Evolving the N-terminal Domain of Pyrrolysyl-tRNA Synthetase for Improved Incorporation of Noncanonical Amino Acids

Abstract

By evolving the N-terminal domain of Methanosarcina mazei pyrrolysyl-tRNA synthetase (PylRS) that directly interacts with tRNA^Pyl, a mutant clone that displays improved amber suppression efficiency for the genetic incorporation of N^ε-(tert-butoxycarbonyl)-L-lysine three-fold more than the wild type was identified. The identified mutations are R19H/H29R/T122S. Direct transfer of these mutations to two other PylRS mutants that were previously evolved for the genetic incorporation of N^ε-acetyl- L-lysine and N^ε-(4-azidobenzoxycarbonyl)- L-δ,ε-dehydrolysine respectively also improved the incorporation efficiency of these two noncanonical amino acids. Since the three identified mutations are in the N-terminal domain of PylRS that is separated from its catalytic domain for charging tRNA^Pyl with a noncanonical amino acid, they can be potentially introduced to all other PylRS mutants for improving incorporation efficiency of their corresponding noncanonical amino acids. Therefore, it represents a general strategy to optimize the pyrrolysine incorporation system-based noncanonical amino acid mutagenesis.

TOC image

By evolving the N-terminal domain of pyrrolysyl-tRNA synthetase, mutations that improve the genetic incorporation of N^ε-(tert-butoxycarbonyl)-L-lysine at amber codon were identified. These mutations can be directly transferred to PylRS mutants for improve incorporation efficiency of their corresponding noncanonical amino acids.

Pyrrolsysine (Pyl) is the 22^nd proteinaceous amino acid that is genetically encoded by amber codon in a number of methanogenic archaea and some bacteria.^[1] The incorporation of Pyl at amber codon during translation in these organisms is procured by the cooridinated action of pyrrolysyl-tRNA synthetase (PylRS) and its cognate amber suppressing tRNA, tRNA^Pyl. PylRS directly aminoacylates tRNA^Pyl with Pyl and the afforded Pyl-tRNA^Pyl recognizes amber codon in a ribosome-bound mRNA for the delivery of Pyl into a nascent protein. Due to their unique structures and distinct recognition of each other,^{[2, 3]} PylRS and tRNA^Pyl have turned out to be orthogonal in naïve organisms such as Escherichia coli, Saccharomyces cerevisiae, Caenorhabditis elegans, and mammals and consequently form one of the most important genetic code expansion tools.^[4] Since the first use of the pyrrolysine incorporation system for the genetic incorporation of noncanonical amino acids (ncAAs) in E. coli, the past decade has witnessed explosive exploration of the genetic incorporation of more than 100 ncAAs.^[5] Applications of these ncAAs span many research areas from understanding basic biology to the production of therapeutic biologics.^{[6, 7]} The most popularly used PylRS-tRNA^Pyl pairs have two origins, Methanosarcina Barkeri and Methanosarcina mazei. The PylRS enzyme from both origins contains two domains (Figure 1A), the C-terminal catalytic domain (PylRS-c) that catalyzes the aminoacylation of tRNA^Pyl and the N-terminal domain (PylRS-n) that is not directly involved in the tRNA^Pyl aminoacylation but appears to interact directly with tRNA^Pyl for improved binding. Although the catalytic domain alone is able to aminoacylate tRNA^Pyl with a substrate ncAA, the N-terminal domain is indispensable for the enzyme’s in vivo activity.^[3] There is also a unique characteristic of PylRS-n. It easily aggregates, leading to a relatively low cytosolic concentration of PylRS in cells.^[8] This insoluble nature makes it not feasible to improve amber suppression efficiency for the genetic incorporation of ncAAs through PylRS overexpression. In order to improve amber suppression efficiency of the pyrrolysine incorporation system, an approach that can be potentially generalized for enhanced incorporation of ncAAs were developed previously. An evolved tRNA^Pyl was used to improve the Ef-Tu mediated delivery of ncAA-tRNA^Pyl to ribosome for increasing the ncAA incorporation efficiency.^[9] Different from this approach, the current work aims to address the same problem from a different angle.

Figure 1.

(A) Two domains of M. mazei PylRS. (B) putative M. mazei tRNA^Pyl nucleotides that potentially interact with two PylRS domains. (C) Our method for the identification of improved PylRS clones. 0.2 mM BocK is provided to the growth medium to drive the production of full-length sfGFP that is visibly fluorescent under blue light.

Krzycki et al. previously reported that a bacterial homolog of PylRS-n directly interacts with nucleotides in the anticodon loop, the anticodon stem, and the short variable arm of tRNA^Pyl (Figure 1B).^[8] Given these interaction sites are distal from the acceptor stem that directly participates in the aminoacylation with a ncAA, we reason that mutations introduced in PylRS-n that are also distal from the active site of PylRS for improved binding to tRNA^pyl potentially improve the aminoacylation of tRNA^Pyl with a ncAA and subsequently increase the ncAA incorporation efficiency. Since most engineered ncAA-specific PylRS mutants have their mutations in the PylRS-c domain, mutations in PylRS-n that can be procured via the engineering of wild-type PylRS to improve the incorporation of N^ε-(tert-butoxycarbonyl)-L-lysine (BocK), a nonnative substrate of wild-type PylRS, can be simply transferred to PylRS mutants for improving incorporation efficiency of their corresponding ncAAs. To test this idea, we employed a screening strategy as presented in Figure 1C. A plasmid pEVOL-PylRS derived from a pEVOL vector developed by Schultz et al.^[10] was constructed to contain genes coding tRNA^Pyl under the control of the proK promoter and M. mazei PylRS under the control of the pBAD promoter. The PylRS gene was codon-optimized for optimal expression in E. coli. Two restriction sites SpeI and BstXI were introduced around DNA coding sites for the N-terminal methionine and residue S130, respectively. Error-prone polymerase chain reaction (PCR) was then carried out to introduce randomized mutations to the PylRS-n coding region from M1 to S130 and the amplified DNA fragments were ligated back to the original plasmid between SpeI and BstXI to afford the first randomized pEVOL-PylRS plasmid library. This plasmid library was used to transform E. coli Top10 cells that also contained a selection plasmid pBAD-sfGFP134TAG. The pBAD-sfGFP134TAG plasmid contains a gene coding superfolder green fluorescent proteins (sfGFP) with an amber mutation at the 134th amino acid residue coding position and a C-terminal 6xHis tag. The transformed cells were then grown on LB-agar plates that was supplemented with 0.2 mM BocK and 0.2% arabinose. A previous test indicated that E. coli Top10 cells transformed with the original pEVOL-PylRS and pBAD-sfGFP134TAG plasmids displayed visible but low green fluorescence when grown in the presence of 0.2 mM BocK. This condition serves as a good starting point for screening. We expected that mutant PylRS clones more active than the wild type would lead to higher fluorescence of expressed sfGFP, therefore leading to more fluorescently visible colonies on plates. 25 clones that displayed enhanced fluorescence of their expressed sfGFP were chosen for sequencing. The R1-7 clone that displayed the highest fluorescence contains the mutation R19H. The pEVOL-R1-7 PylRS plasmid was then harvested. The N-terminal M1-S130 coding region of R1-7 PylRS was randomized again using the approach mentioned above to construct the second randomized pEVOL-PylRS plasmid. This library was subjected to the same screening as carried out for the first library. 25 clones that displayed highest fluorescence were sequenced. In comparison to the R1-7 clone, the R2-23 clone has a slightly better ability to incorporate BocK to generate full-length sfGFP. This clone harbors an additional H29R mutation. The pEVOL-R2-23 PylRS plasmid was extracted and subjected to the same kind of randomization at the N-terminal coding region of R2-23 PylRS and then screening. 25 best performing clones were selected and sequenced. An ever more improved clone R3-11 was identified. Besides mutations R19H/H29R, R3-11 PylRS contains an additional T122S mutation. It is worth mentioning that some selected clones contain additional mutations in the non-randomized region, a phenomenon that has happened a lot in the selection of PylRS mutants specific for ncAAs^[11].

The efficiency of all three mutant clones and wild type PylRS for the genetic incorporation of BocK was further tested and compared in parallel. E. coli Top10 cells containing both pBAD-sfGFP134TAG and one of the four pEVOL plasmids with either a wild type or mutant PylRS gene were grown in LB medium supplemented with 0.2 mM BocK and the expression of full-length sfGFP was induced by the addition of 0.2% arabinose. The expressed sfGFP was then affinity-purified using the Ni-NTA resin and analyzed by SDS-PAGE. Expression levels of sfGFP in four different cells were then determined (Figure 2A). The R3-11 clone lead to an expression level almost four folds of that for wild-type PylRS. For the other two clones, R1-7 and R2-23, expression levels of sfGFP were about two and three folds, respectively, of that for wild type PylRS. This result demonstrates our initial rationale that an improved PylRS clone can be evolved by mutating PylRS-n that interacts directly with tRNA^Pyl. Encouraged by this result, we further examined whether the three mutations R19H/H29R/T122S identified in the R3-11 PylRS clone can be transferred to other PylRS mutants for improving genetic incorporation efficiency of their corresponding ncAAs. We chose to work with AcKRS that was previously evolved from M. mazei PylRS by Soll et al.^[12] for the genetic incorporation of N^ε-acetyl-L-lysine (AcK) and AcdKRS that was evolved in our group^[13] for the genetic incorporation of N^ε-(4-azidobenzoxycarbonyl)-L-δ, ε-dehydrolysine (AcdK) and subsequent installation of N^ε, N^ε-dimethyl-L-lysine into proteins. Two plasmids pEVOL-R3-11 AcKRS and pEVOL-R3-11 AcdKRS that contain mutations R19H/H29R/T122S and additional mutations found in original AcKRS or AcdKRS, respectively were constructed from pEVOL-R3-11 PylRS. These two plasmids were used separately to couple with pBAD-sfGFP134TAG in E. coli Top10 cells for testing incorporation efficiency of AcK and AcdK. Results were then compared to those from using original AcKRS and AcdKRS-containing pEVOL plasmids. Both AcKRS and AcdKRS contain mutations only in PylRS-c for selective binding of their corresponding ncAAs. The introduction of R19H/H29R/T122S to these two proteins is not expected to interrupt their recognition of ncAA substrates. Quite on the contrary, improved incorporation of ncAAs is anticipated. Indeed, in comparison to a pEVOL-AcKRS plasmid that we previously used for the genetic incorporation of AcK, a newly constructed pEVOL-R3-11 AcKRS plasmid led to a significantly improved production level of sfGFP in the presence of 5 mM AcK (Figure 2B). Although Cells transformed with plasmids pEVOL-R3-11 AcdKRS and pBAD-sfGFP134TAG did not produce sfGFP in the presence of 1 mM AcdK dramatically higher than cells transformed with original pEVOL-AcdKRS and pBAD-sfGFP134TAG, the yield is still improved by about 20% (Figure 2C). It is possible that the incorporation of AcdK is already efficient at 1 mM and therefore it leaves little space for improvement. But it is also possible that tRNA aminoacylation processes with BocK and AcdK are different and therefore mutations at the PylRS N-terminus have different effects on the two processes.

Figure 2.

(A) The comparison of three mutant PylRS clones and wild type PylRS in the incorporation efficiency of BocK in E. coli Top10 cells. (B) The comparison of an original AcKRS and R3-11 AcKRS in the incorporation efficiency of AcK in E. coli Top10 cells. (C) The comparison of an original AcdKRS and R3-11 AcdKRS in the incorporation efficiency of AcdK in E. coli Top10 cells. (D) The structures of AcK and AcdK.

There is one concern that mutations R19H/H29R/T122S might induce interactions between PylRS and endogenous E. coli tRNAs. This reduced orthogonality of PylRS might lead to the aminoacylation of endogenous E. coli tRNAs and the subsequent misincorporation of ncAAs at other sensing codon sites. To dissolve this concern, we transformed E. coli BL21 cells with two plasmids pEVOL-R3-11 PylRS and pETduet-Ub48TAG that contains a gene coding ubiquitin with an amber mutation at K48 and a C-terminal 6×His tag and grew the transformed cells in LB media supplemented with 1 mM BocK for the expression of Ub48BocK. The Ni-NTA affinity-purified Ub48BocK was subjected to the mass spectrometry (MS) analysis by an Autoflex MALDI TOF/TOF spectrometer. A single sharp peak at 9488.9 Da was observed (Figure S3). The theoretical molecular weight of a single charged Ub48BocK is 9488.8 Da. No side peaks that indicate misincorporation of BocK at sensing codon sites were detected. Similarly, we transformed E. coli BL21 cells with plasmids pEVOL-R3-11 PylRS and pETduet-Ub that contains a wild type Ub gene and grew the transformed cells in LB media supplemented with 1 mM BocK for the expression of Ub. Ub was extracted from the cell lysate simply by the addition of acetic acid to precipitate the majority of E. coli cytosolic proteins. The extracted Ub was then subjected to the same MS analysis and a major peak at 8565.8 Da was detected (Figure S4). This detected value matches exactly the theoretic molecular weight of a single charged Ub. No nearby side peaks around 8565.8 Da were detected to indicate misincorporation of BocK. The two minor peaks at 7271.9 Da and 9122.3 Da are too far from 8565.8 Da to be resulted from the misincorporation of BocK. They are probably from unprecipitated E. coli proteins.

One important application of the genetic incorporation of AcK is to produce acetyl-histones for functional investigation of histone acetylation in the regulation of chromatin.^[6] Using the originally evolved AcKRS, we previously synthesized histone H3 with AcK incorporated at multiple lysine sites and used these expressed acetyl-H3 proteins to probe histone lysine site selectivity of three NAD-dependent histone deacetylases, SIRT1, SIRT2, and SIRT6.^[14] To demonstrate the application of R3-11 AcKRS in the production of acetyl-histones, pEVOL-R3-11 AcKRS was coupled with another plasmid pETDuet-1-H3K23TAG, that contains a gene coding H3 with amber mutation at the K23 coding position, for the production of H3K23ac (H3 with acetylation at K23). E. coli BL21 cells transformed with these two plasmids produced H3K23ac with a yield of 9.4 mg/L in LB medium supplemented with 5 mM AcK. A control experiment that used the original pEVOL-AcKRS provided only a yield of 3.7 mg/L (Figure 3A). This result implies that R3-11 AcKRS can be generally applied to improve the synthesis of acetyl-histones. We have also used the expressed H3K23ac to assemble a histone tetramer with H4 that was conjugated to SUMO for easy refolding and improved production in E. coli. The SDS-PAGE analysis of the refolded H3K23ac-H4 tetramer, a commonly accepted standard to analyze the folded H3/H4 tetramer,^[15] indicated a correctly folded complex with a 1:1 ratio of H3K23ac to H4 (Figure 3B). This tetramer complex was then subjected to a SIRT1-catalyzed deacetylation assay and then probed by a pan anti-Kac antibody in the Western blot analysis. As shown in Figure 3C, SIRT1 actively removed acetylation at H3 K23.

Figure 3.

(A) The comparison of H3K23ac expression in two conditions. E. coli BL21 cells transformed with two plasmids, pEVOL-AcKRS (original or R3-11) and pETDuet-1-H3K23TAG were grown in LB medium supplemented with 5 mM AcK. The expression of H3K23ac was induced by the addition of 0.2% arabinose. (B) SDS-PAGE analysis of the refolded H3K23ac-H4 tetramer. H4 is fused with SUMO at its N-terminal side. (C) Deacetylation of the H3K23ac-H4 tetramer by SIRT1. 0.4 μM H3K23ac-H4 was incubated with or without 0.2 μM SIRT in the presence of 1 mM NAD for 3 h at 37 °C before they were analysed by SDS-PAGE and probed by pan anti-Kac and anti-H3 antibodies.

In summary, we have developed a general method to improve the ncAA incorporation efficiency of the pyrrolysine incorporation system. By evolving a N-terminal region of M. mazei PylRS that interacts directly with tRNA^Pyl but doesn’t participate in the aminoacylation catalysis, we showed that three mutations R19H/H29R/T122S can significantly improve the incorporation of BocK at amber codon. The direct transfer of these three mutations to two PylRS mutants that were previously evolved for the genetic incorporation of AcK and AcdK also improved the incorporation efficiency of these two ncAAs, implying that the three mutations can be generally applied to improve ncAA incorporation efficiency of PylRS and its mutants. Given the increasing adoption of the pyrrolysine incorporation system for basic and advanced research, we foresee broad application of our current reported method. However, how exactly the three mutations improve the incorporation, whether through improved enzyme solubility or through improved tRNA binding, needs to be further tested. Interestingly, Owens et al. recently developed a two-tier screening platform for the identification of improved PylRS clones.^[16] Although random mutations were introduced to the whole PylRS coding region, identified clones had mutations converged at PylRS-n. similarly as what we show in the current work, these mutations can be transferred to other PylRS mutants for improved ncAA incorporation.