Pyrrolysine Encoded by UAG in Archaea: Charging of a UAG-Decoding Specialized tRNA

See allHide authors and affiliations

Science  24 May 2002:
Vol. 296, Issue 5572, pp. 1459-1462
DOI: 10.1126/science.1069588

This article has a correction. Please see:


Pyrrolysine is a lysine derivative encoded by the UAG codon in methylamine methyltransferase genes of Methanosarcina barkeri. Near a methyltransferase gene cluster is thepylT gene, which encodes an unusual transfer RNA (tRNA) with a CUA anticodon. The adjacent pylS gene encodes a class II aminoacyl-tRNA synthetase that charges the pylT-derived tRNA with lysine but is not closely related to known lysyl-tRNA synthetases. Homologs of pylS and pylT are found in a Gram-positive bacterium. Charging a tRNACUA with lysine is a likely first step in translating UAG amber codons as pyrrolysine in certain methanogens. Our results indicate that pyrrolysine is the 22nd genetically encoded natural amino acid.

In Methanosarcinaspecies, specific methyltransferases initiate methanogenesis and carbon assimilation from substrates such as trimethylamine (TMA), dimethylamine (DMA), or monomethylamine (MMA). The highly expressed, nonhomologous genes encoding these methyltransferases have in-frame UAG (amber) codons that do not stop translation during synthesis of the full-length proteins (1–3). Nearly identical copies of the DMA and MMA methyltransferase genes with conserved single in-frame amber codons occur in the same genome (2,4). There is no evidence for transcript editing (2, 4), and, unlike many other stop codon readthrough events (5), readthrough of the amber codons is highly efficient (3). Amber serves as a sense codon within the methylamine methyltransferase genes (3), previously unknown in any other group of archaeal genes. In the accompanying manuscript, Hao et al. have shown that in intact MtmB the amber-encoded residue is pyrrolysine, whose structure is proposed as lysine with its epsilon nitrogen in amide linkage with (4R,5R)-4-substituted-pyrroline-5-carboxylate. Here, we describe a specialized tRNACUA and lysyl-tRNA synthetase (LysRS) that underlie amber codon translation as pyrrolysine in certain methane-producing Archaea.

An unannotated gene, pylT (Fig. 1), whose predicted tRNA product has a CUA anticodon, was identified in theMethanosarcina barkeri Fusaro genomic database (GenBank accession number NC_002724) using tRNAScanSE (6,7). Sequencing of M. barkeri MS DNA (7) also revealed pylT, as well as the three following open reading frames, pylS, pylB, and pylC (GenBank AY064401). Northern blots of the RNA pool from MMA-grown M. barkeri MS revealed an RNA of the size expected for the tRNACUA product of pylT (Fig. 1C). The predicted secondary structure of tRNACUA has unusual properties compared with typical tRNAs (8). Even though the structure has the expected sizes for the acceptor, D, and T stems and D, T, and anticodon loops, the anticodon stem could form with six, rather than five, base pairs. This would constrain the variable loop to only three, rather than four, bases. However, if the anticodon stem has five base pairs, two bases are found between the D and anticodon stems, also atypical of most tRNA structures. The predicted secondary structure of tRNACUA has only one base, rather than the typical two bases, between the acceptor and D stems. Many of the conserved bases in tRNAs are found, but not the almost universally conserved GG sequence in the D loop or the TψC sequence in the T loop.

Figure 1

The pyl gene cluster and RNA products in M. barkeri. (A) Location of thepyl cluster in M. barkeri Fusaro. The genes were found on contig 1956, GenBank NC_002724. The functions of the variousmtm genes are in Burke et al. (1). TheramM gene was recently found to be involved in activation of methylamine methyltransferase corrinoid proteins (28). The location of in-frame amber codons (TAG) in each mtmBgene copy is indicated. The line underneath the pylT gene cluster indicates the corresponding sequence obtained from M. barkeri MS. (B) The deduced secondary structure of the amber suppressing tRNA from M. barkeri FusaropylT. Arrows indicate base changes found in M. barkeri MS pylT gene product. Bases that are highly conserved (boxed) or semiconserved (circled) in most tRNAs are indicated. The potential sixth base pair in the anticodon stem is denoted by two dashes between the bases. (C) Northern blot of 4 μg of the total tRNA pool from M. barkeri MS. The RNA was electrophoresed through an 8% polyacrylamide-8 M urea gel, then blotted onto Nytran membrane (Schleicher and Schuell, Keene, NH) before probing. (D) Northern blot of total RNA pool made from a 1% agarose electrophoretic gel showing larger transcript frompylT gene cluster. Both northern blots were made by probing with a 72-base synthetic DNA oligonucleotide probe complementary to the predicted RNA transcribed from pylT following the techniques in (2). To the side of each blot, the migration of size standards of indicated base sizes is shown.

A 4.2-kilobase transcript was detectable with probes forpylT, pylS, pylB, and pylC(Fig. 1D), indicating possible cotranscription of these genes. ThepylS gene has a predicted product similar to the core catalytic domains of class II aminoacyl-tRNA synthetase (AARS) enzymes containing motifs 1 through 3 (9, 10). An RPS-BLAST (11) search for conserved domains found in the Pfam database (12) maintained at the National Center for Biotechnology Information produced alignments of different portions of the predicted PylS sequence with class II AARS from three different subclasses (13, 14). To test the activity of PylS as an AARS, the M. barkeri pylS gene was expressed in Escherichia coli (7) as a 49-kD protein with an NH2-terminal hexahistidine tag, which was then isolated by nickel-affinity chromatography (Fig. 2A). The recombinant PylS was tested for LysRS activity, as pyrrolysine is a lysine derivative. PylS ligated [14C]lysine, but not [3H]phenylalanine or [14C]histidine, to tRNA in the cellular pool isolated from M. barkeri (7, 15). Addition of [12C]lysine, but not the other 19 amino acids, diluted [14C]lysine ligation to tRNA (Fig. 2B). A protein with LysRS activity could not be isolated by nickel-affinity chromatography of extracts of E. coli not carrying pylS.

Figure 2

LysRS activity of the pylS gene product from M. barkeri MS. (A) Coomassie-stained SDS-polyacrylamide gel (29) of (lane 1), cell extract (15 μg protein) of E. coli transformed with pXRS carryingpylS (7); (lane 2), recombinant purified PylS (5 μg protein) found in imidazole eluate of the HiTrap column of the same E. coli cell extract; and (lane 3), an equivalent volume of the imidazole eluate fraction, but from the same E. coli strain transformed with plasmid lacking pylS(7). Numbers to left of gel indicate molecular mass of marker proteins in kDa, arrow at bottom indicates location of the dye front. (B) Charging of tRNA in the total tRNA fraction fromM. barkeri MS using [14C]lysine. The complete reaction mixture (100 μl volume) contained 20 nM PylS, 50 mM KCl, 10 mM MgCl2 , 5 mM ATP, 5 mM dithiothreitol (DTT), 20 μM UL-14C-labeled lysine (717 dpm/pmol) in 100 mM Hepes buffer, pH 7.2, with 330 μg crude M. barkeri tRNA preparation. At the indicated timepoints [14C]lys-tRNA formation was assayed by filter assay as described in (18). Reactions illustrated are complete (diamonds), complete minus PylS (open circles), minus tRNA (open squares), plus 1 mM [12C]lysine (open triangles), or plus a mixture of 1 mM each of the 20 canonical nonradioactive amino acids except lysine (filled circles). To illustrate isotope dilution of the product, the specific activity of 717 dpm/pmol lysine was used to calculate pmol lys-tRNA formed in all reactions. (C) Charging of tRNACUA with lysine. The conditions and symbols are the same as in the previous figure, but with cellular tRNA replaced with 40 μM in vitro transcribed tRNACUA.

In the presence of a saturating level of the tRNA pool from M. barkeri , the apparent K m of PylS for lysine was 2.2 μM, and the apparent k cat was 1.6 min−1. Most measured K m values for class II LysRS are equivalent or higher, butk cat values typically have ranged from 12 to 7800 min−1 (16). Thek cat value is closer to those for recombinant class I LysRS from Borrelia burgdorferi andMethanococcus maripaludis that are 4 min−1 and 47 min−1, respectively (17). To test whether PylS could use tRNACUA as a substrate, an 89–base pair double-stranded oligonucleotide with a T7 polymerase promoter spaced in front of the pylT gene was used as a template to produce a tRNACUA transcript. PylS charged tRNACUA with lysine at a rate of 0.3 pmol/min per pmol PylS. Controls also indicated this reaction was specific for lysine (Fig. 2C).

LysRS enzymes are the only type of AARSs with nonhomologous representatives that are either class I or class II enzymes (18). PylS adds to this diversity, because its sequence does not indicate a ready grouping with the known class II LysRS, all of which are in subclass IIb (13). An alignment of the catalytic core of PylS with other class II LysRS (Fig. 3A) reveals that PylS contains those residues binding Mg–adenosine triphosphate (MgATP) in class IIb LysRS, but not those binding solely lysine (19). BlastP alignments with the PylS sequence and the NH2-terminal tRNA binding domain of LysRS from different sources did not produce any significant alignments. If the oligosaccharide/oligonucleotide binding (OB) fold responsible for binding the anticodon of the tRNA in known class II LysRS (20) is present in PylS, the primary sequence is highly divergent. PylS may represent a new subclass of class II AARS and also represents the third LysRS identified in M. barkeri. Class I and class II LysRS genes (GenBank AF337056 andAF337055, respectively), which produce active LysRS enzymes (21), have also been identified.

Figure 3

Sequence alignments with PylS. (A) Alignment by the program ClustalW (30) using the Blossum62 matrix (31) of the core catalytic domains (spanning motif 1 to 3) of LysRS from E. coli (Ec, GenBankaccession number AAG58018.1, LysS), human (AAH04132.1), andSaccharomyces cerevisiae (Sc, NP_010322.1), as well as PylS from M. barkeri MS (MbPylS) and PylSc fromDesulfitobacterium hafniense (dhPylSc). The blue letters indicate positive scores in the Blosum matrix; the red letters indicate identical residues, as individual sequences are compared with the Ec sequence as reference. In order to highlight similarity between only PylS and PylSc, conserved (+) and identical (*) residues between those two sequences are indicated below the alignment. Residues in motif 1, 2, or 3 are indicated by gray shading. Other shadings of residues indicate conserved resi-dues involved in: dimerization(light blue); lysine binding (yellow); ATP binding (green); or both lysine and ATP binding (dark blue) as seen from the E. coli LysRS crystal structures (19). (B) BLASTP alignment of portions of the predicted NH2-terminal sequence of PylS from M. barkeri MS with a portion of D. hafniense PylSn (dhPylSn).

A search of the available genome sequences revealed an mttBhomolog, the gene encoding TMA methyltransferase in M. barkeri (2), in the Gram-positive bacteriumDesulfitobacterium hafniense. The mttB genes from both organisms have corresponding single in-frame amber codons. The genome also encodes homologs of the M. barkeritRNACUA and its cognate LysRS (22). The bacterial tRNACUA shares the unusual features of theM. barkeri tRNACUA, including the ability to form a six–base pair acceptor stem, as well as one base between the D and acceptor stems (fig. S1). However, in this bacterium, two distinct reading frames encode homologs of the NH2- and COOH-termini of PylS (figs. S1 and S2 and Fig. 3). The D. hafniense pylScgene follows the pylT gene and encodes a predicted product homologous to the COOH-terminus of PylS, containing the core catalytic domain of the protein. A putative protein homologous to the NH2-terminus of PylS is encoded by a short open reading frame (pylSn) found downstream of pylSc. This apparent splitting of the NH2-terminus of PylS from its catalytic domain in D. hafniense is reminiscent of the encoding of the two domains of methionyl-tRNA synthetase (23) by separate genes in certain Eucarya and Crenarchaeota (24).

Three predicted D. hafniense genes between pylScand pylSn are homologs of the M. barkeri pylB,pylC, and pylD genes (figs. S1 and S2). The presence of these genes near pylT in two phylogenetically distinct organisms may reflect conserved gene function associated with decoding amber codons. A bioinformatics approach suggests potential roles for pyl genes in the synthesis of pyrrolysine (25).

The presence of tRNACUA and an unusual cotranscribed cognate LysRS indicates a mechanism in which pyrrolysine is inserted into MtmB during translation. Because PylS acts as a high-affinity LysRS, lys-tRNACUA is a likely first intermediate in formation of pyrrolysine. Testing of pyrrolysine itself as a PylS substrate must await the chemical synthesis of this amino acid, but the low apparent K m of PylS for lysine would render the charging of tRNACUA with pyrrolysine problematic under cellular conditions where the two amino acids would be in competition. The lysyl group on tRNACUA could be modified to pyrrolysine either before or during incorporation into the protein. However, the cotranslational modification of lysine following incorporation into MtmB as signaled by a stop codon would be an unprecedented phenomenon, possibly involving recruitment of modification enzymes to the ribosome during translation. Rather, synthesis of pyrrolysyl-tRNACUAcould be achieved by condensation of the epsilon nitrogen of lys-tRNACUA and carboxyl group of (4R,5R)-4-substituted-pyrroline-5-carboxylate, allowing direct translation of the UAG codon as pyrrolysine. This would have precedent in the biosynthesis of asparagine, glutamine, formylmethionine, or selenocysteine on different tRNA species, as recently reviewed in (26). A strong analogy can now be made between pyrrolysine and selenocysteine. Both of these noncanonical amino acids are found at positions encoded by canonical stop codons. Decoding UGA as selenocysteine involves tRNAUCA species with marked deviations from typical tRNA structure, such as an elongated acceptor stem (27). Taken together, the present data suggest that pyrrolysine represents the 22nd genetically encoded amino acid to be identified in nature.

  • * To whom correspondence should be addressed. E-mail: Krzycki.1{at}


View Abstract

Stay Connected to Science

Navigate This Article