Update of the Pyrrolysyl-tRNA Synthetase/tRNA^Pyl Pair and Derivatives for Genetic Code Expansion

ABSTRACT

The cotranslational incorporation of pyrrolysine (Pyl), the 22nd proteinogenic amino acid, into proteins in response to the UAG stop codon represents an outstanding example of natural genetic code expansion. Genetic encoding of Pyl is conducted by the pyrrolysyl-tRNA synthetase (PylRS) and its cognate tRNA, tRNA^Pyl. Owing to the high tolerance of PylRS toward diverse amino acid substrates and great orthogonality in various model organisms, the PylRS/tRNA^Pyl-derived pairs are ideal for genetic code expansion to insert noncanonical amino acids (ncAAs) into proteins of interest. Since the discovery of cellular components involved in the biosynthesis and genetic encoding of Pyl, synthetic biologists have been enthusiastic about engineering PylRS/tRNA^Pyl-derived pairs to rewrite the genetic code of living cells. Recently, considerable progress has been made in understanding the molecular phylogeny, biochemical properties, and structural features of the PylRS/tRNA^Pyl pair, guiding its further engineering and optimization. In this review, we cover the basic and updated knowledge of the PylRS/tRNA^Pyl pair’s unique characteristics that make it an outstanding tool for reprogramming the genetic code. In addition, we summarize the recent efforts to create efficient and (mutually) orthogonal PylRS/tRNA^Pyl-derived pairs for incorporation of diverse ncAAs by genome mining, rational design, and advanced directed evolution methods.

INTRODUCTION

With few exceptions, all living organisms genetically encode 20 amino acids according to the rules of the genetic code. It is now known that variations to the standard genetic code table are present in nature, including, but not limited to, codon reassignment and ambiguous decoding (1). A notable example is site-specific incorporation of Pyl into polypeptides in response to an in-frame UAG stop codon in certain anaerobic archaea and bacteria (2, 3). Inspired by surprising findings of genetic code flexibility, scientists aim to site-specifically install distinct noncanonical amino acids (ncAAs) with unique physicochemical properties into proteins in living cells. Pioneered by Peter Schultz and others, numerous useful ncAAs have been genetically encoded in various model organisms via genetic code expansion technology (4–6) (Fig. 1), which facilitates the investigation of protein structure and function and enables the development of novel methods to study, control, and evolve biological processes, as summarized in other reviews (7–11).

FIG 1.

Genetic encoding of noncanonical amino acids (ncAAs). The schematic diagram showing the orthogonal aaRS/tRNA pair (grayish purple) decodes the amber stop codon to incorporate ncAA (grayish purple diamond) site-specifically into proteins by genetic code expansion. The resulting products include the full-length ncAA-containing proteins of interest and the undesired truncated polypeptide.

Developing orthogonal aminoacyl-tRNA synthetase and tRNA (aaRS/tRNA) pairs, the central biomolecules ensuring accurate genetic code interpretation, is a prerequisite for genetically encoding “designer” ncAAs. The cross talk between the orthogonal aaRS/tRNA pair and the endogenous tRNAs and aaRSs should be minimized. In addition, the introduced tRNA should be able to decode a specific codon. The UAG stop codon is commonly used since it is not assigned to an amino acid and is the rarest codon in many organisms such as Escherichia coli. Finally, by repurposing the amino acid substrate specificity of the orthogonal aaRS, the orthogonal tRNA is acylated with the desired ncAA, and the resulting ncAA-tRNA is installed into the growing polypeptide during protein translation (Fig. 1). As a natural amber suppressor, tRNA^Pyl is specifically charged by its cognate PylRS enzyme to form Pyl-tRNA^Pyl (12, 13), which is then decoded in the same manner as canonical tRNAs during ribosomal protein biosynthesis (Fig. 2A). Despite the PylRS/tRNA^Pyl pair allowing site-specific incorporation of Pyl into proteins in response to the in-frame UAG codon in the Pyl-decoding organisms, the reassignment of UAG codon to Pyl also leads to the extension of many cellular proteins where UAG specifies translation termination. Pioneering studies demonstrated natural PylRS/tRNA^Pyl pairs could enable genetic encoding of Pyl and its analogs in E. coli (12, 14, 15). Since then, the PylRS/tRNA^Pyl-derived pairs have become a popular choice for synthetic biologists to expand the genetic code with ncAAs.

FIG 2.

Biosynthesis and genetic encoding of Pyl in nature and the domain organization of PylRS enzymes. (A) Schematic diagram showing the Pyl biosynthesis pathway. Two lysine molecules (shown as black and cyan, respectively) are used to synthesize the Pyl in three steps by a cascade of enzymes consisting of PylB, PylC, and PylD. The free Pyl is ligated to tRNA^Pyl, a natural amber suppressor, by its cognate PylRS enzyme to form Pyl-tRNA^Pyl. By decoding the in-frame amber codon, Pyl is installed into proteins in Pyl-utilizing organisms. (B) Domain organization of the three classes of PylRS enzymes (PylSn–PylSc fusion, PylSn+PylSc, and ΔPylSn). The C-terminal catalytic domain (PylSc), the N-terminal tRNA-binding domain (PylSn), and linker region are labeled in orange, light blue, and dark blue, respectively. The representative species containing each class of PylRS are indicated. The crystal structures of PylSc of M. mazei (PDB accession no. 2ZIM), D. hafniense (PDB accession no. 2ZNI), and “Ca. Methanomethylophilus alvus” (PDB accession no. 6JP2) are shown.

In this review, we provide essential and updated information about key features of the PylRS/tRNA^Pyl pair that make this tool highly amenable to genetic code expansion. More details on the discovery, ncAA substrate diversity, and structural and functional properties of the native and engineered PylRS/tRNA^Pyl pair can be found in previous reviews (16–19). Particularly, we focus on key developments in improving the efficiency and (mutual) orthogonality of PylRS/tRNA^Pyl for cotranslational and site-specific incorporation of many useful ncAAs in the past 5 years; the relevant new approaches and mechanistic understandings accompanying the development and optimization processes are also discussed.

BIOSYNTHESIS AND GENETIC ENCODING OF Pyl

The discovery of Pyl was initiated by Krzycki et al. They discovered the presence of an in-frame UAG stop codon within a gene encoding the methylamine methyltransferase in Methanosarcina barkeri (20, 21). The chemical identity of Pyl was later determined at the position of an UAG-directed residue based on the high-resolution crystal structures of monomethylamine methyltransferase (3, 22). The pylT and pylS genes encoding tRNA^Pyl and PylRS, respectively, are found in the vicinity of methyltransferase gene cluster (23). Unlike tRNA-dependent pathways for the biosynthesis and genetic encoding of certain amino acids such as selenocysteine (24, 25), the free Pyl is synthesized from two lysine molecules and specifically ligated to tRNA^Pyl catalyzed by the PylRS enzyme in all Pyl-utilizing organisms (Fig. 2A). Biosynthesis of Pyl is mediated by a cascade of enzymes, including PylB, PylC, and PylD enzymes. Specifically, the radical S-adenosyl-methionine enzyme PylB converts one lysine to 3-methylornithine, which is then ligated to the second lysine moiety with the help of the ATP-dependent enzyme PylC. Finally, the intermediate product is oxidized by the amino acid dehydrogenase PylD to produce Pyl (Fig. 2A) (26). Heterogenous expressing of the pylTSBCD gene cassette in E. coli was found to enable cotranslational incorporation of Pyl by amber suppression (27), indicating that all the natural components responsible for biosynthesis and genetic encoding of Pyl have been identified.

Biological function of Pyl.

The Pyl encoding system is concentrated in methylotrophic methanogens (28). In these Pyl-utilizing organisms, biosynthesis and genetic encoding of Pyl are needed to produce the methylamine methyltransferase, an enzyme that allows the cell to utilize methylamines as a carbon source and energy. Deletion of pylT gene in Methanosarcina acetivorans was found to disable the cell growth on methylamines (29). The appearance of Pyl in the active sites of the methylamine methyltransferases and the co-occurrence of the Pyl operon and at least one methylamine methyltransferase in Pyl-utilizing organisms strongly suggest the essential role of Pyl in catalyzing methane formation (3, 30). In contrast, Pyl, in the archaeal tRNA^His guanylyltransferases, is a dispensable residue and is only required to produce the full-length protein via amber suppression (31); thus, the biological function of Pyl is not solely restricted to its catalytic role. Interestingly, deduction of the genetic code from 21 to 20 amino acids in M. acetivorans by deletion of tRNA^Pyl leads to a global change of the proteome such as reduced level of methanol methanogenesis enzymes, which indicates that the genetic code expansion of Pyl is related to the proteome composition and cellular metabolism (32). Overall, the exact biological role of Pyl might differ in distinct native environments and requires further investigation.

PylRS: A UNIQUE aaRS FOR COTRANSLATIONAL INSTALLATION OF Pyl

As the dedicated aaRS for genetic encoding of Pyl, PylRS has been extensively characterized to understand its phylogenetic, structural, and functional characteristics. Most PylRS enzymes consist of the C-terminal catalytic domain (PylSc) that resembles that of the class II aaRSs (33, 34) and the N-terminal tRNA-binding domain (PylSn) that shares no sequence homology to any known RNA-binding proteins (35). Based on the different compositions of PylSn and PylSc, PylRS enzymes are categorized into three classes, PylSn–PylSc fusion, PylSn+PylSc, and ΔPylSn (Fig. 2B). This classification is consistent with recent phylogenetic analyses based on PylSc sequences (28, 36, 37). The two domains of the PylSn–PylSc fusion enzyme are linked by an unstructured loop with different lengths (Fig. 2B). The pylS genes encoding the PylSn–PylSc fusion class enzymes have been identified in a total of 47 archaeal genomes, even including nonmethanogenic archaea (28). Derivatives of Methanosarcina mazei PylRS (MmPylRS) and M. barkeri PylRS (MbPylRS) represent the most widely used PylSn–PylSc fusion enzymes, which enables the site-specific incorporation of more than 100 ncAAs into different proteins so far (16). The PylSn+PylSc class enzymes are expressed as different polypeptides encoded by separate pylS genes (35). This class of PylRS enzymes is commonly observed in Pyl-utilizing bacteria. Interestingly, the co-occurrence of individual genes encoding PylSc and PylSn was also found in seven archaeal genomes, including two Asgardarchaeota lineages (28, 38). As the most recently discovered class, the ΔPylSn enzymes are encoded in the genomes that lack gene homologs encoding PylSn. Overall, these newly identified pylS genes extend our knowledge about the diversity and molecular phylogeny of PylRS enzymes in nature.

The recent discovery of ΔPylSn enzymes.

The discovery of numerous ΔPylSn enzymes opens up new opportunities to expand the PylRS-based toolbox for genetic code expansion. A notable example of ΔPylSn enzyme is that from “Candidatus Methanomethylophilus alvus” (MaPylRS), which displayed robust UAG readthrough activity in E. coli cells and cell-free systems (39, 40). However, a recent biochemical study showed that the aminoacylation activity of MaPylRS is less competent than MmPylRS (28), suggesting other factors might be responsible for the high amber suppression activity of MaPylRS in vivo (for example, the solubility of MaPylRS is much higher than MmPylRS [40]). In addition, MaPylRS and its cognate, MatRNA^Pyl, could be engineered to become mutually orthogonal, with PylRS belonging to the PylSn–PylSc fusion class (39, 41–43). Based on the sequence similarity, two distinct clusters of ΔPylSn enzymes (class A and class B) have been disclosed; measurements of amber suppression activity in E. coli found these ΔPylSn enzymes preferentially work with the tRNA^Pyls belonging to the same class (44) (Fig. 3A). Another surprising discovery about the ΔPylSn class enzyme is the simultaneous presence of two distinct PylRS enzymes encoded in the genome of Methanomassiliicoccus luminyensis B10 and “Candidatus Methanohalarchaeum thermophilum” HMET1 (45). Interestingly, the two wild-type PylRS/tRNA^Pyl pairs from HMET1 share different identity elements and are mutually orthogonal in haloarchaeon Haloferax volcanii (45) (Fig. 3A). The combination of engineered ΔPylSn and PylSn–PylSc fusion enzymes enables the simultaneous incorporation of multiple distinct ncAAs into a single polypeptide (39, 41–44, 46–48) (discussed below).

FIG 3.

Schematic diagram showing the mutually orthogonal PylRS/tRNA^Pyl pairs found in nature and developed by engineering. (A) Mutually orthogonal PylRS/tRNA^Pyl pairs found in nature. The left panel shows the mutually orthogonal pairs identified from HMET1; the unique features (motif 2 loop with a single amino acid deletion in PylRS2 and A73 discriminator base in tRNA^Pyl2) accounting for mutual orthogonality are highlighted in red. The right panel shows two distinct clusters of ΔPylSn enzymes (class A and class B), and some of these ΔPylSn/^ΔPylSntRNA^Pyl pairs display mutual orthogonality. Some unique nucleotides between class A and class B ^ΔPylSntRNA^Pyl are indicated in brown and black, respectively. (B) Mutually orthogonal PylRS/tRNA^Pyl pairs by engineering. The variable loop of ^ΔPylSntRNA^Pyl (such as MatRNA^Pyl) recognized by the PylSn domain could be engineered to become orthogonal with PylRS belonging to the PylSn–PylSc fusion class (such as MmPylRS). Triply orthogonal PylRS/tRNA^Pyl pairs could be further created by evolving the variable loop of class A ^ΔPylSntRNA^Pyl and both the acceptor stem and variable loop of the class B ^ΔPylSntRNA^Pyl with PylSn–PylSc full-length pairs. The evolved sites are shown in cyan.

High tolerance for amino acid substrates.

PylRS harbors a deep catalytic pocket and has low selectivity toward the side chain and α-amine of substrate amino acids (34, 49–51). The remarkably broad polyspecificity of PylRS for the substrate amino acids has been illustrated by crystal structures of the wild-type and engineered PylRS enzymes in complex with Pyl and its analogs (19). A recent study provides new insights into how bulky lysine derivatives are recognized by the MmPylRS variant bearing two mutations (Y306A and Y384F) by determining crystal structures of PylSc in complex with 14 different ncAAs with diverse functional applications (such as click chemistry) (52), which reveals distinct binding modes and additional key residues for the substrate interaction of PylRS enzyme. Only a few modifications of the PylRS catalytic pocket are required to repurpose this enzyme to recognize phenylalanine and histidine derivatives (53–56). In addition, PylRS variants with improved specificity for aromatic ncAA analogs have been recently developed via the semirational design of the enzyme-substrate binding pocket (57). As PylRS weakly recognizes the α-amine group, noncanonical substrates with high backbone diversity, including α-hydroxyacid, N_α-methyl-amino acid, and d-Pyl derivatives, could be recognized by the wild-type and evolved PylRS (58, 59). Although high plasticity in the active site of PylRS is desirable for genetic code expansion of diverse ncAAs, the cross-reactivity between evolved PylRS enzymes and different ncAA substrates should be paid attention to in future applications of simultaneous incorporation of multiple ncAAs with structural similarity into a single polypeptide.

Recognition of tRNA^Pyl by PylRS.

PylRS has evolved to specifically recognize tRNA^Pyl, a noncanonical tRNA bearing several unique features (17). In contrast to the majority of aaRS enzymes that employ the tRNA anticodon as an identity element (60), PylRS does not recognize the anticodon of tRNA^Pyl based on the biochemical and structural data (14, 61, 62). This notable feature allows the facile engineering of tRNA^Pyl to decode diverse codons, such as unnatural codons and four-base codons (63, 64). Early aminoacylation assays discovered that the discriminator base (G73) and the first base pair in the acceptor stem (G1·C72) in tRNA^Pyl serve as the major identity elements (14, 65). The crystal structure of the PylRS/tRNA^Pyl pair from Desulfitobacterium hafniense further illustrates the key role of motif 2 loop in interacting G73 and the G1-C72 base pair in a base-specific manner (61). PylRS also has evolved a specialized tRNA binding surface to accommodate the compact core of tRNA^Pyl while rejecting the bulkier tertiary core of other canonical tRNAs (61). The crystal structure of the PylSn domain of MmPylRS in complex with MmtRNA^Pyl further explains the high orthogonality of the PylSn–PylSc fusion class enzyme (62). Mechanistically, the steric constraint is employed by the PylSn domain to accept the shortened variable loop of tRNA^Pyl but excludes other canonical tRNAs with a larger variable arm (62). Based on this recognition mode, MatRNA^Pyl with an expanded variable loop has been designed to be orthogonal to PylSn–PylSc fusion PylRS while still serving as the substrate for MaPylRS, a ΔPylSn class enzyme (41, 42) (Fig. 3B). In general, recent studies have deepened our understanding of the molecular basis of orthogonality adopted by the PylRS/tRNA^Pyl pair, which guides the design and creation of (mutually) orthogonal tools for genetic code expansion.

PylRS/tRNA^Pyl: AN OUTSTANDING GENETIC CODE EXPANSION TOOL

As mentioned above and discussed in other reviews (16, 17), the PylRS/tRNA^Pyl pair exhibits high orthogonality due to its many unique properties compared to other canonical aaRS/tRNA. For this reason, the PylRS/tRNA^Pyl pair has become the most popular workhorse of synthetic biologists for genetic code expansion. Over the past 5 years, the versatility and robustness of the PylRS/tRNA^Pyl pair for genetic encoding ncAAs have been further demonstrated in more organisms such as Synechococcus elongatus (66), Xenopus oocyte (67), Lactococcus lactis (68), Pseudomonas aeruginosa (69), and Bombyx mori (70). Also, PylRS/tRNA^Pyl pair derivatives have also been heavily employed in the cell-free system (40, 71–73). In this section, we focus on recent progress in the incorporation of diverse ncAAs implemented by PylRS/tRNA^Pyl-derived pairs, their optimization in the aspects of efficiency and substrate specificity, and the development of the mutually orthogonal tools for simultaneously encoding multiple distinct ncAAs.

Genetic encoding of a wide range of function-oriented ncAAs by PylRS/tRNA^Pyl pairs.

The high expansibility of the PylRS substrate spectrum has been employed to create PylRS variants for the incorporation of a variety of “designer” ncAAs with novel chemical and physical properties. These recent efforts have enabled many exciting applications, including cross-linking, regulation of protein function, posttranslational modifications (PTMs), probing, and protein imaging. Detailed information about the chemical names of these ncAAs, application purposes, and the sequence of PylRS variants using MmPylRS as a reference sequence are summarized in Table 1.

TABLE 1.

Information of recently developed PylRS variants that recognize different ncAA substrates and relevant applications in different heterogenous organisms^a

Source	Host	Mutation site	ncAA	Application
MmPylRS	E. coli	A302T/V346A/W348A/L401V/W417T (75)	(S)-2-amino-3-(4-(3-bromopropoxy)-3-ethynylphenyl) propanoic acid (EB3)	Crosslinking
	E. coli	L301M/Y306A/L309A/C348F (128)	Nε-heptanoyl-l-lysine (HepoK)	Crosslinking
	E. coli and HEK293 cells	A302I/N346T/C348I/Y384L/W471K (76)	Fluorosulfate-l-tyrosine (FSY)	Crosslinking
	E. coli and mammalian cells	Y306L/L309A/N346A/C348M/W417T (129)	o-Sulfonyl fluoride-O-methyltyrosine (SFY)	Crosslinking
	E. coli	C348W/W417S (125)	S-allyl cysteine (Sac)	Crosslinking
	E. coli	Y306A/Y384F (80)	o-Nitrobenzyl alcohol lysine (o-NBAK)	Photo-crosslinking
	E. coli	A302T/N346A/C348G/Y384F/W417T (130)	Dibenzo[b,f][1,4,5]thiadiazepine-based alanine (DBTDA)	Photo-crosslinking
	HEK293 cells, HeLa cells, CHO cells	R91K/G131E/Y306A/Y384F (131)	Nε-(para-trifluoromethyl-diazirinyl-benzyloxycarbonyl)-l-lysine (pTmdZLys)	Photo-crosslinking
	E. coli and HEK293 cells	Y306V/L309A/C348F/Y384F (79)	N-methyl pyrroletetrazole-lysine (mPyTK)	Photo-crosslinking
	E. coli and HEK293 cells	L309G/N346A/C348I/V401K/W417I (88)	Thyronine (Thy)	Protein labeling
	E. coli with deletion of release factor 1	Y306A/Y384F (132)	Trans-cyclooct-2-ene-lysine (TCO*-AK)	Protein labeling (fluorescence)
	E. coli and HeLa cells	N346Q/C348S/V401G/W417T/G419G (95)	l-β-(Quinolin-6-yl) alanine (Qui)	Protein labeling
	E. coli	R61K/M300L/A302G/L309F/C348A/E444G (133)	Nε-allyloxycarbonyl-l-lysine (AlockOH)	Protein labeling (self-cleavage)
	E. coli	F271L/F313C, F271L/F313M, and F271N/F313I (84)	Nε-thioacetyl-l-lysine (TAcK)	PTM (acetylation)
	E. coli	Y306L/C348I/Y384F (86)	Azidonorleucine (AznL)	PTM (acetylation)
	E. coli	A302S/L309M/I322L/N346A/W348G/W417T (87)	Phosphortyrosine (pTyr)	PTM (phosphorylation)
	E. coli and HEK293 cells	N346D/C348S/Y384F (134)	o-Chlorophenylalanine (o-ClF)	Substrate expansion
	E. coli	N346S/C348M/V401G (57)	3-Benzothienyl-l-alanine (Bta) and 3-(1-naphthyl)-l-alanine (1-NaA)	Substrate expansion
	E. coli	N346G/C348M/W417L (57)	3-Chloro-l-tyrosine (3-ClY) and 3-bromo-l-tyrosine (3-BrY)	Substrate expansion
MbPylRS	E. coli and HEK293 cells	L274[309]A/C313[348]F/Y349[384]F (78)	o-4-Fluorophenyl lysine (FPheK)	Crosslinking
	HEK293 cells	Y271[306]A/L274[309]A/C313[348]A (74)	(S)-2-amino-6-(6-bromohexanamido) hexanoic acid n lysine (BrCnK)	Crosslinking
	E. coli	Y271[306]C/N311[346]Q/Y349[384]F/V366[401]C (135)	2,3-Diaminopropionic acid (DAP)	Photo-crosslinking
	E. coli	C323[358]W/W382[417]T (136)	Se-allyl selenocysteine (Asec)	Caged amino acid
	E. coli and HEK293 cells	N311[346]Q/C313[348]S (137)	N6-((2-azidoacetyl) glycyl)-l-lysine (AzGGK)	Caged amino acid
	E. coli and HeLa cells	L270[305]F/L274[309]M/N311[346]A/C313[348]G (81)	(2 R)-2-amino-3-fluoro-3-(4-((2-nitrobenzyl) oxy) phenyl) propanoic acid (FnbY)	Photocage
	HEK293 cells	L270[305]G/N311[346]G/C313[348]A (138)	3-(6-Alkyl-s-tetrazin-3-yl) phenylalanine (Tet-v3.0)	Protein labeling
	E. coli and HEK293 cells	N311[346]S/C313[348]G/V366[401]A/W382[417]T (93)	Acridonylalanine (Acd)	Protein labeling (fluorescence)
	E. coli	A267[302]Q/N311[346]S/C313[348]W (139)	4-Thiazoylalanine (4ThzA)	Protein labeling
	E. coli	N311[346]Q/Y349[384]F (139)	3-(3-Thienyl) alanine (3ThiA)	Protein labeling
	E. coli and HEK293 cells	L266[301]M/L270[305]I/L274[309]A/C313[348]F (89)	Nε-trifluoroacetyl-l-lysine (TFAcK)	Protein labeling (NMR)
	E. coli and HEK293 cells	Y271[306]M/C313(348)T (85)	Nε-l-lactyl-l-lysine (LacK)	PTM (lactylation)
	E. coli and HEK293 cells	C313[348)T/Y349[384]F (85)	Nε-β-hydroxybutyryl-l-lysine (BhbK)	PTM (butyrylation)
	E. coli and HEK293 cells	Y271[306]A (85)	Nε-lipoyl-l-lysine (LipoK)	PTM (lipoylation)
	E. coli	T13[48]I/I36[71]V/C313[348]W/W382[417]S (140)	S-allyl-l-cysteine (Sac)	Substrate expansion
MaPylRS	E. coli	Y126[306]G/M129[309]A/V168[348]F/H227[405]T/Y228[406]P/L229[407]I (77)	Fluorosulfonyloxybenzoyl-l-lysine (FSK)	Crosslinking
	E. coli and mammalian cells	Y126[306]L/M129[309]A/N166[346]A/V168[348]M/W239[417]T (129)	o-Sulfonyl fluoride-O-methyltyrosine (SFY)	Crosslinking
	E. coli	L125[305]F/N166[346]A/V168[348]G (82)	o-2-Nitrobenzyl-β-fluorotyrosine (FnbY), o-2-nitrobenzyl-3-fluoromethyl tyrosine (FmnbY)	Photocage
	E. coli-based cell-free system	Y126[306]A (40)	Nε-((((E)-cyclooct-2-en-1-yl) oxy) carbonyl)-l-lysine (TCO*K)	Protein labeling (fluorescence)
	E. coli-based cell-free system	V168[348]G/A223[401]C/Y206[384]F (92)	N6-(((trimethylsilyl) methoxy) carbonyl)-l-lysine (TMSK)	Protein labeling (NMR)
	E. coli HEK293 cells	Y126[306]T/M129[309]R/V168[348]H/H227[405]I/Y228[406]P and Y126[306]T/M129[309]R/V168[348]H/Y206[384]W (47)	Nε-benzoyllysine (BzK)	PTM (benzoylation)
G1PylRS (PylRS from methanogenic archaeon ISO4-G1)	E. coli-based cell-free system	V167[348]G/A221[401]C/Y204[384]F (92)	N6-(((trimethylsilyl) methoxy) carbonyl)-l-lysine (TMSK)	Protein labeling (NMR)
	E. coli	L124[305]G/Y125[306]F/N165[346]G/V167[348]F/Y204[384]W/A221[401]G/W237[417]Y (91)	7-Fluoro-l-tryptophan (7FTrp)	Protein labeling (NMR)
DhPylRS (PylRS from Desulfitobacterium hafniense)	E. coli	N176[367]A/T178[368]G (104)	Para-azido-l-Phenylalanine (AzF)	Crosslinking
ChPylRS (chimeric PylRS consisting of PylSn MbPylRS and MmPylRS PylSc)	E. coli and Bacillus stearothermophilus	C348G/V401C/Y384F, and V401K/Y384F (92)	N6-(((trimethylsilyl) methoxy) carbonyl)-l-lysine (TMSK)	Protein labeling (NMR)

^aThe mutation site corresponding to the residue of MmPylRS is listed in square brackets.

Genetic encoding of ncAAs bearing the cross-linking side chains to selectively form the covalent bond is a valuable approach to identifying weak and transient protein interactions. For instance, using a PylRS variant with the ability to install lysine derivatives that contain structurally flexible bromoalkyl moieties into proteins, a method was developed to capture the stabilized protein complex in a proximity-dependent way (74). A similar method called genetically encode chemical cross-linkers (GECX) was also invented, which is based on covalent bond formation between the alkyl bromide group of the ncAA in the target protein and the cysteine residue in partner proteins (75). Notably, GECX facilitates the subsequent mass spectrometric analysis of partner proteins through biotin enrichment with the help of an additional bio-orthogonally reactive group in the ncAA (75). In addition, by the genetic encoding of fluorosulfate-containing ncAAs with latent bioreactivity to cross-link with diverse natural residues, a novel method called sulfur-fluoride exchange (SuFEx) was recently developed (76, 77). Evolved PylRS variants also allowed the site-specific incorporation of ncAAs with reactive aryl carbamate groups into proteins, which could also be utilized for cross-linking purified proteins under the basic environment (78). To map protein interactions with spatiotemporal resolution in living cells, several lysine-derived photo-cross-linkers were designed and installed into proteins of interest by genetic code expansion using developed PylRS enzymes (79, 80).

To precisely and optically control protein function with residue selectivity, photocaged ncAAs bearing photocleavable protecting groups have been site-specifically incorporated into target proteins. As for the regulation of cross-linking process, an ncAA bearing the photocaged quinone methide was successfully inserted into proteins using an engineered MbPylRS enzyme; the quinone methide could be liberated to cross-link with nearby nucleophilic residues upon photoactivation (81). Genetic code expansion of the photocaged quinone methide could also be employed to endow proteins with diverse functionalities via facile protein labeling (82). Additionally, by genetically encoding a photocaged ncAA that could trap diverse acyl-enzyme complexes upon activation, a powerful method was recently developed to identify unknown substrates of various hydrolases (83).

The capability to site-specifically incorporate ncAAs with designed PTMs into proteins provides a powerful way to elucidate the role of these modifications in protein function. By repurposing the amino acid substrate specificity, many PylRS/tRNA^Pyl-derived pairs were recently developed to genetically encode PTMs on the lysine residue, including a nonhydrolyzable analog of lysine acetylation (84), lipoylation, and newly discovered benzoylation, lactylation, and β-hydroxybutyrylation (47, 85). In addition, site-specific insertion of lysine succinylation into proteins was achieved by combing the PylRS-based genetic code expansion with traceless Staudinger ligation (86). Except for PTMs on the lysine residue, phosphotyrosine could be directly installed into different proteins by the engineered MmPylRS variant in two steps: the phosphotyrosine analog was incorporated in response to the amber stop codon, followed by deprotection to generate the native phosphotyrosine (87).

Genetic code expansion using the PylRS/tRNA^Pyl pair derivatives enables many applications of ncAA-based probes. For example, genetic encoding of a novel ncAA called thyronine could be used for fluorescent detection of peroxynitrite in vivo (88). Various ncAAs with the fluorine atom have been site-specifically installed into proteins by engineered PylRS variants and serve as novel nuclear magnetic resonance (NMR) probes to study conformational changes of these proteins with essential physiological functions (89–91). In addition, a PylRS variant was recently evolved to genetically encode a trimethylsilyl analog on lysine residue as a probe for ¹H NMR spectra, facilitating the investigation of protein structure and function (92).

Compared with fluorescent fusion proteins or self-labeling tags, site-specific incorporation of ncAAs for protein imaging might have some advantages, such as minimal perturbation of protein function and an expanded range of fluorophores (7). A fluorescent ncAA with high photostability and a long lifetime was recently reported to be directedly inserted into proteins by engineered MbPylRS for live cell imaging (93). A PylRS variant was identified to produce proteins with ncAA composed of styrene, an unstrained alkene; the ncAA residue could be subsequently labeled with novel fluorophore via the bio-orthogonal reaction (94). Interestingly, a pH-dependent fluorescent protein was developed using a PylRS derivative-assisted genetic code expansion; this protein exhibits enhanced fluorescence at the acid pH condition through the switch of chromophore charge enabled by a novel quinoline-containing ncAA (95).

Improving the ncAA incorporation efficiency by PylRS/tRNA^Pyl pairs.

Although PylRS/tRNA^Pyl pair derivatives enable widespread applications of genetic code expansion in living organisms, in vitro aminoacylation experiments reveal PylRS enzymes exhibiting significantly lower turnover rates than other canonical aaRSs (18). Also, compared with the native substrate Pyl, many ncAAs are poorly recognized by the wild-type and engineered PylRS enzymes (1 to 2 orders of magnitude lower affinity) (28, 62). Thus, improving the catalytic activity of PylRS enzymes is an important field in the optimization of genetic code expansion tools.

As many key residues in the amino acid binding pocket of PylSc have been determined based on crystal structures, PylRS variants with improved substrate specificity (listed in Table 1) could be obtained by successive rounds of selections from the saturation mutagenesis library corresponding to these residues (96). The strategy of positive selection to select active PylRS variants often relies on cell survival based on antibiotic tolerance, fluorescence intensity measurement, and colorimetric assay (97, 98). Considering the advantages of throughput, sensitivity, and quantifiability, the fluorescence-activated cell sorting (FACS) technology was utilized to discover designed PylRS derivatives with enhanced incorporation efficiency and selectivity toward ncAAs (99, 100). To speed up the process of laboratory evolution, which typically requires a month's effort via successive rounds of positive and negative selections, advanced methods, such as phage-assisted continuous evolution (PACE) (101), phage- and robotics-assisted near-continuous evolution (PRANCE), a high-throughput version of PACE (102), and phage-assisted noncontinuous evolution (PANCE) that does not require specialized equipment (62), were developed for directed evolution of PylRS. The strategy of positive selection employed in the phage-assisted evolution is to link the catalytic activity of PylRS to the production of pIII protein, an essential protein for phage infectivity, via amber suppression (101). One advantage of these new methods is that no prior knowledge about the enzyme structure is needed for directed evolution, which is similar to the error-prone PCR-based laboratory evolution. Using these advanced methods, superior PylRS variants bearing various beneficial mutations were evolved which exhibit improved k_cat and affinity for both tRNA^Pyl and ncAA substrates (62, 101). Employment of these evolved PylRS enzymes achieved a much improved yield of ncAA-containing proteins (62, 101, 103). As the PylSn and PylSc domains of different PylRS enzymes share high sequence similarity, transplantation of the beneficial mutations from evolved PylRS variants to other PylRS derivatives serve as a simplified way to promote the ncAA incorporation efficiency (39, 97, 98, 101, 104). Interestingly, manipulating the length of the linker connecting PylSn and PylSc was also found to be effective to increase the ncAA incorporation efficiency (105), highlighting the feasibility of rational design strategy to optimize the catalytic efficiency of PylRS.

As the protein stability affects its robustness to be engineered (106, 107), enhancing this property of the PylRS enzyme could be employed to promote its activity for ncAA incorporation. For instance, the PylRS enzymes from thermophilic archaea exhibit enhanced thermostability and are more tolerant to mutations in the catalytic pocket than MbPylRS (108); these thermophilic PylRS derivatives outperform the mesophilic counterpart for installing diverse ncAAs into proteins (108). Improving the stability of MmPylRS by site-directed mutations that prevent the enzyme from hydrolysis in E. coli was found to increase the amber suppression activity (109). In addition, enhancing the solubility of PylRS enzyme is also an effective strategy to improve the efficiency of genetic code expansion. For example, by fusing a soluble tag called small metal binding protein (SmbP) to MmPylRS enzyme derivatives, drastic improvement of the enzyme solubility and increased production of ncAA-containing proteins were achieved (110). To facilitate the directed evolution of PylRS enzyme by PACE, a chimeric PylRS with enhanced solubility was utilized (101). Also, employment of MaPylRS exhibits 5-fold-higher solubility than MmPylRS and could enable efficient ncAA incorporation into an antibody fragment using the cell-free system (40).

The introduced tRNA^Pyl, a noncanonical tRNA with many characteristic features, is not optimal for amber suppression in heterologous cells, probably owing to the lack of evolutionary tuning (111, 112). A recent study found that MmPylRS displays 2-fold-higher activity of amber suppression in vivo for MatRNA^Pyl than MmtRNA^Pyl. However, the aminoacylation efficiency for MatRNA^Pyl is worse than that of MmtRNA^Pyl (28). This finding suggests that factors other than aminoacylation efficiency would affect the overall efficiency of ncAA incorporation in heterogenous organisms. An improved tRNA^Pyl variant for ncAA insertion was developed in E. coli by directed evolution of the acceptor and T stem, a key region in tRNA involved in the interaction with elongation factor Tu (113). By rational design, two novel classes of tRNA^Pyl variants were recently developed in mammalian cells which exhibited enhanced intracellular concentration, intracellular stability, and ncAA incorporation efficiency (114). As posttranscriptional modification is widespread in all domains of life and modified nucleosides on MbtRNA^Pyl were found (13, 114), manipulation of the posttranscriptional modification in tRNA^Pyl might open up new directions for its further optimization in heterologous organisms.

Development of mutually orthogonal PylRS/tRNA^Pyl-based pairs.

Simultaneous incorporation of distinct ncAAs into a single protein requires multiple mutually orthogonal aaRS/tRNA pairs. Due to the excellent orthogonality of the PylRS/tRNA^Pyl pair to canonical aaRS/tRNA pairs, the joint use of PylRS/tRNA^Pyl-based pairs and other developed aaRS/tRNA pair derivatives (such as the tyrosyl pair from archaea) allows encoding of two to three distinct ncAAs in one gene (46, 115–117). Increased understanding of key features in PylRS that underly the molecular basis of orthogonality also contributes to the creation of mutually orthogonal PylRS/tRNA^Pyl derivatives. The N-terminal domain of the PylSn–PylSc fusion enzyme such as MmPylRS, which is absent in the ΔPylSn enzyme (e.g., MaPylRS), interacts with the T arm and variable loop of its cognate tRNA^Pyl. Thus, it is possible to engineer these regions in MatRNA^Pyl to interfere with its recognition by the PylSn domain of the PylSn–PylSc fusion class enzyme; the engineered tRNA^Pyl variants still retain the interaction with the cognate ΔPylSn enzyme (Fig. 3B). By rational modification and directed evolution of the variable loop, MatRNA^Pyl variants were developed which exhibit orthogonality to the full-length MmPylRS enzyme and remain as the substrate for MaPylRS (41, 42). Combining these mutually orthogonal pairs allowed the genetic encoding of two distinct ncAAs into one protein in both E. coli and mammalian cells (41–43). The unique properties of the engineered MatRNA^Pyl variant (the acceptor stem and modified variable loop) could be transplanted into another tRNA^Pyl to create a hybrid tRNA with good orthogonality to the MmPylRS/MmtRNA^Pyl pair and high activity in mammalian cells (118). Two classes of ΔPylSn enzymes and their cognate tRNA^Pyls (^ΔPylSntRNA^Pyl) were recently identified (44). Taking advantage of inherent mutual orthogonality between some of these ΔPylSn/^ΔPylSntRNA^Pyl pairs, triply orthogonal PylRS/tRNA^Pyl pairs were further developed by evolving the variable loop of class A ^ΔPylSntRNA^Pyl and both the acceptor stem and variable loop of the class B ^ΔPylSntRNA^Pyl (44) (Fig. 3B). A novel mechanism underlying the natural orthogonality of two novel PylRS/tRNA^Pyl pairs was recently disclosed which provides a compelling example of a single base change at the discriminator position, leading to additional coding capacity (45). The unique features accounting for mutual orthogonality consist of the A73 discriminator base of HMET1 tRNA^Pyl2 (almost all tRNA^Pyl contains the canonical G73 base) and a shorter motif 2 loop in HMET1 PylRS2 (45) (Fig. 3A). Overall, recent works highlight the feasibility of creating mutually orthogonal PylRS/tRNA^Pyl pair derivatives based on distinct tRNA recognition modes.

Genetic encoding of different ncAAs in a single polypeptide also requires additional “blank” codons other than the amber codon, which could be the liberated stop and sense codons created via whole-genome synthesis and new codons with synthetic nucleotides or four bases (119). In contrast to other developed aaRS/tRNA pairs that are mainly restricted to decoding the commonly used amber codon, the PylRS/tRNA^Pyl pair is recognized as a versatile tool to decode different codons by engineering. For instance, the anticodon of tRNA^Pyl could be modified accordingly to decode the UAA stop codon (46, 117), different sense codons (TCG and TCA) (120), the unnatural codons based on the dNaM/dTPT3 pair (63), and various four-base codons (48). These (mutually) orthogonal PylRS/tRNA^Pyl derivatives with the capability to decode a variety of “blank” codons allowed cotranslational installation of multiple distinct (up to four) ncAAs (46, 48, 117, 120, 121).

CONCLUDING REMARKS AND OUTLOOK

Over the past 2 decades, employment and engineering of PylRS/tRNA^Pyl pairs to site-specifically insert a variety of useful ncAAs into proteins have attracted increasing attention. The optimization process of PylRS for ncAA incorporation is accelerated by recent progress in understanding the biochemical and structural properties of PylRS enzymes and the development of advanced methods for the directed evolution of proteins (62, 101, 102). The recent discovery of a novel class of PylRS enzymes, especially the ΔPylSn class enzymes, further enriches the toolbox for genetic code expansion, laying the foundation for creating active and (mutually) orthogonal PylRS/tRNA^Pyl-derived pairs.

Continued efforts on developing more PylRS/tRNA^Pyl-based aaRS/tRNA pairs for genetic encoding of diverse ncAA substrates and improving their catalytic power are worth adhering to. As the orthogonality of engineered PylRS/tRNA^Pyl pairs has been confirmed in a variety of living organisms, transplantation of key orthogonal components from PylRS/tRNA^Pyl pairs to other aaRS/tRNA pairs would serve as a promising strategy to generate additional orthogonal translation systems, as demonstrated in a previous study (122). Recent works also revealed different tRNA recognition modes adopted by natural and engineered PylRS/tRNA^Pyl pairs to maintain mutual orthogonality (41, 42, 44); it might be possible to rationally design more (mutually) orthogonal PylRS/tRNA^Pyl pairs by combining distinct identity elements. To further increase the efficiency of genetic code expansion, efforts need to focus on improving the solubility, especially for pylSn–pylSc fusion enzymes, aminoacylation activity, and compatibility of tRNA^Pyl with the ribosome and other translational components in heterogeneous cells, as suggested in recent works (28, 40). To bypass the potential problem of ncAA uptake and the requirement for a supply of exogenous sources of ncAAs, coupling metabolic engineering of ncAA biosynthesis and genetic code expansion is a promising strategy to evolve PylRS/tRNA^Pyl-derived pairs for enhanced activity (123–126). As the complete Pyl biosynthesis pathway is known and functional in heterologous cells (26, 127), utilization of autonomous cells with the capability to synthesize Pyl, the authentic substrate for PylRS, may facilitate to improve the amber suppression activity of engineered PylRS/tRNA^Pyl pairs in future directed evolution experiments. Once the superior PylRS/tRNA^Pyl pairs are developed, the molecular architecture of amino acid binding pockets, which is highly conserved in all PylRS enzymes (28), could be rapidly redesigned by transplanting key residues from other developed PylRS variants with different ncAA substrate specificity (52). Overall, we envision an increased mechanistic understanding of how PylRS/tRNA^Pyl-derived pairs work in heterogenous organisms, and persistent optimization efforts would further unleash their potential for applications of genetic code expansion.

Biographies

Xuemei Gong is a Ph.D. student at University of Chinese Academy of Sciences, Beijing. Her research is focused on expanding the genetic code of cells for simultaneous incorporation of distinct noncanonical amino acids (ncAAs). She is interested in developing novel aminoacyl-tRNA synthetases/tRNA pairs, especial the pyrrolysyl-tRNA synthetase/tRNAPyl, as tools to install the distinct ncAAs with novel functions.

Haolin Zhang is a Research Associate of Synthetic Biology in BGI-Research. His interests are focused on the application of genetic code expansion (GCE) in the characterization of biological pathways and industrial applications. His research focuses on improving the incorporation efficiency of noncanonical amino acids in yeasts and archaea to explore GCE application in these organisms.

Yue Shen currently serves as the Chief Scientist of Synthetic Biology in BGI-Research, Deputy Director of the Guangdong Provincial Key Laboratory of Genome Read and Write, PI of Shenzhen Synthetic Biology Innovation Institute, and Visiting Professor of Center for Synthetic Genomics at SIAT, CAS. She earned her Ph.D. in Molecular Biology from the University of Edinburgh. She is one of the key members in the synthetic yeast consortium (Sc2.0). Her research focuses on the development of DNA synthesis technologies and instruments, synthetic genomics and its downstream applications, and DNA-based data storage. She has published over 40 papers in prestigious journals, including Science, Genome Research, Nature Communications, and Nature Computational Science, and was selected as one of the Top Ten Scientific Advances of China in 2017.

Xian Fu is a principal investigator at the synthetic biology platform of BGI-Research at Shenzhen. He earned his Ph.D. from University of Florida in 2016, mentored by Dr. Julie Maupin-Furlow. His Ph.D. thesis uncovered key components involved in targeted proteolysis and ubiquitin-like protein modification in archaea. He finished the postdoctoral training in Dr. Dieter Söll’s lab at Yale University to engineer aminoacyl-tRNA synthetase and tRNA pairs for genetic code expansion (GCE) and selenoprotein biosynthesis. By combining synthetic genomics, genome mining, and advanced directed evolution methods, his main research interests focus on developing an efficient GCE system in yeasts and archaea to genetically encode a variety of “designer” noncanonical amino acids (ncAAs) for investigation of fundamental biology and industrial applications. So far, he has published 24 peer-reviewed articles. He also serves on the editorial board of Frontiers in Genetics.