|Home | About | Journals | Submit | Contact Us | Français|
In Methanosarcina spp., amber codons in methylamine methyltransferase genes are translated as the 22nd amino acid, pyrrolysine. The responsible pyl genes plus amber-codon containing methyltransferase genes have been identified in four archaeal and five bacterial genera, including one human pathogen. In E. coli, the recombinant pylBCD gene products biosynthesize pyrrolysine from two lysine and the pylTS gene products direct pyrrolysine incorporation into protein. In the proposed biosynthetic pathway, PylB forms methylornithine from lysine, which is joined to another lysine by PylC, and oxidized to pyrrolysine by PylD. Structures of the catalytic domain of pyrrolysyl-tRNA synthetase (archaeal PylS or bacterial PylSc) revealed binding sites for tRNAPyl and pyrrolysine. PylS and tRNAPyl are now being exploited as an orthogonal pair in recombinant systems for introduction of useful modified amino acids into proteins.
Methanogenesis is a process unique to the Archaea. Befittingly, many unusual enzymes, cofactors, and metabolites were first encountered in methanogens and only later in Bacteria . One such find is pyrrolysine, the 22nd amino acid to be encountered in the natural genetic code [2,3].
In Archaea, the pyrrolysyl residue is known to occur only in the family Methanosarcinaceae. Unlike most other methanogens, members of this group can use methylamines as precursors to methane. Metabolism of trimethylamine, dimethylamine, or monomethylamine is respectively initiated by the pyrrolysine-containing proteins MttB, MtbB, or MtmB (Fig. 1) [4–6]. These proteins each methylate a cognate corrinoid protein (MttC, MtbC, or MtmC) in the Co(I) state. Adventitious oxidation of the corrinoid proteins to Co(II) can inactivate the methyltransferase reactions, but the iron-sulfur protein RamA can reactive the corrinoid proteins via ATP dependent reduction . The methyl-Co(III) corrinoid proteins are substrates of MtbA which methylates coenzyme M (CoM) [8,9]. From here, methyl groups can be converted directly to carbon dioxide, and cell carbon.
MttB, MtbB, or MtmB have no significant sequence similarity, but each of their genes contains a single in-frame amber codon [10,11] that is translated [12,13]. The crystal structure of MtmB revealed pyrrolysine as the UAG-encoded residue [3,14]. Mass spectral studies demonstrated that the UAG-encoded residues of MttB and MtbB are also pyrrolysine . Pyrrolysine was observed in the crystal structure to bind ammonia at the carbon of the imine bond [3,14], and it is hypothesized that a pyrrolysine-methylammonium adduct serves to activate and orient methylamines as substrates for nucleophilic attack by the Co(I) corrinoid protein (Fig. 1) [3,15].
Pyrrolysine is synthesized and incorporated into the methylamine methyltransferases through the combined actions of the products of the pyl genes (Fig. 1). The pylT gene encodes tRNAPyl whose CUA anticodon allows for amber codon translation [2,16]. The pylS gene produces the pyrrolysyl-tRNA synthetase that charges tRNAPyl directly with pyrrolysine [17,18]. The synthesis of pyrrolysine is carried out by the pylBCD gene products .
Incorporation of pyrrolysine into protein under the direction of amber codons does not require specific signals in the gene, as in-frame amber codons inserted into the E. coli uidA gene for β-glucuronidase are translated as pyrrolysine at 20–30% efficiency in M. acetivorans [20,21]. Translation of UAG as pyrrolysine even in foreign genes with an introduced amber codon suggests this level of translation occurs by a mechanism analogous to that underlying amber suppression, with tRNAPyl acting as a suppressor tRNA. On the other hand, UAG translation in the methanogen with mtmB1 transcripts appears much more efficient, with little UAG-termination product detectable. Substitution of sequence immediately downstream of the UAG codon dramatically increases the UAG-termination product, but with still a relatively high level of UAG translation as pyrrolysine. While the deleted region may act as a sequence enhancing translation over termination with mtmB1 transcripts, previous suggestions from several laboratories of an obligate pyrrolysine insertion element or tRNAPyl specific translation factors have not been borne out [20–24].
At the time of the discovery of pyrrolysine, only Methanosarcina spp. and the bacterium Desulfitobacterium hafniense were known to possess pyl genes . Recent sequencing has now expanded the count to six bacterial and six archaeal species belonging to nine genera (Fig. 2). In the archaea, pyl genes are still limited to members of the Methanosarcinacea, but now also include psychrotrophic methanogen Methanococcoides burtonii, and two halophilic methanogens, Methanohalophilus mahii and Methanohalobium evestigatum. The pyl genes are also found in select species of the bacterial groups Clostridia or the Deltaproteobacteria. The latter group includes symbionts of multicellular organisms, such as a gut inhabitant of a marine worm identified in a metagenomic study [25,26]. Most recently, pyl genes were annotated in the genome of the human intestinal bacteria, Bilophila wadsworthia (Fig. 2). The unannotated pylT gene was identified by the authors some distance from the other pyl genes (Fig. 2, Fig. S1). This organism has been a common isolate in cases of gangrenous appendicitis and abscesses in a variety of bodily locations , and represents the first known human symbiont, and pathogen, to have pyrrolysine genes.
While amber codons have been identified in Thg1 and transposase genes of one or two Methanosarcina spp., these are likely to be mutations stable in the context of tRNAPyl and PylS, as homologs of these genes without amber codons exist in other Methanosarcinacea [21,28,29]. In sharp contrast, each of the organisms possessing the five pyl genes also possess one or more homologs of mtmB, mtbB, and/or mttB with conserved in-frame amber codons, as well as the genes for associated proteins such as the corrinoid proteins and RamA (Fig. 2). Each organism is an anaerobic respirer found where methylamines or their precursors are available. The correlation runs both ways, as no amber-codon containing methylamine methyltransferase is found in any genome lacking the five pyl genes. The Desulfobacterium autotrophicum genome is interesting in this regard, as pylSn, pylT, and pylB, are present, but not pylSc, pylC, or pylD. A transposase gene homolog within the remaining pylTSnB cluster suggests the pyl gene cluster was disrupted . Accordingly, although many mttB homologs are found in this genome, all lack an in frame amber codon.
Such homologs of methyltransferase genes without amber codons are found in many genomes. Particularly numerous are mttB non-amber homologs. A recent BLASTP search by the authors in ~1500 microbial genomes kept at NCBI revealed 278 mttB homologs in sequenced genomes. A similarity tree shows the bacterial and archaeal MttB homologs whose genes have amber codons form a single clade not following the rRNA phylogenetic tree (J. Krzycki, unpublished data). The pyrrolysine-containing MttB are well separated from the great majority of MttB homologs whose encoding genes lack the pyrrolysine amber codon. However, one clade of MttB homologs without pyrrolysine branches very close to the pyrrolysine-containing MttB family. This raises the question as to whether such MttB homologs without pyrrolysine are actually TMA methyltransferases, or homologs evolved to use, for example, other N-methyl substrates.
The archaeal pyrrolysyl-tRNA synthetase is a homodimer whose subunits have two functional domains. The C-terminal domain contains the three motifs of the class-II aminoacyl tRNA synthetases [2,30,31], while the N-terminal domain is not similar to proteins with known function . In bacteria, N- and C-terminal domains of archaeal PylS are respectively represented by two independent proteins, PylSn and PylSc (Fig. 2) [2, 32].
Crystal structures of the C-terminal domain of archaeal PylS and bacterial PylSc have revealed the catalytic site that accommodates pyrrolysine and ATP [30,31,33–35]. The pyrrolysine ring is accommodated by a hydrophobic pocket closed by a mobile loop bearing a tyrosine which may H-bond the imine nitrogen of pyrrolysine , and/or provide stability to the formed pyrrolysyl-adenylate prior to tRNA binding . While pyrrolysine has the most favorable kinetics for amino acid activation, analogs having an oxygen atom replacing the imine nitrogen are favored over those with carbon, suggesting H-bonding to the imine nitrogen could play a role in substrate binding . Loss of the loop tyrosine does not inactivate the enzyme, but kinetic parameters have not yet been determined .
Even with the limited sample size of pyl containing organisms, considerable diversity is apparent in tRNAPyl with only 26 residues universally conserved (Fig. 3). While the sequences of the four archaeal tRNApyl homologs from the Methanosarcinaea are 87% identical, the five bacterial tRNAPyl molecules are more divergent, exhibiting only ~ 45% sequence identity. All tRNAPyl examples share unusual secondary structure features, including a three-base variable loop, small D-loop, long anticodon stem, and a single base between the D- and acceptor stems [2,35,37]. Nonetheless, tRNAPyl assumes the typical tRNA L-shaped tertiary structure, albeit with a smaller core [35,37]. In the co-crystal structure , PylSc approaches tRNAPyl from the major groove interacting with the D-stem residues in the core as well as the acceptor stem (Fig. 3).
PylSc does not have direct interaction with residues in the anticodon stem or T-stem . However, mutations in the T-stem and residues flanking the anticodon have relatively strong effects on in vitro aminoacylation of the M. barkeri tRNAPyl by PylS, in the apparent absence of a structural role for the targeted residues . In further contrast to archaeal PylS, PylSc has no detectable in vivo activity in E. coli in all but the most sensitive assays for amber suppression [35,39], and PylSc binds cognate tRNAPyl with relatively low affinity. A possible resolution to this conundrum is that PylSc lacks the N-terminal domain of the methanogen enzyme, represented by bacterial PylSn [2,32]. As a homolog of the archaeal PylS N-terminal domain, PylSn may be responsible for strengthened interaction with tRNAPyl.
Amber codons in E. coli bearing pylTS can translate UAG using exogenous pyrrolysine . This suggested the possible utility of PylS and tRNAPyl as an orthogonal pair, and this potential has now been demonstrated in bacterial , yeast , and mammalian cells [42,43]. The hydrophobic pocket in PylS can accommodate other moieties, so long as they are in amide linkage to εN of lysine, leading to notable substrate flexibility . Either wild type or mutated PylS have been employed to insert unnatural residues such as: acetyl-Nε-lysine, with potential applications in histone research ; Nε-(o-azidobenzyloxycarbonyl)-L-lysine, allowing fluorescent tagging ; tetrahydrofuran-εN-lysine  modified with an alkyne group, allowing modification for FRET analysis ; D-cysteinyl-εN -lysine, allowing site-specific chemical ubiquitination of protein , and finally α–hydroxy acid derivatives, inserting a backbone ester bond for easy hydrolysis . Most recently, PylS was used to incorporate a photocaged lysine derivative into p53 in a mammalian cell line, giving control over the timing of a specific protein’s transport into an organelle .
The relationships of PylB, PylC, and PylD to enzymes with known metabolic functions fueled the idea they functioned in pyrrolysine biosynthesis . This has proved the case, as E. coli expressing archaeal pylTSBCD can incorporate biosynthesized pyrrolysine into protein, with pylBCD required to make the PylS substrate .
Hypothetical pathways for pyrrolysine biosynthesis generally considered amide formation between the εN of lysine and a ring or precursor derived from ornithine, glutamate, isoleucine, or proline [15,20,47]. Stable isotope labeling experiments using the E. coli recombinant pyl system did indeed show that the acyl portion of pyrrolysine derives from lysine . However, the methylated pyrroline ring also derives from lysine. All six carbons of two molecules of lysine are retained in pyrrolysine. One εN is lost from the two lysine molecules, presumably as one lysine becomes the ring precursor .
How is lysine converted into the methylated pyrroline ring? An important clue is that addition of D-ornithine allows pyl transformed E. coli to make desmethylpyrrolysine (dmPyl), a pyrrolysine analog lacking the ring methyl group . Formation of dmPyl requires only PylC and PylD. PylC carries out ligation of D-ornithine to the terminal amine of lysine, as cells transformed with pylC produce D-ornithyl-εN-lysine dependent on exogenous D-ornithine. PylC has similarity to members of the carbamoyl phosphate synthetase family, including D-ala, D-ala ligase, and amide formation is in keeping with this phylogeny (Fig. S2) . PylD is similar to several different dehydrogenases binding FAD or NAD (Fig. S3), and may carry out oxidation of the terminal amine of the PylC product, leading to dehydration and formation of the dmPyl ring .
However, synthesis of pyrrolysine itself is not D-ornithine dependent, but instead requires PylB, in addition to PylC and PylD. This suggests that PylB produces a D-ornithine derivative (Fig. 1). PylB is related to members of the radical S-adenosyl-L-methionine (SAM) family (Fig. S4) , and possesses the characteristic CXXXCXXC motif binding the Fe4S4 cluster that initiates reactions through the reductive cleavage of SAM to generate a 5′deoxyadenosine radical . Subsequent hydride abstraction from the substrate with radical SAM enzymes can lead to difficult reductions or mutase reactions, depending on the enzyme. For PylB, we hypothesize that exchange of the glycyl radical and a hydride between the beta and gamma carbons of lysine could create R,R-3-methylornithine, which is ligated to another lysine by PylC then oxidzed by PylD to form pyrrolysine (Fig. 1) . The proposed lysine mutase reaction of PylB would be the first time a radical SAM enzyme has been implicated in this type of mutase reaction involving carbon backbone rearrangement. Analogous reactions are performed by coenzyme B12 enzymes, such as glutamate mutase, or methylmalonyl-CoA mutase .
Much has been learned about pyrrolysine since it discovery, but certain areas remain relatively unexplored. How pyrrolysine interacts with PylS for ligation to tRNAPyl is now better understood, but the role of the N-terminal domain of PylS (and PylSn) in in vivo charging of tRNAPyl remains unknown. The recent proposal of the first empirically based pathway of pyrrolysine biosynthesis must be coupled to in vitro demonstration of the reactions and properties of PylB, PylC, and PylD. Such studies may provide a new route to useful genetically encoded unnatural residues produced from modified biosynthetic precursors. Finally, the function of pyrrolysine in methylamine methyltransferase remains largely a hypothesis, and direct tests of the proposed mechanism and alternatives are highly desirable. In a similar vein, investigation of homologs of the methanogen methylamine methyltransferases lacking pyrrolysine may help to determine if pyrrolysine is essential for the function as a methylamine methyltransferase. The answers to such questions as these may provide insight into the forces that have brought and maintained pyrrolysine in the genetic code.
The work in our laboratory is supported by National Institutes of Health Grant GM070663 and Department of Energy Grant DE-FG0202-91ER200042. The authors wish to thank Charles Daniels and John Reeve for helpful discussions.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.