Orthogonality of Pyrrolysine tRNA in the Xenopus oocyte

Chemical aminoacylation of orthogonal tRNA allows for the genetic encoding of a wide range of synthetic amino acids without the need to evolve specific aminoacyl-tRNA synthetases. This method, when paired with protein expression in the Xenopus laevis oocyte expression system, can extract atomic scale functional data from a protein structure to advance the study of membrane proteins. The utility of the method depends on the orthogonality of the tRNA species used to deliver the amino acid. Here, we report that the pyrrolysyl tRNA (pylT) from Methanosarcina barkeri fusaro is orthogonal and highly competent for genetic code expansion experiments in the Xenopus oocyte. The data show that pylT is amendable to chemical acylation in vitro; it is then used to rescue a cytoplasmic site within a voltage-gated sodium channel. Further, the high fidelity of the pylT is demonstrated via encoding of lysine within the selectivity filter of the sodium channel, where sodium ion recognition by the distal amine of this side-chain is essential. Thus, pylT is an appropriate tRNA species for delivery of amino acids via nonsense suppression in the Xenopus oocyte. It may prove useful in experimental contexts wherein reacylation of suppressor tRNAs have been observed.

Pyrrolysine, the so-called "22 nd amino acid, " is encoded by methanogenic archaea and bacteria using a tRNA that naturally recognizes the amber codon TAG 23,24 . This unique tRNA (pylT) displays exceptional orthogonality in bacteria and in mammalian cells and has been used to encode ncAAs via evolved aminoacyl-tRNA synthetases that recognize an amino acid of interest 25 . Overall, tRNA-synthetase pairs have enabled the encoding of more than 100 ncAAs in diverse environments including cell free translation 17,26 , mammalian cells 27 , bacteria 28 , and even whole animals 29 . However, the evolution of aminoacyl-tRNA sythetases to discern derivatized isosteric analogs from their natural amino acids counterparts (often necessary for atomic-level insights) has proven challenging. Here, we assayed the orthogonality of pylT via in vitro chemical aminoacylation and injection into Xenopus oocytes (Fig. 1), using high-resolution ion channel function as a sensitive and quantitative readout of rescue and readthrough.

Results
The sodium channel conducting state is strongly coupled to the transmembrane potential, thus one can precisely measure the flow of sodium conductance through the channel through standard electrophysiological approaches, Fig. 1. Voltage-gated sodium channel activation is characterized by transient inward, rapidly inactivating ionic currents. When cells are bathed in physiological recording solutions, e.g. 140 mM extracellular sodium, inward sodium currents appear as downward deflections in response to transient depolarization 30,31 ; thus the level of recorded negative current is directly indicative of the number of full-length functional channels at the cell surface. To begin to scrutinize the orthogonality of the pylT, we chose amino acid position S571 in the human cardiac sodium channel hNa v 1.5 as a model site for encoding. Serine 571 is located in an unstructured intracellular loop between two domains of the channel and importantly has been mutated to other amino acids with diverse side-chain chemistries via conventional mutagenesis without compromising channel function 32 . We co-injected Xenopus laevis oocytes hNa v 1.5 S571 cRNA (complementary RNA generated via in vitro transcription) with acylated and non-acylated tRNAs, and measured currents using the Two-Electrode Voltage Clamp technique after 24 hours 33 . When THG73 was used as the carrier tRNA for this position, significant hNa v 1.5 current was generated in the presence or absence of an appended tyrosine amino acid (i.e., whether we ligated the pdCpA, which lacks an amino acid, or the pdCpA-Tyrosine substrate) (Fig. 2, top panels). There was, in fact, no significant difference in the two conditions after 24hrs (Table 1), eliminating the prospect of adjusting injection conditions to abrogate readthrough while sparing rescue. Injection of S571TAG hNa v 1.5 cRNA alone did not elicit significant current, signifying that this site lacks appreciable 'intrinsic' bleedthrough at the level of the cRNA (Table 1) [19][20][21] . Therefore, in these conditions, it is possible that THG73-CA is acylated by an endogenous aminoacyl sythetase, and the resultant acylated tRNA supports the incorporation of an amino acid at position S571. By contrast, co-injection of pylT and hNa v 1.5-S571TAG yielded sodium currents that were strictly dependent on tRNA acylation, (Fig. 2, lower panels). Specifically, no sodium currents were detected for the condition for hNa v 1.5-S571TAG with non-acylated pylT (pylT-CA). In contrast, robust voltage-dependent sodium currents were seen when channel cRNA was co-injected with an tyrosine-acylated pylT (pylT-Tyr).
To confirm the in vitro enzymatic ligation of pdCpA and pdCpA-amino acid substrates to pylT, ligated and unligated tRNA samples assayed via denaturing TBE-Urea gels. Ligation of the pdCpA substrate is indicated by a gel shift corresponding to a two nucleotide increase in tRNA length. As indicated in Fig. 3, both pdCpA and pdCpA-amino acid substrates were efficiently ligated by the T4 RNA ligase. Therefore, the lack of observed reacylation-based readthrough in the oocyte expression system was not due to a lack of ligation of the pdCpA dinucleotide to truncated pylT.  The encoding fidelity of pylT was evaluated independently by rescuing a stop codon at position K1237 which resides in the sodium ion-selectivity filter of the rat skeletal muscle sodium channel, rNa v 1.4. This unique structural feature promotes the selective passage of sodium through the channel over other monovalent cations, namely potassium. Mutagenesis and functional studies demonstrate that, while this site when mutated produces functional channels, the lysine amino acid at this position is absolutely necessary to support selectivity of sodium ions through the pore 34,35 . As a consequence of this functional prerequisite, any other amino acid encoded at this site, even the charged congener arginine, results in altered channel selectivity. Using standard electrophysiological approaches, ion channel selectivity can easily be quantified. Altered sodium ion selectivity of rNa v 1.4-K1237 is evidenced by a shift in the so-called reversal potential from +60 mV, the Nernst potential for sodium, to near 0 mV, the voltage where electrochemical gradients are balanced for a non-selective pore 36 . To demonstrate the encoding fidelity of the pylT, we coinjected rNa v 1.4-K1237TAG cRNA and pylT-lysine and observed voltage-dependent currents of size −4.9 ± 1.1 µA at −20 mV (N = 5) (Fig. 4A). Currents resulting from co-injection of rNa v 1.4 K1237TAG and full length (pdCpA-ligated) pylT were negligible (−0.12 ± 0.13 µA, N = 10, Fig. 4B). Importantly, the rescued channels displayed a reversal potential of +64.6 ± 3.9 mV (N = 5, Fig. 4). This value is in close agreement with that of WT rNa v 1.4 recorded in parallel (+66.6 ± 2.1 mV, N = 4, p = 0.72 between conditions), confirming the strict encoding of lysine at K1237TAG.

Discussion
Taken together, our results demonstrate that pylT is orthogonal in the Xenopus oocyte, and that it is useful for genetic code expansion experiments. We used voltage-gated sodium channels as exemplar proteins for this purpose, but we have also begun to enlist pylT as a tRNA for delivery of ncAAs into voltage-gated potassium channels, chloride channels, and enzymatic pumps, and these efforts have thus far been met with similar results. Thus, pylT may be used as an alternative to THG73 in cases where reacylation of THG73 poses a significant challenge.  However, regardless of the species being used, we regard it advantageous to test for bleedthrough via co-injection of cRNA of interest with the pdCpA-ligated tRNA as there is no substitute for the empirical support provided by this negative control. It is also highly advantageous that pylT has been shown to be amenable to recoding of its anticodon from TAG to TGA and TAA 37 . Therefore, it may be used in future studies wherein site-specific dual suppression is desired 16 . Finally, it bears mentioning that there are additional tRNA species that have been shown to be orthogonal as part of coevolved tRNA-aminoacyl-tRNA synthetase pairs. Two such pairs have been sucessfully used in Xenopus oocytes 6,7 . These suppressor tRNAs therefore represent good candidates to test for amenability for chemical aminoacylation in future studies.

Methods
Molecular Biology. The S571TAG mutation was made into a pcDNA3.1 human Na v 1.5 construct 38 and the K1237TAG mutation was made into a pBSTA-based rat Na v 1.4 construct 22 using standard methods. For direct comparison of THG73 and pylT in hNa v 1.5, tRNA was generated and purified using the exact same procedure described by our group in detail in a recent report 26 . The transcription of template oligonucleotides generated tRNA with the following sequences: for THG73: For experiments assaying the fidelity of encoding in the selectivity filter of rNa v 1.4, the pylT 70mer tRNA was synthesized by Integrated DNA Technologies (Coralville, IA). In all cases tRNA was reconstituted in 10 mM HEPES pH 7.2 and 3 mM MgCl 2 and it was refolded in a thermocyler using a protocol with a denaturation step (94 °C for 3 min), followed by a linear ramp down to 4 °C over 20 minutes. Ligation reaction conditions, purification of acylated tRNA, and reconstitution of acylated tRNA were done as recently described 26 . Denaturing TBE-Urea gels were run as described in 4 except that precast Mini-Protean gels were used (Biorad, Hercules CA).
Electrophysiology. Xenopus laevis oocytes were obtained through Ecocyte, Inc (Austin TX USA). For rescue of hNa v 1.5 and rNa v 1.4, we injected 25 nl of 1 ng/nl Na v cRNA and 25 nl of 25 μg of a tRNA pellet resuspended in 2.5 μl 3 mM cold NaOAc. Recordings were made approximately 24 hours later. For Na v 1.5 experiments only, we also injected 12.5 ng of the rat β1 cRNA. For WT Na v 1.4, we injected 2 ng of cRNA and recorded currents 24 hours later. All recordings were in oocyte Ringer's essentially as described previously 22 . Currents were analysed via Clampfit 9.2, Molecular Devices, (Sunnyvale, CA). Reversal potentials were derived by fitting the linear component of the current-voltage relationship (+20 to +40 mV), and solving for the y-intercept. Statistical analysis was by unpaired student's t-test, and exact p-values are noted in the text or tables. Data availability. The raw data pertinent to this study is available from the authors upon reasonable request.