Pyrrolysine is not hardwired for cotranslational insertion at UAG codons

Abstract

Pyrrolysine (Pyl), the 22nd naturally encoded amino acid, gets acylated to its distinctive UAG suppressor tRNA^Pyl by the cognate pyrrolysyl-tRNA synthetase (PylRS). Here we determine the RNA elements required for recognition and aminoacylation of tRNA^Pyl in vivo by using the Pyl analog N-ε-cyclopentyloxycarbonyl-l-lysine. Forty-two Methanosarcina barkeri tRNA^Pyl variants were tested in Escherichia coli for suppression of the lac amber A24 mutation; then relevant tRNA^Pyl mutants were selected to determine in vivo binding to M. barkeri PylRS in a yeast three-hybrid system and to measure in vitro tRNA^Pyl aminoacylation. tRNA^Pyl identity elements include the discriminator base, the first base pair of the acceptor stem, the T-stem base pair G51:C63, and the anticodon flanking nucleotides U33 and A37. Transplantation of the tRNA^Pyl identity elements into the mitochondrial bovine tRNA^Ser scaffold yielded chimeric tRNAs active both in vitro and in vivo. Because the anticodon is not important for PylRS recognition, a tRNA^Pyl variant could be constructed that efficiently suppressed the lac opal U4 mutation in E. coli. These data suggest that tRNA^Pyl variants may decode numerous codons and that tRNA^Pyl:PylRS is a fine orthogonal tRNA:synthetase pair that facilitated the late addition of Pyl to the genetic code.

Incorporation of noncanonical amino acids into proteins is an exciting and active research field. To date, >30 unnatural amino acids have been placed into proteins with high fidelity mostly directed by the amber codon UAG (1, 2). The other two termination codons as well as enlarged codons (with 4–6 bases) have also been used (e.g., refs. 3 and 4). The key step in this process is the introduction of an orthogonal tRNA:aminoacyl-tRNA synthetase pair into the host protein synthesizing system. Such an orthogonal tRNA should not be recognized by any endogenous aminoacyl-tRNA synthetase, whereas the orthogonal synthetase should acylate solely the orthogonal tRNA with the unusual amino acid.

Less attention was devoted to incorporation of the noncanonical amino acids selenocysteine (Sec) and pyrrolysine (Pyl) that arose from natural expansion of the genetic code (5). Sec, the 21st cotranslationally inserted amino acid (6), is not suitable, because many organisms need this amino acid for viability and because its UGA-directed insertion requires additional RNA and protein components. Pyl, the 22nd cotranslationally inserted amino acid, appears more suited for this purpose, because it is restricted to a small number of organisms, where it accomplishes a special function (7). The Methanosarcinaceae contain a devoted UAG-recognizing suppressor tRNA^Pyl (8) and a pyrrolysyl-tRNA synthetase (PylRS) dedicated to forming Pyl-tRNA^Pyl (9, 10). Initial studies indicated that Lys-tRNA^Pyl can be recognized by bacterial EF-Tu (11) and that in Escherichia coli tRNA^Pyl acts like an amber suppressor (10, 12).

A thorough investigation in E. coli of archaeal tRNA^Pyl should uncover the structural determinants that may make tRNA^Pyl and PylRS an ideal orthogonal pair when used in bacterial protein synthesis. Here we investigate the interaction of Methanosarcina barkeri PylRS and M. barkeri tRNA^Pyl and explore the fitness and coding response of this orthogonal tRNA for translation in E. coli.

Results

Nucleotides that Determine Fitness of tRNA^Pyl for Translation in E. coli.

To screen a large number of M. barkeri Fusaro tRNA^Pyl variants generated by mutagenesis, we made use of this tRNA's ability to suppress the lac amber mutation A24 in a lacI–lacZ fusion system (12, 13). Therefore, we transformed E. coli strain XAC/A24 with plasmid-borne copies of M. barkeri pylS and 42 mutant pylT genes and grew the transformants in the presence of the Pyl analog N-ε-cyclopentyloxycarbonyl-l-lysine (Cyc), because Pyl is not commercially available (12). Suppression was quantitated by measuring β-galactosidase activity (Table 1). The mutations covered all regions of the tRNA^Pyl; however, the anticodon was not altered because the suppression assay depended on the integrity of amber codon recognition.

Table 1.

Suppression efficiency of M. barkeri tRNA^Pyl variants

Mutations	Suppression efficiency, %
Wild type
Fusaro	100
MS (C44U/A3G)	95
Discriminator base
G73→C	88
G73→A	33
G73→U	43
Acceptor stem
G1→A	26
C72→U	30
G1:C72→C:G	70
G2:C71→A:U	52
A3→G	58
A5:U68→C:G	100
D-stem and loop
U15→A	39
G18→U	87
U20→A	43
A21→U	95
G10:C25→U:A	50
A11:U24→C:G	88
U12:A23→G:C	112
C13:G22→A:U	25
Insert at G18	47
TψC-stem and loop
U54→A	80
A56→U	33
A58→U	40
U60→A	108
C50:G64→A:U	79
G51:C63→U:A	29
G52:C62→U:A	63
G53:C61→U:A	63
Variable loop
C44→U	56
A45→G	23
G48→U	17
Insert A at C44	17
Anticodon stem/loop
A27:U43→G:A	56
U29a→C	40
Δ(U29aG41b)	12
G29b:C41a→U:A	80
G30:C40→ U:A	74
A31:U39→C:G	46
C32→A	31
U33→G	16
A37→C	13
A38→C	49

Suppression efficiency was measured by β-galactosidase activity in E. coli strain XAC/A24. A percentage of 100 corresponds to 3,800 Miller units (13). In the absence of Cyc, background suppression in all cases was <5%.

Discriminator Base and Acceptor Helix.

Systematic mutation of the nucleotides in the acceptor stem revealed that the discriminator base G73 and the first base pair of the acceptor stem are major tRNA^Pyl identity elements. The G73A and G73U mutations decreased suppression efficiency markedly. The base pair G1:C72, conserved in all known tRNA^Pyl species, could be flipped with some loss of activity, yet replacement with a weaker pair (G1:U72) or loss of the base pair (A1:C72) resulted in severe loss of suppression efficiency. These results suggest that the primary role of the first base pair is to ensure the proper productive placement of recognition elements, such as the discriminator base and the terminal adenosine. Mutation of the G2:C71 base pair to A2:U71 reduced suppression efficiency by ≈50%. Conversion of the A3:U70 base pair into a G3:U70 base pair resulted in a drop of suppression efficiency. This was unexpected, as this G3 is found in wild-type M. barkeri MS tRNA^Pyl. However, the MS tRNA also contains a C44U mutation in the first nucleotide of the variable loop. Because both independent mutations result in a decrease in suppression efficiency (Table 1) and because the wild-type tRNA^Pyl species from both M. barkeri strains are equally good substrates for suppression, the two variations compensate their respective negative impacts on the suppression efficiency.

D- and T-Stem/Loop Regions.

Losses in suppression observed upon mutation of nucleotides in these regions of the tRNA molecule are probably related to their role in maintaining D-loop/T-loop interactions, crucial for the ability of the tRNA to maintain its L-shape tertiary structure. Mutations of U20A in the D-loop and A56U in the T-loop resulted in loss of suppression efficiency. Similarly, the negative effect of the D-loop mutation U15A supports a role of this nucleotide in D-loop/T-loop interaction via nucleotide U59 (14). One of the most severe decreases in suppression efficiency was the T-loop mutation A58U, because it would disrupt the putative T-loop upon interaction with U54 (14). However, the U54A mutation resulted only in a minor reduction of suppression efficiency, suggesting that the integrity of the T-loop interaction is not critical for tRNA^Pyl activity and that A58 could be directly involved in PylRS binding and recognition. This idea is supported by the fact that mutation of A58 in Desulfitobacterium hafniense tRNA^Pyl resulted in a >1,000-fold loss of in vitro aminoacylation efficiency (15).

Base pair mutations G10:C25 to U:A, C13:G22 to A:U in the D-arm, and G51:C63 to U:A in the T-arm caused significant reduction of suppression efficiency. Although the observed effect of mutation of G10:C25 can be attributed to a role in maintaining the core structure of tRNA^Pyl via possible interactions with a nucleotide in the variable loop, base pairs C13:G22 and G51:C63 do not have obvious structural roles (14), suggesting a direct contribution of these nucleotides to tRNA^Pyl fitness in the translation machinery.

Variable Loop.

The short variable loop (only three nucleotides instead of the normal five) is one of the distinctive features of tRNA^Pyl. Mutation of any of these nucleotides resulted in strong reduction of suppression efficiency; the most dramatic effects were observed in mutants A45G and G48U, and when an additional A was inserted to make a 4-nt variable loop. Such dramatic effects are consistent with the function of the variable loop nucleotides in ensuring the proper relative positioning of the two stacked helices that make up the tRNA L-shape. The effect of the A45G mutation may also be considered in light of the reduction observed upon mutating the first base pair of the D-arm (G10:C25) because they make a potential tertiary interaction (14).

Anticodon Stem and Loop.

The elongated anticodon stem is another striking feature of tRNA^Pyl. Although the mutations of the anticodon stem base pairs resulted in only moderate suppression loss, three mutations are nevertheless worth noticing. Disruption of the first base pair of the anticodon stem (A27:U43 to G:A) lowered the suppression efficiency. Conversion of the wobble pair U29a:G41b to a C:G pair or of A31:U39 to C31:G39 also caused significant reductions of tRNA^Pyl fitness as a UAG suppressor. Deletion of the U29a:G41b pair, resulting in a canonical 5-bp anticodon stem was particularly detrimental to suppression efficiency. Although mutations in any of the four anticodon loop bases that flank the anticodon led to loss of suppression efficiency, the major effects were noticed upon mutation of the conserved nucleotides U33 and A37, directly adjacent to the anticodon triplet. From these results, we can infer that the length of the anticodon stem is critical for tRNA^Pyl activity because it places the important anticodon loop nucleotides at an appropriate distance from the 3′ terminal adenosine.

Effect of Mutations in tRNA^Pyl on in Vitro Charging by and in Vivo Binding to M. barkeri PylRS.

Because suppression efficiency of tRNA reflects the sum of this molecule's properties from aminoacylation to codon recognition and through the later steps of protein synthesis, we wanted to determine the aminoacylation and PylRS binding properties of the mutant tRNAs most affected in suppression. Therefore, we determined the in vitro kinetic parameters for acylation of the corresponding tRNA^Pyl molecules with Cyc by the purified recombinant M. barkeri PylRS, and we measured the effect of point mutations in tRNA^Pyl on binding to PylRS by using the in vivo yeast three-hybrid method (Table 2). Both methods unambiguously demonstrated the lack of recognition of the tRNA^Pyl anticodon by PylRS. Generally, the anticodon is one of the major identity elements in tRNA synthetase recognition. Seryl-tRNA synthetase (SerRS), leucyl-tRNA synthetase (LeuRS), histidyl-tRNA synthetase (HisRS), and alanyl-tRNA synthetase (AlaRS) are the four other synthetases that do not, at least in E. coli, rely on anticodon recognition for aminoacylation of their cognate tRNA. For SerRS and LeuRS, this is explained by the number of tRNA isoacceptors, because Ser and Leu are the only two amino acids that can be incorporated in response to six different codons. For HisRS and AlaRS, anticodon recognition has been replaced by unique decisive features in their cognate tRNAs, such as G-1 for tRNA^His and the acceptor stem pair G3:U70 for tRNA^Ala (16).

Table 2.

Aminoacylation and in vivo binding of M. barkeri tRNA^Pyl variants by M. barkeri PylRS

tRNA	Mutations	KM, μM	kcat, s−1	kcat/KM	L	Binding, %
tRNAPyl	Wild type	0.23 ± 0.04	(4.0 ± 1.2) × 10−2	1.74 × 10−1	1	100
tRNAPyl	G73A	0.48 ± 0.14	(1.5 ± 0.2) × 10−3	3.00 × 10−3	58.0	55
tRNAPyl	C72U	0.10 ± 0.03	(1.6 ± 0.3) × 10−4	1.60 × 10−3	108.7	76
tRNAPyl	C13:G22→A:U	0.16 ± 0.04	(1.9 ± 0.3) × 10−2	1.16 × 10−1	1.5	95
tRNAPyl	A45G	0.51 ± 0.08	(3.2 ± 0.9) × 10−2	6.31 × 10−2	2.8	92
tRNAPyl	G48U	0.39 ± 0.04	(9.2 ± 3.1) × 10−3	2.36 × 10−2	7.4	79
tRNAPyl	G51:C63→U:A	0.97 ± 0.23	(8.2 ± 1.4) × 10−3	8.49 × 10−3	20.5	69
tRNAPyl	U33G	0.12 ± 0.02	(1.2 ± 0.2) × 10−3	1.00 × 10−2	17.4	67
tRNAPyl	A37C	0.20 ± 0.04	(1.3 ± 0.3) × 10−3	6.63 × 10−3	26.2	55
tRNAPyl	Anticodon CUA→CAA	0.09 ± 0.02	(2.0 ± 0.4) × 10−2	2.19 × 10−1	0.8	82
tRNAPyl	Anticodon CUA→AUA	0.67 ± 0.10	(1.7 ± 0.4) × 10−2	2.57 × 10−2	6.8	94
tRNAPyl	Anticodon CUA→CUU	0.17 ± 0.02	(3.2 ± 0.6) × 10−2	1.88 × 10−1	0.9	86
tRNASer	Wild type	ND	ND	ND	ND	ND
tRNASer/Pyl	Chimera 1	0.26 ± 0.04	(2.1 ± 0.4) × 10−2	8.15 × 10−3	2.13	78.3
tRNASer/Pyl	Chimera 2	0.29 ± 0.05	(2.3 ± 0.5) × 10−2	7.93 × 10−3	2.19	82.6

In vivo binding of tRNA^Pyl to PylRS was determined by the yeast three-hybrid method as described in Materials and Methods. L represents loss of aminoacylation efficiency and is calculated as (k_cat/K_M mutant)/(k_cat/K_M wild type). ND, the activity could not be detected.

The discriminator base G73, the nucleotides flanking the anticodon (U33 and A37), and the T-stem base pair G51:C63 are significant identity elements in tRNA^Pyl. Furthermore, destabilization of the first acceptor stem pair (G1:C72) by conversion to G1:U72 resulted in significant decrease of aminoacylation efficiency. The A45G, G48U, and C13:G22→A:U mutations had no impact on PylRS binding or charging, which suggests that the marked loss in translational fitness of these mutant tRNA^Pyl species (Table 1) is due to their inability to perform steps in protein synthesis that are downstream of aminoacylation. Additionally, the suppression data (Table 1) show the existence of tRNA^Pyl mutations (U60A and U12:A23→G:C) that increase translational fitness of this tRNA.

Transplantation of Cyc Acceptor Activity into Bovine Mitochondrial tRNA^Ser.

As noted earlier (9, 11, 15), tRNA^Pyl has an unusual structure shared with only one Bos taurus mitochondrial tRNA^Ser isoacceptor (17). Neither tRNA_CUA^Ser nor a stabilized version of this RNA (see Materials and Methods) were substrates for any endogenous E. coli synthetase in vivo or for M. barkeri PylRS in vitro. Therefore, we attempted to transplant the tRNA^Pyl identity elements into this bovine tRNA^Ser scaffold. We constructed two slightly different tRNA^Ser/Pyl chimeric molecules (Fig. 1) that included the tRNA^Pyl discriminator base G73, base pairs G1:C72, C13:G22 in the D-arm, G51:C63 in the T-stem, and A31:U39 in the anticodon stem. However, addition of some nucleotides from the tRNA^Pyl core structure (U8, G10:C25, G26, C44, A45, and G48) were needed to generate tRNA^Ser/Pyl chimeras with efficient Cyc charging properties (Table 2). The yeast three-hybrid data also showed acquisition of in vivo tRNA binding by PylRS when wild-type tRNA^Ser is compared with the two chimeric tRNAs (Table 2).

Fig. 1.

M. barkeri tRNA^Pyl and bovine mitochondrial tRNA^Ser L-shape structures and scheme of tRNA numbering. The circled nucleotides indicate tRNA positions that, upon mutation, resulted in a significant decrease in in vivo suppression activity, in vivo binding, and in vitro aminoacylation. The boxed nucleotides indicate tRNA positions that, upon mutation, only affected in vivo suppression activity.

A more complex pattern is seen when translational fitness of the chimeric tRNAs was tested by lacZ amber mutant suppression in the E. coli XAC/A24 strain. Some chimeras (data not shown) allowed efficient suppression in the absence of Cyc, suggesting that the resulting tRNAs were charged by endogenous E. coli tRNA synthetases. Edman degradation (data not shown) of a reporter protein based on the E. coli dihydrofolate reductase (DHFR) system containing a UAG codon in position 3 (18) revealed the presence of glutamine (80%), threonine (12%), and serine (8%). The selectivity of the tRNAs toward PylRS could be reestablished by keeping the tRNA^Ser nucleotides A8 (Fig. 1, chimera 1) or A26 and G44 (Fig. 1, chimera 2) in the chimeric tRNAs, even though their fitness in protein synthesis was somewhat lowered (Table 3). Maintaining all three tRNA^Ser nucleotides at these positions yielded an inactive tRNA (data not shown). Taken together, these data show that Pyl identity elements could be successfully transplanted into the tRNA^Ser scaffold.

Table 3.

Suppression efficiency of tRNA^Ser/Pyl chimeras

tRNA	Suppression efficiency, %
tRNACUASer	<5
Stabilized tRNACUASer	<5
tRNASer/Pyl chimera 1	62
tRNASer/Pyl chimera 2	69
tRNAPyl	100

Suppression efficiency was measured by β-galactosidase activity in E. coli strain XAC/A24. In the absence of Cyc, background suppression in all cases was <5%.

tRNA^Pyl Can Be Converted into an Efficient UGA Suppressor.

Because the aminoacylation data (Table 2) clearly showed that the nature of the anticodon sequence is not critical for tRNA^Pyl charging, we wanted to test whether introduction of a UCA anticodon would generate an efficient UGA suppressor tRNA^Pyl. Again, we used the strategy of suppressing lac nonsense mutations, either the U4 opal or the A24 amber mutant, in the lacI–lacZ fusion system (13). Upon transformation of the E. coli strains XAC/U4 and XAC/A24 with plasmids carrying M. barkeri pylS and either pylT (with a UCA anticodon) or wild-type pylT, the cells were grown in the presence of Cyc, and suppression efficiency was assessed by β-galactosidase activity (Table 4). The UGA suppressor tRNAUCAPyl is 81% as efficient as the native tRNACUAPyl amber suppressor. Thus, Cyc can also be inserted into proteins in response to a UGA stop codon in vivo. A similar change in coding response was reported for tRNA^Sec (19); anticodon changes led to efficient selenocysteine incorporation in response to all three termination codons (Table 4), even though the requirements for codon context are very different for Sec (UGA) and Pyl (UAG). Compared with the synthesis of wild-type LacZ protein from a lacZ⁺ construct, the absolute suppression efficiency of tRNAUCAPyl is 20%, which is the same level as we observed (data not shown) and had been reported for the E. coli tRNA_UCA^Arg suppressor (20).

Table 4.

Decoding properties for UAG and UGA of M. barkeri tRNA^Pyl and E. coli tRNA^Sec

tRNA	Anticodon	lacZ codon (mutation)	Suppression efficiency, %
tRNAPyl	CUA	UAG (A24)	100
tRNAPyl	CUA	UGA (U4)	<1
tRNAPyl	UCA	UAG (A24)	<1
tRNAPyl	UCA	UGA (U4)	81
tRNASec	CUA	UAG	68*
tRNASec	CUA	UGA	<1*
tRNASec	UCA	UAU	<1*
tRNASec	UCA	UGA	100*

Suppression efficiency was measured by β-galactosidase activity in E. coli strain XAC/U4 (for opal) and in E. coli strain XAC/A24 (for amber). In the absence of Cyc, background suppression in all cases was <5%.

*Suppression efficiencies for tRNA^Sec are from ref. 19.

Discussion

Sec and Pyl are the 21st and 22nd genetically encoded amino acids. Unlike any of the 20 canonical amino acids in protein synthesis, these two unusual amino acids share the property of being cotranslationally inserted in-frame at UGA and UAG, respectively; these codons normally specify termination. However, the mechanism of reassigning these codons to sense appears to differ between Sec and Pyl (21). Although the insertion of Sec at UGA requires a specific mRNA context provided by the presence of the SECIS stem/loop structure as well the selenocysteine-specific elongation factor SelB (6), Pyl can be efficiently inserted into proteins in an anonymous context and appears not to depend on the presence of additional proteins. This characteristic is well demonstrated by the fact that Pyl or Cyc could be efficiently incorporated into E. coli reporter proteins lacking any specific RNA structures in the vicinity of the UAG codon when tested in Methanosarcina acetivorans (22) or in E. coli (12). A stem/loop structure located just downstream of the UAG codon and therefore termed PYLIS by analogy to the established SECIS element, was proposed to stimulate Pyl incorporation (22, 23). However, the lack of sequence and structure conservation between the predicted PYLIS structures from the three Pyl-containing methylamine methyltransferase genes is a clear argument against a functional role for such an RNA motif (21).

The difference in recoding strategies used for Pyl and Sec incorporation may reflect different evolutionary histories. A compelling body of data indicates that Sec was already present at the time of the last common ancestor and, therefore, already was using contextualized UGA as a sense codon for Sec insertion. Although Sec is present in organisms from all three domains of life, it is absent from many bacteria and most archaea and present in higher eukaryotes with the exception of plants (5, 6, 24, 25). Although the widespread presence of this unusual amino acid in organisms from all three domains is a clear sign of vertical inheritance from the last common ancestor, its absence from many organisms is attributed to the loss of Sec coding capability due to environmental factors (25). Based on the known genomes, the organismal distribution of PylRS suggests that this enzyme is a late archaeal invention designed to meet the specific physiological needs of a particular archaeal lineage. PylRS would then be another example of genetic code evolution after the era of the last common universal ancestor (5). The late apparition of PylRS implies that the Methanosarcinaceae progenitor added Pyl to its genetic code. The fact that the tRNA^Pyl anticodon is not recognized by PylRS may then have conferred a significant advantage because it allowed testing of different codons, eventually selecting the UAG codon as the least disruptive for protein synthesis. In M. barkeri, for instance, the UAG codon is used at a much lower frequency (0.05%) than UAA (1.8%) or UGA (1.2%). Furthermore, when UAG codons are used as genuine stop codons, they are almost immediately followed by either UAA or UGA codons, thus reducing the negative impact that the unintended read-through of the UAG termination signal may have on the integrity of the Methanosarcinaceae proteomes (25). Our data suggest that tRNA^Pyl anticodon variants may respond to any codon, a possibility that may be tested in an organism that does not use all sense codons [e.g., Micrococcus luteus (26)]. The nonessentiality of the tRNA^Pyl anticodon sequence, combined with the absence of competition by any other aminoacyl-tRNA, and the low usage of UAG codons may have been determining factors for the successful insertion of Pyl into the Methanosarcinaceae genetic code.

Considerable efforts have been made in devising orthogonal aminoacyl-tRNA synthetase:tRNA pairs solely specific for recognition of nonnatural amino acids and in assuring their subsequent incorporation at termination codons into E. coli and eukaryotic proteins (1, 3). Because Pyl-tRNA^Pyl formation is independent of the tRNA anticodon sequence, because the PylRS structure is being solved (27), because tRNA^Pyl is a strong suppressor (at least as tested for UAG and UAA) in E. coli (10, 12), and because additional RNA or protein factors are not required for Pyl or Cyc incorporation, the potential use of the PylRS:tRNA^Pyl orthogonal pair, optimized through natural or man-made evolutionary processes, is particularly appealing.

Materials and Methods

General.

Oligonucleotide synthesis, DNA sequencing, and Edman degradation were performed by the Keck Foundation Biotechnology Resource Laboratory at Yale University. [γ-³²P]ATP (3,000 Ci/mmol) and [α-³²P]ATP (3,000 Ci/mmol) (1 Ci = 37 GBq) were from Amersham Biosciences (Uppsala, Sweden). Cyc was purchased from Sigma (St. Louis, MO). The yeast strain L40 was a gift from Marvin Wickens (University of Wisconsin, Madison, WI).

tRNA Mutants.

Mutant tRNA genes were constructed by PCR (QuikChange kit; Qiagen, Hilden, Germany) using as templates the wild-type M. barkeri pylT, cloned into pTECH (12) for in vivo expression and pUC18 (28) for production of in vitro T7 RNA polymerase transcripts. The sequence of each pylT mutant construct was confirmed by dideoxy-automated sequencing. The M. barkeri tRNA^Pyl sequence has two variants: the M. barkeri MS tRNA^Pyl contains a G in position 3 and a U in position 44, whereas the tRNA of the Fusaro strain has an A and a C at the respective positions (8). The bovine mitochondrial tRNA^Ser was shown to be rather unstable (14). Therefore, a stabilized tRNA^Ser was obtained by replacing the four wobble pairs present in the molecule (Fig. 2) by standard Watson–Crick pairs: G5:U69 was replaced with G5:C69; G50:U64 and G51:U63 were replaced with the corresponding A:U pairs, and finally the G28:U42 pair was replaced with a G:C pair. Given the unusual structure of tRNA^Pyl, the nucleotide numbering is defined in Fig. 2.

Fig. 2.

Transplantation of M. barkeri tRNA^Pyl identity elements into the bovine mitochondrial tRNA^Ser scaffold. L-shape structures of stabilized tRNA^Ser (see Materials and Methods) and of two tRNA^Ser/Pyl chimeric molecules (see Results). Filled circles refer to nucleotides different in the stabilized tRNA^Ser and tRNA^Pyl; circled nucleotides refer to positions mutated in the stabilized tRNA^Ser required to obtain Cyc accepting activity. The other indicated nucleotides are common to the stabilized tRNA^Ser and tRNA^Pyl species.

Suppression of E. coli XAC/A24 and XAC/U4 Nonsense Mutations.

E. coli strains XAC/A24 and XAC/U4 carry an inactivating mutation in the lacI–lacZ fusion system: Trp codon position 658 to UAG nonsense and leucine codon position 565 to UGA, respectively (13). Both strains have been used in the past for insertions of a number of different amino acids. For both strains, the presence of an in-frame stop codon results in the inability of this strain to degrade the chromogenic 2-nitrophenyl β-d-galactopyranoside. Cells were then cotransformed by electroporation with plasmids carrying the M. barkeri Fusaro pylS (pCBS) and M. barkeri Fusaro pylT (pTECH) wild-type and mutant genes. The transformants were grown in Luria broth containing 10 mM Cyc, and β-galactosidase activity was measured as previously mentioned (12). In cases of low suppression, Northern blots revealed that tRNA^Pyl was not limiting in the experiment (data not shown).

Suppression of a UAG (Codon 3) in an E. coli folA Reporter Gene.

The E. coli folA gene (encoding DHFR) containing a UAG triplet in place of codon 3 was amplified from genomic DNA, cloned into pCR 2.1-TOPO plasmid vector (Invitrogen, Carlsbad, CA), sequenced, and subcloned into pRSFDuet-1. E. coli BL21(DE3)-competent cells were cotransformed with a tRNA^Ser/Pyl chimera cloned into pTECH and the folA reporter construct and grown in LB at 37°C, in the presence of ampicillin and kanamycin. Production of recombinant DHFR was induced with 1 mM isopropyl β-d-thiogalactoside when cells reached an A₆₀₀ of 1.0. The cells were harvested after 17 h; DHFR was purified by using standard Ni-NTA technology and then blotted onto a PVDF membrane. The first eight amino acids were identified by Edman degradation.

In Vivo Binding Affinity of tRNA^Pyl for PylRS by Using the Yeast Three-Hybrid System.

The yeast three-hybrid experiments were performed as described (29). The effect of point mutations in tRNA^Pyl on PylRS binding was measured by using the in vivo yeast three-hybrid method. Yeast strain YBZ-1, auxotrophic for histidine and carrying a genomic copy of the first hybrid molecule (MS2 coat protein fused to the GAL4-DB DNA binding protein), was transformed with two vectors encoding the RNA hybrid and protein hybrid molecules. The shuttle plasmid pIIIA/MS2–1 with a URA3 marker was used to express the RNA hybrid, an RNA hairpin with two MS2 binding sites for the coat protein fused to the 5′ end of tRNA^Pyl, as well as yeast tRNA^Lys and tRNA^Gln for specificity controls. The third hybrid is a fusion between the M. barkeri PylRS gene and the GAL4 activation domain (768–881) and was expressed in a 2-μm LEU2 vector pGADT7 under ADH1 promoter. When tRNA^Pyl interacts with its cognate PylRS through the chain-interaction of the three-hybrid protein and RNA molecules, GAL4-dependent transcription of a histidine biosynthetic gene and lacZ reporter genes is activated. Qualitatively, weak and strong interactions can be discriminated phenotypically by using the growth rate on minimal media lacking histidine and a quantitative assessment by measurement of β-galactosidase activity in the yeast cell extract. tRNA^Lys and tRNA^Gln did not result in any binding.

Purification of Recombinant M. barkeri PylRS and in Vitro Transcription of tRNA Genes.

The recombinant M. barkeri Fusaro PylRS was overexpressed and purified as reported (9). The M. barkeri Fusaro wild-type, mutants tRNA^Pyl, and tRNA^Ser/Pyl chimeric molecules were prepared as reported (28).

tRNA Aminoacylation Assays.

Aminoacylation assays were adapted from a recently described procedure (30). Aminoacylation reactions (10 μl) were carried out at 37°C (unless otherwise noted) in 100 mM Na-Hepes (pH 7.2)/25 mM MgCl₂/60 mM NaCl/5 mM ATP/1 mM DTT/10 mM Cyc/tRNA 3′-labeled with [α-³²P]ATP ranging in concentration from 0.25 to 5 times K_M concentrations. PylRS concentrations ranged from 5 to 90 nM. Reactions were stopped by removing 1 μl of the reaction and adding it to 3 μl of 2.5 units/μl nuclease P1 (where 1 unit liberates 4 μmoles of orthophosphate from 3′AMP per minute at 37°C; American Bioanalytical, Natick, MA) and incubated at 25°C for 30 min. Nuclease P1 digests were spotted on polyethyleneimine-cellulose plates (Baker, Phillipsburg, NJ), and [α-³²P]AMP (from uncharged tRNA) and aminoacyl-[α-³²P]AMP (from charged tRNA) were separated in a running buffer of 0.1 M sodium acetate and 5% acetic acid (31) and visualized by using a phosphoimager system. Kinetic parameters represent the results of at least three independent experiments.

Acid Urea Gel Electrophoresis of Aminoacyl-tRNA and Northern Hybridization.

This method (32) allows the separation of charged and uncharged tRNAs based on the electrophoretic mobility difference seen between the two species. Hybridization to a ³²P-labeled sequence-specific probe permits the determination of the identity of the aminoacylated tRNA. The aminoacylation level of tRNA^Pyl mutants isolated from E. coli strain XAC/A24 grown in the presence of 10 mM Cyc was verified. Unfractionated tRNA was isolated from each strain under acidic conditions [0.3 M sodium acetate (pH 4.5)/10 mM EDTA], dissolved in 2× loading buffer [7 M urea/0.3 M sodium acetate (pH 4.5)/10 mM EDTA/0.1% bromophenol blue/0.1% xylene cyanol] and were loaded (20 μg) on a 6.5% polyacrylamide gel containing 7 M urea and 0.1 M sodium acetate (pH 5.0). Northern blot was performed as described earlier (28).

Abbreviations

Pyl

pyrrolysine

PylRS

pyrrolysyl-tRNA synthetase

DHFR

dihydrofolate reductase.