|
Author
|
Topic: Response to Jason Rosenhouse
|
charlie d.
Member
Member # 159
|
posted 05. September 2004 18:46
quote:
First of all, my reading of the article didn't seem to indicate how many species were examined. As I showed with the bilateral symmetry of the nematode expressed through to different develomental pathways, we can't make the rote generazation that symmetries are automatically generated the same way in every organism even though such homology are reasonbly inferred to exits.
The discovery of more counter examples, such as in the nematode, in terms of pentadactyl development would strengthen Denton's argument. Here we have a case of areas ripe for novel investigation. But even if we find that the homology is universally controlled by the same sets of genes, it raises a potenially worse problem for natural selection as now evolution of features require selection pressure to act on several features simultaneously.
I think you are missing the point: "convergence" and "homology" have little if anything to do with the pentadactyl pattern of fore- and hind-limbs. The reason why the follow the same pattern is that they use the same developmental pathways because they are expressions of the same segmental properties.
As for the species analyzed, that was of course just mouse (you can't make human knock-outs). This actually has nothing to do with Denton's point (the supposed evolutionary conundrum of the origin of similar developmental plans for the hind- and fore-limbs in the same organism, not between organisms), but anyway, all we know of these basal developmental pathways (including hox genes in limb development) shows that they are highly conserved.
Regarding what the nematode vulva shows, you have not given sufficient detail to make any conclusion or "show" anything about "rote generalizations" and such. From what little you say, it seems to be a case of redundancy, but maybe I am misinterpreting. Provide some basic info and reference, please. quote:
In Proceedings of National Academy of Science article on developmental biology, we have the following rather curious fact regrading expression of the pentadactyl form:
quote:
--------------------------------------------------------------------------------
A similar dose-response is observed in the morphogenesis of the penian bone, the baculum, which further suggests that digits and external genitalia share this genetic control mechanism.
--------------------------------------------------------------------------------
The article goes on to relate the fact that the number of toes (ok, Digits) is tied to penis characteristics. How in the world from a scientific standpoint do we formulate selectionist pressures to account for this curious fact? This brings to bear the rather difficult issues of morphological convergence when genes have such deeply integrated diverse, non-modular roles. Denton in Evolution a Theory in Crisis explores the problems posed by pleiotropy to Darwinian Evolution in "The Failure of Homology". He further elaborates the problem in Nature's Destiny. I believe exploration of these topics will lead to further fruitul areas of pure observational data gathering.
Charlie pointed to the fact that convergence of the pentadactyl is attributable to the same genes, but this is complicated by the fact that the same genes affect other parts of the organism. Simply arguing for effeciency of re-use is problematic, because then how can one account for the evolution of genes that are so deeply integrated into the biology of the organism.
Again, I am not arguing that one can explain "convergence" between hind- and fore-limbs, but that their similar developmental patterns have nothing to do with "convergence" (at least in the evolutionary sense of the term, which I think you are using, though I am not sure).
Also I fail to see what the problem is with the baculum thing. If anything, the sharing of developmental pathways and programs between unrelated organs is another example of nature as bricoleur. (If there are deep engineering reasons for having baculum size correlate with digit development, they escape me).
As for Denton, this is what the IDists at ARN had to say about Nature's Destiny vs Evolution: a Theory in Crisis quote:
In August 1998, Denton’s eagerly-awaited second book arrived: Nature’s Destiny: How the Laws of Biology Reveal Purpose in the Universe (Free Press, 1998). Readers expecting a continuation of the arguments of Evolution: A Theory in Crisis, however, found a line of argument markedly different from the earlier book. Although much of Denton’s skepticism about neo-Darwinism remained, gone were the challenges to the theory of universal common descent--i.e., the common ancestry of all terrestrial organisms--which had made Evolution especially controversial with mainstream biologists. In their place was an unstinting advocacy of common descent, and a notion of “directed evolution” in which the historical unfolding of life on earth was “built into” the universe from the start.
Oh, yeah, and tenascin-C-deficient mice have a phenotype (several, in fact). But that's the problem with running behind the data, rather than with them.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. September 2004 03:56
Charlie:
You wrote:
quote:
The problem of course is that codon reassignments bring global proteome changes. So, while it is a good rule of thumb to say that a single or few conservative substitutions will not disrupt the structure or function of a given protein, one may expect that at least some protein will be affected if all the residues encoded by a certain codon are replaced.
This is not just a off-the-cuff prediction, by the way - people have been playing around with genetic codes for many years. To my knowledge (but you may want to verify this, since this is not my field), even in organisms with limited proteomes, such as E. coli, artificial changes to the code by tRNA manipulation result in decreased viability (just like suppressor of amber mutations). Thus, a code variant would be expected to evolve, like suppressor of amber, only in conditions in which it can effectively counterbalance some other selection pressure. Those conditions should be quite rare.
Thanks, however, I suspect what you are referring to is the work done at expanding the code (i.e., introducing non standard amino acids outside the usual set of 20). I don't know of any work at merely swapping codons. If your memory is jogged I'd appreciate any information you might have.
Of course you are correct that any such codon reassignment is going to be proteome wide. But anyone familiar with proteins knows how trivial a few Leu --> Ile, for instance, swaps would be. For this and the other reasons I mentioned, I think the state of the art indicates that there are many codon reassignments that could easily be tolerated.
Certainly, this conclusion could turn out to be wrong. But, at this point, we don't have reason to deny it, aside from the presumption of evolution.
How do we explain the near universality of the code from an evolutionary perspective? It seems incredible that it would be so stable. And furthermore, if it is found to be so stable, then how did it evolve in the first place. Indeed, the view that the code did evolve seems to be, itself, at odds with the idea that it is so stable. This seems to be a tension within evolution.
[ 06. September 2004, 11:56: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 06. September 2004 10:41
quote:
Thanks, however, I suspect what you are referring to is the work done at expanding the code (i.e., introducing non standard amino acids outside the usual set of 20). I don't know of any work at merely swapping codons. If your memory is jogged I'd appreciate any information you might have.
No, I am talking about codon reassignment. These were mostly old papers, from the '80s I'd say. Here and here are a couple of recent ones; try walking backwards from those. quote:
Of course you are correct that any such codon reassignment is going to be proteome wide. But anyone familiar with proteins knows how trivial a few Leu --> Ile, for instance, swaps would be. For this and the other reasons I mentioned, I think the state of the art indicates that there are many codon reassignments that could easily be tolerated.
As I said, the individual conservative amino acid substitution is highly likely to have no impact on a protein's structure, but all it takes is one particularly sensitive position in one protein (out of the many thousands positions that would be altered by a code change even in a simple organism like E. coli) to effectively select against global codon reassignment. I think therefore your assumption of neutrality is far from proven (that is, until you find a way to prove it experimentally - there's a good research project for you!). quote:
First, how to explain the near universality of the code from an evolutionary perspective. It seems incredible that it would be so stable. And furthermore, if it is found to be so stable, then how did it evolve in the first place. Indeed, the view that the code did evolve seems to be, itself, at odds with the idea that it is so stable. This seems to be a tension within evolution.
This is just an issue of perception, frankly. You are skeptical, most scientists don't see any problem. There are examples in nature of mutant and alternative codes, there are examples of "wobbly" codons, etc. The observational and experimental data indicate that limited code evolution can occur in certain, rare circumstances. If you think the code can't evolve, or should evolve more readily (I can't understand which one you think is the case) I think the burden is on you. quote:
Second, this seems to be case of evolution strongly influencing research. That is, given the presumption of evolution and the evidence of the (near) universal DNA code, we had to assume that the code was highly stable ("frozen").
That may, in part, be true, but it is understandable and appropriate. Once a theory has strong evidentiary support, it is taken for granted in most cases and does not need to be re-examined in every instance. NASA guys don't feel compelled to show that our physical laws and geological processes apply on Mars before discussing how martian rock formations could have arisen. That's how science works.
[ 06. September 2004, 10:41: Message edited by: charlie d. ]
IP: Logged
|
|
Salvador T. Cordova
Member
Member # 959
|
posted 06. September 2004 11:18
Hi Charlie D,
Thank you for responding. I'd like to address the points you brought up, however, for the time being, I'll let Cornelius take the lead on his thread, and I may pursue your points much much later. I don't want to be a distraction, and I want Cornelius to have the chance to field and take questions especially from scientists like you.
respectfully,
Salvador
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. September 2004 12:31
Thank you Charlie for those citations, I appreciate it. I take back my second point (and have edited my post accordingly). There are a few papers in there that are quite relevant to this discussion. But I think the best one was the first that you cited:
Doring V, Marliere P., "Reassigning cysteine in the genetic code of Escherichia coli," Genetics. 1998 Oct;150(2):543-51.
They reassigned four codons, AUU, AUC, AUA, and AUG to cysteine, and found it to be "deleterious but well tolerated." I was thinking of something less aggressive than this, yet this one worked.
One reason they chose Cys was so they would not introduce a bigger amino acid, but otherwise Cys is fairly unique. Typically, after Trp it is the second most rare amino acid. With this reassignment Cys goes from less than 2% abundance to something more average in abundance. Met is fairly similar to Cys, but Cys is quite a bit less hydrophobic than Ile, and of course can introduce disulfide bonds.
This work shows that a code change results in a workable system. Evolution looks for mutations that, while not improving fitness, can be tolerated. It can subsequently be fixed, and perhaps even become part of an improved design.
Obviously, reassignments that are less aggressive would have even less effect than this aggressive experiment. This supports my earlier point that the code is not at a sort of spike in the fitness landscape where any change cannot be tolerated, but rather that there are several reasons to think code evolution (to some extent) is quite feasible.
[ 24. September 2004, 12:06: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 06. September 2004 12:49
Cornelius:
those changes were deleterious. The strains could not be propagated in the absence of a counterbalancing selection (suppressor of missense).
I suggest you try to read the relevant literature first, before commenting further.
[ 06. September 2004, 12:55: Message edited by: charlie d. ]
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 06. September 2004 22:38
Charlie:
There are two different issues at hand here: the pathway for codon reassignment and fitness of an organism given a codon reassignment. My point dealt with the latter. That is, given the facts that
(1) most amino acids have multiple codons so codon reassignment does not necessarily mean eliminating any amino acid altogether,
(2) many amino acids are similar so conservative reassignments are conceivable,
(3) proteins generally can tolerate a substantial fraction of sequence substitution;
then it seems that certain DNA code changes should be insignificant. Nonetheless, you doubted this because a global codon reassignment would probably cause problems somewhere in the proteome. You now have provided references that provide experimental evidence for my point [1-3], though it appears that only a fraction of the potential codon reassignments have been experimented with. As I noted earlier, it is a bit surprising that cysteine has been found to be a reasonably viable substitute for other amino acids. Nonetheless, "based on a number of studies it was suggested that cysteine, possibly because of its small size and amphiphilic character, is an acceptable replacement at most protein positions without deleterious effect on cell viability." [3] I noted that [2] shows some interesting and significant results for substituting cysteine for isoleucine and methionine.
You did recall correctly, however, that there are cases where codon reassignment degrades fitness, such as in using Phe instead of Leu for codon CUC in E. coli [4], or Ser instead of Leu for codon UUG [5]. Your contention, as I understand it, is that the standard code is pretty much at a local optimum, and any change, however innocuous it may seem, will likely confer some sort degradation to make it selected against. Not surprisingly to me, that is not well supported.
The other issue at hand is the pathway for evolving the code. This is where your comment about the required counterbalancing selection comes in. One particular model of codon assignment evolution was tested in [2].
They used a plasmid with the gene for Thymidylate synthase (ThyA). Thymidylate synthase is the enzyme for the final step in the synthesis of deoxy-thymidine monophosphate (dTMP). An E. coli cell needs only about a hundred copies of Thymidylate synthase. Its active site has a cysteine which is known to be required for activity. If you inactivate the enzyme then E. coli needs exogenous thymine (or the thymidine nucleoside).
So, here's the trick. They replace the cysteine codon (UGC), of the plasmid ThyA gene with each of the AUN codons (AUU, AUC, AUA, and AUG); they use a strain of E. coli that otherwise lacks the normal ThyA gene; and they supply exogenous thymidine. So they have four strains in all, one for each of the four AUN codons. At this point, the plasmid is not needed.
They next transform with another plasmid with a gene for a cys tRNA which contains a missense, using the anti codons for the AUN family. That is, a gene that changes the DNA code so that cysteine is used for any of the four AUN codons. So now there are 16 strains in all, since there are 4*4 combinations. They propagate these in the absence of thymidine.
Because there is no exogenous thymidine, the ThyA gene must be expressed. But to be active, it requires the the DNA code change that is supplied on the second plasmid (which is going to supply cysteine at the needed sequence location). Of the 16 different combinations, clearly there are 4 that are expected to work (the ones with the corresponding AUN codons and anti-codons in the ThyA gene and new amino-acyl tRNA synthetase, respectively). These 4 strains, for the most part, grew just about as fast as the strains that had exogenous thymidine, demonstrating that the plasmid-supplied genes were being expressed and, importantly, the DNA code change was active. Cysteine was now being used where isoleucine or methionine was being used.
There are more details of course. For instance, when they were able to get cysteine to replace just about all the isoleucine, then growth rates were hindered (I didn't catch that on my first reading, though I'm not surprised). But the bottom line is that there are DNA code changes that do not hinder growth rates.
Charlie is right that they used a selection technique to "activate" the code change. But this is entirely aside from the question of whether proteome-wide codon reassignments can work. They can. The selection technique they used was due to the particular model they are using for codon reassignment. Since code variations exist in nature, no one questions that such reassignments are possible via evolution. Certainly, the question of pathway and fixation is interesting, and there are several ideas out there. No one knows how it actually happened, say when yeast species switched the leucine codon, CUG, to serine. [6] But this is a different question than the question of whether proteome-wide reassignments are feasible.
Notes
1. Doring, V., et. al., "Enlarging the Amino Acid Set of Escherichia coli by Infiltration of the Valine Coding Pathway," Science, 292:501-504, 2001.
2. Volker Döring, V., Marlière, P., "Reassigning Cysteine in the Genetic Code of Escherichia coli," Genetics, Vol. 150, 543-551, October 1998.
3. Ahel, I., et. al., "Cysteine Activation Is an Inherent in Vitro Property of Prolyl-tRNA Synthetases," J. Biol. Chem., 277:34743-34748, 2002.
4. Pages, D., "Suppression of a double missense mutation by a mutant tRNA(Phe) in Escherichia coli," Nucleic Acids Research, Vol 19:867-869, 1991.
5. Thorsteinsdottie, S., et. al., "Escherichia coli supH suppressor: temperature-sensitive missense suppression caused by an anticodon change in tRNASer2," J. Bacteriol., 161:207-211, 1985.
6. Ohama, T., et. al., "Non-universal decoding of the leucine codon CUG in several Candida species," Nucleic Acids Res., 21:4039-4045, 1993.
[ 11. September 2004, 19:14: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
Scott
Member
Member # 1222
|
posted 07. September 2004 00:23
quote:
charlie d:
If you think the code can't evolve, or should evolve more readily (I can't understand which one you think is the case) I think the burden is on you.
I don't mind being a distraction. ![[Smile]](smile.gif)
Which does evolutionary theory predict? Or does evolutionary theory predict anything at all with regard to this topic?
It seems to me that evolutionary theory does not, and could not, even predict the existence of the genetic code, much less any particular genetic code. So how is it that it can be said to predict that the genetic code would be "universal"?
Now if the "universality" of the genetic code is not a prediction of evolutionary theory, how is it that it should be taken as evidence in favor of the theory?
Actually, rather than being a distraction, I think this gets back to one of the main issues raised in the OP. What is the evidentiary value of the genetic code?
Essentially, any shared character could be asserted to be a prediction of common descent, but why should that be the case?
quote:
The place of prediction in the scientific method
In the scientific method,
- "'observe', 'wonder, react and guess', 'predict', 'test' and finally 'review'",
- a prediction is a logical consequence of some hypothetical explanation of an observation.
Prediction - Wikipedia
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 07. September 2004 00:55
Cornelius:
read the paper again, then answer these questions. In Figure 1, why did they overexpress the CysS gene in some of their mutant Cys-tRNA expressing strains? Why did that overexpression result in reduced growth? In Figure 2, why did the number of chloramphenicol-resistant cells go down 100-fold over 500 generations in cells bearing wild-type thyA (b5301)? Why do they say "A single missense mutation thus seems sufficient to perpetuate an ambiguity in the genetic code"?
Now, do you still think that the paper "provide[s] experimental evidence for [your] point"?
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 07. September 2004 03:34
Here I respond to Jason Rosenhouse's blog entry of September 5, 2004, at:
http://evolutionblog.blogspot.com/
regarding my chapter in Uncommon Dissent.
Regarding the DNA code, Rosenhouse admits that he is not able to describe to what degree the code could have evolved:
quote:
The question then becomes just how many divergences the common descent hypothesis can tolerate. It would take someone more knowledgeable about evolutionary genetics than I am to give a good answer to that …
Rosenhouse is not alone here, as this is a difficult question. What is clear, however, is that it would be difficult, as I earlier pointed out, to place a tight limit on the evolution of the code without presupposing evolution. Amino acids tyically have multiple codons so single codon reassignment does not mean a complete absence of the corresponding amino acid. Furthermore, many amino acids are similar, so conservative code changes are conceivable. And finally, proteins are quite robust to amino acid substitutions. Perhaps someday we'll find evidence for the code's rigidity, but at this point we have only speculation.
So this invalidates Rosenhouse's claim that the near universality of the code is strong evidence for common descent. Greater differences in the code, at this point anyway, could be explained by evolution and common descent.
Rosenhouse next turns to an erroneous idea from Ken Miller. Miller says that there is "something remarkable" in the patterns of the variations of the DNA code. Miller concludes:
quote:
What this means is that these slight variations of the code provide powerful -- and unexpected -- confirmation of the evolution of the code from a single common ancestor.
Actually, there is no such evidence. Miller's powerful confirmation is, in reality, scattered across various types of organisms. For example, the UAR codon is observed to switch from “stop” to “Gln” in green algae, various ciliates, and some diplomonads. Likewise, the UGA codon is observed to switch from “stop” to “Trp” in other various ciliates and two firmicutes. There is no powerful, unexpected confirmation of evolution here. [1]
Unfortunately Rosenhouse accepts Miller's tale unequivocally, concluding that "the pattern of divergences that are known are also consistent with the idea that they are derived from the standard code via descent with modification."
Regarding the fossils, Rosenhouse mistakenly thinks that "The fossil record reveals a history of life that is consistent with evolutionary expectations." Actually, evolution does not expect for phenomenal complexity to appear abruptly. In fact, it does not expect phenomenal complexity at all. Also, if the earth was full of nothing but bacteria that could just as easily be described as being consistent with evolutionary expectations.
Regarding comparative anatomy, a few posts back Rosenhouse agreed that massive convergence would be a problem for evolution, now he finds it to be vindication:
quote:
But why is convergence a problem for evolution? If the various genetic modifications required to produce saber teeth occur with reasonably high probability, and if there is selection pressure in favor of such teeth, then the convergence of these structures is easy to explain in terms of standard mechanisms. In fact, under such circumstances, a lack of convergence would be puzzling. In this sense, many known convergences can be viewed as vindications for evolutionary theory.
Evolutionists generally do not acknowledge evidential problems. Hence, massive convergence becomes a vindication for evolution. Rosenhouse denies evolution invokes ad hoc explanations, but this is ad hoc.
Although he agreed that massive convergence is a problem for evolution, Rosenhouse now asks why evolution is incapable of explaining convergences. That is a convenient way of framing the problem, but it misses the point. Above, Rosenhouse claimed the pentadactyl pattern strongly suggests common descent. They are obvious similarities right? So they strongly suggest common descent because they must have come from a common ancestor.
But now, with similarities that could not conceivably have been inherited from a common ancestor, this too is a vindication because it must have been caused by similar mechanisms responding to similar selective pressures. This is ad hoc. If a similarity can conceivably be ascribed to a common ancestor then it is viewed as a homology and strong evidence. If it cannot be, then it is viewed as an analogy, and again strong evidence.
The point is that if similar designs are present in distant species, where common descent cannot be used to explain those similarities, then common descent need not be invoked to explain similarities in species that are not so distant. Homologies, such as the pentadactyl pattern, were a key argument for Darwin. He viewed them as a mandate for common descent. But this argument is contradicted by the convergences. It is not scientific to say that the pentadactyl similarity mandates common descent when there are other such similarities that do not mandate common descent. The argument is arbitrary. It is not a question of whether the theory allows for similar adaptations to evolve, it is question of whether or not evolution is supported by the evidence.
Regarding the question of origin of life (OOL) Rosenhouse continues to maintain that the creation of the DNA code is outside of evolution and an OOL problem. But he avoided answering my question. Does he think the creation of the DNA code is a serious problem for OOL? If not then this is merely a rhetorical dodge since he would believe the code evolved (somehow). Furthermore, Rosenhouse fails to understand that DNA code evolution falls into the category of Darwinian evolution, even according to his own definition:
quote:
The universal common ancestor possessed a genetic code, and that is the point from which evolution is considered to begin. To even discuss anything like a Darwinian evolutionary process, you need a collection of imperfect replicators competing for resources. The first replicators were likely far simpler than the first thing that was unambiguously alive, but the fact remains that evolutionary theory takes for granted a certain minimal level of complexity. The distinction between the origin of life, and the subsequent development of life once it appeared, is not complicated.
We can agree with all this, and it means the DNA code evolution falls into the Darwinian evolutionary process. The code had to have evolved from simpler codes. How strange that the first cells to appear with the extant code constitute "a collection of imperfect replicators" but in going back just one step in the code's evolution we no longer have "a collection of imperfect replicators."
Rosenhouse's argument may seem confused here, but he is following a common strategy. For evolutionists, the near universality of the DNA code is just too tempting to turn down as powerful evidence. Evolutionists have always argued that the DNA code is frozen, or nearly so. Minor variants may be possible, but nothing more. At some point the cellular system just became too complex to sustain any more code changes.
As noted above, this claim is problematic, but ignoring those problems for the moment, a frozen code that is essentially universal among the species is too good to turn down. Unfortunately, there is another thorny problem to deal with; namely, how did the code arise in the first place? If the system is so complex so as to prohibit code evolution, then how could the code have evolved in the first place? Indeed, evolutionists have only a variety of vague speculations about how the code could have arisen. The answer to this question, and here is the strategy, is to insist that the evolution of the code must be outside of evolution. The universality of the code mandates its use as strong evidence, so the evidential problem of its creation must be removed from evolution. Moving the code's evolution from evolution to OOL is arbitrary. It is a rhetorical move, not a scientific move.
Finally, Rosenhouse finishes up with a misinterpretation of Ridley's statement that if the species had independent origins then they would not have the same code. Rosenhouse erroneously thinks Ridley is using a random chance model for independent origins:
quote:
The quote from Ridley does nothing to alter that assessment. He is assuming only that it is asking too much of chance to argue that precisely the same genetic code and translation apparatus evolved more than once.
Rosenhouse need only read Ridley more carefully to clear that up.
Notes
1. Knight, D., "Rewiring the keyboard: evolvability of the genetic code," Nature Reviews-Genetics, 2:49-59, 2001.
[ 07. September 2004, 10:56: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 08. September 2004 21:52
Well, since Cornelius is not answering my questions, I'll answer them myself.
In the Doering article, Cys-tRNAs with altered codon specificity (for one of three isoleucine codons, or a methionine codon) were overexpressed in bacteria, so that cysteines would now be occasionally inserted into proteins in place of Ile or Met encoded by the relevant codons. However, the replacement was not complete, even in the presence of overexpression of the mutated tRNA, becase of various factors, chiefly stochastic competition with wild-type, normal tRNAs in the bacteria. The replacement effect could be however potentiated by overexpression of the enzyme that "loads" cysteine onto the tRNA (so that a large frction of the mutant tRNAs would be loaded at all times).
The experiment in Figure 1 just looks at rapid, short term growth in liquid medium to establish bacterial viability. Because proteins with misincorporated aminoacids are more sensitive to certain structural stresses, including increased temperature, they also checked for growth differences at 30, 37 and 42 degrees C. The results show that bacteria overexpressing all tRNAs grow fine in most cases (the only exception being one of the mutant tRNA-expressing strains at 42C). However, when misincorporation of cysteine was increased by overexpression of the loading enzyme, 2 of the strains displayed reduced growth at 37C, and all of them did at 42C.
This shows that substitution of Cys for some Ile's or Met (even when not as complete as it would be in a bona fide code mutant), is in fact sufficient to detectably alter protein stability and cell growth, even in rich, glucose-supplemented medium.
But this experiment provides just a gross estimate of the effect of Cys replacement on bacterial fitness. More indicative is the next experiment. The authors took one of the tRNAs (Cys-tRNA/UAU) that had essentially no impact on rapid growth rates (and the least impact at 42C with loading enzyme overexpression), and expressed it into two bacterial strains: one wild-type, and one in which a mutant gene could be "rescued" by misincorporation of cysteine at a critical position by Cys-tRNA/UAU. Then they grew the 2 strains in parallel under selection against the mutant gene. Both strains grew fine, but when the authors checked after ~500 generations, they found that the wild-type strain, which did not require Cys-tRNA/UAU, had lost the genetic element that encoded the mutant tRNA (100-fold decline in 530 generations). At the same time, all cells from the strain that required the mutant tRNA for the selection gene to be functional did, of course, retain mutant tRNA expression.
This shows that, in the absence of selection, even partial cysteine replacement (remember, these guys do not over-express the loading enzyme) has enough detrimental effect on fitness that the cells expressing the mutant tRNA are quickly outcompeted by non-expressing counterparts.
Again let me remind you that this is only a partial codon shift, not a complete one as would occur in a variant genetic code. Yet, contrary to Cornelius' claims, it is counterselected.
The authors note finally that presence of a single muation that can be suppressed by Cysteine misincorporation is sufficient for genetic stability of the mutant Cys-tRNA. This of course suggests a way by which, in particular circumstances, such codon shifts can provide a selective advantage and become fixed in a population.
[ 08. September 2004, 21:55: Message edited by: charlie d. ]
IP: Logged
|
|
Darel R. Finley
Member
Member # 100
|
posted 10. September 2004 08:50
Dr. Hunter,
I think that it might be legitimate for Rosenhouse et al to cite the universality of the DNA code as powerful evidence of common ancestry, while keeping OOL (origin of life) separate. This is because DNA code universality could illustrate the common descent of all existing organisms from the first microbe to use the DNA code, without regard to how that microbe came to be.
That said, I still agree with pretty much everything you are saying, and I think that the evolutionary community is obligated to address OOL as long as they insist on a purely naturalistic worldview in general.
And while we're on the subject, I would also like to stress that Behe's irreducible complexity is sometimes dismissed as an OOL issue, but it is much more, since the first microbes would not have any use for blood clotting, immunity, vision, and other systems Behe examines.
Darel
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 10. September 2004 12:48
Charlie:
I'm sorry for the delay in responding. I was going to answer your earlier post. Now that you have posted your own response, let me respond to both posts at once.
As I pointed out earlier, there are two issues here: the pathway for codon reassignment and fitness of an organism given a codon reassignment. My point is that given that a codon has been reassigned, there are many reassignments that are likely to be rather neutral because:
(1) most amino acids have multiple codons so codon reassignment does not necessarily mean eliminating any amino acid altogether,
(2) many amino acids are similar so conservative reassignments are conceivable,
(3) proteins generally can tolerate a substantial fraction of sequence substitution;
I pointed out that the Doring paper [1] provides experimental support for my point. You disagreed, saying that in [1] they required a selection technique to activate the code change. I agreed, and pointed out that there are many models for codon reassignment and that they were testing out one model. Also, that the use of selection hardly disqualifies the method they used anyway. And finally, that no one questions that such reassignments are possible via evolution.
Now in your most recent two posts, you continue to say that [1] does not support my point. Your refer to Figures 1 and 2 to make the point, in two different ways, that their codon reassignment experiments reveal sufficient levels of detrimental effect on fitness to render the DNA code changes untenable.
I agree with you that detrimental effects were found in some of the Figure 1 experiments, but only some. Furthermore, I think the Figure 2 experiment does not support your conclusion as you describe.
First, for Figure 1. For others reading this who are not familiar with the paper, these researchers began by deleting a necessary gene from E. coli (the gene is ThyA which codes for Thymidylate synthase). The interesting thing about this gene, in addition to the fact that it is required, is that codon 146 needs to code for cysteine if the gene is to be active. The researchers supplied an inactive version of the gene which had a different codon at position 146. Actually, they did this four times, testing out four different codons, one for methionine and three for isoleucine.
Obviously, these cells are not going to survive in this state. But the researchers then provided another gene that provides for an alternate DNA code with a single codon change (the gene codes for the cysteine tRNA, but the translator codon is switched from cysteine to methionine or isoleucine). Specifically, in each case, the codon change was precisely what was needed to activate the necessary gene (ie, make a cysteine go at position 146 in Thymidylate synthase). In other words, they changed the methionine codon and each of the three isoleucine codons, in turn, to be cysteine. This would activate the necessary gene so the bacteria could survive; however, at the same time, this codon reassignment would be happening in other proteins as well. Overexpression of the DNA-code-changing gene helped to make the change occur as much as possible.
If people have trouble following all that, the bottom line is that the DNA code was altered, though not absolutely 100%, so that 1 out of the 64 codons was now switched to cysteine. Across the E. coli's proteome the cysteine amino acid would now be showing up where there was supposed to be a methione or isoleucine, in four separate experiments. This is not exactly a conservative change, as cysteine is, nominally, quite rare. And it is reactive compared to isoleucine. Also, it introduces the possibility of a disulfide bond. Swapping cysteine for isoleucine is not like swapping two similar amino acids.
So what happened? The growth rates of the four DNA code variants were indistinguishable from the control cases at lower temperatures. They had reduced growth rates at high temperatures. And two of the four variants did fine at medium temperatures. Charlie concludes that these results demonstrate that these DNA code variants aren't doing so well. Well, that's true at high temperatures but not at low temperatures, and not in two of the cases at medium temperatures. My point was not that all codon reassignments are workable, nor that codon reassignments are going to always work in all environments. Indeed, what surprised me was that these non conservative reassignments (not 100% but pretty high) worked as well as they did. Six out of the 12 experiments showed the DNA code variants doing quite well. These results support my original point that there are reassignments that are likely to leave the organism working just fine, even though the change was implemented proteome-wide.
Now on to Figure 2. In this experiment, the researchers took the DNA code variant that did the best and grew it for 80 days (or 530 generations). Once a day they had to sample the strain and restart it in a fresh medium to keep things going (serially subculturing). Also, they inserted markers so they could track the inserted genes. Specifically, they provided the gene to change the DNA code and gave it one marker (chloramphenicol-resistance, call this marker CH). And they provided the necessary ThyA gene mutated to work only with the altered DNA code, and gave it another marker (carbenicillin-resistance, call this marker CA).
They also created a control case, where the only difference was that the ThyA gene was not mutated. Therefore, the the gene to change the DNA code would not be needed. Everything else was the same, including the markers.
So what happened? The DNA code variant strain propagated just fine. It grew just as well as the control case, and both markers continued to give signal. In other words, the DNA code change did not go away. They even did an independent check at the end to ensure that the DNA code change had not reverted. It hadn't. The authors concluded that "A single missense mutation thus seems sufficient to perpetuate an ambiguity in the genetic code."
The control case showed that the growth rates were the same between the two strains. Also, in the control case while the CA marker persisted, the CH marker gradually went away. This means that the presence of the gene to change the DNA code gradually decreased. How did this happen? Well, what if at the beginning of the experiment that gene (and its CH marker) failed to make it into every E. coli cell? The population would be mixed, with most E. coli having both inserted genes and a few with only the the ThyA inserted gene (any E. coli that did not have the ThyA gene inserted would have quickly died off).
If those few E. coli that lacked the gene to change the DNA code had even an ever so slight higher growth rate, then eventually they would take over and dominate the population. And that is apparently what happened.
So what does this mean. First, recall that the Figure 1 results and the Figure 2 experiment tell us that the growth rates are comparable between the two strains. In fact, the effect of the code change is even less than indicated in Figure 1. This is because the code change is happening only to a minor extent (~50%) in the Figure 2 experiment because the code change has not been selected for. So the code change that produced no significant change on growth rate is now having even less effect in the control case.
Charlie suggests that the fact that the code change gene (and CH marker) went away indicates that even a partial code change has enough detrimental effect on fitness that those cells are quickly outcompeted. Well, that may be true, but it may not be. We're talking about a minor change in fitness. Those cells were not "quickly" outcompeted, and the minor code change is not the only difference between the strains. Remember, the code change strain is also carrying an additional gene now compared to the other strain that beat it out.
In any case, even if the minor code change is the cause of the minor fitness reduction, it is still not clear what to make of that. Even known code variants are recognized to have been non adaptive. Remember, the Figure 2 experiment is a highly artificial environment where one of the two strains is likely to become dominant eventually. Consider a one-on-one basketball game. One player is going to win and the other is going to lose. The losing player may, nonetheless, be very good. He may be good enough to be a pro and even to get into the All-Star game. But he isn't as good as the other player. In this artificial environment the minor code change strain eventually goes away, but that doesn't mean it won't survive in natural environments.
Bottom line is this. The Figure 1 results do not indicate that all the codon reassignments are less fit. The Figure 2 results show that the code change is stable. The fact that the CH marker went away in the control case is not easy to interpret for our purposes and does not prove that the code change variant could not survive, as other code variants must have.
Notes
1. Doring, V., et. al., "Enlarging the Amino Acid Set of Escherichia coli by Infiltration of the Valine Coding Pathway," Science, 292:501-504, 2001.
Edit: Deleted "which was never used" in Figure 2 discussion and minor editing.
[ 11. September 2004, 19:12: Message edited by: Cornelius G. Hunter ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 10. September 2004 14:05
I honestly now think you do not understand this paper - there are many mistakes in your latest post. You may want to revise it. I will have some time to answer tonight, either way.
IP: Logged
|
|
Cornelius G. Hunter
Member
Member # 81
|
posted 10. September 2004 15:54
Charlie:
Well I wouldn't be terribly shocked if I have misinterpreted the paper. I look forward to your comments on that. I do see that I made an unintended mistake in discussing Figure 2, regarding the expression of the DNA code variant gene (I said it wasn't expressed which is an obvious mistake). It is obvious that this was unintended from the subsequent discussion, and this does not change my conclusions.
IP: Logged
|
|
|
Printer-friendly view of this topic