|
Author
|
Topic: The nature of proteins: Sub-optimal design?
|
Grape Ape
Member
Member # 399
|
posted 02. January 2003 01:28
double post
[ 02. January 2003, 01:30: Message edited by: Grape Ape ]
IP: Logged
|
|
Grape Ape
Member
Member # 399
|
posted 02. January 2003 01:28
Hi Josh. Thanks for your reply.
quote:
But this just begs the question of why it's so tough to get them to fold right in the first place. If we found a system where the catalytic and structural units always folded fine (or never needed to fold) then we would have every reason to see design there too, because this would be elegant simplicity.
--I see this as equivalent to speculating on the number of alternative universes that exist but are undetectable. It would be a lot more productive to simply focus on how biology looks to us now.
I agree, but that's been my whole point. Speculating about the "design" of cells is just that -- speculation. This is true whether you're talking about suboptimality or optimality. Without some sort of theory about how and why the designs were actualized, there is nothing that can be tested by appealing to the evidence.
quote:
--Levinthal's paradox clearly indicates that as far as we can tell, the current system is exquisitely optimal, beyond any and all imagination, since we cannot at all understand HOW PROTEINS FUNCTION AT ALL, not to mention "how they function optimally." The fact that proteins function (within the scope of our current knowledge) indicates strongly that the system is well optimized.
From my understanding, Levinthal's paradox simply means that proteins do not reach their native structures by randomly sampling structure space, but rather do so deterministically. The intermediate molten globule state for example constrains the structural space which gets searched from that point on. This page discusses some of the issues with Levinthal's "paradox". Here's an excerpt:
quote:
Some have tried to resolve Levinthal's paradox by hypothesizing a reduction in the number of states that the protein has to search through. Typically it is argued that the protein first collapses rapidly to a compact conformation and that the subsequent search is through the greatly reduced space of these compact states[232,233,234]. Although in some cases this might so reduce the number of states that Levinthal's paradox no longer applies, it cannot in general provide a complete answer. For instance, the number of minima corresponding to the liquid-like state of a 55-atom Lennard-Jones cluster is of the order of 1012 (§3.6). This is much less than the total number of minima on the PES but the ease with which the global minimum can be found is still incompatible with a random search through this reduced configuration space. In fact, when the cluster has an energy or temperature in the melting region, it can switch back and forth between the Mackay icosahedral global minimum and the liquid-like states on a time scale of the order of nanoseconds (using parameters appropriate for argon)[110,111].
Seeking to resolve Levinthal's paradox by reducing the search space may be unproductive in a more fundamental sense, for it ignores a basic fallacy in Levinthal's paradox: namely the assumption that each point in configuration space is equally likely. Levinthal was effectively assuming that the PES is totally flat. However, the topography of a protein PES is far from flat, involving many `mountain ranges' and `valleys', thus making some configurations more likely than others. (In the canonical ensemble the low potential energy configurations have larger Boltzmann weights, and in the microcanonical ensemble they have a larger momentum density of states.)
Furthermore, I'm not sure what argument you're trying to make here. If I take your statement "we cannot at all understand HOW PROTEINS FUNCTION AT ALL" literally, I would have to say that there is an army of biochemists and biophysicists who strongly disagree with you. We know lots about how proteins function, but we don't yet know everything. On the other hand, if you're saying that the fact that we don't know some things about proteins is evidence for their design, I would say that your conclusion does not logically flow from your premises. Proteins are difficult to study because they're small, unstable, dynamic, and they cannot be easily produced in vitro. This has nothing to do with whether or not they're designed.
quote:
Until contrary evidence appears, basically resolving levinthal's paradox, I think we can assume that the proteome is optimal.
Why simply assume that? Why not assume the reverse instead? (And for that matter, which of the many zillions of proteomes are you referring to? ![[Wink]](wink.gif) )
quote:
(And I think any assessment of the situation taking levinthal's paradox into consideration would strongly argue pro-optimum.)
Again, I don't see a logical connection between Levinthal's paradox and optimality. Unless you're arguing that the "impossiblity" of proteins to fold into their native structures means that each one individually requires an intelligent designer just to fold. But I don't think you're arguing that.
quote:
Both evolutionists and design theorists should say the same: several billion years of refinement by selection/ intelligent construction should lead to very little suboptimal design in current systems. I find it a little inverted when someone argues opposite as evidence for evolution because they found some sloppiness. Shouldn't a billion years be enough time for exquisite refinement?
First of all, I'm not arguing for suboptimality. I'm arguing that it's nothing more than metaphysical speculation to say that the proteome is or is not optimal, because there is nothing to compare it to, nor is there even an ID hypothesis to explain how it came about, so we have nothing to go on. There are other situations in biology that are a little more questionable, since we do have other species with which to compare alternative "designs". And I would certainly say that much of genomics, if not proteomics, makes little sense except from an evolutionary perspective. I can't imagine why a designer would have included pseudogenes, for example, but we can watch evolutionary forces creating them.
The fact of the matter is that evolution has certain constraints on it that ID presumably wouldn't, so if you're looking for evidence one way or the other, seeing if life tends to conform to those constraints is how you go about doing it. Evolution, despite tons of time, is slow. Not every possible mutation will be realized. And most importantly, things like proteins tend to come from other proteins, so they should conform to a pattern of duplication/divergence and fit into nested heirarchies in the same way that species do. This is why protein can be arranged into families and superfamilies. Whether or not this leads to suboptimality is something I don't know (as per above, I don't think one can make that determiniation) but these tendencies are what we expect if evolution is true.
quote:
If we imagine evolution is solely responsible for the origin of the proteome, we are no more familiar with the type of "complexity" exhibited by the proteome than if we "imagine a designer."
But this isn't true. If evolution is responsible for the proteome, then we expect most of its pieces to have originated by duplication and divergence from existing pieces.
quote:
Imagining evolution or design still leaves the mystery of protein folding and function unresolved. This argument is incredulity at its best. Again, compare what we know about how/why proteins fold so well with how well they actually fold, and optimization rules the day.
Well, what we do know about proteins is that misfolding is very common. I don't have any statistics off-hand and I'm too lazy to search, but there are a number of cellular systems whose job it is to correct this problem, so we know that it's frequent enough to warrant doing something about it. Most commonly, the misfolded proteins just get destroyed, and the energy that went into making them is lost. Is this suboptimal? If one could make a system where misfolding was either rare or nonexistent, then I guess it would be fair to say that what we have is suboptimal, if we assume that there is a premium on energy efficiency. But of course there's no way to know that, so we can't say that it's suboptimal. But the converse holds too: we can't say that it's optimal either, and so any argument about "gee, look how well designed it is!" doesn't fly.
quote:
Now if you compare with some abstract and unknowable perfect reality (which we can't currently say isn't true to some extent of the proteome) then perhaps you might say that we can't know the exact degree of sub-optimal protein construction. But none of this gives great support for either evolution or design.
Well, I agree, but you seem to be arguing both that the proteome is optimal, and that we can't know if it's optimal. IDists can't have it both ways.
quote:
I recently attended a lecture by a protein folding expert who related some of the importance of these examples to us. The take home was that this scenario allows for the ability of a protein to acquire extremely high specificity without unwanted excess of affinity. In this scenario, the unstructured domain becomes structured upon binding and then reversed when unbound and the energetic difference between the folded states offsets the affinity necessary to acquire high specificity. This seems to be an ingenious example of design, incorporating very clever construction for a very specific purpose. There are other types of unfolded domains though, for example domains that allow flexibility so a protein can twist or fold back. If you consider the first class of unfolded proteins, the sequence must be very specific to accomplish the given function, however, for the second class of unfolded proteins I mentioned sequence restrictions would be very minimal.
I didn't say that disordered proteins had no function! Of course they do, or else we wouldn't expect them to be here, nor would we expect many of them to be highly conserved like they are. The point is simply that proteins don't always need to fold to be functional, and so therefore the unlikelihood of generating a folded protein de novo is of questionable relevancy.
quote:
In fact, there is evidence that disordered proteins outnumber ordered proteins, but that the ordered ones represent more resolved structures in the PDB simply because (big shock) they’re easier to crystallize.
What is the evidence for this? What are the statistics of the proteome for folded verses unfolded proteins/domains? (Since "unfolded" is doubtfully a recognized domain due to the probably large number of ways to achieve this state, there should be little sequence homology to give such a prediction of percent unfolded proteins/domains without direct testing on each individual protein, so I'd be interested in knowing how such a calculation has been currently made.)
One of the papers I linked to earlier (I think) was a study of sequence data that picked out recognizable folds and recognizable disordered regions. So the evidence is based on sequence data of unknown protein structures compared to the known ones. We won't know for sure of course until we resolve the structures of every protein.
quote:
Also, there are many reasons a protein won't crystallize besides the fact that they are unfolded.
Oh, I'm well aware of how hard it is to get a protein to crystalize, believe me. But this has nothing to do with the fact that it's even harder for disordered proteins. Without a regular structure, how do you get a crystal? With the continued improvement of NMR, this hopefully won't be a problem for much longer.
quote:
These other factors are so important for "why proteins crystallize," that implying that a large percentage of the proteome is unfolded because the structure of many proteins has not been resolved is completely fallacious (I would hope that is not the argument you are trying to make, although it appears to be.)
Er no, that was not the argument I was making. I was just pointing out that many proteins are disordered, and so therefore the unlikelihood of a random sequence having a good fold isn't relevant. I assume that so much of the proteome is disordered (if indeed it is) because folding is frequently unnecessary or undesirable for obtaining function.
quote:
So one possible way for folded proteins to come about is by evolving from functional yet disordered proteins, and in this case there would never be a period of time when there was not a selectable function.
This is a computational issue which relates directly to what I said about speculations. Basically you speculate that deriving function will be relatively easy in all cases of protein formation because function very loosely follows sequence identity. Thus you argue that function can arise very easily from within the sequence landscape. This however needs to be supported by something other than the isolated case of proteins with low sequence complex regions that are unfolded yet functional (of which you do not distinguish between types of unfolded domains and relate their sequence complexity to the functions they perform.)
I agree, but I doubt that anyone has undertaken such a study. I'm just pointing it out as a plausibility. That's really all it takes to defeat "evolution is impossible" type ID arguments. If you guys insist on bringing up various things in biology and saying, "how could this have evolved?" you're going to get some speculations. It's just not reasonable to expect us to know absolutely everything about evolution at this point in time. Perhaps you take a different tact, but when most IDists talk about the "evidence for ID", they are almost exclusively talking about the supposed impossibility of evolution. If there is a way to gain positive evidence for or against ID, that does not rely on the validity of evolution, then I have yet to hear it.
quote:
So both of these papers support the idea that an evolutionary process not only can account for the emergence of protein folds, but that the distribution of folds is a predicted consequence of evolution.
This actually does not follow. The two papers actually show the parity of the evolution of proteins, assuming evolution is true, for the following reason. Paper one discusses the fact that proteins fit into discrete relationships and many proteins utilize similar folds. The second indicates that the proteome only utilizes 10,000 folds. Now, if evolution has an easy time deriving sequences and function is so easy to generate, why is a particular domain saved and used over and over and over? Novel sequences should be so common and easy to generate that there is hardly any relationship between proteins at all, following the previous logic.
Who said that function was easy to generate de novo? Function is much easier to generate by duplicating existing proteins and having them diverge to perform a related or overlapping function. Assuming this happens, we should see a pattern whereby existing proteins fit into particular networks. Those papers I referenced show that this pattern exists for protein folds, which indicates that they did indeed come about by duplication and divergence. Of course, this still leaves unresolved where the first protein fold(s) came from, but if there only needed to be one or a few, then it's not much of a problem even if it's highly unlikely. Here's what Dokholyan et al say about their results:
quote:
The discovery of the scale-free character of the protein domain universe is striking and represents the main result of this paper. It has immediate evolutionary implications by pointing to a possible origin of all proteins from a single or a few precursor folds, a scenario akin to that of the origin of the universe from the Big Bang. An alternative scenario, whereby protein folds evolved de novo and independently, would have resulted in random PDUG (similar to the one shown in Fig. 3b) rather than that observed in the scale-free one.
If you think that this doesn't follow from their study, then I guess you'll have to take it up with the authors.
quote:
I admit that it is reasonable based upon this fact alone to say that evolution found what was needed for the first cell and then did the minimal amount of work to generate larger diversity. However, distinguishing this from a designer who carefully generated sequences that uniquely fit each organism in a maximal capacity for optimized function is impossible based upon this fact alone.
Of course it's impossible to distinguish a designer's actions from any natural process. A designer, if one does not define it, could have done absolutely anything it wanted. It could have even mimicked the evidence for evolution down to the most minute detail. How does this help us understand anything? That's why we say that ID isn't science, because it isn't doing what scientific theories are meant to do, which is to explain why things are the way they are instead of being different. What we can say is that postulating a unknown designer is completely superfluous if we have plausible mechanisms that can be observed.
quote:
Adding other arguments generates different conclusions, but I wanted to harp on this (since others have also brought up basically the samepoint on the phylogenies thread.. see I think John Brachts first few posts.) Secondly, the fact that only 10,000 domains are used indicates that evolution somehow had to select through a much larger set of theoretical protein sequences and domains to only end up with 10,000 (which is infintesimal compared to the possible sequences available.) This requires a "theory of discovery" for creating proteins, because only 10,000 are used.
There are different theories to account for the fact that proteins cluster into so few folds, such as the designability principle. But the most likely theory IMO is that protein folds form only rarely de novo, and that all or almost all of the existing folds came from one or a few precursors. Why the lack of folds should be a problem for evolution is puzzling to me. If there were millions of folds, you guys would be using that as evidence against evolution. (And would probably have a good point too, if indeed they are hard to generate.)
quote:
However, once you have a functioning cell, I don’t see any problem with the ability of current theory to account for protein evolution.
I don't see how this follows? Is function somehow easier to come by/ folding easier for proteins once inside a cell? How do cells narrow the relationship between sequence, structure and function?
Once you have an existing and functioning cell, then you have the raw material for duplication and divergence. It's more likely for a novel function to come about from an already functional protein than it is for a more or less random sequence to generate a function. (Hence, why Dokholyan et al liken the formation of the first fold(s) to the Big Bang.) That's why proteins can be found as members of families and superfamlies, or as chimeras. The hard part is getting that existing cell to begin with; that would seem to require numerous functions being generated de novo from random sequences, but of course there are theories (like the RNA world) which break this up into manageable steps.
quote:
Actually, there is good reason for it in relation to what I was saying. The fact that they lumped all proteins that displayed any type of disorder "sometimes existing in the native state" and found that these proteins displayed "significantly overlapping" order distribution says a lot.
Well, what it says to me is that "folded" or "unfolded" is somewhat of a false dichotomy. There seems instead to be a gradual cline between order and disorder. This would reinforce the point that ordered proteins could evolve from disordered proteins, because they wouldn't have to become tightly folded all at once. But as before, I think the current evidence shows that most protein folds come from other folds.
quote:
I think the analysis would be better if function of the disordered region was taken into consideration. Perhaps then a more general conclusion can be made about unfolded domains and how they widen the relationship between function and structure and sequence.
I don't know why function should be relevant to sequence complexity, but we do know that many disordered and loopy proteins are highly conserved, so that indicates that they have an important function. (Well, if you're an evolutionist it does.) It may be that their functions are distinct from those of folded proteins, which is certainly true at least some of the time. But whether or not this can stand as a reasonable argument against one evolving into the other is questionable IMO. There may be reasons to think that ordered proteins cannot come from disordered ones, but I'm not aware of what those reasons might be.
IP: Logged
|
|
Josh
Member
Member # 405
|
posted 02. January 2003 12:03
GrapeApe-
We are beginning to talk past each other and move from discussing the evidence into debating. I do not enjoy debating so I will decline the opportunity. However, I am still interested in discussing the evidence, which requires more effort and closer scrutiny of evidence and facts presented here. Also more investigation into these issues will promote a useful discussion rather than a debate. That being said, I need time to investigate and look further in depth on some issues to clarify some of the points I have made. This serves as notice that I may take a couple of weeks to respond. Please keep checking back, however
Josh
IP: Logged
|
|
|
|
|
Printer-friendly view of this topic