Directed Evolution of Heterologous tRNAs Leads to Reduced

Publication Date (Web): April 25, 2018 ... We suspected that reduced expression likely minimized nonorthogonal interactions with host cell machinery. ...
47 downloads 4 Views 3MB Size
Research Article Cite This: ACS Synth. Biol. XXXX, XXX, XXX−XXX

pubs.acs.org/synthbio

Directed Evolution of Heterologous tRNAs Leads to Reduced Dependence on Post-transcriptional Modifications Kevin C. Baldridge,† Manasses Jora,‡ Andre C. Maranhao,§ Matthew M. Quick,∥ Balasubrahmanyam Addepalli,‡ Jennifer S. Brodbelt,∥ Andrew D. Ellington,§ Patrick A. Limbach,‡ and Lydia M. Contreras*,† †

McKetta Department of Chemical Engineering, The University of Texas at Austin, Austin, Texas 78712, United States Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 45221, United States § Department of Molecular Biosciences, The University of Texas at Austin, Austin, Texas 78712, United States ∥ Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712, United States ‡

S Supporting Information *

ABSTRACT: Heterologous tRNA:aminoacyl tRNA synthetase pairs are often employed for noncanonical amino acid incorporation in the quest for an expanded genetic code. In this work, we investigated one possible mechanism by which directed evolution can improve orthogonal behavior for a suite of Methanocaldococcus jannaschii (Mj) tRNATyr-derived amber suppressor tRNAs. Northern blotting demonstrated that reduced expression of heterologous tRNA variants correlated with improved orthogonality. We suspected that reduced expression likely minimized nonorthogonal interactions with host cell machinery. Despite the known abundance of post-transcriptional modifications in tRNAs across all domains of life, few studies have investigated how host enzymes may affect behavior of heterologous tRNAs. Therefore, we measured tRNA orthogonality using a fluorescent reporter assay in several modification-deficient strains, demonstrating that heterologous tRNAs with high expression are strongly affected by some native E. coli RNA-modifying enzymes, whereas low abundance evolved heterologous tRNAs are less affected by these same enzymes. We employed mass spectrometry to map ms2i6A37 and Ψ39 in the anticodon arm of two high abundance tRNAs (Nap1 and tRNAOptCUA), which provides (to our knowledge) the first direct evidence that MiaA and TruA post-transcriptionally modify evolved heterologous amber suppressor tRNAs. Changes in total tRNA modification profiles were observed by mass spectrometry in cells hosting these and other evolved suppressor tRNAs, suggesting that the demonstrated interactions with host enzymes might disturb native tRNA modification networks. Together, these results suggest that heterologous tRNAs engineered for specialized amber suppression can evolve highly efficient suppression capacity within the native post-transcriptional modification landscape of host RNA processing machinery. KEYWORDS: genetic code expansion, amber suppressor tRNA, orthogonal translation system, post-transcriptional modification, TruA, MiaA

F

tRNA:aminoacyl tRNA synthetase (tRNA:aaRS) pairs for site-specific incorporation of noncanonical amino acids into a protein of interest (recently reviewed by Reynolds et al.).7 A commonly used strategy for noncanonical amino acid incorporation in vivo is to reassign a stop codon using an engineered “suppressor tRNA” to incorporate the desired ncAA. Often, the amber stop codon UAG is chosen due to its low abundance in the E. coli genome.11 For these tRNA:aaRS pairs, orthogonality has multiple important aspects. For example, the aaRS must not errantly mischarge host tRNAs, resulting in potentially toxic ncAA incorporation throughout the host organism.10 Similarly, the heterologous tRNA must

or decades, molecular engineering efforts have sought to expand the potential functionality of proteins through the use of noncanonical amino acids (ncAAs).1,2 In addition to expanding the genetic lexicon of organisms for synthetic biology, these efforts can introduce a variety of useful properties into recombinant proteins such as selectively reactive amino acids for “click” labeling.1−3 To that end, a variety of approaches for expanding the standard 20-amino acid genetic code have been attempted.4−9 The fitness defects observed in many of these studies emphasize the importance of orthogonality, which is the ability of the engineered components to operate independently of the native cellular machinery and vice versa.7,10 One of the most successful and broadly used approaches for genetic code expansion has been the engineering of © XXXX American Chemical Society

Received: November 22, 2017

A

DOI: 10.1021/acssynbio.7b00421 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Research Article

ACS Synthetic Biology

hosting both high and low abundance heterologous expressed tRNA variants, demonstrating that high abundance (but not low abundance) tRNA variants depend on tRNA-modifying enzymes for ideal behavior. Amber suppression activity assays in modifying enzyme deletion strains hosting heterologous tRNAs along with the cognate Methanocaldococci jannaschii tyrosyl-tRNA synthetase (MjYRS) suggest that low abundance tRNAs retain higher activity in deletion strains as compared to the two high abundance tRNAs. Analysis of tRNA modifications by mass spectrometry identified ms2i6A37 (2-methylthioN6-isopentenyladenosine) and Ψ39 (pseudouridine) in the high abundance heterologous tRNAs. Together, these results suggest that our previous directed evolution experiments selected for tRNA sequences with improved orthogonal behavior by reducing expression and minimizing interference with host RNA processing machinery.

not be erroneously mischarged by host synthetases, resulting in unexpected behavior such as decreasing host fitness due to nonspecific effects of amber suppression even in the absence of the cognate aaRS. In fact, a number of studies have focused on the effects of errant amber suppression in the host genome, which led to genome engineering efforts that successfully reduced fitness issues often observed in cells hosting orthogonal translation systems (OTSs).12−14 These efforts are a powerful example for how minimizing points of interference with host machinery can be a powerful approach for improving OTS performance. Few studies have focused on the mechanisms by which evolved OTS behave more ideally than their progenitors. In a similar vein to the reduction of amber codons in the host genome, some efforts have sought to optimize expression of heterologous tRNAs, aaRSs, or both.15,16 In light of the evidence that optimized tRNA abundance can indeed improve behavior, a clear understanding of the potential mechanisms by which abundance affects the host may provide further insight into the myriad aspects of orthogonality related to incorporation of heterologous OTSs for genetic code expansion. Thus, the focus of this work is another important aspect of orthogonality, i.e., the examination of other interactions with host machinery beyond only host aaRS enzymes. Identifying these potential points of interaction where heterologous tRNAs might interfere with the host cell machinery may provide avenues for rationally improving OTS behavior. Given the abundance of post-transcriptional modifications in host translational machinery (especially tRNAs) and their variegated functions, it is possible that interactions with host RNA-modifying enzymes contribute to nonorthogonal behavior of heterologous tRNAs. This is especially likely given the variation in modification profiles of tRNAs and rRNAs across all domains of life,17−20 which results in a new environment to which a heterologous tRNA must adapt. In native host translational machinery, PTMs play important roles in maintaining RNA stability and structure,21 responding to stress,19,20 and can even affect tRNA identity.22,23 Notably, one of the few studies with heterologous tRNAs in E. coli that investigated the role of modifications in orthogonality found that an inosine modification in the anticodon led to off-target effects of their Mj tRNATyr-derived (histidine) sense suppressor tRNACAU. This result demonstrated that modifications can similarly lead to alterations in heterologous tRNA fidelity.24 Importantly, modifying enzymes themselves can play important roles in cellular metabolism. For example, dihydrouridine synthase B (DusB) that reduces uridine to dihydrouridine in the D-loop of E. coli tRNAs has a regulatory effect on Fis, an enhancer that regulates up to 20% of the E. coli genome.25,26 Thus, it is conceivable that heterologous tRNA interactions with host modification machinery might result in subtle but meaningful changes to host metabolism. In this work, we surveyed a set of suppressor tRNA variants in an effort to understand how directed evolution led to improved tRNA behavior. We first gauged the abundance of the selected tRNA variants by Northern blotting, revealing that excessive amber codon readthrough in the absence of the cognate aaRS (i.e., poor orthogonality) correlated with high stable expression. Examination of mutations corresponding to the most significant improvements in orthogonality led us to gauge the effects of several E. coli post-transcriptional modification enzymes. We conducted this study by assaying orthogonality in RNA-modifying enzyme deletion strains



RESULTS Selection of Candidate Amber Suppressor tRNA Variants for Surveying Determinants of Orthogonality. In our previous work,27 we aimed to improve heterologous tRNA:aaRS pairs using a directed evolution scheme incorporating a tunable negative selection for orthogonality. In our systematic survey, we chose to examine the parent tRNA for these previous directed evolution experiments (tRNAOptCUA15). We also examine another tRNA from the literature (Nap128) relative to nine other variants developed by our selection: Ac.02, Ac.10, AcT.05, S2, S7, S9, L2, L6 and L9, where the first six but not the last three have been previously published.27 These 11 tRNAs span a range of levels of amber codon readthrough in the absence of cognate aaRS and provide an attractive case study for characterizing important features of heterologous suppressor tRNAs that improve orthogonal behavior. As a starting point, we examined the stable expression level of the 11 tRNAs by Northern blotting. As seen in Figure 1 (and in Supplementary Figures S1−S9), there are strong differences among the selected tRNA variants; on average, the nine evolved tRNAs show approximately 20-fold decrease in their stable expression level relative to the parent strain. The most abundant tRNAs also tended to display the least orthogonality (i.e., most amber readthrough in absence of cognate aaRS) among the tested variants in our previous work.27 This relationship suggests that reduced stability or expression level may be important for the improved orthogonality observed in the evolved tRNAs. In further examining the suite of suppressor tRNA variants, we looked to the sequence characteristics of the chosen variants (see Table 1).15,27,29 An important observation is that relatively small differences (5−6 mutations) in sequence from tRNAOptCUA to the next generation of tRNAs (Ac.02, Ac.10, AcT.05) can significantly alter tRNA behavior and abundance. Thus, we further examined the particular sequence changes that occurred at this evolutionary step, with the reasoning that these mutations might point to interactions with the host cellular network that are relevant to orthogonality. Notably, the most obvious mutations that seem to correspond to significant improvements in orthogonality occur at the anticodon stemloop, clustering largely around the sites of several known posttranscriptional modifications in E. coli tRNA (Table 1). Furthermore, we suspected that modifications may have an effect on heterologous tRNA stability since this is known to be one role for modified nucleosides in native tRNAs.30 Thus, we B

DOI: 10.1021/acssynbio.7b00421 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Research Article

ACS Synthetic Biology

speculated that perhaps the highly abundant tRNAs might interact significantly with post-transcriptional modification enzyme networks, some of which can have widespread effects on cellular metabolism. In particular, the anticodon domain of the high abundance tRNAs present several likely modification target sites, indicated in Figure 2a. One of these modification enzymes, TruA (which pseudouridylates at U38, U39, and U40 in a number of native tRNAs) affects growth rate and plays a role in translational fidelity under low histidine conditions.31,32 Moreover, another modification enzyme, MiaA that creates N6-isopentenyladenosine at A37 in a number of E. coli tRNAs, is known to control regulation of several proteins with global regulatory impacts (Figure 2b), including at least one other tRNA modification enzyme.33,34 Given the potential for anticodon modifying enzymes to also affect cellular regulation, we hypothesized that modification of heterologous tRNAs by anticodon targeting PTM enzymes might mediate interference with host machinery by the highly expressed tRNA variants which contributes to their less orthogonal behavior. Analysis of High Abundance Suppressor tRNAs Reveals That Their Orthogonality Is Impacted by Anticodon Modifying Enzymes. To assess how E. coli post-transcriptional modification enzymes might play a role in the poor orthogonal behavior associated with the high abundance tRNAs, we assayed the orthogonality of the two high abundance tRNA suppressors (Nap1 and tRNAOptCUA) by measuring GFP fluorescence of the sFastGFP-1×Amb reporter in the absence of the cognate heterologous aaRS.27 It is expected that completely orthogonal tRNAs display no fluorescence; in this case, if the orthogonal tRNA is truly not interfering with amber termination, translation termination should occur at the amber codon in the absence of the cognate

Figure 1. Orthogonality of evolved tRNAs correlates strongly with stable tRNA expression level by Northern blotting. Northern blotting analysis suggests that high stable expression level may explain the poor orthogonality behavior observed with the Nap1 and tRNAOptCUA variants. Comparison of the relative orthogonality (y-axis, see also Figures 3 and 4) to the normalized blot intensity (x-axis) of each tRNA demonstrates a strong correlation (R = 0.98, p