tRNA Pair Research Articles

Genetically encoded alkenyl–pyrrolysine analogues for thiol–ene reaction mediated site-specific protein labeling

A series of alkene-bearing pyrrolysine analogues were synthesized and subsequently incorporated into proteins at two sites by a mutant PylRS–tRNA pair with excellent efficiency. This strategy allowed the site-specific labeling of proteins carrying single or double genetically encoded alkene handles via bioorthogonal thiol–ene ligation reactions.

Genetically Encoding Bioorthogonal Functional Groups for Site-selective Protein Labeling

Site-selective protein labeling is an indispensible approach in the currently intense chemical biology research area. Studies involving site-selective protein labeling span from the protein dynamic analysis in vitro to the proteinprotein interaction investigation in living cells. In the past decade, multiple methods have been introduced to achieve site-selective protein labeling. These include genetic fusion of green fluorescent protein and its derivatives, selective chemical labeling of proteins with fusion tags, and site-specific modification of noncanonical amino acids that are genetically encoded. Using evolved orthogonal aminoacyl-tRNA synthetase-nonsense suppressor tRNA pairs, noncanonical amino acids with bioorthogonal functional groups such as azide, alkyne, tetrazine, alkene, keto, phenylhalide, etc. have been genetically incorporated into proteins in E. coli, yeast, and mammalian cells. Genetic encoding of these noncanonical amino acids enables multiple ways for site-selective protein labeling both in vitro and in vivo, allowing diverse strategies to interrogate protein functions. This review intends to provide a brief introduction to the genetic noncanonical amino acid incorporation technique and recent progresses in applying this technique to achieve site-selective protein labeling.

Rational design of an evolutionary precursor of glutaminyl-tRNA synthetase

The specificity of most aminoacyl-tRNA synthetases for an amino acid and cognate tRNA pair evolved before the divergence of the three domains of life. Glutaminyl-tRNA synthetase (GlnRS) evolved later and is derived from the archaeal-type nondiscriminating glutamyl-tRNA synthetase (GluRS), an enzyme with relaxed tRNA specificity capable of forming both Glu-tRNA(Glu) and Glu-tRNA(Gln). The archaea lack GlnRS and use a specialized amidotransferase to convert Glu-tRNA(Gln) to Gln-tRNA(Gln) needed for protein synthesis. We show that the Methanothermobacter thermautotrophicus GluRS is active toward tRNA(Glu) and the two tRNA(Gln) isoacceptors the organism encodes, but with a significant catalytic preference for tRNA(Gln2)(CUG). The less active tRNA(Gln1)(UUG) responds to the less common CAA codon for Gln. From a biochemical characterization of M. thermautotrophicus GluRS variants, we found that the evolution of tRNA specificity in GlnRS could be recapitulated by converting the M. thermautotrophicus GluRS to a tRNA(Gln) specific enzyme, solely through the addition of an acceptor stem loop present in bacterial GlnRS. One designed GluRS variant is also highly specific for the tRNA(Gln2)(CUG) isoacceptor, which responds to the CAG codon, and shows no activity toward tRNA(Gln1)(UUG). Because it is now possible to eliminate particular codons from the genome of Escherichia coli, additional codons will become available for genetic code engineering. Isoacceptor-specific aminoacyl-tRNA synthetases will enable the reassignment of more open codons while preserving accurate encoding of the 20 canonical amino acids.

Designing and Engineering of a Site‐specific Incorporation of a Keto Group in Uricase

Urate oxidase is a potential therapeutic protein in the prevention and treatment of tumor lysis syndrome and hyperuricemia. However, its severe immunogenicity limits its clinical application. In our work, several strides have been made toward engineering site-specific modifications of keto groups in urate oxidase by using evolved Methanocaldococcus jannaschii aminoacyl-tRNA synthetase(s)/suppressor tRNA pairs to reduce its antigenicity. Our approach, described here, consisted of designing a M.jannaschii tyrosyl-tRNA synthetase library based on the homology modeling and molecular docking model of the species-specific TyrRS-Tyr complex. The active mutation was picked, and pBR-RS series vectors were constructed to define the relationship between the expression of aaRS and the efficiency of the orthogonal amber suppressor tRNA/synthetase system. Two sites based on the 3D structure of the Candida utilis uricase, Lys21 and Lys248, were substituted for p-acetyl-l-phenylalanine, and the yields were optimized. The products were purified, and their enzyme activities and antigenic properties were analyzed. The mutated uricase exhibited decreased antigenic properties, while its catalytic activities remained unchanged. This method imparts new insights into structure-function relationship research and provides a means by which site-specific modifications may be achieved by using PEG derivates to improve pharmacological properties of urate oxidase.

The pyrrolysine translational machinery as a genetic-code expansion tool

The discovery of pyrrolysine not only expanded the set of the known proteinogenic amino acids but also revealed unusual features of its encoding mechanism. The engagement of a canonical stop codon and a unique aminoacyl-tRNA synthetase-tRNA pair that can be used to accommodate a broad range of unnatural amino acids while maintaining strict orthogonality in a variety of prokaryotic and eukaryotic expression systems has proven an invaluable combination. Within a few years since its properties were elucidated, the pyrrolysine translational machinery has become a popular choice for the synthesis of recombinant proteins bearing a wide variety of otherwise hard-to-introduce functional groups. It is also central to the development of new synthetic strategies that rely on stop-codon suppression.

Heat Maps for Intramolecular Communication in an RNP Enzyme Encoding Glutamine

Allosteric signaling within large ribonucleoproteins modulates both catalytic function and biological specificity, but the spatial extent and quantitative magnitudes of long-distance free-energy couplings have yet to be well characterized. Here, we employ pre-steady-state kinetics to generate a comprehensive mapping of intramolecular communication in the glutaminyl-tRNA synthetase:tRNA(Gln) complex. Alanine substitution at 29 positions across the protein-RNA interface reveals distinct coupling amplitudes for glutamine binding and aminoacyl-tRNA formation on the enzyme, respectively, implying the existence of multiple signaling pathways. Structural models suggest that long-range signal propagation from the tRNA anticodon is dynamically driven, whereas shorter pathways are mediated by induced-fit rearrangements. Seven protein contacts with the distal tRNA vertical arm each weaken glutamine binding affinity across distances up to 40 Å, demonstrating that negative allosteric coupling plays a key role in enforcing the selective RNA-amino acid pairing at the heart of the genetic code.

Rational design of an orthogonal tryptophanyl nonsense suppressor tRNA

While a number of aminoacyl tRNA synthetase (aaRS):tRNA pairs have been engineered to alter or expand the genetic code, only the Methanococcus jannaschii tyrosyl tRNA synthetase and tRNA have been used extensively in bacteria, limiting the types and numbers of unnatural amino acids that can be utilized at any one time to expand the genetic code. In order to expand the number and type of aaRS/tRNA pairs available for engineering bacterial genetic codes, we have developed an orthogonal tryptophanyl tRNA synthetase and tRNA pair, derived from Saccharomyces cerevisiae. In the process of developing an amber suppressor tRNA, we discovered that the Escherichia coli lysyl tRNA synthetase was responsible for misacylating the initial amber suppressor version of the yeast tryptophanyl tRNA. It was discovered that modification of the G:C content of the anticodon stem and therefore reducing the structural flexibility of this stem eliminated misacylation by the E. coli lysyl tRNA synthetase, and led to the development of a functional, orthogonal suppressor pair that should prove useful for the incorporation of bulky, unnatural amino acids into the genetic code. Our results provide insight into the role of tRNA flexibility in molecular recognition and the engineering and evolution of tRNA specificity.

Chemical Approaches for Studying Histone Modifications

Histones form the protein core around which genomic DNA is wrapped in eukaryotic chromatin. Numerous genetic studies have established that the structure and transcriptional state of chromatin are closely related to histone post-translational modifications. Further elucidation of the precise mechanistic roles for individual histone modifications requires the ability to isolate and study homogeneously modified histones. However, the highly heterogeneous nature of histone modifications in vivo poses a significant challenge for such studies. Chemical tools that have enabled biochemical and biophysical studies of site-specifically modified histones are the focus of this minireview.

Cell-specific differences in the requirements for translation quality control

Protein synthesis has an overall error rate of approximately 10(-4) for each mRNA codon translated. The fidelity of translation is mainly determined by two events: synthesis of cognate amino acid:tRNA pairs by aminoacyl-tRNA synthetases (aaRSs) and accurate selection of aminoacyl-tRNAs (aa-tRNAs) by the ribosome. To ensure faithful aa-tRNA synthesis, many aaRSs employ a proofreading ("editing") activity, such as phenylalanyl-tRNA synthetases (PheRS) that hydrolyze mischarged Tyr-tRNA(Phe). Eukaryotes maintain two distinct PheRS enzymes, a cytoplasmic (ctPheRS) and an organellar form. CtPheRS is similar to bacterial enzymes in that it consists of a heterotetramer in which the alpha-subunits contain the active site and the beta-subunits harbor the editing site. In contrast, mitochondrial PheRS (mtPheRS) is an alpha-subunit monomer that does not edit Tyr-tRNA(Phe), and a comparable transacting activity does not exist in organelles. Although mtPheRS does not edit, it is extremely specific as only one Tyr-tRNA(Phe) is synthesized for every approximately 7,300 Phe-tRNA(Phe), compatible with an error rate in translation of approximately 10(-4). When the error rate of mtPheRS was increased 17-fold, the corresponding strain could not grow on respiratory media and the mitochondrial genome was rapidly lost. In contrast, error-prone mtPheRS, editing-deficient ctPheRS, and their wild-type counterparts all supported cytoplasmic protein synthesis and cell growth. These striking differences reveal unexpectedly divergent requirements for quality control in different cell compartments and suggest that the limits of translational accuracy may be largely determined by cellular physiology.

Functional replacement of the endogenous tyrosyl-tRNA synthetase-tRNATyr pair by the archaeal tyrosine pair in Escherichia coli for genetic code expansion.

Non-natural amino acids have been genetically encoded in living cells, using aminoacyl-tRNA synthetase–tRNA pairs orthogonal to the host translation system. In the present study, we engineered Escherichia coli cells with a translation system orthogonal to the E. coli tyrosyl-tRNA synthetase (TyrRS)–tRNATyr pair, to use E. coli TyrRS variants for non-natural amino acids in the cells without interfering with tyrosine incorporation. We showed that the E. coli TyrRS–tRNATyr pair can be functionally replaced by the Methanocaldococcus jannaschii and Saccharomyces cerevisiae tyrosine pairs, which do not cross-react with E. coli TyrRS or tRNATyr. The endogenous TyrRS and tRNATyr genes were then removed from the chromosome of the E. coli cells expressing the archaeal TyrRS–tRNATyr pair. In this engineered strain, 3-iodo-l-tyrosine and 3-azido-l-tyrosine were each successfully encoded with the amber codon, using the E. coli amber suppressor tRNATyr and a TyrRS variant, which was previously developed for 3-iodo-l-tyrosine and was also found to recognize 3-azido-l-tyrosine. The structural basis for the 3-azido-l-tyrosine recognition was revealed by X-ray crystallography. The present engineering allows E. coli TyrRS variants for non-natural amino acids to be developed in E. coli, for use in both eukaryotic and bacterial cells for genetic code expansion.

The transition from noncoded to coded protein synthesis: did coding mRNAs arise from stability-enhancing binding partners to tRNA?

BackgroundUnderstanding the origin of protein synthesis has been notoriously difficult. We have taken as a starting premise Wolf and Koonin's view that "evolution of the translation system is envisaged to occur in a compartmentalized ensemble of replicating, co-selected RNA segments, i.e., in an RNA world containing ribozymes with versatile activities".Presentation of the hypothesisWe propose that coded protein synthesis arose from a noncoded process in an RNA world as a natural consequence of the accumulation of a range of early tRNAs and their serendipitous RNA binding partners. We propose that, initially, RNA molecules with 3' CCA termini that could be aminoacylated by ribozymes, together with an ancestral peptidyl transferase ribozyme, produced small peptides with random or repetitive sequences. Our concept is that the first tRNA arose in this context from the ligation of two RNA hairpins and could be similarly aminoacylated at its 3' end to become a substrate for peptidyl transfer catalyzed by the ancestral ribozyme. Within this RNA world we hypothesize that proto-mRNAs appeared first simply as serendipitous binding partners, forming complementary base pair interactions with the anticodon loops of tRNA pairs. Initially this may have enhanced stability of the paired tRNA molecules so they were held together in close proximity, better positioning the 3' CCA termini for peptidyl transfer and enhancing the rate of peptide synthesis. If there were a selective advantage for the ensemble through the peptide products synthesized, it would provide a natural pathway for the evolution of a coding system with the expansion of a cohort of different tRNAs and their binding partners. The whole process could have occurred quite unremarkably for such a profound acquisition.Testing the hypothesisIt should be possible to test the different parts of our model using the isolated contemporary 50S ribosomal subunit initially, and then with RNAs transcribed in vitro together with a minimal set of ribosomal proteins that are required today to support protein synthesis.Implications of the hypothesisThis model proposes that genetic coding arose de novo from complementary base pair interactions between tRNAs and single-stranded RNAs present in the immediate environment.ReviewersThis article was reviewed by Eugene Koonin, Rob Knight and Berthold Kastner (nominated by Laura Landweber).

An Enhanced System for Unnatural Amino Acid Mutagenesis in E. coli

We report a new vector, pEVOL, for the incorporation of unnatural amino acids into proteins in Escherichia coli using evolved Methanocaldococcus jannaschii aminoacyl-tRNA synthetase(s) (aaRS)/suppressor tRNA pairs. This new system affords higher yields of mutant proteins through the use of both constitutive and inducible promoters to drive the transcription of two copies of the M. jannaschii aaRS gene. Yields were further increased by coupling the dual-aaRS promoter system with a newly optimized suppressor tRNA CUA opt in a single-vector construct. The optimized suppressor tRNA CUA opt afforded increased plasmid stability compared with previously reported vectors for unnatural amino acid mutagenesis. To demonstrate the utility of this new system, we introduced 14 mutant aaRS into pEVOL and compared their ability to insert unnatural amino acids in response to three independent amber nonsense codons in sperm whale myoglobin or green fluorescent protein. When cultured in rich media in shake flasks, pEVOL was capable of producing more than 100 mg/L mutant GroEL protein. The versatility, increased yields, and increased stability of the pEVOL vector will further facilitate the expression of proteins with unnatural amino acids.

Aminoacyl-tRNA Synthesis and Translational Quality Control

Translating the 4-letter code of RNA into the 22-letter alphabet of proteins is a central feature of cellular life. The fidelity with which mRNA is translated during protein synthesis is determined by two factors: the availability of aminoacyl-tRNAs composed of cognate amino acid:tRNA pairs and the accurate selection of aminoacyl-tRNAs on the ribosome. The role of aminoacyl-tRNA synthetases in translation is to define the genetic code by accurately pairing cognate tRNAs with their corresponding amino acids. Synthetases achieve the amino acid substrate specificity necessary to keep errors in translation to an acceptable level in two ways: preferential binding of the cognate amino acid and selective editing of near-cognate amino acids. Editing significantly decreases the frequency of errors and is important for translational quality control, and many details of the various editing mechanisms and their effect on different cellular systems are now starting to emerge.

One ancestor for two codes viewed from the perspective of two complementary modes of tRNA aminoacylation.

BackgroundThe genetic code is brought into action by 20 aminoacyl-tRNA synthetases. These enzymes are evenly divided into two classes (I and II) that recognize tRNAs from the minor and major groove sides of the acceptor stem, respectively. We have reported recently that: (1) ribozymic precursors of the synthetases seem to have used the same two sterically mirror modes of tRNA recognition, (2) having these two modes might have helped in preventing erroneous aminoacylation of ancestral tRNAs with complementary anticodons, yet (3) the risk of confusion for the presumably earliest pairs of complementarily encoded amino acids had little to do with anticodons. Accordingly, in this communication we focus on the acceptor stem.ResultsOur main result is the emergence of a palindrome structure for the acceptor stem's common ancestor, reconstructed from the phylogenetic trees of Bacteria, Archaea and Eukarya. In parallel, for pairs of ancestral tRNAs with complementary anticodons, we present updated evidence of concerted complementarity of the second bases in the acceptor stems. These two results suggest that the first pairs of "complementary" amino acids that were engaged in primordial coding, such as Gly and Ala, could have avoided erroneous aminoacylation if and only if the acceptor stems of their adaptors were recognized from the same, major groove, side. The class II protein synthetases then inherited this "primary preference" from isofunctional ribozymes.ConclusionTaken together, our results support the hypothesis that the genetic code per se (the one associated with the anticodons) and the operational code of aminoacylation (associated with the acceptor) diverged from a common ancestor that probably began developing before translation. The primordial advantage of linking some amino acids (most likely glycine and alanine) to the ancestral acceptor stem may have been selective retention in a protocell surrounded by a leaky membrane for use in nucleotide and coenzyme synthesis. Such acceptor stems (as cofactors) thus transferred amino acids as groups for biosynthesis. Later, with the advent of an anticodon loop, some amino acids (such as aspartic acid, histidine, arginine) assumed a catalytic role while bound to such extended adaptors, in line with the original coding coenzyme handle (CCH) hypothesis.ReviewersThis article was reviewed by Rob Knight, Juergen Brosius and Anthony Poole.

Recognition of Non-α-amino Substrates by Pyrrolysyl-tRNA Synthetase

Pyrrolysyl-tRNA synthetase (PylRS), an aminoacyl-tRNA synthetase (aaRS) recently found in some methanogenic archaea and bacteria, recognizes an unusually large lysine derivative, l-pyrrolysine, as the substrate, and attaches it to the cognate tRNA (tRNAPyl). The PylRS–tRNAPyl pair interacts with none of the endogenous aaRS–tRNA pairs in Escherichia coli, and thus can be used as a novel aaRS–tRNA pair for genetic code expansion. The crystal structures of the Methanosarcina mazei PylRS revealed that it has a unique, large pocket for amino acid binding, and the wild type M. mazei PylRS recognizes the natural lysine derivative as well as many lysine analogs, including Nɛ-(tert-butoxycarbonyl)-l-lysine (Boc-lysine), with diverse side chain sizes and structures. Moreover, the PylRS only loosely recognizes the α-amino group of the substrate, whereas most aaRSs, including the structurally and genetically related phenylalanyl-tRNA synthetase (PheRS), strictly recognize the main chain groups of the substrate. We report here that wild type PylRS can recognize substrates with a variety of main-chain α-groups: α-hydroxyacid, non-α-amino-carboxylic acid, Nα-methyl-amino acid, and d-amino acid, each with the same side chain as that of Boc-lysine. In contrast, PheRS recognizes none of these amino acid analogs. By expressing the wild type PylRS and its cognate tRNAPyl in E. coli in the presence of the α-hydroxyacid analog of Boc-lysine (Boc-LysOH), the amber codon (UAG) was recoded successfully as Boc-LysOH, and thus an ester bond was site-specifically incorporated into a protein molecule. This PylRS–tRNAPyl pair is expected to expand the backbone diversity of protein molecules produced by both in vivo and in vitro ribosomal translation.

High‐level cell‐free synthesis yields of proteins containing site‐specific non‐natural amino acids

We describe an E. coli-based cell-free system for the production of proteins with a non-natural amino acid (nnAA) incorporated site-specifically (modified protein). The mutant Methanococcus jannaschii tyrosyl-tRNA synthetase (mTyrRS) and tRNA(Tyr) pair were used as orthogonal elements. The mTyrRS experienced proteolysis and modified protein yields improved with higher synthetase addition (200-300 microg/mL). Product yields were also improved by increasing levels of total protein to 20 mg protein/mL and available vesicle surface area to 0.5 m(2)/mL. This new E. coli-based cell-free procedure produced up to 400 microg/mL of eCAT109pAz, 660 microg/mL of eDHFR10pAz, and 210 microg/mL of mDHFR31pAz with p-azido-L-phenylalanine (pAz) incorporated site-specifically at the amber nonsense codon. O-methyl-L-tyrosine and p-acetyl-L-phenylalanine were incorporated by similar protocols. The desired specificity for incorporation of the nnAA by the cell-free system was confirmed. Additionally, the modified proteins were enzymatically active and reactive for copper(I)-catalyzed (3 + 2) cycloadditions (click chemistry).

Transplantation of a tyrosine editing domain into a tyrosyl-tRNA synthetase variant enhances its specificity for a tyrosine analog

To guarantee specific tRNA and amino acid pairing, several aminoacyl-tRNA synthetases correct aminoacylation errors by deacylating or "editing" misaminoacylated tRNA. A previously developed variant of Escherichia coli tyrosyl-tRNA synthetase (iodoTyrRS) esterifies or "charges" tRNA(Tyr) with a nonnatural amino acid, 3-iodo-l-tyrosine, and with l-tyrosine less efficiently. In the present study, the editing domain of phenylalanyl-tRNA synthetase (PheRS) was transplanted into iodoTyrRS to edit tyrosyl-tRNA(Tyr) and thereby improve the overall specificity for 3-iodo-l-tyrosine. The beta-subunit fragments of the PheRSs from Pyrococcus horikoshii and two bacteria were tested for editing activity. The isolated B3/4 editing domain of the archaeal PheRS, which was exogenously added to the tyrosylation reaction with iodoTyrRS, efficiently reduced the production of tyrosyl-tRNA(Tyr). In addition, the transplantation of this domain into iodoTyrRS at the N terminus prevented tyrosyl-tRNA(Tyr) production most strongly among the tested fragments. We next transplanted this archaeal B3/4 editing domain into iodoTyrRS at several internal positions. Transplantation into the connective polypeptide in the Rossmann-fold domain generated a variant that efficiently charges tRNA(Tyr) with 3-iodo-l-tyrosine, but hardly produces tyrosyl-tRNA(Tyr). This variant, iodoTyrRS-ed, was used, together with an amber suppressor derived from tRNA(Tyr), in a wheat germ cell-free translation system and incorporated 3-iodo-l-tyrosine, but not l-tyrosine, in response to the amber codon. Thus, the editing-domain transplantation achieved unambiguous pairing between the tRNA and the nonnatural amino acid in an expanded genetic code.

Polyphasic evidence delineating the root of life and roots of biological domains

Twenty different lines of polyphasic evidence obtained from tRNA and protein sequences, anticodon usages, gene contents, metabolism and geochemistry have made possible the identification of a Last Universal Common Ancestor (LUCA) phylogenetically located proximal to the hyperthermophilic methanogenic archaeon Methanopyrus. Combined with analysis of high-similarity cross-domain tRNA pairs, the evidence also suggests a Thermotoga-proximal Last Bacterial Common Ancestor (LBACA) that originated from Crenarchaeota close to Aeropyrum, and a Plasmodium-proximal Last Eukaryotic Common Ancestor (LECA) derived from Ferroplasma through endosymbiosis.

Origin of the Genetic Code: First Aminoacyl–tRNA Synthetases Could Replace Isofunctional Ribozymes When Only the Second Base of Codons Was Established

Analysis of the updated compilation of more than 8,000 tRNA gene sequences confirmed our previously reported finding that in pairs of consensus tRNAs with complementary anticodons, their second bases in the acceptor stems are also complementary. This dual complementarity points to the following: (1) the operational code embodied in the acceptor stem, and the classic genetic code embodied in the anticodon could have had the same common ancestor; (2) new tRNAs most likely entered primitive translation in pairs with complementary anticodons; and (3) this process of code expansion was directed by the primordial double-strand coding. However, we did not find the dual complementarity when testing all tRNA pairs in which anticodons were complementary only at the central position, but not complementary at least at one of the flanking two positions. This observation, together with certain additional evidence, suggests that both codes were still being shaped (with only the second base established at the time) when the first protein aminoacyl-tRNA synthetases could have already started replacing their ribozymic precursors.

Complete set of orthogonal 21st aminoacyl-tRNA synthetase-amber, ochre and opal suppressor tRNA pairs: concomitant suppression of three different termination codons in an mRNA in mammalian cells.

We describe the generation of a complete set of orthogonal 21st synthetase-amber, ochre and opal suppressor tRNA pairs including the first report of a 21st synthetase-ochre suppressor tRNA pair. We show that amber, ochre and opal suppressor tRNAs, derived from Escherichia coli glutamine tRNA, suppress UAG, UAA and UGA termination codons, respectively, in a reporter mRNA in mammalian cells. Activity of each suppressor tRNA is dependent upon the expression of E.coli glutaminyl-tRNA synthetase, indicating that none of the suppressor tRNAs are aminoacylated by any of the twenty aminoacyl-tRNA synthetases in the mammalian cytoplasm. Amber, ochre and opal suppressor tRNAs with a wide range of activities in suppression (increases of up to 36, 156 and 200-fold, respectively) have been generated by introducing further mutations into the suppressor tRNA genes. The most active suppressor tRNAs have been used in combination to concomitantly suppress two or three termination codons in an mRNA. We discuss the potential use of these 21st synthetase-suppressor tRNA pairs for the site-specific incorporation of two or, possibly, even three different unnatural amino acids into proteins and for the regulated suppression of amber, ochre and opal termination codons in mammalian cells.