Your starting assumption is wrong, aaRS are much bigger than 18-20 aa. See for example the human aaRS for Ala on https://www.uniprot.org/uniprot/P49588.

Engelbert Buxbaum my question is not about the overall number of amino acids in an aaRSs but about the number of different amino acids. With a script I went through all aaRSs that are listed on the following page: https://www.uniprot.org/docs/aatrnasy and counted the number of different amino acids.

About the sentence you wrote: "The more complex an enzyme the more substrates can be recognized?"

I don't think this can be true. My conception is that enzymes, in the beginning, would be much more non-specific to diferente class of substrates, and them, as complexity increases, a preference for specificity should be acquired.

The aaRS is a good example, there are 20 to recognize each of the 20 a.a. In earlier times, I believe that some aaRS begun to recognize 2 or more.

May you can find more of this discussion in this paper:

Enzyme evolution: innovation is easy, optimization is complicated

Article Enzyme evolution: innovation is easy, optimization is complicated

Do a 3D superposition of the two similar enzymes and determine where they differ in aa (corresponding) positions. This would be similar to doing a 3D superposition of the same enzyme but from different species, e.g., human vs. mouse enzyme. They are not 100% identical and have similar but not identical specificities, kinetics, etc.

BTW there are only 20 amino acids. The fact that aaRS #1 contains only 18 or 19 of the 20, while aaRS #2 contains all 20 amino acids is, I believe, immaterial due to the semi-degeneracy of amino acids (e.g., leucine and isoleucine).

Vicente Mendoza Reyes thanks for your answer.

I chose randomly two aaRS from the same organism, one with 19 and one with 20 amino acids. The one with 19 has a sequence length of 511 and the other one has a sequence length of 822. I speculate that using less different amino acids leads to shorter sequence lengths and shorter sequence lengths mean less specific aaRSs. Current and older research assumes that evolution didn't start with the full set of 20 amino acids and that the first aaRSs have been less specific than todays aaRSs. I think less amino acids and shorter sequences make the aaRSs more robust against changes like mutations or when an amino acid no longer gets used in the translation process. So it is likely that aaRSs with less different amino acids appeared earlier in evolution and took a greater part in genetic code formation than the other aaRSs.

If any of these assumptions are wrong or there is interesting research about this, please point it out.

Felipe Akihiro Otsuka thanks for your answer and the link to the paper :) .

"as complexity increases"

Do you measure the complexity by number of different amino acids used in the amino acid sequence and sequence length? Or does "complexity" include more (or other) factors for you?

I think that there has been high ambiguity in early aaRS evolution as well.

Todays aaRSs still recognize two (I'm not sure if more) amino acids. In many cases it is considered as failure but there are also cases where an ambiguous translation has a positive effect on cell fitness. I read that loading tRNA more often with the "wrong" amino acid can be a regulatory reaction on stress. And @Przemyslaw Plonka introduced me to the following really interesting paper: http://emboj.embopress.org/content/16/5/1122 .

It would be interesting to see a code table that includes also the "wrong" translations with their frequency of appearance but I didn't find one yet.

You can also look at the size of the respective tRNAs that they recognize. Perhaps the larger aaRS recognizes a larger tRNA than does the smaller one. In other words, look at both the enzyme and its substrate because they might have coevolved.

I have the impression that you are looking at this problem from a too reductionistic perspective, check for instance this reference:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4938167/

Now there are rules, but no simple ones like the one you refer in your question.

Larger proteins have more cavities and may interact with more ligands, yet the true question is :

How many of these possible interactions reach the specificity required to do some defined physiological role? - that is assuming this putative ligand can be found in the same biological medium where the protein is found.

Many proteins do moon-lightening, meaning they perform different physiological roles under different cellular contexts, and larger proteins could be expected to be more likely to do so, but size is also related to stability and intracellular proteins tend to be larger than extracellular proteins, particularly if these are in an unregulated fluid, like the soil, or a natural water body.

Proteins can also be very large and have a repetitive structure, like silk fibroin, Titin, and collagen, but even then, some proteins of this kind have notorious versatility, and a large number of biological partners, such as tubulin.

So in order to define rules, you need first to circumscribe a biological context.

Best wishes

Rogelio