ER as an intermediate. The nuclear membrane starts to form additional stack-like extensions when certain proteins are added. If vesicles are then generated from this proto-ER/Golgi, a plasma membrane can be formed on an existing cytoskeleton, a process similar to the formation of the cellular blastoderm in Drosophila and similar to the addition of vesicle to the current plasma membrane. It is interesting to note that certain algae can regenerate their own plasma membrane. The cytoskeleton may have formed when the translation machinery moved out of the nucleus through the pores, or maybe have extended out of the nucleus through the pores.
My design approach is based on a careful study of the existing functional modules and their interaction, while paying special attention to the conserved interfaces. After such an analysis, the pieces of the evolutionary puzzle are assembled into a logical scenario. Apart from this theory on the origin of the nucleus, I have published an article on the origin of introns and have a manuscript for the origin of Life based on the same design paradigm. When this approach proves to be fruitful for evolution in general, it may be the first design hypothesis for evolution.
October 27th, 2006 at 11:52 pm
Albert, what are your thoughts about the evolution of the NPC? In your text you describe them as "large simple pores in them that keeps macromolecules such as DNA, RNA and proteins inside, while nutrients and waste is free to diffuse in and out". That, of course, hardly fits the description of an NPC.
Interestingly, it appears that structural analysis of the NPC indicates that it evolved from a simpler coating module that allowed to sharply curve membranes. This suggests that indeed the nucleus was not a free living organism, but is rather a secondary compartment.
I was wondering how you reconcile such analysis with your hypothesis.
Comment by hrun — October 27, 2006 @ 11:52 pm
October 28th, 2006 at 12:38 am
Albert, I have additional questions:
1) Could you explain why in the karyogenic model the changes would be more disruptive to, let's say, translation, than in your model. It appears to me that in the karyogenic model the NPCs develop gradually and become ever more sophisticated in transporting the appropriate substrates into the appropriate compartment. The same is required for you model. The pores need to become ever more sophisticated in order to allow translation to occur independently from transcription in a different compartment.
2) In what sense differs the 'Design by Contract as a Framework' from 'darwinistic gradualism as a framework'? It appears to me that both postulate the same thing.
Comment by hrun — October 28, 2006 @ 12:38 am
October 28th, 2006 at 10:47 am
It does? It would seem to me that Albert could easily use these data to support his hypothesis.
Comment by MikeGene — October 28, 2006 @ 10:47 am
October 28th, 2006 at 11:01 am
Hi Albert,
First of all, thank you for taking the time to write about your recently published article.
Like Hrun, I also have a question: What type of free-living cell was it that gave rise to the nucleus? I seem to remember that you have previously been very critical of the traditional endosymbiotic hypothesis for the origin of the nucleus (where the nucleus arose from an archaeon engulfed by a bacterium). So do you instead have something in mind like Hartman's and Fedrov's "chronocyte"?
Comment by Krauze — October 28, 2006 @ 11:01 am
October 28th, 2006 at 12:31 pm
Mike, there are three reason why this would support a karyogenic model, rather than Albert's.
1) It is believed that the invagination of the plasma membrane to increase surface area is an evolutionary old adaptation. This, of course, may be wrong.
2) Curving the membrane alone does not have any useful function for a pore in the way Albert describes it (keep proteins in, allow for diffusion of 'waste'). This proto-pores would leak proteins, mRNA and waste into the environment. Yet, curving the membrane would be all the functionality that is needed for invaginations of an outer membrane to increase surface area.
3) Finally, not just functionally, the coatamere is also physically a much less complex structure than the NPC. Thus, it is likely that the coatamere evolved first and the NPC added additional proteins to that primitive coatamere.
Comment by hrun — October 28, 2006 @ 12:31 pm
October 29th, 2006 at 12:52 pm
Hi hrun,
Maybe I'm just not getting it, but it seems to me that Albert would argue that the nuclear envelope was at one time the plasma membrane. If so, all the data from that paper would be easily incorporated into his hypothesis.
Comment by MikeGene — October 29, 2006 @ 12:52 pm
October 29th, 2006 at 2:11 pm
Yes, Mike, that's the whole point that Albert argues that the NE was at some point the plasma membrane (i.e. the outermost membrane of the organism). However, this does not square with the data from nuclear pore development.
For one, the 'simple pores' that Albert describes, ones that would let waste escape but retain proteins, are in fact extremely complex structures (their appearance was preceded by the much simpler coatamer as described above).
In addition, even if this proto-eukaryote already had a highly complex NPC, one that was able to retain proteins and RNA, while allowing for free difusion of waste, the function of such an NPC is still dependent on a gradient of transport factors over the mebrane with the pores. That means, without an outer layer, we know of no possible pore that would conform to Albert's description.
This problem goes exactly counter to Albert's idea of keeping existing interfaces intact. In Albert's scenario, the original pore in the proto-eukaryote has properties that we can't explain. This pore would need to be replaced by a completely different pore (namely the moder NPC) after the proto-eukaryote formed the outer membrane.
While this scenario is complicated (and counter to 'keeping existing interfaces intact') it is not impossible. But, in addition to being complicated, it also runs counter to what we know about the evolution of the NPC, namely, development of coatemere to for invaginations in the outer membrane and subsequently and diversification in function of the coatamere that lead to the formation of a nuclear envelope with simple pores (allowing for diffusion of proteins, RNA and 'waste') to the formation of a true nuclear pore complex that selectively allows for the transport of proteins, mRNA and mRNPs.
Comment by hrun — October 29, 2006 @ 2:11 pm
October 29th, 2006 at 2:33 pm
hrun, thanks for the questions. The current NPC is indeed not simple at all concerning macromolecular transport, but as a large non-selective pore it is functionally quite simple. It also makes a good step from an semi-open system (e.g. DNA coated and surrounded by proteins) to a semi-closed system (membrane with nuclear pores). The big problem with live originating in lipid vesicles is that they are impermeable and would thus need channels that could not have evolved in an abiotic environment (i.e. a classical irredcuible complex system which is abviously contradictory to gradual evolution). From an open to a semi-closed system would not have these problems. The same problem would arise, however, if a closed double membrane would have evolved around my proto-cell made of DNA/nuclear matrix. The pores would have to appear concomitantly with the nuclear membrane. The intimate link between DNA/matrix and nuclear pores would suggest that this is feasible and that the newly generated membrane would have holes when in contact with the nuclear matrix.
I could also think of a scenario in which the pore proteins would actually guide the formation of the double membrane. The DNA/matrix would first interact with lipids to form vesicle complexes, which turned later into a complete membrane. My thoughts are in the line of a complete self-assembly of Life.
The article from Rout is from the unfounded assumption that internal membrane systems evolve from the plasma membrane through the coated vesicle pathway. Mine is the other way around: the coated vesicles arise from the internal membrane components (the ER feeds the plasma membrane, and not the other way around. His results are compatible with mine, his conclusions are different.
Comment by AdR — October 29, 2006 @ 2:33 pm
October 29th, 2006 at 2:56 pm
In all karyogenic models, translation and transcription were once in the same compartment, but transcription and ribosome assembly was moved to the nucleus. I don't see how you can evolve the complex multistep nuclear export system of RNA and the multistep export of the ribosomes (which does not fit through the nuclear pore to make things complicated and thus needs to be de-assembled and then exported) at the same time as the generation of the nuclear membrane. All these export systems are irreducibly complex when you start from the cytoplasm and work your way back. They only function when complete.
Design-by-contract could be considered the technical framework or architecure how to build Life from a the gradualistic concept. In my opinion, mainstream evolutionary science has misunderstood darwinism by not looking at the engineering aspects of an 'evolving molecular machine', and they somehow replaced it by a goal-oriented fitness approach.
Comment by AdR — October 29, 2006 @ 2:56 pm
October 29th, 2006 at 3:29 pm
Hi Krauze. In my view, the 'nucleus' was the first free-living cell and evolved from DNA surrounded by a nuclear matrix and possibly in some sort of other microenvironment like porous rock.
The endosymbiotic theory for the nucleus is very speculative. It doesn't pay any attention the the huge mechanistic and functional aspects of such a scenario. Fusion of three bacteria would even be more extravagant. IMHO, the endosymbiotic theory of chloroplasts and mitochondria remains speculative as well. For simplistic textbook figures, a bacterium swallowing another bacterium and becoming a mito or cholorplast is very nice. The molecular evidence for an endosymbiotic scenario is also very thin. It only gives evidence for common descent and close relatedness between eukaryotes and mitochondria. A scenario in which an ancestor genome would split off its energy-producing compartment would also be compatible with the data, but would not need extensive horizontal gene transfer. But I digress.
Comment by AdR — October 29, 2006 @ 3:29 pm
October 29th, 2006 at 5:58 pm
I have no idea what you mean by 'as a large non-selective pore it is functionally quite simple'. Apart from that, your argument from an irreducible complex system (lack of channels) must be applied to your scenario as well. The newly acquired outer membrane must appear complete with channels, which in your words would be a irreducible complex system.
As I have explained to Mike already, the assumption (that the coatamere evolutionary preceded the NPC) is not unfouded. It actually has factually support: specifically, that a coatamere that mere bends membranes to form invaginations is both functionally and physically less complex that a complex that creates pores of any function at all.
Albert, I don't see the difference. In the karyogenic model, a nucleus with non-selecitive pores comes first. From there, translation is moved into a different compartment.
In your model, a proto-eukaryote with simple non-selective pores acquires a second, outer membrane. From there, translation is moved into a different compartment.
Why do you think this process is plausible in one model, but not plausible in another model?
That's puzzling to me. The mainstay of darwinian evolution is a gradualistic concept and I have not observed that is replaced by goal-oritented approaches. In fact, the key to evolution is that it is not goal-oriented. I wonder how you arrived at the conclusion that gradualism is replaced by a goal-oriented approach?
Comment by hrun — October 29, 2006 @ 5:58 pm
November 1st, 2006 at 6:20 am
hrun, thanks for the questions.
No matter how complex the pore now is, the first protopore did not needed to be complex, because its only function was to be a non-selective pore. Later in evolution, the pore became more complex since mRNA, ribosomes etc were transported out of the nucleus. In the beginning, they just stayed inside.
You are right, the acquired membrane should have channels or some form of communication in it. Again, large aselective channels may do the trick and they may have been derived or are identical to existing channels in the ER. Another possibility is that the vesicle trafficking may have provided an early transport system. Either way, IC systems are not possible in evolution. (That's why the efforts of proving Behe wrong should be focused on providing alternative scenario's like mine).
Since the nuclear pores are formed by bending of membranes, for a simple proto nuclear pore, these membrane proteins are the only nes that are needed. I don't think you can derive the direction of evolution from your examples. It can just as well be the other way around (membrane curvature leads to vesicle budding).
In this respect, Gaspar Jekely has done large scale phylogenetic research on trafficking which suggest that secretory function appeared first in evolution, which is in line with my hypothesis.
Moving mRNA and ribosomes out of the nucleus (as happens in normal cell function) is simpler than moving the entire machinery inside, since in the latter case, a complete transport system out of the nucleus had to be in place. In my scenario, translation outside the nucleus is an added feature and normal translation and transcription in the nucleus can stillgo on during the gradual evolution of the transport system.
In essence, the current evolutionary approach is goal-oriented, namely the putative fitness advantage of future generations is taken as the goal. Read PZ Myers for example, he can't write a single piece in which he doesn't refer the cell or organism as being intelligent and goal-oriented.
The biggest mistake IMHO is that the implicit goal-orientation of MET is denied and can therefore not be questioned. I thnk the fitness goal-orientation should be replaced by for instance an intrinsic drive for greater complexity. An active exon-shuffling mechanism would fall into that category.
What I also try to confer is that you don't need divine interaction when talking about designs or goal-orientation.
Comment by AdR — November 1, 2006 @ 6:20 am
November 1st, 2006 at 10:37 am
Albert, if this "˜protopore' used the same proteins to curve membranes as the current one, then this protopore would leak proteins through diffusion at a rate at an alarming rate. If you disrupt transport in yeast cells, for example, fluorescently tagged proteins leak out of the nucleus (by diffusion through the pore) in seconds. I have no idea how you envision this to work if the protopore was in the outermost membrane of an organism. This means that your envisioned "˜protopore' must be fundamentally different from the current NPC; thus, not in line with your views on 'design by contract'.
If you say that I am right, that the acquired membrane needs large aselective channels, then I don't see you your scenario is an alternative that gets around the IC argument? In both cases an outer membrane has to appear with large aselective channels. How do you get around the IC argument?
(As an aside, which definition of IC are you using? There are so many different ones floating around that it is hard to know exactly what you mean.)
Albert, you are mistaken. Bending of membranes is only a precursor to forming a pore. Bending membranes is sufficient to form invaginations, useful for vesicle formation. However, membrane bending does not make a pore in a membrane.
Albert, there is no "˜moving of the machinery inside' in the karyogenic model. As the nuclear membrane and simple pores form, all required components can still freely diffuse throughout the whole cell. As the nuclear envelope closes, the NPC must develop it's selective transport function. So, the machinery is ubiquitous at the beginning of the process and is separated only as the pore and the nuclear envelope evolve. Why would this be more complicated than your scenario?
Again, what's the difference? You are replacing one "˜goal' (increased fitness) with another "˜goal' (greater complexity). And for one "˜goal' (increased fitness) there is actually evidence that natural selection favors it. However, the other goal (greater complexity) is not universal, as the evolution of S. cerevisiae nicely demonstrates.
Aside from that, I do not agree that there is an implicit goal-orientation in MET. A goal is "˜the purpose toward which an endeavor is directed' or an "˜intention' (a course of action that one intends to follow). The selection for greater fitness is not an intended plan. It is an involuntary necessity. When dump a million E. coli onto a plate with an antibiotic and only a few fortuitous ones survive because of the presence of the right mutation in the right place, there was not intention behind the whole process. But, let's not discuss this. It will lead us down a path of a philosophical discussion that I don't have sufficient time for.
—–
In summary, I think you are inconsistent with your application of your arguments. Your scenario is (just like the karyogenic models) at odds with your "˜design by contract' hypothesis. Your model (just like the karyogenic model) requires the sudden appearance of IC systems (large non-selective channels in a membrane). Finally, in the karyogenic model (just like in your model) there is no need to "˜move the machinery inside the nucleus'.
Comment by hrun — November 1, 2006 @ 10:37 am
November 1st, 2006 at 12:47 pm
As discussed in the article, I go from a semi-open system (e.g. porous rock) to a semi-closed system (large pores). Remember, we are not talking about fully functional cells as we know them, but the first living organisms. Also, in my model (which is open to revision by the way as any scientific model) the nuclear matrix couls have provided quite a lot of potential for a microenvironment.
I think in a different way, in functions (as most software architects do) and not in implementation. Since the original proto-cell was semi-open in my model, this functionality of letting waste en and food diffuse more or less freely in and out of the cell, this functionality should remain intact. Pores just do this trick. In interface terms, the inside of the nucleus expects delivery of food and removal of waste, it is not important how this is accomplished. Just as my fridge expects 220V AC delivered with a two-prong plug (in Europe), but it is unimportant if this comes from nuclear energy or solar energy.
Your comments are valid for my theory, but also for any other theory about the origin of cellular life. My definition of an IC system is that it cannot be formed gradually because the individual steps are interdependent, for example you cannot assume a fucntional nuclear pore to appear in a closed membrane, if the only way to get such a pore is the need of a closed membrane. The same holds for my plasma membrane: it cannot be first formed as a closed system and afterwards the channels are evolved. Two ways out of this problem: a) the first plasma membrane was not closed at all but also semi open and could only be closed after the evolution of aselective pores, b) the closing of the membrane happened concomitant with the generation when vesicles that bud off the ER contained already channels or the vesicle trafficking served as pinocytosis and waste disposal. This requires more research, but I don't see a fundamental problem.
The process I am referring to is similar to the formation of nuclear pores after telophase. It is a self-assembly process where indeed more proteins are involved. An interaction between DNA, nuclear matrix and membrane proteins shoudl do the trick.
I think there is. Pre: translation ans transcription in the cytoplasm, Post: translation separated from transcription. You write 'as the nuclear envelope closes, the NPC must develop its selective function'. But what is the driving force for that if it can still freely diffuse. Also it is a multistep process, where only the last step makes it functional.
As Aitchison and Rout put it: "More mysterious still is the question of how this translocative [nuclear export] pathway was retrofitted onto the ribosome during the evolution of eukaryotes from their protoplasmically unsegregated prokaryotic ancestors. Ribosomes may be old, but they've certainly not gone simple." A nice way to end an article, but the question should be posed first. How was it retrofitted? Since I see retrogradely fitting as an IC system, my first conclusion is that it is impossible. My model doesn't need retrofitting.
Comment by AdR — November 1, 2006 @ 12:47 pm
November 1st, 2006 at 2:11 pm
[...] See also discussion on Telic Thoughts [...]
Pingback by origin of the eukaryotic cell at Albert de Roos — November 1, 2006 @ 2:11 pm