Reply To Kenneth Miller On The Genetic Code
Contact: Mark Edwards 206.292.0401 x107 / email@example.com
On Tuesday, September 25, 2001, Professor Kenneth Miller of Brown University issued a press release entitled "A Dying Theory Fails Again," available here:
In this document, Miller claims that the Discovery Institute (DI) tried to "smear" PBSs Evolution series when the DI charged that program with making a false statement about the universality of the genetic code. Miller also claims that the DI failed to tell the public that "the very discoveries they cite provide elegant and unexpected support for Darwins theories."
These claims are false. Millers press release, however, provides an excellent teaching opportunity for the DI, not only to show why Millers claims are false, but also to amplify on our original objection. We shall explain why statements such as "the genetic code is universal" not only harm science by creating what Charles Darwin called "false facts" but also cheat the public, by concealing the real puzzles facing evolutionary theory. We conclude by touching on some of the deeper issues raised by patterns of evidence such as the genetic code.
We begin with the errors and misrepresentations in Millers press release.
Lets consider each problem in more detail:
1. Miller completely misrepresents Knight et al.s composite phylogeny of genetic codes.
In his press release, Miller writes:
Look closely at the figure from this paper, and youll see something remarkable. The variations from the standard code occur in regular patterns that can be traced directly back to the standard code, which sits at the center of the diagram.
This is false. The variant codes do not "occur in regular patterns," but appear independently in unrelated lineages. Knight et al. explain this pattern of convergent (i.e., non-homologous) appearance in the article itself:
The genetic code varies in a wide range of organisms (FIG. 2 [reproduced in Millers press release], some of which share no obvious similarities. Sometimes the same change recurs in different lineages: for instance, the UAA and UAG codons have been reassigned from Stop to Gln in some diplomonads, in several species of ciliates and in the green alga Acetabularia acetabulum (reviewed in Ref. 5). Similarly, animal and yeast mitochondria have independently reassigned AUA from Ile to Met.
In their caption to Figure 2, Knight et al. note explicitly that variant codes have arisen "repeatedly and independently in different taxa." This pattern of convergent variation has generated much discussion in the primary literature. If these are indeed convergent changes, they do not provide evidence of common descent at all, but rather would be misleading similarities that, taken by themselves, generate a false history of the organisms in question.
In short, Miller completely misrepresents the Knight et al. composite phylogeny. There is no "regular pattern" to the variant codes that maps congruently onto phylogenetic trees from other data. Thus, far from providing what Miller calls "unexpected confirmation of the evolution of the code from a single common ancestor," the pattern of variant codes represents a puzzle for a single tree of life.
2. Variant genetic codes are not analogous to the differences between dialects of the same language.
In his press release, Miller writes:
As evolutionary biologists were quick to realize, slight differences in the genetic code are similar to differences between the dialects of a single spoken language. The differences in spelling and word meanings between the American, Canadian, and British dialects of English reflect a common origin. Exactly the same is true for the universal language of DNA.
This is at best a wildly inaccurate analogy. From context and other clues, English speakers can discern that the words "center" and "centre," or "color" and "colour," refer to the same object. Meaning is preserved by context, and the reader moves along without a hitch.
But a gene sequence from a ciliated protozoan such as Tetrahymena (for instance), with the codons UAA and UAG in its open reading frame (ORF), cannot be interpreted correctly by the translation machinery of other eukaryotes having the so-called "universal" code. In Tetrahymena, UAA and UAG code for glutamine. In the universal code, these are stop codons. Thus the translation machinery of most other eukaryotes, when reading the Tetrahymena gene, would stop at UAA or UAG. Instead of inserting glutamine into the growing polypeptide chain, and continuing to translate the mRNA, release factors would bind to the codons, and the ribosomes would halt protein synthesis. The resulting protein would be truncated in length, and very possibly non-functional. Unlike variant spellings of "center," therefore, context cannot preserve meaning. With the codons UAA and UAG, no shared context exists.
Knight et al. present a much better analogy for code changes:
Any change in the genetic code alters the meaning of a codon, which, analogous to reassigning a key on a keyboard, would introduce errors into every translated message.
Indeed, for two decades (see below), it was exactly this deeply-embedded feature of the genetic code that led to strong predictions about its necessary universality across all organisms. It was widely thought that any change to the genetic code of an organism would affect all the proteins produced by that organism, leading to deleterious consequences (e.g., truncated or misfolded proteins) or lethality. Once the code evolved in the progenitor of all life, it "froze," and all subsequent organisms would carry that code.
In any case, the differences between genetic codes are not properly analogous to minor differences among dialects of a single language.
3. Millers references to biotechnology do not accurately represent the experimental literature on variant genetic codes.
In his press release, Miller writes:
...the entire biotechnology industry is built upon the universality of the genetic code. Genetically-modified organisms are routinely created in the lab by swapping genes between bacteria, plants, animals, and viruses. If the coded instructions in those genes were truly as different as the critics of evolution would have you believe, none of these manipulations would work.
But some manipulations namely, those involving organisms with variant codes do not work, unless the researchers themselves intervene to ensure function.
Consider, for instance, the release factor from the ciliate Tetrahymena thermophila. Release factors (in eukaryotes, these proteins are abbreviated as "eRF" to distinguish them from prokaryotic release factors) catalyze the separation of completed polypeptide chains (nascent proteins) from the ribosomal machinery. Unlike other eukaryotic release factors, however, that recognize all three stop codons (UAA, UGA, and UAG), the Tetrahymena thermophila release factor recognizes only the UGA codon as "stop."
In 1999, Andrew Karamyshev and colleagues at the University of Tokyo isolated the release factor (Tt-eRF1) from Tetrahymena thermophila. But in order to express and purify the protein, Karamyshev et al. had to manipulate it genetically first. Why? The Tetrahymena thermophila gene for Tt-eRF1 contains 10 codons in its open reading frame that would be interpreted as "stop" by other organisms whereas Tetrahymena thermophila reads these codons as glutamine:
To express and purify the recombinant Tt-eRF1 protein under heterologous expression conditions [i.e., in a cell other than Tetrahymena Karamyshev et al. used yeast cells], 10 UAA/UAG triplets within the coding sequence were changed to the glutamine codon CAA or CAG by site-directed mutagenesis.
Furthermore, Tt-eRF1 would not function when employed in combination with ribosomes (translation machinery) from other species:
In spite of the overall conservative protein structure of Tt-eRF1 compared with mammalian and yeast eRF1s, the soluble recombinant Tt-eRF1 did not show any polypeptide release activity in vitro using rat or Artemia ribosomes.
Thus, when using an organism with a variant code (Tetrahymena thermophila), researchers found that
Experiments to change the identity of transfer RNA (tRNA) another possible mechanism by which genetic codes might reassign codon "meanings" have shown that the intermediate steps must be bridged by intelligent (directed) manipulation. In one such experiment, for instance, Margaret Saks, John Abelson, and colleagues at Caltech changed an E. coli arginine tRNA to specify a different amino acid, threonine. They accomplished this however only by supplying the bacterial cells (via a plasmid) with another copy of the wild-type threonine tRNA gene. This intelligently-directed intervention bridged the critical transition stage during which the arginine tRNA was being modified by mutations to specify threonine. Indeed, in reporting on an earlier experiment to modify tRNA, Abelson and colleagues noted that "if multiple changes are required to alter the specificity of a tRNA, they cannot be selected but they can be constructed" constructed, that is, by intelligent design.
We stress here that, in contrast to Millers blithe dismissal of the difficulties raised for biotechnology by variant genetic codes, experts in the field caution that assuming a "universal" code may lead to serious problems. In a recent article on the topic entitled "Codon reassignment and the evolving genetic code: problems and pitfalls in post-genome analysis," Justin OSullivan and colleagues at the University of Kent observe:
The emerging non-universal nature of the genetic code, coupled with the fact that few genetic codes have been experimentally confirmed, has several serious implications for the post-genome era. The production of biologically active recombinant molecules requires that careful consideration be given to both the expression system and the original host genome. The substitution of amino acids within a protein encoded by a nonstandard genetic code could alter the structure, function or antibody recognition of the final product.
Thus, Millers statements on biotechnology are highly misleading. Variant codes are not a minor matter easily overcome in experiments using different organisms.
We conclude by considering some of the deeper issues raised by Millers press releas
A little history and some basic logic
Not so very long ago, the universality of the genetic code was widely regarded as an important prediction (or confirmation) of the theory of common descent. Consider, for instance, an evolutionary biology textbook by the zoologist Mark Ridley, entitled The Problems of Evolution (Oxford University Press, 1985). In his first chapter, "Is Evolution True?" Ridley argues that common descent predicts a universal genetic code. His formulation of this argument mirrors dozens of similar arguments present in the biological literature from the mid-1960s to the mid-1980s:
The outstanding example of a universal homology is the genetic code....The universality of the code is easy to understand if every species is descended from a common ancestor. Whatever code was used by the common ancestor would, through evolution, be retained. It would be retained because any change in it would be disastrous. A single change would cause all the proteins of the body, perfected over millions of years, to be built wrongly; no such body could live. It would be like trying to communicate, but having swapped letters around in words; if you change every a for an x, for example, and tried talking to people, they would not make much sense of it. Thus we expect the genetic code to be universal if all species have descended from a common ancestor.
Shortly after Ridleys argument was published in The Problems of Evolution, the evolutionary biologist Brian Charlesworth reviewed the book. He cautioned that Ridley was "less sound on the more modern aspects" of evolution, including the genetic code. Ridleys genetic code argument, Charlesworth worried,
provides an opening for the creationists by asserting that the genetic code is universal, whereas it is now known that slight deviations from the standard code occur in mitochondria and in Mycoplasma.
But how did Ridley create "an opening for the creationists," if the genetic code variants are as insignificant as Kenneth Miller suggests?
Here we should consider a basic feature of the logic of scientific prediction. If a theory, T, strongly predicts a particular outcome, O, but O is not observed, then one has grounds for doubting T. Of course, this logical schema greatly oversimplifies how scientists may actually behave when met with a failed prediction. One can shift or broaden the prediction "T didnt really predict O, but actually O plus something else" or one can throw doubt onto some theory other than T, and blame it, rather than T, for the failed prediction.
The problem is that both of these solutions weaken ones case for the theory T. Any theory that predicts an observational outcome and its negation is a theory without much empirical power. "It will rain today and it wont rain today" tells one everything and therefore nothing. If common descent predicts that the genetic code will be universal, except when it is not universal, then common descent does not actually specify any observations about the code.
One might also say that some other theory, linked conceptually to common descent, is responsible for the failed prediction of universality. In this move, the truth of common descent is preserved while another part of our biological knowledge pays the cost. Most biologists working on the evolution of the code have taken this route; Niles Lehman of SUNY-Albany, for instance, writes:
Once thought universal, the specific relationships between amino acids and codons that are collectively known as the genetic code are now proving to be variable in many taxa. While this realization has been disappointing to some the genetic code was often hailed as the ultimate evolutionary anchor in that is universality was perhaps the indisputable piece of evidence that all life shared a common ancestor at some point it has also opened up a rich field of evolutionary analysis by forcing us to consider what sequence of molecular events in a cell could possibly allow for codon reassignment.
Again, however, this move weakens the case for common descent. One preserves the truth of common descent only by cashing in one of the theorys predictions, namely, the universality of the code. "It seems we were wrong, after all, about the genetic code not being able to vary. So lets figure out how variant codes arise."
Well, how do variant code arise? Kenneth Miller doesnt say, but that is not surprising. No one really knows, although that is not for a lack of theories. Here we refer the curious reader to the superb review article by Knight, Freeland, and Landweber (2001), who list several different theories explaining codon change, none of which (they note) is unequivocally supported by the evidence.
Is it possible that the variant codes derived from a single common ancestor? Yes.
It is also possible, of course, that they did not. Miller assumes that a single origin is the case, but there is a world of difference between assumptions and real knowledge.
These are matters for legitimate debate. What is not a matter for debate are the following facts:
1. Robin D. Knight, Stephen J. Freeland, and Laura F. Landweber, "Rewiring the Keyboard: Evolvability of the Genetic Code," Nature Reviews GeneticsII (2001): 49-58. 2 (2001):49-58; p. 49
2. Catherine A. Lozupone, Robin D. Knight and Laura F. Landweber, "The molecular basis of nuclear genetic code change in ciliates," Current Biology 11 (2001):65-74; Patrick J. Keeling and W. Ford Doolittle, "Widespread and Ancient Distribution of a Noncanonical Genetic Code in Diplomonads, Molecular Biology and Evolution 14 (1997):895-901; A. Baroin-Tourancheau, N. Tsao, L.A. Klobutcher, R.E. Pearlman, and A. Adoutte, "Genetic code deviations in the ciliates: evidence for multiple and independent events," EMBO Journal 14 (995):3262-3267.
3. Robin D. Knight, Stephen J. Freeland, and Laura F. Landweber, "Rewiring the Keyboard: Evolvability of the Genetic Code," Nature Reviews Genetics 2 (2001):49-58; p. 49.
4. Andrew L. Karamyshev, Koichi Ito, and Yoshikazu Nakamura, "Polypepetide release factor eRF1 from Tetrahymena themophila: cDNA cloning, purification and complex formation with yeast eRF3," FEBS Letters 457 (1999):483-488; p. 485.
5. Ibid., p. 487.
6. Margaret E. Saks, Jeffrey R. Sampson, and John Abelson, "Evolution of a Transfer RNA Gene Through a Point Mutation in the Anticodon," Science 279 (13 March 1998):1665-1670.
7. Jennifer Normanly, Richard C. Ogden, Suzanna J. Horvath & John Abelson, "Changing the identity of a transfer RNA," Nature 321 (15 May 986):213-219.
8. Justin M. O’Sullivan, J. Bernard Davenport and Mick F. Tuite, "Codon reassignment and the evolving genetic code: problems and pitfalls in post-genome analysis," Trends in Genetics 17 (2001):20-22; p. 21.
9. Mark Ridley, The Problems of Evolution (Oxford: Oxford University Press, 1985), pp. 10-11.
10. Brian Charlesworth, "Darwinism is alive and well," review of The Problems of Evolution, New Scientist 11 July 1985 pg 58 11 July 1985, p. 58.
11. Niles Lehman, "Please release me, genetic code," Current Biology 11 (2001):R63-R66; p. R63.