Engineered mRNA–ribosome fusions for facile biosynthesis of selenoproteins

Significance

Selenoproteins are proteins that contain the rare amino acid selenocysteine (Sec), typically found in their catalytic centers. Because Sec exhibits superior catalytic properties and can resist irreversible oxidative damage, many efforts have been made to improve natural enzymes by replacing their catalytic cysteine residues with Sec. While the current production of useful selenoproteins primarily relies on relatively expensive chemical synthesis, ongoing efforts aim to simplify their production by genetically encoding engineered selenoproteins. In this study, we present a simple engineering solution that “hijacks” the natural machinery of selenoprotein synthesis, enabling the site-specific insertion of Sec in target proteins. Therefore, our work provides a new basis for the development of robust systems for the facile synthesis of designer selenoproteins.

Abstract

Ribosomes are often used in synthetic biology as a tool to produce desired proteins with enhanced properties or entirely new functions. However, repurposing ribosomes for producing designer proteins is challenging due to the limited number of engineering solutions available to alter the natural activity of these enzymes. In this study, we advance ribosome engineering by describing a novel strategy based on functional fusions of ribosomal RNA (rRNA) with messenger RNA (mRNA). Specifically, we create an mRNA–ribosome fusion called RiboU, where the 16S rRNA is covalently attached to selenocysteine insertion sequence (SECIS), a regulatory RNA element found in mRNAs encoding selenoproteins. When SECIS sequences are present in natural mRNAs, they instruct ribosomes to decode UGA codons as selenocysteine (Sec, U) codons instead of interpreting them as stop codons. This enables ribosomes to insert Sec into the growing polypeptide chain at the appropriate site. Our work demonstrates that the SECIS sequence maintains its functionality even when inserted into the ribosome structure. As a result, the engineered ribosomes RiboU interpret UAG codons as Sec codons, allowing easy and site-specific insertion of Sec in a protein of interest with no further modification to the natural machinery of protein synthesis. To validate this approach, we use RiboU ribosomes to produce three functional target selenoproteins in Escherichia coli by site-specifically inserting Sec into the proteins’ active sites. Overall, our work demonstrates the feasibility of creating functional mRNA–rRNA fusions as a strategy for ribosome engineering, providing a novel tool for producing Sec-containing proteins in live bacterial cells.

In organisms like Escherichia coli, ribosomes can produce an average-sized protein in under 15 s while using over 20 distinct substrates and making just one error per approximately 3,000 amino acids (1). Due to these catalytic properties, many successful strategies were developed to harness ribosomes for the synthesis of chemically modified proteins and completely new protein-like polymers (2–12). For example, by mutating the anti-Shine–Dalgarno sequence of E. coli ribosomes, it became possible to create so-called orthogonal ribosomes that avoid cross-reacting with housekeeping mRNAs and instead recognize engineered mRNAs with matching Shine–Dalgarno sequences (13). Through mutations in the catalytic center, ribosome variants were produced with improved compatibility with unnatural substrates, including D-amino acids, β-amino acids, long-chain carbon amino acids, and dipeptides (14–16). Furthermore, tethering the two ribosomal subunits made it possible to express engineered ribosomes in bacteria without cross-reacting with natural ribosomes (17, 18), enabling the further ribosome engineering that would otherwise be toxic or lethal for a cell (19–21).

Among the current directions of ribosome engineering, one is aiming to produce a better system for the synthesis of selenoproteins (15). Selenoproteins, also known as Sec-containing proteins, are an important group of natural proteins that are characterized by the presence of the genetically encoded amino acid Sec. Sec is structurally similar to cysteine but bears a selenium atom instead of sulfur. This difference endows Sec with lower pK_a and higher nucleophilicity. Consequently, selenoproteins display higher catalytic rates and resistance to irreversible oxidation, making Sec-containing proteins work faster and longer compared to their cysteine-containing counterparts (22–24). Therefore, synthesis of designer selenoproteins is an attractive direction of synthetic biology that aims to produce proteins with artificial or enhanced catalytic functions.

However, our ability to produce designer selenoproteins is limited by the intricacies of the natural mechanism of selenoprotein synthesis which is distinct from the mechanism for other cellular proteins. In nature, Sec is encoded by UGA codons. To ensure that these UGA codons are translated as Sec instead of stop codons, they are 3′-flanked by a specialized 20 to 60 nucleotide-long sequence known as the SECIS element in mRNAs of selenoproteins (25). The SECIS element allows mRNAs of selenoproteins to recruit the elongation factor SelB, which in turn binds to the specialized tRNA^Sec that delivers Sec to the active site of the ribosome (26–28). Thus, the SECIS element fulfills a dual role in selenoprotein synthesis: it serves not only as a SelB-binding signal but also as a protein-coding segment of the mRNA. As a result, designing mRNAs to encode artificial selenoproteins can be challenging, unless the incorporation of Sec is very close to the C-terminus of the protein molecule (29–31).

To address this limitation, several strategies have been developed to encode selenoproteins by mRNAs that lack SECIS elements. For example, the engineered tRNA^UTu was designed to recognize the elongation factor Tu (EF-Tu), instead of the Sec-specific elongation factor SelB (SelB), and translate UAG stop codons, instead of UGA stop codons. Using tRNA^UTu allowed the insertion of Sec into cellular proteins in a SECIS-independent manner by redefining the rules of the genetic code. In this strategy, Sec-bound tRNA^UTu would allow for the insertion of Sec in response to every UAG codon in cellular mRNAs (32). As a result, despite their value, the currently existing strategies of selenoprotein biosynthesis exhibit inherent shortcomings as they require costly and, at times, impossible genome-wide alterations of the genetic code (33). Hence, alternative approaches for the biosynthesis of selenoproteins are needed.

In this work, we have addressed this need by reporting a novel approach to ribosome engineering for the synthesis of designer selenoproteins. By mutating the anti-Shine–Dalgarno sequence of the ribosome and by fusing the 16S rRNA of the ribosome with SECIS, we designed orthogonal ribosomes that translate the UAG stop codon as a Sec codon in reporter SECIS-free mRNAs. We used these engineered ribosomes, which we termed RiboU, to produce two active selenoproteins, namely formate dehydrogenase FDH_H and glutathione peroxidase Grx1. Our work provides the proof of principle for the general use of RiboU for recombinant synthesis of designer selenoproteins in live bacterial cells. On a broader level, our study describes a conceptually new approach to ribosome engineering based on creating chimeric molecules between rRNA and mRNAs.

Results

Rational Design of Engineered Ribosomes for Selenoprotein Biosynthesis.

In this study, we set out to engineer ribosomes capable of incorporating Sec at a specific site of a target protein in live bacterial cells. Because SECIS sequences recruit the machinery of selenoprotein synthesis to natural SECIS-mRNAs (Fig. 1A), we hypothesized that inserting the SECIS element into rRNA could produce ribosomes that are permanently associated with the selenoprotein synthesis machinery. Consequently, we anticipated that this association would enable ribosomes to translate UAG codons as Sec rather than stop codons (Fig. 1A).

Fig. 1.

Engineering RiboU, ribosome–mRNA fusions for synthesis of selenoproteins in live bacterial cells. (A) Schematic comparison of natural and engineered machinery for site-specific insertion of Sec in proteins. (B) Structure illustrating the close proximity of SECIS and the ribosomal helix h16 during the synthesis of selenoproteins in E. coli (pdb id 5LZD). (C) Sequence and secondary structures of the 16S rRNA helix h16 and the minimal SECIS. Two alternative insertion sites of SECIS in helix h16 are indicated by arrows. The inserted SECIS and h16 are separated by short linkers comprising 2 to 4 random nucleotides (N). (D) Schematic illustrating the modified rRNA-coding construct used to create RiboU.

To test this idea, we used an available structure of the E. coli ribosome bound to protein SelB and a SECIS-containing mRNA (34). This structure revealed that, during natural selenoprotein synthesis, both SECIS and SelB bind the ribosome in close proximity (approximately 11Å) to the helix 16 (h16) of the 16S rRNA (Fig. 1B). Therefore, we used h16 as an insertion site for a SECIS sequence to produce SECIS–16S rRNA fusions.

We then created a library of SECIS–16S rRNA fusion variants in which the h16 is covalently attached to the minimal functional SECIS from the mRNA of the E. coli selenoprotein FDH_H. This version of SECIS consists of 19 bases that are necessary and sufficient for SelB binding (Fig. 1C) (26). We inserted this SECIS sequence into the tip of the helix h16 (between the residues 420 and 421 or 424 and 425), using 2 to 4 nucleotides-long randomized linkers for each insertion site. Thus, we produced a library of genetically encoded variants of SECIS–16S rRNA fusions.

To ensure that these engineered SECIS–16S rRNA molecules do not bind natural mRNAs and rather associate with engineered mRNAs of interest, we mutated their anti-Shine–Dalgarno sequence to a previously characterized “orthogonal” anti-Shine–Dalgarno sequence known as OR2 (35). We have termed the resulting ribosome–SECIS fusions RiboU variants to show that this generation of engineered ribosomes recodes stop codons as Sec, which is designated as “U” in a single-letter amino acid abbreviation. We then cloned each SECIS–16S rRNA variant into a reporter plasmid under an inducible promoter for functional characterization and improvements (Fig. 1D).

Establishing an Experimental System for Testing RiboU Activity.

To experimentally test whether RiboU can produce selenoproteins in E. coli, we created a system based on a Sec-specific reporter gene and a genetically engineered E. coli strain.

To create the reporter gene, we used the E. coli fdhF gene as a starting point. This gene encodes the natural selenoprotein formate dehydrogenase, FDH_H, that bears a catalytically active Sec at position 140 (36). We have selected FDH_H because the activity of this protein, and hence the presence of Sec, can be readily monitored in vivo through colorimetry. In the wild-type mRNA of FDH_H, this residue is encoded by a UGA codon located adjacent to the SECIS sequence. To use the FDH_H as a bona fide reporter, we have introduced the previously characterized mutations that disrupt the activity of the natural SECIS sequence but preserve the fdhF open reading frame (26). We then replaced the natural UGA Sec codon with a UAG stop codon and then altered the Shine–Dalgarno sequence to match the OR2 sequence in RiboU. The resulting FDH_H gene was cloned under the control of its natural promoter into a plasmid containing SelC—the gene for tRNA^Sec in which the wild-type UCA anticodon was mutated to CUA anticodon to make tRNA^Sec recognize UAG codons instead of canonical UGA codons. As a result, this reporter gene allowed us to monitor FDH_H biosynthesis in the derivative of E. coli strain C321, in which all natural UAG codons have been changed to synonymous UAA termination codon (37).

To adapt E. coli for testing the RiboU activity, we further altered the C321 strain by deleting the genomic copies of fdhF and selC to avoid background FDH_H activity or a competition between RiboU and FDH_H mRNA for SelB binding (FDH_H mRNA is the only SECIS-containing mRNA in the wild-type E. coli). We co-transformed the resulting strain with plasmids encoding RiboU and the reporter FDH_H and then tested whether RiboU can govern FDH_H synthesis. We observed that 24 of the 32 tested RiboU-coding constructs could support the production of FDH_H activity with different efficiencies (Fig. 2A). These results suggested that the production of a functional selenoprotein from a SECIS-less mRNA is possible with the aid of the SECIS-containing ribosomes.

Fig. 2.

Calorimetric tests confirm the ability of RiboU to produce a reporter selenoprotein. (A) A benzyl viologen-based assay detects the activity of the Sec-containing protein FDH_H produced by RiboU. The 32 RiboU variants (labeled as 16S v1-v32) are compared in their ability to mediate selenoprotein synthesis in vivo, as measured by the activity of the selenoprotein FDH_H. (B) The same assay illustrates the impact of the SECIS insertion site in the helix h16 on the RiboU activity. Each of the six tested insertion sites is labeled by the 16S rRNA residue number (419 to 425) at the 5′-end of the SECIS insertion site (e.g., “419” means that the SECIS sequence was inserted between the bases 419 and 420 of the 16S rRNA). (C) Comparison of RiboU mutants illustrating that the 16S rRNA mutations U531G and U534A improve RiboU activity approximately 10-fold. (D) The same assay illustrates the further improvement of RiboU by using different combinations of the Shine–Dalgarno sequence in RiboU and the anti-Shine–Dalgarno sequence in mRNA of the reporter protein, FDH_H. Each tested variant was created using variant RiboU-v24-419, where the SECIS sequence was inserted between the bases 419 and 420 of the 16S rRNA.

RiboU Activity Can Be Improved by Changing the SECIS Insertion Site in the 16S rRNA.

We next tested whether the RiboU activity could be increased by optimizing the insertion site for the SECIS in the 16S rRNA. We reasoned that the activity of our initial RiboU variants could, in part, suffer from the suboptimal location of the SECIS in the ribosome structure. To test this idea, we compared six insertion sites for the SECIS sequence in h16. Using one of our most active RiboU variants, RiboU-v24-424-OR2 (Fig. 2A and SI Appendix, Table S1), in which the SECIS element was inserted between the bases G424 and G425 of the 16S rRNA with the linkers (5′-TTAA-SECIS-AAA-3′), we tested several alternative insertion sites that followed the bases 419, 420, 421, 422, 423, and 425 in the h16. When the SECIS was inserted between bases 419 and 420, or 420 and 421, RiboU variants displayed approximately fivefold higher activity compared to the original construct (Fig. 2B). We, therefore, selected one of these new RiboU variants (RiboU-v24-419-OR2) for further use and improvements. Overall, these results showed that RiboU activity can be markedly improved by optimizing the SECIS insertion site in the ribosome structure.

RiboU Is Compatible with Other Strains of E. coli, Albeit with Lower Activity Levels.

While we were encouraged by the selenoprotein synthesis observed in our model C321 strain that was designed for UAG decoding, our ultimate goal was to enable the use of RiboU in other laboratory strains with more advantageous fitness attributes. Therefore, we have tested RiboU activity in the E. coli strains B-95.ΔA strain (38) and BW25113 ΔfdhF from the Keio collection (39), which are commonly used to express recombinant proteins. Notably, the levels of FDH_H activity mediated by the RiboU-v24-419-OR2 variant were two to three orders of magnitude lower in these strains compared to C321 (compare the RiboU-v24-419-OR2 variants in the Fig. 2 B–D panels).

RiboU Activity Can Be Improved by Additional Mutations in the 16S rRNA.

Previously, certain 16S rRNA mutations were shown to enhance either the natural selenoprotein synthesis (C1100U and G1516U) (15) or the ribosomal ability to interpret UAG codons as sense codons (U531G, U534A) (40). Furthermore, the mutations U531G and U534A were shown to reduce ribosome’s affinity to the release factor RF1 (release factor 1), making RF1 less competitive with tRNAs for UAG recognition (40). We inserted these mutations into the RiboU-v24-419-OR2 construct and examined their influence on selenoprotein synthesis. We found that the mutation G1516U (and C1100U) did not enhance selenoprotein production, but the mutations U531G and U534A increased the RiboU activity approximately 10-fold, based on cell dilution measurements (Fig. 2C). Interestingly, this result was consistent even in the strain lacking RF1, suggesting that the effects of these mutations on UAG recoding occur at least in part in an RF1-independent manner.

Optimizing the SD/Anti-SD Sequences Increases RiboU Activity.

We also tested whether we could improve the production of selenoproteins by optimizing the recruitment of RiboU-v24-419-OR2 to its target mRNAs. We compared three alternative and previously described orthogonal SD/anti-SD pairs (35) and found that changing OR2 to the orthogonal SD/anti-SD sequence OR1 results in three orders of magnitude higher FDH_H activity (Fig. 2D). Thus, we produced a superior variant of RiboU-v24-419-OR1.

Increased RiboU Expression Enables the Efficient Synthesis of Recombinant Selenoproteins.

Because the colorimetry-based assay detects even trace quantities of FDH_H, we conducted additional testing of RiboU activity using a system that more accurately represents the production of recombinant selenoproteins. We used intein-based reporters, which produce an inactive precursor of a reporter protein that is converted into its active form through an autocatalytic self-splicing (41). For our assay, we employed a previously developed intein DnaB that requires an N-terminal cysteine or Sec residue for its self-splicing (Fig. 3A) (41, 42). We inserted this intein into the kanamycin resistance protein KanR, which allowed us to monitor Sec’s presence in KanR-DnaB protein by measuring kanamycin resistance (Fig. 3).

Fig. 3.

RiboU can be further improved by optimizing the 16S rRNA-SECIS expression system. (A) Schematic showing the intein cleavage mechanism. Cleavage is dependent on the Cys residue at the first position of the intein, allowing the surrounding extein sequences to be spliced together and form a functional reporter protein. Replacement with Sec, but not Ser, still allows cleavage and reconstitution of a functional reporter protein KanR. (B) KanR intein assay with Cys or UAG at the first position of the intein comparing growth of BW25113 cells harboring the original RiboU expression plasmid (low-copy) in LB-agar plates containing 50 μg/mL kanamycin (Kan). (C) KanR intein assay with Cys or UAG in B-95.ΔAΔfabR cells expressing the upgraded RiboU expression plasmid (higher copy) in LB-agar plates containing 50 or 75 μg/mL Kan. RiboU refers to RiboU-sec420-OR1 and o-WT indicates wild-type 16S rRNA with OR1 Shine–Dalgarno.

Our initial experiments did not reveal any measurable changes in kanamycin resistance in response to RiboU expression, suggesting that our RiboU construct was insufficient for large-scale selenoprotein synthesis (Fig. 3B). We, therefore, attempted to improve our technology by changing three parameters. First, we recloned our rRNA-SECIS fusion into a plasmid with a higher copy number, to increase from 10 to 15 to 20 copies per cell, according to previous studies (42). Second, we replaced the original promoter [the aTc (anhydrotetracycline)-inducible promoter P_LtetO-1] with an arabinose-inducible promoter to induce rRNA-SECIS transcription with a less toxic molecule. Third, we used a different E. coli strain. We reasoned that the initial strain, BW25113, may not be the most suitable for recoding the UAG codon. This is because this strain expresses RF1, which may compete with RiboU for the recognition of UAG codons. We, therefore, used the E. coli strain that has been genetically modified to knock out RF1 (B-95.ΔAΔfabR) (38). After implementing these changes, we observed a readily detectable emergence of kanamycin resistance (up to 75 µg/mL) in response to RiboU expression, indicating a significant expression of an active reporter selenoprotein (Fig. 3C). We have observed similar results using another reporter protein in which the intein DnaB was inserted into the superfolder GFP (SI Appendix, Fig. S1). Overall, these data show that RiboU can support the production of reasonably high levels of selenoproteins that can be readily detected through changes in antibiotic resistance or fluorescent-based assays.

Mass-Spectrometry Confirms Site-Specific Sec Insertion into a Recombinant Target Protein by RiboU.

Finally, we tested the ability of the RiboU-v24-419-OR1 variant to insert Sec residues in an accurate, site-specific manner. As a model selenoprotein, we used the small protein glutathione oxidoreductase from E. coli (Grx1). As a catalytic residue, Grx1 bears Cys11 in some species, including E. coli, or Sec11 in other species. Cys11 or Sec11 are required for Grx1 activity as their mutation to Ser11 inactivates Grx1 (36). Based on these Grx1 properties, we created three Grx1 variants, including Grx1(C11) as a positive control, Grx1(S11) as a negative control, and Grx1(UAG11) as a reporter protein to measure RiboU-mediated Sec insertion.

We then produced each Grx1 variant in B-95.ΔAΔfabR cells expressing RiboU-v24-419-OR1 and isolated the reporter proteins for biochemical analyses (SI Appendix, Fig. S2). The estimated protein yield of Grx1(UAG11) was 0.15 mg/mL for 1 L of cell culture. Our enzymatic activity assays showed that the Grx1(Ser11) variant was catalytically inactive, whereas both Grx1(C11) and Grx1(U11) displayed comparable activity (Fig. 4A). Our mass-spectrometry analysis confirmed the site-specific Sec insertion in the Grx1(UAG11), indicating Sec11 presence in 34.4% of the Grx1(UAG11), whereas the remaining 65.6% of UAG11 was translated as Gln, Tyr, and Ser, but not as Cys (Fig. 4B). The remaining Grx1(UAG11) variant contained Ser11, which is formed by the incomplete conversion of Ser-tRNA^Sec to Sec-tRNA^Sec (36). Together, these experiments showed that RiboU, although imperfect, enables accurate site-specific incorporation of Sec into recombinant proteins produced in live E. coli cells.

Fig. 4.

Mass-spectrometry analysis confirms site-specific insertion of Sec in a target protein by RiboU. (A) The glutathione oxidoreductase activity assay assesses the enzymatic activity of Grx1 variants produced using RiboU. Sec-containing Grx1 is denoted as Grx1 (UAG11). Grx1(S11) is used as a negative control, and Grx1(C11) is used as a positive control. (B) Mass-spectrometry analysis confirmed the accurate site-specific incorporation of Sec in position 11 of Grx1(U11).

How Does Protein Synthesis Efficiency by RiboU Compare to Those by Wild-Type Ribosomes?

Previously, mutations in h16 were shown to reduce the rate of translation (43). First, we compared the growth curves of E. coli cells expressing RiboU-v24-419-OR1 and wild-type orthogonal ribosomes. This experiment (SI Appendix, Fig. S3A) showed that RiboU-v24-419-OR1 had no detectable impact on cell fitness, illustrating that RiboU is non-toxic for E. coli cells.

To test whether SECIS insertion affects the rate of ribosomal protein synthesis (as measured by the overall yield of the reporter protein), we compared the amounts of the reporter protein GFP produced by either wild-type ribosomes or RiboU (SI Appendix, Fig. S3B). The results reveal that GFP synthesis was ~14-fold less efficient in cells expressing RiboU relative to wild-type ribosomes, showing that SECIS insertion does significantly affect the translation efficiency by the ribosome, at least for this reporter system (SI Appendix, Fig. S3B).

Discussion

In this study, we describe a new strategy of ribosome engineering to enable site-specific insertion of Sec in target proteins in live bacterial cells. By transplanting SECIS from the mRNA of a selenoprotein into 16S rRNA, we created ribosomes that differ from natural ribosomes in their ability to recognize UAG codons in mRNAs. Instead of reading these codons as stop codons, our engineered ribosomes, named RiboU, translate UAG codons as Sec, leading to a site-specific incorporation of Sec in a target protein. Furthermore, because RiboU ribosomes are designed to recognize only engineered mRNAs rather than natural mRNAs, they insert Sec in target proteins without requiring the genome-wide repurposing of UAG codons.

Our work expands our current ability to harness the activity of natural ribosomes for the purpose of synthetic biology. Specifically, we show that such a conserved mRNA segment as SECIS can be transposed into the ribosome without the loss of function, allowing us to create unnatural mRNA-ribosome hybrids for the synthesis of recombinant selenoproteins. While the molecular mechanism of RiboU activity remains to be confirmed, our data imply that fusing SECIS with helix h16 in the 16S rRNA of E. coli ribosomes creates ribosomes that are permanently associated with SECIS-binding partners like SelB and selenocysteinylated tRNA^Sec. As a result, these ribosomes appear to effectively translate UAG codons as Sec, leading to the site-specific incorporation of Sec in a target protein. Thus, our work demonstrates how adding 20 extra bases to ribosomal RNA can alter the ribosome's ability to interpret the genetic code, resulting in an identity switch of UAG codons from stop to sense codons.

Furthermore, our engineering of RiboU expands the repertoire of currently existing systems for producing target selenoproteins in live bacterial cells. Previously, strategies for site-specific insertion of Sec were developed through engineering of tRNAs (producing such tRNAs as tRNA^UTu, tRNA^UTuX, and tRNA^SecUX) and the elongation factor EF-Tu (36, 44, 45). In these previous studies, site-specific insertion of Sec was achieved in a SECIS-independent manner through a genome-wide reassignment of UAG codons from stop codons to Sec codons. As a result, these technologies were limited to their use in engineered E. coli strains lacking UAG codons. By contrast, using our engineered ribosomes with mutated Shine–Dalgarno sequences allows us to reassign UAG codons only in a single mRNA of interest, leading to the selective insertion of Sec into just a single target protein in wild-type E. coli. As a result, our approach produces functional selenoproteins without requiring the modification of genomes or the natural machinery of selenoprotein synthesis.

It is important to note that current RiboU variants exhibit limited efficiency, as evidenced by the low yields of Grx1 production (SI Appendix, Fig. S2). Specifically, the co-expression of RiboU with the Grx1(UAG11) reporter resulted in a production of 0.15 mg/mL of Grx1 per 1 L of cell culture. In contrast, Grx1 with Ser11 and Cys11 codons yielded higher protein amounts of 1.29 mg/mL and 1.18 mg/mL, respectively, highlighting the limited efficiency of RiboU in producing target selenoproteins. Moreover, current RiboU variants show limited efficiency at the level of Sec insertion in Grx1 protein. In our experiments, only 34.4% of the UAG11 codon was translated as Sec, with the remaining 65.6% corresponding to the incorporation of Ser, Gln, and Tyr. The incorporation of these non-Sec amino acids likely stems from the fact that their corresponding tRNAs can mistranslate UAG codons through near-cognate codon:anticodon base pairing (46). It is also possible that the incorporation of Ser stems from an incomplete transformation of Ser-tRNA^Sec to Sec-tRNA^Sec during the natural synthesis of Sec-tRNA^Sec.

Future optimization of RiboU should also consider the accuracy of ribosomal translation since some mutations in h16 slightly decreased the enzymatic activity of a ß-galactosidase reporter protein due to missense errors (43).

While RiboU technology currently is of limited efficiency, it provides evidence that fusing rRNA with mRNA regulatory elements can indeed result in functional rRNA–mRNA hybrids with altered capacity to read the genetic code of a living cell. We therefore believe that, with further optimization through directed evolution or fine-tuning the expression levels of the natural Sec machinery (e.g., tRNA^Sec), our system may provide a valuable tool for selenoprotein synthesis for biotechnology and fundamental science. In this regard, it is important to note that the present ribosome engineering strategy offers a means to use a similar approach for recoding other codons through subsequent engineering of the SECIS-bound molecules, such as Sec-tRNA^Sec and SelB.

Materials and Methods

Plasmid Assembly, Site-Directed Mutagenesis, and Sequencing.

The plasmids produced in this study are listed in SI Appendix, Table S1, and their sequences were confirmed using Sanger sequencing (QuintaraBio). The RiboU-coding constructs were generated using site-directed mutagenesis with PfuUltra II Fusion HS DNA polymerase (Agilent). The initial SECIS–16S ribosome construct RiboU-v1-420-OR2, where the minimal SECIS sequence (ATCGGTTGCAGGTCTGCACC) was inserted after the position 420 within the 16S rRNA gene, was created using the previously engineered rRNA-encoding plasmid pZA31-OR2 as a scaffold (35). Subsequent library variants were created by performing site-directed mutagenesis of RiboU-v1-420-OR2 with primers listed in SI Appendix, Table S2. All the reporter DNA constructs were assembled using the In-Fusion® HD Cloning Kit (Takara Bio).

The wild-type 16S rRNA (from pZA-OR1) and RiboU-v24-419-OR1-U531G-U534A, which contains the anti-SD sequence OR1 (35), were cloned into the high-copy plasmid pBAD under the arabinose-inducible promoter. To create this plasmid, the ribosome-coding DNA was amplified with PfuUltra II Fusion HS DNA Polymerase (Agilent), and the pBAD vector was amplified with Herculase II Fusion DNA Polymerase (Agilent). Then, the PCR products were purified from a 1% agarose gel using Monarch® DNA Gel Extraction Kit (New England Biolabs, NEB) and ligated using NEBuilder® HiFi DNA Assembly Master Mix (NEB). The ligation reaction was transformed into E. coli DH5α competent cells and successful clones were confirmed using Sanger sequencing (QuintaraBio). The kanR gene in the resulting plasmid was replaced with the specR gene to allow co-transformation of this plasmid with the KanR-M86-coding constructs.

The FDH_H reporter plasmid pAM69 was created using the vector pTrc99a encoding the fdhF gene with a UGA codon at position 140 and a C-terminal His6-tag (47). Site-directed mutagenesis was used to replace the UGA codon at position 140 to UAG and introduced recognition sites for restriction endonucleases NotI and BamHI between the rrnB transcriptional terminators T1 and T2. The natural fdhF SECIS element was replaced by the previously described scrambled sequence (26) using site-directed mutagenesis (primers 1,166, 1,176, 1,177, SI Appendix, Table S2), and the fdhF natural promoter together with the orthogonal SD sequence OR2 (35) was inserted upstream of the fdhF gene. The reporter variants with alternative orthogonal SD sequences were generated by site-directed mutagenesis. The E. coli selC gene (encoding tRNA^Sec bearing a CUA anticodon) under a constitutive lpp promoter was inserted using the NotI and BamHI cut sites. All PCR reactions were performed with PfuUltra II Fusion HS DNA Polymerase (Agilent) and ligated with In-Fusion® HD Cloning Kit (Takara Bio).

The three E. coli codon-optimized sfGFP-M86 (Cys, Ser, or UAG) reporter variants (SI Appendix, Table S1), each with a C-terminal His₆-tag, were individually amplified with a P_trc promoter and the OR1 SD sequence (35). These variants were then inserted into the FDH_H reporter plasmid, replacing the fdhF gene and its promoter. The M86 DnaB mini-intein was inserted after position 204 of GFP, and the variants contained the TGC (Cys), AGC (Ser), or TAG (Sec) codon at position 1 of the intein. The three KanR-M86 variants with a C-terminal His₆-tag were also amplified with a P_trc promoter and the SD sequence of OR1 (35), replacing the fdhF promoter and gene on the FDH_H reporter plasmid. The M86 DnaB mini-intein was inserted after residue 153 of KanR, and each variant contained either TGC (Cys), AGC (Ser), or TAG (Sec) codon at position 1 of the intein (41, 42).

Similarly, E. coli Grx1(11UAG/14Ser) with a C-terminal His6-tag was amplified with a P_trc promoter and the SD sequence of OR1 (35) and inserted into the FDH_H reporter plasmid where it replaced the fdhF promoter and gene. Position 11 of Grx1 was mutated to TGC (Cys) or AGC (Ser) through site-directed mutagenesis. All PCR reactions were performed with PfuUltra II Fusion HS DNA Polymerase (Agilent), and the products were ligated using NEBuilder® HiFi DNA Assembly Master Mix (NEB).

Bacterial Strains.

For the initial tests of RiboU activity, the E. coli C321.ΔA.opt strain (48) was genetically modified by replacing the chromosomal fdhF gene with Zeocin resistance gene (zeo) and deleting the selC (tRNA^Sec) gene to generate the strain C321.ΔA.opt.fdhF::zeo.ΔselC (AS374). B-95.ΔAΔfabR and B-95.ΔA were described previously (38, 49). BW25113 ΔfdhF is part of the Keio collection (39).

FDH_H Activity Assays.

Electrocompetent cells were co-transformed with plasmids encoding the ribosome construct and the reporter vector. Cells were recovered after electroporation at 30 °C in Super Optimal Broth and were plated on Luria-Bertani (LB)-agar plates (100 µg/mL ampicillin, 35 µg/mL chloramphenicol). Plates were left at 37 °C for 24 h, and single colonies were picked and grown overnight in liquid cultures (100 µg/mL ampicillin, 34 µg/mL chloramphenicol) at 37 °C. Cultures were diluted in 10-fold serial dilution increments, and 7.5 µL of each dilution were spotted on LB-agar plates (100 µg/mL ampicillin, 34 µg/mL chloramphenicol, 1.5% agar, 50 mM sodium formate, 5 µM sodium selenite, 1 µM sodium molybdate, 200 ng/mL aTc). The plates were incubated under anaerobic conditions (90% N₂, 5% H₂, 5% CO₂) at 30 °C for 24 h, and the top agar layer (0.75% agar, 1 mg/mL benzyl viologen dichloride, 250 mM sodium formate, 25 mM potassium dihydrogen phosphate) was poured over the plates in an anaerobic chamber (90% N₂, 5% H₂, 5% CO₂) and left for 30 min for the spots to develop before imaging.

Intein-Based Kanamycin Resistance Assay.

Assays were performed as previously described (42). Briefly, plasmids encoding the engineered ribosome (RiboU-v24-419-OR1-U531G & U534A_spec) and KanR-M86 variant were electroporated into BW25113 or B-95.ΔAΔfabR E. coli cells (49) and plated on LB-agar plates [50 µg/mL ampicillin (KanR-M86) and 50 µg/mL spectinomycin (RiboU-v24-419-OR1-U531G & U534A_spec) or 25 µg/mL chloramphenicol (RiboU-v24-419-OR1-U531G & U534A)]. Plates were incubated at 37 °C overnight. Single colonies were picked the following day and grown in 5 mL LB cultures [50 µg/mL ampicillin, 50 µg/mL spectinomycin, or 25 µg/mL chloramphenicol, 0.1% arabinose, 10 µM sodium selenite, 1 mM isopropyl β-D-a-thiogalactopyranoside (IPTG)] at 37 °C overnight. Cultures were diluted to absorbance at 600 nm of 2.0, and 5 µL from each condition were spotted on LB-agar plates (50 µg/mL ampicillin, 50 µg/mL spectinomycin, or 25 µg/mL chloramphenicol, 0.1% arabinose, 10 µM sodium selenite, 1 mM IPTG) containing 50, or 75 µg/mL kanamycin. Plates were left at 37 °C overnight and imaged the next day using the ChemiDocTM MP Imager (Bio-Rad).

Intein-Based sfGFP Fluorescence Assay.

Assays were performed as previously described (42). Briefly, plasmids encoding the orthogonal WT ribosome and sfGFP-M86 variant were transformed into electrocompetent B-95.ΔAΔfabR E. coli cells (49) and plated on LB-agar plates (50 µg/mL ampicillin, 50 µg/mL spectinomycin). Plates were left at 37 °C overnight and four single colonies from the plate were grown separately in 0.5 mL LB (50 µg/mL amp, 50 µg/mL spec, 0.1% arabinose, 10 µM Na₂SeO₃, 1% glucose) at 37 °C for 8 h. Cultures were mixed 1:1 with fresh media (50 µg/mL amp, 50 µg/mL spec, 0.1% arabinose, 10 µM Na₂SeO₃, 1% glucose) and transferred to a 96-well black plate with clear bottoms for a final volume of 150 µL per well with and without 1 mM IPTG. Fluorescence (Ex. 485 nm, Em. 528 nm) and A600 readings were taken every 15 min for 24 h at 37 °C on the Synergy HTX Plate Reader (BioTek). The fluorescence readings were divided by the cell densities (A₆₀₀) at 24 h and were graphed using GraphPad Prism 9.

Translation Efficiency assay.

E. coli BW25113 cells harboring plasmids encoding the orthogonal WT ribosome (OR1) or RiboU-v24-419-OR1-U531G & U534A together with pAM181 (WT GFP-OR1). Cells were grown with antibiotics (100 µg/mL ampicillin and 35 µg/mL chloramphenicol), 0.1 mM IPTG, 150 ng/µL aTc, 20 µM Na₂SeO₃. Cultures were started by mixing 2 µL of overnight cultures with 123 µL of fresh media. Fluorescence (Ex. 485 nm, Em. 528 nm) and A600 readings were taken every 1 h for 24 h at 37 °C on the Synergy HTX Plate Reader (BioTek). The fluorescence readings were divided by the cell densities (A₆₀₀) at 24 h and were graphed using GraphPad Prism 9.

Grx1 Expression and Purification

Expression and purification of E. coli Grx1 variants were performed as previously described with minor changes (44). Briefly, electrocompetent E. coli B-95.ΔAΔfabR cells were co-transformed with pCZC1 and one of the pGrx1 variants and plated on LB-agar plates (50 µg/mL ampicillin, 50 µg/mL spectinomycin). Plates were incubated at 37 °C overnight, and single colonies were picked the following day and grown in 10 mL LB cultures (50 µg/mL ampicillin, 50 µg/mL spectinomycin) at 37 °C overnight. The overnight culture was used to inoculate 1 L LB (50 µg/mL ampicillin, 50 µg/mL spectinomycin, 0.1% arabinose, 10 µM sodium selenite), and the culture was grown at 37 °C until reaching an A₆₀₀ of 1.2. The temperature was reduced to 20 °C until an A₆₀₀ of 1.5 was reached, and protein expression was induced by adding 100 µM IPTG for 18 h at 20 °C.

All protein purification steps were performed under anaerobic conditions (90% N₂, 5% H₂, 5% CO₂) in an anaerobic tent (Coy Laboratories). Each cell pellet was resuspended in 18 mL of lysis buffer (50 mM sodium phosphate [pH 8.0], 300 mM sodium chloride, 10% glycerol, 30 mM imidazole) and 2 mL BugBuster® 10× Extraction Reagent (EMD Millipore) supplemented with 2 mM β-mercaptoethanol and 0.05 mg/mL lysozyme. The resuspended cells were incubated at room temperature for 20 min with occasional mixing, and the lysate was centrifuged at 38,000 rpm (~235,000 × g) for 45 min at 4 °C. The supernatant was loaded onto 2 mL of nickel resin (Ni-NTA Agarose, Qiagen) pre-equilibrated with lysis buffer. The beads were washed with 40 mL of the lysis buffer and eluted in 1 mL fractions using the elution buffer (50 mM sodium phosphate [pH 8.0], 300 mM sodium chloride, 10% glycerol, 230 mM imidazole). Protein elution was monitored using the Bio-Rad Protein Assay Dye. Eluates were concentrated and buffer-exchanged to the storage buffer (50 mM sodium phosphate [pH 8.0], 300 mM sodium chloride, 10% glycerol) using Amicon® Ultra Centrifugal Filters (Merck Millipore). The samples were then stored at −80 °C.

Grx1 Oxidoreductase Activity Assay.

The assay was performed as previously described (44). Briefly, 35 µL of a 20 mM β-hydroxyethylene disulfide solution was preincubated with 950 µL of a buffer (1 mM reduced glutathione, 100 mM Tri-HCl [pH 8.0], 2 mM Ethylenediaminetetraacetic acid (EDTA), 1 mg/mL bovine serum albumin, 0.4 mM of reduced nicotinamide adenine dinucleotide phosphate (NADPH) and 6 µg/mL glutathione reductase) for 2 min at room temperature. The reaction was initiated by adding 10, 25, or 50 nM Grx1 variant (in triplicates), and the consumption of NADPH was measured at 340 nm every 30 s for 3 min. The change in absorbance over time was calculated from 1 to 2 min and averaged between the three measurements, followed by buffer subtraction. The values were then plotted against the concentration of Grx1 used. The slope and SE of this line were used to calculate the activity of the enzyme (in µmol of NADPH/min/mg protein) using the molecular weight of Grx1 (10,804 Da) and NADPH extinction coefficient (ε = 0.00622 µM⁻¹cm⁻¹).

Mass-Spectrometry.

Protein samples were prepared for liquid chromatography-mass spectrometry/mass spectrometry (LC–MS/MS) analysis at Bioinformatics Solutions Inc. (Waterloo, ON, Canada), as previously described (50). Briefly, samples were reduced with 10 mM dithiothreitol (DTT) (Sigma-Aldrich) alkylated with 20 mM iodoacetamide (Sigma-Aldrich), acetone precipitated, and digested overnight with MS grade trypsin (Promega). Digested samples were lyophilized. Lyophilized samples were resuspended in 0.2% trifluoroacetic acid (TFA) and desalted using a C18 spin column. C18 desalted samples were resuspended in 12 µL buffer A (0.1% TFA in water). Six µL of each sample was injected into the timsTOF Pro (Bruker Daltronics) by nanoflow liquid chromatography using a Bruker NanoElute chromatography system (Bruker Daltronics). Liquid chromatography was preformed using a constant flow of 300 µL/min and a 15 cm reversed-phased column with a 75 µm inner diameter filled with Reprosil C18 (PepSEP). Mobile phase A was 0.1% formic acid and mobile phase B was 99.9% acetonitrile, 0.1% formic acid. Peptide separation was carried out over 30 min as follows: linearly 2% A to 35% B over 30 min with an increase to 95% B over 30 s and held constant for 2.62 min to clean the column. Column equilibration was done prior to automatic sample loading at a temperature of 50 °C.

The timsTOF Pro was outfitted with a Captivespray source (Bruker Daltronics), operated in PASEF mode. Trapped ion mobility separation was achieved by using an accumulation time of 100 ms in the first TIMS region and ramps of the TIMS region from 0.85 Vs/cm² to 1.30 Vs/cm², with each ramp lasting 100 ms.

MS and MS/MS scans were limited to 100 m/z to 1,700 m/z, and a polygon filter was applied to the m/z and ion mobility dimensions to select for multiple charged ions most likely to be peptide precursors. Collision energy was applied as a function of ion mobility with a linear regression using the following parameter settings: 0.85 Vs/cm² → 27 eV, 1.30 Vs/cm² → 45 eV. TIMS voltage was calibrated using ions from the Agilent Tune Mix (m/z 622, 922, 1,222). Active Exclusion of MS/MS scans was enabled at a setting of 0.40 min. Quadrupole isolation was set to 2 m/z for m/z less than 700, and 3.0 m/z for ions with an m/z greater than 800. A linear regression calculation was done automatically for ions between m/z 700 and 800. All mass-spectrometry experiments were completed at the mass-spectrometry lab of Bioinformatics Solutions Inc.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All data are included in the article and/or SI Appendix.