Virus-assisted directed evolution of enhanced suppressor tRNAs in mammalian cells

Abstract

Site-specific incorporation of unnatural amino acids (Uaas) in living cells relies on engineered aminoacyl-transfer RNA synthetase-tRNA pairs borrowed from a distant domain of life. Such heterologous suppressor tRNAs often have poor intrinsic activity, presumably due to suboptimal interaction with a non-native translation system. This limitation can be addressed in Escherichia coli using directed evolution. However, no suitable selection system is currently available to do the same in mammalian cells. Here we report virus-assisted directed evolution of tRNAs (VADER) in mammalian cells, which uses a double-sieve selection scheme to facilitate single-step enrichment of active yet orthogonal tRNA mutants from naive libraries. Using VADER we developed improved mutants of Methanosarcina mazei pyrrolysyl-tRNA, as well as a bacterial tyrosyl-tRNA. We also show that the higher activity of the most efficient mutant pyrrolysyl-tRNA is specific for mammalian cells, alluding to an improved interaction with the unique mammalian translation apparatus.

Figure 1.. The VADER selection scheme.

a, Mammalian cells are infected with AAV2 encoding the tRNA library at low MOI. Plasmids encoding TAG-mutant of Cap, other genetic components needed for AAV replication, and the cognate aaRS are provided in trans by transfection in the presence of a suitable azido-Uaa. Active and orthogonal tRNA mutants facilitate generation of packaged progeny AAV2 incorporating the Uaa into their capsid, which are isolated by chemoselective biotin conjugation followed by streptavidin pulldown. b, Two AAV2 vectors, encoding i) E. coli tRNA^Tyr and EGFP (Tyr-EGFP), and ii) tRNA^Pyl and mCherry (Pyl-mCherry), were mixed in a 10⁴:1 ratio and subjected to the VADER selection scheme using MbPylRS and its substrate AzK. FACS analysis of the surviving population shows a >30,000-fold cumulative enrichment of Pyl-mCherry. Data shown as mean ± s.d. (n = 3 independent experiments)

Figure 2.. Directed evolution of tRNA^Pyl.

a, The sequences randomized to create four different libraries (A1, A2, T1, T2) of tRNA^Pyl are highlighted in four different colors. b, Degree of base pairing observed in the enriched mutants (through NGS analysis) upon the selection of each library. c, For each library, the wild-type sequence is shown on top and analysis of the 1% most-enriched sequences (n = 41) is shown below, revealing the relative abundance of each base at each randomized position. d, Efficiency of TAG suppression for unique tRNA^Pyl selectants measured using the EGFP-39TAG reporter. The tRNA encoded in the pAAV plasmid (also harboring a wild-type mCherry reporter) was co-transfected into HEK293T cells with MbPylRS and EGFP-39TAG in the presence or absence of 1 mM AzK. Expression of EGFP-39TAG was measured in cell-free extract, normalized relative to wild-type mCherry expression and plotted as a percentage of the normalized activity of wild-type tRNA^Pyl. Data shown as mean ± s.d. (n = 6 independent experiments for hits A1-GGG/CCC, A1-GGG/UCC, A1-AGG/CCU, A1-GCU/AGC, A2-AGG/UCU, A2-AAC/GUU, A2-ACU/AGU, A2-GGG/GCC, A2-ACU/GGU, A2-GUG/UGC, and T2-AG/UGCU; n = 8 independent experiments for hits A1-GGC/GCC, A1-GGG/CUC, A2-GGG/UCC, A2-GGG/UCU, and T1-CUG/CAG; n = 10 independent experiments for hits A1-GGG/CCC, A2-AGC/GCU, and A2-UGG/UCA; n = 18 independent experiments for A2-GGG/CCU; and n = 4 independent experiments for all others). *sequences identified through NGS; ^†enriched sequences containing a G:G mispair.

Figure 3.. Characterization of tRNA^Pyl-A2.1 activity.

a, Sequences of wild-type (WT) and A2.1. b, Structures of AcK and AzK. c, Expression of EGFP-39TAG using the WT or A2.1 tRNA with MbPylRS, in the presence (+) and absence (−) of AzK. d, Expression of EGFP-39TAG and EGFP-39TAG-151TAG using the WT or A2.1 tRNA with AcK-selective MbPylRS, in the presence and absence of AcK. e, Expression of EGFP-39TGA and EGFP-39TAA using tRNA_UCA^Pyl and tRNA_UUA^Pyl (for both WT and A2.1), respectively, and MbPylRS in the presence and absence of AzK. f, Expression of EGFR-127TAG (with a C-terminal EGFP fusion) using the WT or A2.1 tRNA with MbPylRS, in the presence and absence of AzK. g, Mammalian cell-optimized baculovirus (BacMam) vectors and the scheme of the experiment for h-i. h, Expression of EGFP-39TAG as increasing MOI of tRNA-BacMam vector is used. i, Expression of mCherry in the same experiments as h shows equivalent delivery of both tRNAs, which increases proportionately with higher MOI. Expression of the fluorescent reporters is measured in HEK293T cell-free extract. Also see Supplementary Figure 13 for the corresponding images. For c-f and h, EGFP-39TAG expression is reported relative to wild-type EGFP expression from the same vector. Data shown as mean ± s.d. (n = 3 independent experiments in 3c, 3d, 3e, 3h, and 3i; n = 12 independent experiments in 3f).

Figure 4.. Origin of the improved activity of tRNA^Pyl-A2.1.

a, Northern blot analysis shows lower expression of A2.1 relative to WT, when plasmids encoding a tRNA^Pyl and a wild-type mCherry reporter was transfected into HEK293T cells. The mCherry reporter was used to ensure comparable transfection efficiency. b, Sequence of tRNA^Pyl-Opt (optimized in E. coli) showing key mutations in red. c, Activity of tRNA^Pyl-WT, A2.1, and Opt in E. coli measured using the sfGFP-151-TAG reporter. d, Activity of tRNA^Pyl-WT, A2.1, and Opt in HEK293T cells measured using the EGFP-39-TAG reporter. For both c and d, a mutant MbPylRS selective for AcK was used and the activities were measured in the presence and the absence of AcK. Expression of the fluorescent reporter was measured in cell-free extract and reported as the % of WT-tRNA^Pyl activity. Data shown as mean ± s.d. (n = 3 independent experiments).

Figure 5.. Directed evolution of tRNA^Tyr.

a, A modified VADER scheme lacking the bioorthogonal capture step. Instead, the selective amplification is performed either in the presence or the absence of the cognate synthetase. Active and orthogonal tRNA mutants are identified by their selective enrichment in the presence of the cognate aaRS. b, The randomization scheme to create the E. coli tRNA^Tyr mutant library. c, The consensus corresponding to 172 homologous bacterial tRNA^Tyr sequences, which was used to guide the tRNA^Tyr randomization scheme. d, Sequence of the tRNA^Tyr mutants exhibiting the highest average enrichment following the VADER scheme shown in panel a in the presence of EcTyrRS. e, Observed enrichment of each mutant in the tRNA^Tyr library upon subjecting them to the VADER selection scheme either in the presence of a cognate EcTyrRS mutant that charge OMeY, or MbPylRS (does not recognize tRNA^Tyr). Each selection was performed in duplicate, and the normalized enrichment factors observed from each were plotted against each other. Nine out of the ten most enriched sequences in the presence of EcTyrRS did not show strong enrichment when MbPylRS was used instead (shown in green; sequences shown in panel d); the cross-reactive mutant is shown in red. f, Efficiency of TAG suppression for these nine tRNA^Tyr mutants using the EGFP-39TAG. The tRNA encoded in the pAAV plasmid (also harboring a wild-type mCherry reporter) was co-transfected into HEK293T cells with OMeY-selective EcTyrRS and EGFP-39TAG in the presence or absence of 1 mM OMeY. Expression of EGFP-39TAG was measured in cell-free extract, normalized relative to wild-type mCherry expression and plotted as a percentage of the normalized activity of wild-type tRNA^Tyr. Data shown as mean ± s.d. (n = 3 independent experiments).