The ability to genetically introduce an unnatural (i.e., non-canonical) amino acid with unique chemical and physical properties into a defined position of a target protein has provided a new avenue to investigate protein function (Brown et al. 2018; Chin 2017; Nodling et al. 2019; Young and Schultz 2018). This approach known as genetic code expansion relies on an orthogonal aminoacyl-tRNA synthetase/tRNA pair to direct the site-specific incorporation of an unnatural amino acid in response to a blank codon. The amber stop codon (UAG) is usually chosen as the blank codon because it does not encode a canonical amino acid and is the least used codon in most organisms (Athey et al. 2017). Here, it is critical that the orthogonal synthetase only acylates the orthogonal tRNA with the designated unnatural amino acid, and neither the orthogonal tRNA nor the unnatural amino acid is a substrate of endogenous synthetases. The pyrrolysyl-tRNA synthetase (PylRS) from archaea Methanosarcina barkeri (MbPylRS) or Methanosarcina mazei (MmPylRS) and its cognate tRNA (Pyl tRNA) are arguably the most widely used orthogonal pair for genetic code expansion in bacteria and eukaryotes (Brown et al. 2018; Chin 2017; Nodling et al. 2019). The two organisms have identical sequences for Pyl tRNA (Srinivasan et al. 2002; Suzuki et al. 2017) that naturally decodes the amber codon. MbPylRS and MmPylRS also have high sequence identity (74%). While wild-type MbPylRS and MmPylRS can recognize some unnatural amino acids, protein engineering of these homologous enzymes has enabled incorporation of over 100 different unnatural amino acids with diverse chemical and physical properties (Brown et al. 2018; Chin 2017; Nodling et al. 2019).
Despite these advances, the use of the amber codon to encode an unnatural amino acid is not without limitation. The amber codon can still be recognized by the release factor, causing translation termination and production of the truncated protein product. To circumvent this problem, different approaches have been explored. In one instance, an orthogonal ribosome has been engineered that has much lower affinity to the release factor (An and Chin 2009; Barrett and Chin 2010; Neumann et al. 2010). In fact, the orthogonal ribosome recognizes an alternative Kozak sequence, the amber codon on the reporter gene can be preferentially decoded as the unnatural amino acid, minimizing unwanted premature termination. In other approaches, the elongation factor can be engineered or the release factor can be removed to increase the amber suppression efficiency (Gan et al. 2017; Johnson et al. 2011; Schmied et al. 2014). Nevertheless, the general applicability of these approaches (An and Chin 2009; Barrett and Chin 2010; Gan et al. 2017; Johnson et al. 2011; Neumann et al. 2010; Schmied et al. 2014) is limited because the translational systems in prokaryotic cells and eukaryotic cells are vastly different and not readily interchangeable. Hence, an alternative approach that would function in both E. coli and mammalian cells is highly desirable, and this may be achieved by directly engineering the non-substrate binding part of the orthogonal synthetase to improve its catalytic activity (Owens et al. 2017; Sharma et al. 2018).
To improve unnatural amino acid incorporation, both MbPylRS and MmPylRS need to be thoroughly investigated. These homologs contain two domains, the C-terminal and N-terminal domains. The C-terminal domain possesses the catalytic site and binds to ATP and the unnatural amino acid. On the other hand, the N-terminal domain is important for the enzymatic activity in cells through interaction with the tRNA though is not directly involved in unnatural amino acid recognition (Suzuki et al. 2017). Often discovered through rational design, directed evolution or a combination of both, mutations in the C-terminal domain can expand the substrate scope, enabling incorporation of structurally diverse unnatural amino acids. On the other hand, though not directly involved in catalysis, mutations in the N-terminal domain have been demonstrated to affect the efficiency of unnatural amino acid incorporation in variants of MmPylRS, MbPylRS and their chimera (Bryson et al. 2017; Owens et al. 2017; Sharma et al. 2018). In a study to increase the incorporation of crotonyl lysine by a MbPylRS variant, six mutations (V8E, T13I, I36V, H45L, S121R, I355T) predominantly localized in the N-terminal domain were identified through random mutagenesis, and about threefold increase in crotonyl lysine incorporation was observed (Owens et al. 2017). When transferring these six mutations to another MbPylRS variant for incorporating Nε-[(2-propynyloxy)carbonyl]-L-lysine, a similar beneficial effect was observed. Another study found that R19H/H29R/T122S mutations in the N-terminal domain improved the capability of three MmPylRS variants for unnatural amino acid incorporation, and up to sixfold increase in the yields of recombinant proteins containing an unnatural amino acid was observed (Sharma et al. 2018). These studies indicate that mutations in the N-terminal domain can affect the incorporation efficiency. However, while C-terminal mutations are transferable between MbPylRS and MmPylRS to recognize specific unnatural amino acids (Brown et al. 2018; Chin 2017; Nodling et al. 2019), it remains elusive if the N-terminal mutations are also transferable between these two homologs.
Provided that MbPylRS and MmPylRS are often used interchangeably for unnatural amino acid incorporation (Brown et al. 2018; Chin 2017; Nodling et al. 2019), beneficial mutations found in the N-terminal domain of one homolog are likely to be transferable to that of the other. As the MmPylRS variant R19H/H29R/T122S was reported to have improved activity, we propose that these mutations may have beneficial effect for the incorporation of unnatural amino acids by MbPylRS. As the corresponding position of MmPylRS T122 in MbPylRS is already a serine residue (Fig. 1), whereas R19 and H29 are conserved in the two homologs. Thus, we set out to probe the effect of R19H/H29R mutations in unnatural amino acid incorporation by MbPylRS variants.
Here, we examined four MbPylRS variants (i.e., wild-type, AcKRS, PrKRS, PCCRS) for incorporating unnatural amino acids (Fig. 2) under two different temperatures (37 °C or 25 °C). Five different unnatural amino acids Nε-Boc-lysine (BocK), acetyl lysine (AcK), thioacetyl lysine (TAcK), propionyl lysine (PrK) and photocaged cysteine (PCC) were used here for our examination. BocK can be incorporated by wild-type MbPylRS and is often used as a model unnatural amino acid in proof-of-principle studies. Both AcK and PrK are commonly used to investigate the effect of lysine post-translational modifications (i.e., lysine acetylation and propionylation, respectively) in bacteria and eukaryotic cells (Drazic et al. 2016; Ju and He 2017; Lin et al. 2012). TAcK is a stable analog mimicking lysine acetylation, but the carbonyl functionality is replaced with a thiocarbonyl group thereby preventing enzymatic hydrolysis (Venkat et al. 2017; Xiong et al. 2016). Due to the structural similarity, AcK and PrK can be incorporated by both AcKRS and PrKRS, and TAcK is likely to be accepted by these PylRS variants as well. On the other hand, PCC has a distinct structure when compared to the lysine analogs. PCC has a photolabile protecting group on the cysteine side chain, useful in light-controlled protein activation of active cysteines (Nguyen et al. 2014).
Overall, we found that R19H/H29R mutations have negligible benefits to MbPylRS variants. In two cases (i.e., PrKRS and PCCRS), the N-terminal mutations greatly decreased the unnatural amino acid incorporation. Our results highlight the difference between MbPylRS and MmPylRS, two closely related homologs, and raise concerns of using these homologs interchangeably in genetic code expansion.