To what extent is the genetic code more than just a code?

Question

It is worth specifying the exact meaning of "code" in this question. A code is a map from one space to another space with which it has no algorithmic connection. Thus representing 321 as 0x141 is not an instance of a code, because there is an algorithm; but representing 321 as the letter Ł is a code (Unicode, in this case) because there is no intrinsic connection between the number and the symbol. The only way of knowing which number matches which symbol is to look it up in a book.

Popular presentations of the way cells work speak of the DNA/RNA code as a code. There is no algorithmic transcription of RNA to proteins. The transcription from codons to amino acids is a code which human beings interpret by looking it up in a table. Cellular machinery interprets it by the fact that GUN→Val transfer RNA is present in the cell but a GUN→Glu transfer RNA is not. During protein synthesis, each transfer RNA molecule, having acquired a molecule of its designated amino acid, pairs with the codon it is designed to pair with. The code table is thus defined by the set of tRNAs that happen to be floating around in the cell.

(There may well have been some over-simplifications in the above picture).

On the other hand, the DNA code is not a code in the sense that the DNA molecule (and still more RNA) is not a passive repository of code sequences but a reactive entity, and those reactions depend on the nucleotide sequence in a way different from the tRNA-mediated encoding. Coding regions; non-coding regions; long strings of single nucleotides; initiation sites; 3-D structure of the molecule – all these depend on the nucleotide sequences in a way separate from their role as codons.

This is why this is a "to what extent?" question.

A thought experiment

To the extent that the genetic code is a code, an experiment of this kind ought to be possible, and perhaps even not far from practicability. To the extent that the code isn't a code, it wouldn't work.

1. Elucidate the structure of GAU-to-Glu tRNA and GAA-to-Asp tRNA, neither of which exist in nature.
2. Synthesise RNA that codes for them and insert it into a suitable cell, making it a machine for synthesizing these two tRNAs. Manufacture a large quantity of these tRNAs.
3. Taking the DNA of a target cell (not necessarily the same species as before), replace the genes coding for GAU-to-Asp and GAA-to-Glu tRNA with genes coding for GAU-to-Glu tRNA and GAA-to-Asp tRNA. Such a cell would not live, since it would mistranscribe most of its proteins.
4. Before reinserting the DNA, make a copy of it, replacing every occurrence of GAT with GAA and every occurrence of GAA with GAT.
5. Just before reinserting the DNA, flood the cell with the new tRNAs manufactured in step 2. This changes the code.
6. Reinsert the DNA.

The cell should now be viable. Everywhere that GAT used to code for Asp in the original cell, GAA is now found; but since the only tRNA available to the synthesis process is GAA-to-Asp, this means that Asp is still inserted exactly where it should be inserted. The change in DNA sequence is exactly compensated for by the change in available tRNAs - that is, the change in the code.

Purpose of the thought experiment

On the one hand, postulate success. The modified cell, using a different genetic code from anything else on the planet, has the advantage of being immune to all viruses. Or will being immune to viruses turn out not to be an advantage after all? The modified cell will also be very much inbred, since it cannot exchange genetic material with anything else. Advantage, or disadvantage?

On the other hand, consider the possibility of failure. Failure would appear to be because GAT and GAA (or GAU and GAA) are subtly different in their chemical properties quite apart from their role as codons. DNA and RNA might adopt different conformations, ones sufficiently different to change the nature and kinetics of their reactions as macromolecules as opposed to information repositories.

Which brings us exactly to the title of this question. If the genetic code is a pure code, then the experiment will work - just as I could make a variant Intel x86 chip in which the code for Add does subtraction and the code for Subtract does addition, and compensate for it by modifying all my programs accordingly. To the extent that the code is more than a code, the experiment will work less well.

But to what extent?

Answer 1

You call it a thought experiment but something like this has actually been done. Not entirely similar as they don't switch 2, but still they replace a codon.

An overview: https://en.wikipedia.org/wiki/Expanded_genetic_code

Big thing: in the two articles leading up to this one they replaced all 314 UAG stop codons in E.coli K12 and used the now unused UAG codon for some fancy stuff: https://www.ncbi.nlm.nih.gov/pubmed/25607366

Answer 2

Hasn’t your question already been answered by those organisms (and organelles) that have a different genetic code from the standard genetic code (originally known as ‘universal’)? Essentially they have performed the experiment for you by developing machinery to decode mRNA differently (transfer RNAs with appropriately different anticodon/amino-acid accepting ability — that’s what determine how the code is interpreted).

And I think DNA is a red herring here. The genetic code is a code to decipher information in regions of mRNA that specify proteins. It does not — and was never intended to — apply to anything else. Thus, it does not apply to the 5' and 3' untranslated regions of mRNAs, where there may be other combinations of nucleotides that are interpreted differently (e.g. ribosome-binding sequences in prokaryotes, polyA-addition signals in eukaryotes). Use of a modified (T for U) genetic code is only used with DNA to make predictions about regions that specify the protein-encoding parts of mRNAs (i.e. whether they can, and if so, what the amino acid sequence of the protein will be).

[Apologies if you were aware of all this molecular biological background and I have misunderstood your argument.]

Answer 3

On one hand, designing an experiment which would kill (and resurrect) a cell is not possible: once a cell malfunctions, it's likely damaged beyond repair. However, other than that I don't think this experiment kind of cannot actually be done. You would just have to simultaneously expose different cultures (grown in the same conditions) to different inserted DNA or RNA. This would include exposures you would predict cause both cell death and survival or growth and "control" exposures you know would kill or spare cells. This is done routinely with bacteria, yeast, and mammalian cancer cells.

Depending on your interpretation of a "true code" this would be testable. Clearly a DNA ATG maps to AUG in a particularly well defined way. mRNA AUG is also decoded to Met but the mechanism here is important: the tRNA is a "decoding" class of molecules, these have a complementary recognition site for an mRNA codon, e.g, UAC, and a binding site for an amino acid such as Met on the other end. These decoding molecules are otherwise identical in structures and interchangeable, indeed there are some bacterial species that use "non-canonical" decoding tRNAs different to other species and "genetic re-coding" where additional amino acids or frameshifts are encoded by our STOP codons.

I don't know if tRNA here fits your analogy as an algorithm or compiler but evolution has produced a vast range of exceptions we could test the genetic "code" on. Basically we work with a table that we're familiar with and some bacteria use a different one with additional amino acids like selenocysteine. Recent experiments have even added DNA/RNA bases not occurring in nature and "non-canonical" amino acids not used in naturally occurring cells, others are attempting synthetic biology to engineer their own cells as you've proposed.

However, it's worth noting gene expression isn't a simple one-to-one mapping. The amount and variant of gene expression is also important, many regulatory sequences affect expression of DNA, splicing of mRNA, or ribosomal binding. Once a protein is produced it is often cleaved, phosphorylated, or glycosylated and can form complexes with identical or different proteins. Whether the "exon" DNA encodes a "raw" protein sequence is intriguing and as much as I would hypothesise that DNA to protein is a true code, in the context of a complex biological system other factors beyond the raw sequence affect gene function.