Expansion And Modification of The Genetic Code

The genetic code of all living organisms is near-universally conserved and was long concidered to be restricted to the 20 naturally occurring amino acids. In 1986 two workgroups independently discovered that the nonstandard amino acid selenocysteine (Sec) is directly incorporated into proteins in response to in-frame opal stop codons (UGA), instead of being created by posttranslational modification^[29,30]. Afterward, this system was regarded as an expansion of the genetic code and selenocysteine was titled the 21^st amino acid^[31,32]. Sixteen years later pyrrolysine (Pyl) was found to be the 22^nd genetically encoded amino acid, this time in response to the amber stop codon (UAG)^[33–35]. Whereas Sec is present in prokaryotes and eukaryotes^[36], the distribution of Pyl appears limited to the Methanosarcinacea and Gram-positive Desulfitobacterium hafniense^[37]. Furthermore, both amino acids differ in their aminoacylation mechanism. Sec is made via an enzymatically modified serine that was charged to a special selenocysteinyl-tRNA. In contrast, Pyl is directly paired to pyrrolysyl-tRNA (PylT) by the cognate aminoacyl-tRNA synthetase PylS^[37–

39].

In addition to Sec and Pyl, even more deviations have been found to the standard genetic code. Genome analyses revealed ten codon reassignments in prokaryotic and eukaryotic nuclear codes which all are a subset of 16 changes occurring in mitochondrial codes^[40]. Moreover, some methanogenic archaea compensate the lack of a canonical cysteinyl-tRNA synthetase by a particular pathway using O-phosphoserine that is enzymatically converted to cysteine prior to incorporation into a nascent protein^[40]. All these modifications show a certain flexibility of the genetic code towards evolutionary novelties, giving the potential for additional genetically encoded nonstandard amino acids that might exist in still-uncharacterized genomes^[40]. However, the search for the 23^rd amino acid has not yet been successful, making the appearance of further widely spread amino acids improbable^[41]. At the end of the last century, scientists began to exploit the degeneracy of the genetic code, in order to artificially expand it for the genetically encoded incorporation of amino acids with new functionalities into proteins. These “unnatural” amino acids (UAAs) bear many different functional groups, such as posttranslational modifications, UV-inducible crosslinkers, spectroscopic probes and chemical handles that can be modified chemically, even in living cells^[42]. This required the introduction of exogenous tRNAs and their cognate aaRSs into the host cell, which have to work completely orthogonal to the endogenous components. That means, the introduced tRNAs should not be charged with any canonical amino acid by the host’s aaRSs and, in turn, the orthogonal aaRS should not aminoacylate

any endogenous tRNA with UAAs. The anticodon of the orthogonal tRNA is typically complementary to blank (nonsense, frameshift, or otherwise unused) codons, especially the rarely used amber stop codon, allowing the reassignment of the appropriate codon to the amino acid used as a substrate by the orthogonal aaRS. The feasibility of this method was proven in 1998 by Furter who was able to site-specifically incorporate p-fluoro-phenylalanine (p-F-Phe) into dihydrofolate reductase (DHFR) in E. coli expressing a yeast amber suppressor tRNA/phenylalanyl-tRNA synthetase (PheRS) pair^[43].

Figure 2.1: Schematic view for genetic code expansion.

Desired unnatural amino acids (UAAs) are taken up by the cell using endogenous transporters. These UAAs are used by evolved aminoacyl-tRNA synthetases (aaRSs) to charge corresponding evolved tRNAs. The tRNAs are then used by the ribosomes to decode (mostly) amber codons introduced in the mRNA, to incorporate the UAAs at predetermined sites on the protein of interest.

Whereas the yeast PheRS accepted the substrate analogue p-F-Phe without further modifications, advances in this system necessitated the adaption of aaRS’ specificities towards specific UAAs^[44]. Furthermore, the original standard amino acid phenylalanine was still a substrate for the PheRS, resulting in non-homogenously labeled protein. This

disadvantage was overcome by a large excess (up to 30-fold) of p-F-Phe supplementation in the growth medium^[43]. The first tRNA/aaRS pair that was truly orthogonal and only recognizing the desired UAA was derived from the tyrosyl pair from Methanococcus jannaschii (M. jannaschii) in the workgroup of Peter Schultz. This pair was evolved to incorporate O-methyl-L-tyrosine into DHFR in response to an amber codon^[45] (Figure 2.1).

To achieve this, they developed a systematic approach to alter the specificity of a synthetase for a certain UAA. First, all active-site residues interacting with the actual tyrosine substrate were randomly mutated, yielding a large library of synthetase variants which were passed through multiple rounds of stringent positive and negative selection.

The positive selection was based on a reporter plasmid containing an antibiotic resistance gene with amber mutations at permissive sites, in this case a chloramphenicol acetyltransferase (CAT). E. coli cells simultaneously transformed with this reporter and the aaRS library plasmids were only able to survive in media containing chloramphenicol and the UAA, if they harbored a functional synthetase variant recognizing either the UAA or a natural amino acid (also see Ch. 3.2.4.1). The subsequent negative selection eliminated undesired aaRSs suppressing amber codons with endogenous amino acids in the absence of the UAA. Therefore, active synthetase clones from the first round were combined with a reporter plasmid comprising an amber mutant of the toxic barnase gene. Clones that produce full-length barnase protein using canonical amino acids will die because of its ribonuclease activity, which is toxic to cells without its specific inhibitor barstar^[46] (also see Ch. 3.2.4.3). Multiple rounds of these two selections were performed, leading to an orthogonal, highly specific and amber suppressing aaRS variant^[44] (Figure 2.2).

Figure 2.2: General selection scheme for the evolution of aminoacyl-tRNA synthetase (aaRS) substrate specificities for unnatural amino acids (UAAs)^[44].

First, non-functional aaRS library variants are removed in a positive selection in the presence of the UAA and chloramphenicol (Cm). Functional variants suppress the amber codon within the chloramphenicol acetyltransferase (CAT) gene with both natural and unnatural amino acids. Synthetases specific for the UAA are isolated in a negative selection in the absence of the UAA, while suppression of the amber codons in the barnase gene leads to cell death.

The success in E. coli led to the development of a very similar selection approach in yeast, driven by amber suppression in the transcriptional activator GAL4. The production of full-length GAL4 in the presence of the UAA activates the expression of GAL4-responsive HIS3, URA3, and lacZ reporter genes, allowing for survival on media lacking histidine or uracil.

Negative selections to remove unspecific synthetases are based on the conversion of the protoxin 5-fluoroorotic acid (5-FOA) to its toxic product by the URA3 gene product on media without the UAA^[44,47].

The evolution of tRNA synthetases in cells of higher eukaryotes, like mammalians, is more difficult due to technical issues concerning transformation efficiency, slow doubling times, and growth conditions. In this direction, a shuttle approach was applied with tRNA/aaRS pairs, which are orthogonal both in E. coli or Saccharomyces cerevisiae (S. cerevisiae) and in mammalian cells. Pairs evolved for UAAs in the easier to handle bacterium or yeast can then be transferred into mammalian cells while keeping orthogonality^[48].

Using the aforementioned techniques a variety of orthogonal tRNA/aaRS pairs from various organisms have been used to add up to 100 unique unnatural amino acids to the genetic code of prokaryotes, including E. coli and some mycobacteria, and eukaryotes, like the yeasts S. cerevisiae and Pichia pastoris and even the multicellular organisms Caenorhabditis elegans and Mus musculus^[42,48]. However, the majority of all genetic code expansion approaches were performed with only four different tRNA/aaRS pairs, each suitable for a particular model organism. First, the already mentioned M. jannaschii tyrosyl pair (MjYRS/MjYT) is orthogonal only in E. coli and other bacteria. The second and third are two synthetases from E. coli decoding for tyrosine (EcTyrRS) and leucine (EcLeuRS) in combination with their cognate tRNAs and can be utilized only in yeast, mammalian and other eukaryotic cells. Lastly, the aforesaid PylS/PylT pair from Methanosarcina species provides the advantage of being orthogonal in both bacteria and eukaryotic cells, showing no cross-reactions with endogenous synthetases or tRNAs. Additionally, the natural aaRS substrate specificity must not be destroyed before the evolution for a new UAA, since it decodes pyrrolysine and none of the 20 canonical amino acids^[49].