Functional integration of a semi-synthetic azido-queuosine derivative into translation and a tRNA modification circuit

Abstract Substitution of the queuine nucleobase precursor preQ1 by an azide-containing derivative (azido-propyl-preQ1) led to incorporation of this clickable chemical entity into tRNA via transglycosylation in vitro as well as in vivo in Escherichia coli, Schizosaccharomyces pombe and human cells. The resulting semi-synthetic RNA modification, here termed Q-L1, was present in tRNAs on actively translating ribosomes, indicating functional integration into aminoacylation and recruitment to the ribosome. The azide moiety of Q-L1 facilitates analytics via click conjugation of a fluorescent dye, or of biotin for affinity purification. Combining the latter with RNAseq showed that TGT maintained its native tRNA substrate specificity in S. pombe cells. The semi-synthetic tRNA modification Q-L1 was also functional in tRNA maturation, in effectively replacing the natural queuosine in its stimulation of further modification of tRNAAsp with 5-methylcytosine at position 38 by the tRNA methyltransferase Dnmt2 in S. pombe. This is the first demonstrated in vivo integration of a synthetic moiety into an RNA modification circuit, where one RNA modification stimulates another. In summary, the scarcity of queuosinylation sites in cellular RNA, makes our synthetic q/Q system a ‘minimally invasive’ system for placement of a non-natural, clickable nucleobase within the total cellular RNA.


INTRODUCTION
Post-transcriptional modification of tRNAs is a ubiquitous yet idiosyncratic feature with versatile chemical structures contributing to stability and folding, as well as fidelity of decoding and translational control (1)(2)(3). The largest variety of chemical structures in RNA is found in the anticodon loop which directly interacts with the mRNA during decoding in the translating ribosome. The chemical variety of the more than 170 modifications known to date is dominated by tRNA anticodon modifications occurring at positions 34 and 37, ranging from simple methylations to highly complex structures of which queuosine (Q) is a particular case (4,5). In both, prokaryotes and eukaryotes, this hypermodified 7-deazaguanosine is exclusively found in the anticodon wobble position 34 of tRNAs containing a G 34 U 35 N 36 motif and therefore specific for a selected group of four tRNAs, namely tRNA Asn , tRNA Asp , tRNA His and tRNA Tyr (6,7). In an intricate multi-step process involving various enzymes and co-factors, Escherichia coli and other prokaryotes are capable of first synthesising the modified precursor base 7-aminomethyl-7-deazaguanine (preQ 1 ) de novo. GTP is converted to preQ 1 via five enzymatic steps successively catalysed by the GTP cyclohydrolase I (GCH I), QueD, QueE, QueC and QueF (8)(9)(10)(11). As a rare type of post transcriptional modification, the noncanonical nucleobase structure is then introduced into tRNA in an exchange reaction. During this transglycosylation step, the bacterial tRNA guanine transglycosylase (bTGT) replaces a guanosine in the anticodon wobble position of cognate tRNAs with the precursor base preQ 1 (12,13) which is then further enzymatically modified by QueA and QueG to yield the final queuosine structure (14,15). In contrast to prokaryotes, eukaryotes salvage the nucleoside queuosine and the corresponding nucleobase queuine (q) from environmental sources including the gut microbiota (reviewed in (5)). Queuosine is hydrolyzed by a queuosine nucleoside glycosylase to release q (16). The incorporation of the salvaged q into tRNA is catalysed by eukaryotic TGT (eTGT), which is a heterodimeric enzyme (17)(18)(19) composed of a catalytic queuine tRNA-ribosyltransferase subunit 1 (QTRT1) and a noncatalytic queuine tRNAribosyltransferase subunit 2 (QTRT2) (20).
Despite its suggestive positioning at position 34 of the tRNA anticodon, molecular details of the physiological relevance of Q remain scarce. It is generally accepted that Q impacts the decoding process on the translating ribosome, with cumulative evidence pointing to pivotal interactions at the A-site. The specific occurrence of Q in GUN anticodons is consistent with a general concept by Grosjean and Westhof (21), wherein modifications at position 34 compensate for the lower stability of codon-anticodon interactions including 2 or more base pairs with less than three hydrogen bonds. This concept receives support from computational modelling, which characterised a stabilising effect of Q on the overall tRNA-mRNA complex involving additional hydrogen bonds (22). In vivo and in cellulo studies did not reveal any strong phenotypes under Q deficiency or in strains lacking TGT. However, the presence of Q improved viability under stress conditions and affected translation accuracy in E. coli (23,24). In vivo studies in eukaryotes likewise reported an impact on the decoding process, enabling decoding of synonymous codons by wobble base pairing (22,25,26), and affecting translation speed and accuracy (27,28). Queuosine's multifaceted involvement in the cellular machinery was reported to be associated with cancer (29)(30)(31)(32), neuronal disorders (33)(34)(35) as well as bacterial and parasitic infection (36,37). Consequently, the perception of therapeutic potential associated with its biogenesis has consistently increased, in keeping with a general trend in epitranscriptomics So far, the only demonstrated molecular interaction affected by Q outside the ribosome is a so-called modification circuit with 5-methylcytidine (m 5 C) in position 38 of Schizosaccharomyces pombe tRNA Asp , stimulating its formation by the Dnmt2 homologue Pmt1 (38,39). Structural analysis suggested that the presence of Q34 leads to optimal positioning of the interacting substrates in the active site of Dnmt2, enhancing the catalytic efficiency of the methyltransferase (40).
Arguably, approaches to a deeper understanding of the molecular action of Q in living cells would need to involve manipulations of details of the structure of Q, e.g. via an incorporation of q-derivatives through transglycosylation. Apart from their natural substrates, both bTGT and eTGT have been shown to tolerate a certain variety of synthetic analogues harbouring large functional groups in vitro (41,42). Leveraging the short hairpin recognition motif of the bTGT installed on different RNA transcripts, Devaraj and co-workers developed a method called RNA-TAG (transglycosylation at guanosine), allowing to site-specifically incorporate analogues in vitro, which contained large fluorophores or affinity labels for pulldown experiments. This method was also applied to visualize mRNA transcripts containing the recognition motif in a fixed cell environment in a direct one-stepreaction (42) and extended to the development of a lightactivated mRNA translation system (43). Furthermore, RNA-TAG was used on modified mRNAs in a twostep-approach, incorporating a preQ 1 -derivative bearing a bioorthogonal tetrazine moiety in the first step, and thus enabling further derivatization by IEDDA click chemistry in a second step (44). However, labelling with clickcompetent compounds in vivo or in cellulo has not yet been achieved in the queuosine field. Indeed, there is strongly suggestive, albeit indirect evidence of successful in vivo incorporation of a non-natural q-analogue as published by Kelly and co-workers in the context of an animal model of multiple sclerosis (34). In addition to concerns about cell permeability of a q-derivative, important aspects to determine for in vivo labelling studies would include the physiological impact of an artificial chemical structure in a functioning tRNA, which would primarily be expected on the level of translation.
In this study, we metabolically label tRNA with a preQ 1 derivative functionalized with an azide group, allowing for further derivatization by click reaction and thus facilitating the proof of successful incorporation as well as the isolation of accordingly tagged RNAs. The latter was combined with RNAseq, in order to re-investigate the RNA substrates of the TGT, which turned out to be specific for the previously reported tRNAs Asn, Asp, His and Tyr. While in previous studies the transglycosylation step was performed in a fixed cell environment, we herein focus on the incorporation of the analogues by the natively expressed TGT in vivo and the physiological consequences in the natural environment. Polysome preparations revealed an enrichment of Qcontaining tRNAs in the polysomal fraction, indicating a targeted selection for modified tRNA to be integrated in the translational process. Moreover, our data demonstrate that the semi-synthetic tRNA modification replaces Q34 and is functionally integrated into the translational process, as well as in the modification circuit with m 5 C38 in tRNA Asp in S. pombe.

MATERIALS AND METHODS
S. pombe strains used in this study are given in Supplementary Table S1, Plasmids used in this study are given in Supplementary Table S2, oligonucleotides used in this study are given in Supplementary Table S3. The names and versions of all software used are provided in Supplementary Table S4.

Recombinant expression and purification of bTGT
The pASK-IBA13plus vector expressing the Zymomonas mobilis TGT (bTGT) with a N-terminal Strep-tag II was kindly provided by Prof. Dr Klaus Reuter (Philipps-University, Marburg). Expression and purification were carried out as previously described with minor changes (46). Briefly, the TGT was expressed in E. coli BL21-CodonPlus (DE3)-RIPL cells, grown in 2× YT medium and protein production was induced using anhydrotetracycline to a final concentration of 0.2 mg/l. After growing the cells for 14 h at 15 • C, cells were harvested and the cell pellets were stored at -80 • C until further processing. To purify the bacterial TGT, cells were thawed in lysis buffer (100 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8.0, 2 mM PMSF, 1 g/ml leupeptin, 1 g/ml aprotinin, 1 g/ml pepstatin and 25 U of DNase I and RNase I, respectively). After sonication (60% amplitude, 6 min, 0.5 s on, 2 s off; Sonifier 250 D, Branson), soluble proteins were isolated by centrifugation at 20 000 g for 1 h, 4 • C. Affinity chromatography was then used to purify the Strep II-tagged TGT. For this purpose, the lysate was incubated with Strep-Tactin ® Superflow Plus resin (Qiagen) for 3 h at 4 • C, 15 rpm. After washing with washing buffer (100 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8.0), the protein complex was eluted in 100 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8.0 and 2.5 mM desthiobiotin. Further purification was achieved by Superdex S200 (GE Healthcare) size exclusion chromatography (10 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8.0). The purified bTGT was stored at -80 • C in 10 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8.0 with 50% glycerol.

Isolation of total tRNA from E. coli
To isolate total tRNA, E. coli cells were grown to an OD 600 of 1 in 50 ml cultures and harvested by centrifugation (10 min, 10 000 g, 4 • C). The RNA was extracted by using the RNA isolation reagent TRI Reagent C (Sigma-Aldrich, Germany) following the manufacturer's instructions and dissolved in MQ-water.

Polysome preparations from E. coli
For polysome preparations the E. coli cells were grown in M9 medium in 150 ml culture volume until they reached an OD of 0.6, chloramphenicol was added to final concentration of 100 g/ml and after further incubation of 3 min the cells were harvested by centrifugation (10 min, 10 000 g, 4 • C). For cell lysis, cell pellets were resuspended in buffer (100 mM NH 4 Cl, 10 mM MgCl 2 , 20 mM Tris, pH 7.5), lysozyme was added and freeze-thaw cycles in liquid nitrogen were performed. Subsequent to this 10% deoxycholate was added to complete lysis, remaining cell wall debris were separated by centrifugation (12 000 g, 10 min, 4 • C). Sucrose gradients from 5 to 40% were generated using a Biocomp gradient station model 108 (settings: time 1.24 min, angle 81.5 • , speed 21 rpm) and lysate was loaded on top of the gradient. After ultracentrifugation (Beckman Ultracentrifuge Optima LE-80K, SW40 Ti rotor from Beckman Coulter) at 150 000 g and 4 • C for 2.5 h, gradients were fractionated by measuring the absorbance at 280 nm (Biocomp Gradient Station model 108 in combination with Gilson Fraction Collector FC203B). Total RNA was isolated from the respective fractions using TRI reagent C (Sigma-Aldrich).

Purification of total tRNA from collected fractions by gel elution
Total RNA extracted from polysomal fraction was separated on a 10% denaturing PAGE gel, stained with GelRed (Biotium) and the bands were visualized on Typhoon 9400 at an excitation wavelength of 532 nm. According to the resulting image, bands of interest were excised from the gel and mashed with a scalpel. The mashed gel pieces were frozen for 1 h and 300 l of 0.5 M ammonium acetate were added. Subsequently, the samples were shaken at 25 • C and 750 rpm overnight. The gel suspension was filtered through NanoSep® centrifugal filters and the filtrate was precipitated with three volumes of 100% ethanol.

S. pombe strains, plasmids and growth conditions
The S. pombe strains and plasmids used in this study are shown in Supplementary Table S1. Cells were cultured in YES (5 g/l yeast extract, 30 g/l glucose, 250 mg/l adenine, 250 mg/l histidine, 250 mg/l leucine, 250 mg/l uracil, 250 mg/l lysine) which did not contain queuosine or queuine. Synthetic queuine (kindly provided by Hans-Dieter Gerber and Gerhard Klebe (Universität Marburg) (47)) and preQ 1 derivatives were added to 0.1 M to the culture.

Isolation of total RNA and small RNAs from S. pombe
To isolate total RNA, S. pombe cells were grown to an optical density at 600 nm (OD 600 ) of 1 in 50 ml cultures. 50 OD of cells were harvested and 1 ml of phenol, glass beads were added. After vigorous shaking for 5 min, samples were centrifuged at 20 000 g for 5 min to clear the cell debris. Equal volume of phenol/chloroform/isoamylalcohol was added to the aqueous phase and centrifuged at 20 000 g for 5 min. After mixing the upper phase with an equal volume of chloroform followed by centrifugation at 20 000 g for 5 min, the RNA was precipitated at -80 • C for 1 h using 0.7 volume of isopropyl alcohol. Following precipitation, total RNA was washed with 70% ethanol and eluted in DEPCtreated water.
Isolation of small RNAs was performed using the PureLink™ miRNA Isolation Kit (Invitrogen) according to the manufacturer's instructions. Yeast cells were grown to an OD 600 of 1 in 5 ml cultures. After harvesting 1 OD of cells, RNAs were isolated using 1 ml TriFast reagent (Peqlab), 0.2 ml chloroform and glass beads. After vigorous shaking for 2 min, samples were centrifuged at 16 000 g, 4 • C for 15 min. After adding 215 l ethanol to the aqueous phase, the samples were transferred to a spin cartridge followed by centrifugation at 12 000 g for 1 min. 700 l ethanol was added to the flow-through and the sample was transferred to a new spin cartridge. Following centrifugation at 12 000 g for 1 min, the cartridge was washed and small RNAs were dissolved in DEPC-treated water. Northern blot-acryloyl aminophenylboronic acid (APB) gels were performed as previously described (28).

Removal of ribosomal RNA
Depletion of ribosomal RNA was performed as previously described (48). Oligonucleotides specific for 5.8S and 5S rRNA were ordered with a 5'-biotin tag from Metabion (see Supplementary Table S3). The oligonucleotides were diluted to 100 M each in nuclease-free water and equal volumes of the 100 M stock were combined to generate the rRNA depletion mix.
For hybridization, 8 g of small RNAs were incubated with 9.92 l of the 100 M rRNA depletion mix in reaction buffer (10 l of formamide, 2.5 l of 20× SSC (3 M NaCl, 0.3 M sodium citrate, pH 7.0) and 5 l of 0.005 M EDTA, pHnbsp;8). Reactions were carried out in a total volume of 50 l with the following thermocycling: 80 • C for 5 min, ramp down to 25 • C at intervals of 5 • C per minute. Following hybridization, 2 l of RNase-OUT (Invitrogen) and 50 l of 1× SCC containing 20% formamide were added. Removal of rRNA/oligonucleotide hybrids was performed using Dynabeads TM MyOne TM Streptavidin C1 (ThermoFisher) according to the manufacturer's instructions. 500 l streptavidin coated magnetic beads were washed as instructed for immunoprecipitation of RNA and added to the hybridization reaction. After incubation for 15 min at room temperature with mild agitation and bead separation on a magnetic rack, the supernatant was once more incubated with 500 l of washed beads for 15 min at room temperature under mild agitation followed by bead separation. Subsequently, the supernatant containing the 5S/5.8S rRNA-depleted RNA was precipitated with 1/10 volume of ammonium acetate and three volumes of 100% ethanol.

RNA substrates for in vitro modification
The S. pombe tRNA Asp substrate was prepared as previously described (49). Briefly, the pJET1 vector carrying the tRNA Asp sequence was linearized with NcoI, and 2.5 g of the linear vector was used for in vitro transcription using the TranscriptAid T7 High Yield Transcription Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Following an 8 h incubation at 37 • C with nucleotides and the T7 RNA polymerase and subsequent DNase I treatment, the respective tRNA was purified from the reaction using phenol/chloroform extraction followed by gel filtration with Sephadex G50 (GE Healthcare).

Recombinant expression and purification of hTGT
The pCDF-Duet1 vector co-expressing the human TGT (hTGT) heterodimer QTRT1 and QTRT2 with a cleavable N-terminal 6xHis tag to QTRT1 was kindly provided by Prof. Dr. Ralf Ficner (GZMB, Göttingen). Expression and purification were carried out as previously described with minor changes (20). Briefly, the heterodimer QTRT1/QTRT2 was co-expressed in E. coli (DE3) Rosetta cells, and protein production was induced using autoinduction. After growing the cells for 50 h at 18 • C, cells were harvested and the cell pellets were stored at -80 • C until further processing. To purify the human TGT, cells were thawed in lysis buffer (50 mM HEPES pH 7.5, 100 mM NaCl, 10 mM imidazole, 2 mM PMSF, 1 g/ml leupeptin, 1 g/ml aprotinin, 1 g/ml pepstatin and 25 U of DNase I and RNase I, respectively). After sonification (60% amplitude, 6 min, 0.5 s on, 2 s off; Sonifier 250 D, Branson), soluble proteins were isolated by centrifugation at 20 000 g for 1 h. Affinity chromatography was then used to purify the 6xHis tagged QTRT1/QTRT2 complex. For this purpose, the lysate was incubated with Talon ® Superflow™ resin (Cytiva) for 3 h at 4 • C, 15 rpm. After washing with washing buffer (50 mM HEPES pH 7.5, 100 mM NaCl, 10 mM imidazole and 1 M LiCl), the protein complex was eluted in 50 mM HEPES pH 7.5, 100 mM NaCl and 500 mM imidazole. Further purification was achieved by Superdex S200 (GE Healthcare) size exclusion chromatography (20 mM HEPES pH 7.5, 100 mM NaCl). The purified hTGT was stored at -80 • C in 20 mM HEPES pH 7.5, 100 mM NaCl with 50% glycerol.

In vitro labelling of tRNA with preQ 1 derivatives
For in vitro labelling of tRNA with the preQ 1 derivatives, 10 M of in vitro transcribed tRNAs or alternatively 10 g of total RNA from S. pombe was incubated with 200 nM hTGT (QTRT1:QTRT2) and 5 M queuine in reaction buffer (50 mM Tris-HCl pH 7.5, 20 mM NaCl, 5 mM MgCl 2 and 2 mM dithiothreitol) for 5 h at 37 • C. The RNA was purified using phenol/chloroform extraction and precipitated with 1/10 volume of ammonium acetate and three volumes of 100% ethanol.

HeLa cells growth conditions and in vivo labelling with preQ 1 -L1
HeLa cell lines were obtained from ATCC and authenticated by multiplex human cell line authentication test (Multiplexon). Cells were grown in Dulbecco's modified Eagle's medium (DMEM) (Thermo Fisher Scientific). The cultures were supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine and a commercial cocktail of antibiotics (Thermo Fisher Scientific). For minus-Q conditions, ultraculture serum-free medium (Lonza) was supplemented with 2 mM L-glutamine and 100 units/ml Penicillin/Streptomycin. PreQ 1 -L1 derivative was added at a concentration of 0.1 M for 72 h to the culture.
Nucleic Acids Research, 2022, Vol. 50, No. 18 10789 HeLa cell polysome profiling 10 7 cells were treated with 100 g/ml cycloheximide for 5 min at RT to stabilize existing polysomes before washing with ice-cold PBS and harvesting by scraping in 400 l polysome lysis buffer (20 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 150 mM NaCl, 1 mM DTT, 1% Triton X-100, 100 g/ml cycloheximide, 1 × Complete Protease Inhibitors (Roche)). Lysates were rotated end-over-end for 10 min at 4 • C and cleared by at 10 000 rpm for 10 min at 4 • C. 40 l of supernatant lysate was saved as input before loading the lysates to linear 17.5 to 50% sucrose gradients in 20 mM Tris-HCl (pH 7.4), 5 mM MgCl 2 , 150 mM NaCl. Centrifugation was carried out at 35 000 rpm for 2.5 h at 4 • C in a Beckmann SW60 rotor. Gradients were eluted with an ISCO UA-6 gradient fractionator, and polysome profiles were recorded by continuously monitoring the absorbance at 254 nm using PeakTrak software. During gradient elution, fractions of ∼300 l were collected every 14 s. For RNA isolation, 300 l urea buffer (10 mM Tris, pH 7.5, 350 mM NaCl, 10 mM EDTA, 1% SDS, and 7 M urea) and 300 l phenol:chloroform:isoamylalcohol (25:24:1) were added to each fraction. After phase separation, RNA was isolated from the aqueous phase and precipitated using isopropanol and GlycoBlue (Thermo Fisher Scientific).

Detection of queuine and preQ 1 modification of RNAs
Labelled RNA that had been CuAAC-clicked with AlexaFluor 594 alkyne was analyzed by denaturing PAGE. Up to 10 g of labelled RNA was separated in 10% polyacrylamide gels (acrylamide/ bisacrylamide (19:1), urea 8 M in 1× TBE buffer). Detection was carried out on the Typhoon 9500 (GE Healthcare) using 532 nm for excitation. As a loading control, gels were stained with Sybr Gold nucleic acid gel stain (Thermo Fisher Scientific) or GelRed (Biotium) for 10 min followed by detection using 495 or 532 nm, respectively, for excitation.
To detect the queuine modification, 300 ng of total RNA or small RNAs from S. pombe WT and qtr2Δ strains were separated in a 10% polyacrylamide gel (acrylamide/ bisacrylamide (19:1), urea 8 M) supplemented with 5 mg/ml 3-(acrylamido)-phenylboronic acid as described previously (51). The separation was performed at room temperature in 1× TBE. The electrophoresed gels were transferred to a Biodyne B Nylon membrane (0.45 M). Selected RNAs were detected using a 5 -biotin-labeled probe at a final concentration of 0.3 M and the Chemiluminescence Nucleic Acid Detection Module Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. The first blocking step was carried out using the DIG Easy Hyb buffer (Roche), and hybridization was performed overnight at 45 • C.

Detection of Q and Q-L1 by LC-MS/MS analysis
Up to 5 g of total tRNA was digested to nucleoside level using 0.6 U nuclease P1 from P. citrinum (Sigma-Aldrich), 0.2 U snake venom phosphodiesterase from C. adamanteus (Worthington), 2 U FastAP (Thermo Fisher Scientific), 10 U benzonase (Sigma-Aldrich), 200 ng Pentostatin (Sigma-Aldrich) and 500 ng Tetrahydrouridine (Merck-Millipore) in 25 mM ammonium acetate (pH 7.5; Sigma-Aldrich) overnight at 37 • C. 1 g of total tRNA was analyzed via LC-MS using an Agilent 1260 series LC with a Synergi Fusion column (4 M particle size, 80Å pore size, 250 × 2.0 mM; Phenomenex) and an Agilent 6460 Triple Quadrupole mass spectrometer equipped with an electrospray ion source (ESI). The elution started with 100% solvent A (5 mM ammonium acetate buffer, pH 5.3) with a flow rate of 0.35 ml/min at 35 • C, followed by a linear gradient to 8% solvent B (LC-MS grade acetonitrile; Honeywell) at 10 min and 40% solvent B after 20 min. Initial conditions were regenerated with 100% solvent A for 10 min. The UV signal at 254 nm was recorded via a multiple wavelength detector (MWD) detector at 254 nm to monitor the main nucleosides. The following ESI parameters were defined for the measurement: gas temperature 350 • C, gas flow 8 l/min, nebulizer pressure 50 psi, sheath gas temperature 350 • C, sheath gas flow 12 l/min, capillary voltage 3000 V, nozzle voltage 0 V. The MS was operated in the positive ion mode using Agilent MassHunter software in the dynamic MRM (multiple reaction monitoring) mode.

Identification of preQ 1 -L1-modified RNAs by HTS
Metabolically labelled and biotin-clicked RNA was purified from total RNA or isolated small RNAs using Dynabeads™ MyOne™ Streptavidin C1 (Thermo Fisher Scientific) according to the manufacturer's instructions. Streptavidin coated magnetic beads were washed as instructed for immunoprecipitation of RNA. 20 g of biotin-labelled RNA was incubated with the beads for 1 h at room temperature with light agitation. After washing the beads, they were resuspended in nuclease-free water, and bound RNA was dissolved from the beads by incubating the samples at 95 • C for 10 min. Library preparation of immunoprecipitated RNAs for deep sequencing was done using the NEBNext Small RNA Library Prep Set for Illumina (Multiplex Compatible; New England Biolabs). 300 ng of RNA per library as starting material was used, and ligation was performed with undiluted adaptors. Adaptor-ligated cDNA was amplified with 15 cycles of PCR reaction using barcoded primers and purified using the Monarch PCR & DNA Cleanup Kit (5 g) (New England Biolabs). Libraries were eluted in nuclease-free water, multiplexed in equimolar ratios and sequenced on one lane of the Illumina MiSeq platform using paired-end 150 bp sequencing.

RT-qPCR quantification of tRNA Asp , snoR38 and snoR69
For quantification of preQ 1 -L1-labelled tRNA Asp , snoR38 and snoR69 from metabolically labelled and immunoprecipitated (IPed) RNAs, quantitative RT-PCR was performed using a stem-loop primer (see Supplementary Table S3). cDNA was synthesized using IPed RNAs from S. pombe WT and qtr2Δ and a sequence specific stem-loop primer. First strand synthesis was carried out using the SuperScript™ III First-Strand Synthesis System (Invitrogen) according to the manufacturer's protocol. Synthesized cDNA was subsequently used for qPCR using the PerfeCTa SYBR Green SuperMix (QuantaBio). 4 l of cDNA was used in a reaction mix containing 12.5 l Master Mix (Quanta, 2×) and 250 nm primers. Reactions were carried out in a total volume of 25 l with the following thermocycling: 95 • C for 2 min, followed by 40 cycles of 95 • C for 10 s, 58 • C for 15 s and 72 • C for 20 s.

RNA bisulfite sequencing
Bisulfite sequencing of tRNA Asp was performed as previously described (38). Briefly, bisulfite-treated tRNAs were reverse transcribed using tRNA Asp 3'-specific stemloop primer followed by amplification with primers binding only to the deaminated sequence at their 5' end. Primer sequences are listed in Supplementary Table S3. Library preparation of the PCR products was performed with the NEXTflex R qRNA-Seq™ Kit v2--Set C (Bioo Scientific) according to the manufacturer's instructions and sequenced on a MiSeq platform using paired-end 150 bp sequencing. Reads were processed using in-house R scripting and the Bioconductor package ShortRead (52). Following the processing, including trimming of PCR primers, selection of high-quality reads and sorting of the reads based on the sequence in the degenerate region of the RT-primer, the reads were analyzed for bisulfite conversion using BISMA (53).

HTS data processing
The sequencing data was adapter-trimmed using Skewer version 0.2.2 (54) and aligned to S. pombe non-coding RNAs (main and mitochondrial) excluding rRNA sequences from Pombase (https://www.pombase.org/) using Salmon version 14.0 (55) and HISAT2 version 2.1.1 (56), as a splice-site sensitive alignment program. The conversion of sam to bam files was performed using SAMtools (57). Aligned sequences were analyzed using custom R scripts and the Bioconductor package DESeq2 (58). Parameters were set to analyze only regions with a minimum of 10 reads and the adjusted P-value was set to <0.1. Additionally, independent hypothesis weighting was conducted using the Bioconductor package IHW (59,60) with an adjusted P-value of <0.1. Furthermore, peak calling was performed using the Bioconductor package exomePeak2 (61). Plots were generated using the integrative genomics viewer version 2.11.1 (IGV) (62).

In vitro incorporation of synthetic preQ 1 analogues in prokaryotes
To assess the substrate properties of synthetic preQ 1ligands, their incorporation into tRNA by bacterial TGT (bTGT) was tested in vitro ( Figure 1). For this purpose, preQ 1 -ligands 1-3 (preQ 1 -L1-3, Figure 1A), each harbouring an azide group, were incubated with tRNA Asp in the presence of recombinant bTGT from Z. mobilis. Taking advantage of the terminal azide group, the successful in vitro incorporation of preQ 1ligands was visualized by copper(I)-catalyzed azide alkyne cycloaddition (CuAAC) click reaction of tRNA Asp with the fluorescent AlexaFluor 594-alkyne in the presence of CuSO 4 , sodium ascorbate and THPTA (tris-((1-benzyl-1H-1,2,3-triazol-4-yl)methyl)amine) ( Figure 1B and Supplementary Figure S1). As shown by fluorescence scan, all of the tested preQ 1 -ligands were incorporated to the same extent ( Figure 1C), indicating that the side chains attached to the preQ 1 structure do not hinder the recognition and turnover by bTGT. This indicates a tolerance of the bTGT active site for large ligands, similar to what was previously described for eTGT in vitro (42).

In vivo incorporation of synthetic preQ 1 analogues in prokaryotes
After the successful in vitro application of synthetic preQ 1ligands with bTGT, we proceeded to metabolic labelling of RNAs in vivo in E. coli. First experiments were performed with the smallest preQ 1 ligand in the series, i.e. preQ 1 -L1. In a feeding experiment, where an E. coli wild-type (WT) strain was grown in medium supplemented with preQ 1 -L1, total tRNA was isolated and enzymatically digested to the nucleoside level for separation on an RP-C18 HPLC column and subsequent analysis of the Q levels by MS/MS. Of note, queuosine exhibits a fragmentation pattern differing from the standard nucleosides. Instead of the exclusive fragmentation at the N-glycosidic bond, cleavage of the ribose in combination with cleavage of the amino linker with a mass shift m/z 410 to m/z 163 was determined as the most abundant product ion (Supplementary Figure S2a) eluting at a retention time of 12.2 min in the WT sample. Using a fragmentation pattern for the incorporated synthetic nucleoside (Q-L1) that was inferred from that of native queuosine, additional signals for the expected transitions were detected at 16.9 min ( Figure  2). Since the product ion m/z 163 was the most prevalent species, it was chosen as diagnostic ion in subsequent LC-MS/MS experiments. Monitoring this product ion produced a strong signal for queuosine and only a weak signal for Q-L1 (Supplementary Figure S3c). We concluded that preQ 1 -L1 was indeed incorporated, but also that it was a weak competitor against the endogenous bacterial preQ 1. Consequently, we reasoned that abrogating preQ 1 biosynthesis would facilitate the incorporation of the supplied preQ 1 -ligands. Considering the various steps of Q de novo synthesis in E. coli (Figure 2A), four different gene deletions, namely ΔqueD, ΔqueE, ΔqueC and ΔqueF, were tested for generation of preQ 1 by monitoring the presence of Q at position 34 of tRNAs. The deletion strains showed no significant growth defects compared to the wild type ( Figure  2C). To validate the absence of Q de novo biosynthesis, tRNA was isolated and analyzed by LC-MS as before; none of the deletion strains generated measurable levels of Q (Supplementary Figure S2b-e). Since all deletions were on a par regarding growth and absence of Q in the isolated total tRNA, ΔqueD was chosen for further experiments, given that the absence of QueD prevents de novo synthesis of Q in its earliest stages and avoids synthesis of any precursor form (e.g. preQ 0 in ΔqueF) that was reported to be incorporated by the TGT and might thus compete with preQ 1 -L1 (Supplementary Figure S3c) (13).
To determine a suitable feeding concentration, the ΔqueD cells were supplemented with increasing amounts of preQ 1 -L1 ( Figure 2B), which did not impair the bacterial growth compared to the control without preQ 1 -L1 feeding (Supplementary Figure S3b). Total tRNA isolated from thus treated ΔqueD cells was labelled via CuAAC click chemistry and subsequently analysed by denaturing urea PAGE. The fluorescence scan revealed clearly visible fluorescent bands after feeding with preQ 1 -L1 in a dose-dependent manner, providing evidence for the enzymatic incorporation of the analogue into E. coli tRNA in vivo ( Figure 2D). The signal intensity was quantified using ImageJ software and plotted against the feeding concentration of preQ 1 -L1 ( Figure 2E). To validate these observations, total tRNA isolated from the ΔqueD strain treated with preQ 1 -L1 was further analysed via LC-MS/MS. The MS-based analysis of total tRNA isolated from the ΔqueD fed with increasing concentrations of preQ 1 -L1 (0.1-10 M) confirmed the results obtained from denaturing urea PAGE analysis and related quantification of the fluorescence signal ( Figure  2E). Both plots indicate a beginning saturation around 10 M. Based on the above, a ligand concentration of 5 M for further feeding experiments was identified as a viable compromise between ligand material consumption and labelling efficiency. Supplementation of a Δtgt strain with the optimized concentration of 5 M preQ 1 -L1 resulted in no signal for Q-L1 in LC-MS/MS measurements, thus confirming that the incorporation was catalysed by bTGT (Supplementary Figure S3c).
Since preQ 1 -L1 was successfully incorporated into RNA in bacteria, the in vivo experiments were expanded to preQ 1 -L2 and preQ 1 -L3. However, in contrast to the previously  Figure S3a), indicating that the incorporation efficiency of preQ 1 -L2 and preQ 1 -L3 into tRNA in vivo was drastically lower compared to preQ 1 -L1. Since the in vitro results indicate indifference of the TGT enzyme towards the alkyl-modified preQ 1 -ligands, the low incorporation in vivo suggests lower bioavailability of preQ 1 -L2 and preQ 1 -L3 for the bacteria.

In vivo interactions of synthetic preQ 1 analogues in prokaryotes
To investigate possible changes in the ensemble of molecular interactions undergone by Q-L1-carrying tRNA under physiological conditions, we turned to the analysis of polysomes. Given that these consist of actively translating ribosomes, their components, including tRNA, can be considered functional in interactions with essential molecular factors involved in translation. We thus aimed at determining the ratio of Q-L1-carrying tRNAs from polysomes versus that in the remainder of tRNAs.
For this purpose, cell lysates from E. coli WT and ΔqueD cells supplemented with 10 M preQ 1 -L1 were applied to a sucrose gradient (5-40%), enabling the separation of different fractions according to their size after ultracentrifugation. As schematically shown in Figure  3A, free RNAs including tRNAs and some mRNAs were located in fraction F0 at the top of the gradient (5% sucrose), while polysomes accumulated in fraction F3 at a sucrose concentration of ∼40%. This separation was monitored by UV absorbance at 260 nm, and the different fractions were collected. Subsequent to fractionation, total RNA was extracted from these fractions and applied to denaturing PAGE for purification of tRNA via gel elution. Digested tRNA samples were subjected to LC-MS/MS analysis, and the respective abundances of Q and Q-L1 were compared between the free RNA fraction F0 and the polysomal fraction F3 ( Figure 3B). Interestingly, in WT cells, endogenous Q was more abundant in tRNAs isolated from the polysomal fraction compared to fraction F0. This suggests that queuosinylated tRNAs are selectively enriched in polysomes that are actively engaged in translation.
In contrast, the Q-L1 level in polysomal tRNA (F3 fraction) from preQ 1 -L1 fed ΔqueD cells reached a similar amount compared to its level in the respective F0 fraction. This may reflect either a deficit in the aforementioned selection, or a cumulation of minor detrimental effects at the different steps of translation. However, the data clearly illustrate that Q-L1-containing tRNAs actively engage in protein biosynthesis and are able to sustain it at a high enough level to not cause any perceivable growth phenotype.

In vitro incorporation of synthetic preQ 1 analogues in eukaryotes
In a next step, the investigations were extended from bacteria to eukaryotes. Of note, eukaryotes do not possess the enzymes to synthesize queuosine de novo, but salvage it from external sources for incorporation into tRNA (16). S. pombe is a particularly well-suited single cell eukaryotic model organism, because salient features of queuosine have already been elaborated in this yeast, and queuosine levels can easily be manipulated by supplementation of the growth medium with queuine (38).
We next tested the ability of eTGT to incorporate the preQ 1 -ligands into RNA in vitro. As substrates for this reaction, total RNA was isolated from S. pombe wild-type cells or qtr2Δ cells cultured in the presence of queuine. In WT cells, this results in Q-modification of the tRNAs, whereas qtr2Δ cells lack the essential Qtr2 subunit of S. pombe eTGT, therefore maintaining a guanosine in position 34 of the respective tRNAs. Total RNA preparations of these strains were incubated with the preQ 1 ligands in presence of hTGT, and subsequently labelled by click reaction. The incorporation of all three ligands into tRNA from both S. pombe strains was measured by fluorescence scan ( Figure 4A and Supplementary Figure S4b). In comparison to the fluorescence signals of the tRNA from WT cells, the respective signals of the qtr2Δ tRNAs showed significantly higher intensities. This indicates that more tRNAs unmodified at position G34 are available for in vitro modification with preQ 1 -L1 in the qtr2Δ sample. In contrast, in WT cells only guanosines that were not replaced by Q despite the presence of a functional enzyme remained for the in vitro reaction. Unlike observed for the bTGT, the hTGT incorporated the preQ 1 -ligands to differing degrees, indicating a higher ability to distinguish between these analogues in accordance with previously published results by Kelly and co-workers (41). Additionally, incubation of in vitro transcribed tRNAs Asp, His, Tyr and Asn with human TGT and preQ 1 -L1 showed successful incorporation of the analogue into all of the four tRNAs (Supplementary Figure  S4a).

In vivo incorporation of synthetic preQ 1 analogues in eukaryotes
Subsequent to the successful in vitro experiment, the in vivo incorporation of the synthetic preQ 1 analogues was examined in S. pombe. To this end, S. pombe WT and qtr2Δ cells (as a control), were cultured in the presence of preQ 1ligands in medium that otherwise lacked Q or q, and RNA was isolated and analysed as before. After click reaction, a fluorescence signal was detected in the RNA isolated from the WT cells treated with preQ 1 -L1, but not qtr2Δ ( Figure  4B), showing that the presence of Q-L1 in tRNA in vivo depended on functional TGT. As in bacteria, preQ 1 -L1 did not negatively affect cell growth (Supplementary Figure  S4c), and no labelling was observed with preQ 1 -L2 and -L3, again indicating that their derivatives are not bioavailable for incorporation into tRNAs in vivo.
Collectively, the above experiments indicate that preQ 1 -L1 can readily be employed as a proxy for Q from the perspective of synthetic biology. To further develop this compound for the investigation of the epitranscriptome, we made use of the click chemistry feature of preQ 1 -L1 to identify RNAs into which it was incorporated in vivo by eTGT.
For this purpose, total RNA isolated from S. pombe wild-type or qtr2Δ cells that were cultured in the presence of preQ 1 -L1 was bio-conjugated in vitro with alkyne-functionalized biotin. Subsequent to affinity purification using streptavidin-coated magnetic beads, the biotinlabelled RNA was subjected to reverse transcription and high-throughput sequencing ( Figure 5A, termed Q-RIP-Seq). The analysis showed that the known cytosolic Q-tRNAs tRNA Asn , tRNA Asp , tRNA His and tRNA Tyr were significantly enriched from WT, but not qtr2Δ cells (n = 3, P adj < 0.1, Figure 5B, C and Supplementary Figure S5). Other enriched signals from snoR38 and snoR69 were scrutinised as potential substrates of TGTmediated incorporation of preQ 1 -L1. However, neither APB Northern blotting nor quantification by q-RT-PCR substantiated this hypothesis (Supplementary Figure S6). Interestingly, mitochondrial tRNA Asn , when analysed for q content by APB-northern blot, was queuosinylated to about 50% (Supplementary Figure S5b). The fact that no mitochondrial tRNA sequences were found in Q-RIP-Seq could mean that they are too low in abundance. An alternative explanation would be that preQ 1 -L1 is not incorporated into mitochondrial tRNA. The above findings indicate that the four known Q-tRNAs are the only cytosolic RNAs that are Q-modified in S. pombe, which is congruent with crosslinking-based studies in human cells (41). These results establish that any major metabolic influence resulting from feeding preQ 1 -L1 would be mediated through the four classical tRNA substrates of TGT. It should, however, be noted that an early study reported preQ1-modification in vitro of larger RNA species in E. coli (63).

In vivo interactions of synthetic preQ 1 analogues in eukaryotes
Having established that preQ 1 -L1 is actively incorporated into native tRNAs, we next investigated a particularly interesting effect of queuosine, namely a so-called tRNA modification circuit, where the formation of one modification is enhanced by the presence of another modification (64). The particular circuit involving queuosine was first identified in S. pombe. We had shown earlier by RNA bisulfite sequencing that the formation of m 5 C38 in tRNA Asp by the Dnmt2 tRNA methyltransferase is strongly enhanced by the presence of queuosine at position 34 ( Figure 6A) (38,49).
We therefore asked whether Q-L1 can serve as a biologically active surrogate for queuosine in this circuit. Figure 6B shows the m 5 C38 levels of tRNA Asp in response to increasing concentrations of preQ 1 -L1 in medium otherwise free of queuosine derivatives. A clear dose-dependent increase of the C38 methylation level was observed, indicating that the incorporated preQ 1 -L1 is functionally integrated into this modification circuit, efficiently replacing queuosine in its capability of triggering Dnmt2 activity in S. pombe. Considering the direct functional connection of Q/Q-L1 and m 5 C38, the increase of the C38 methylation level from 15% in non-treated culture up to 60% in cultures supplemented with 100 nm preQ 1 -L1 points to its incorporation in significant amounts in S. pombe. However, it is important to mention that the effect of preQ 1 -L1 incorporation on tRNA Asp methylation is less efficient compared to the known effect of Q under normal conditions, as we previously reported (38). The incorporation efficiency in E. coli can only be gauged even more indirectly, namely by comparison of fluorescent signals after click (Supplementary Figure S4d). Lastly, we were also able to demonstrate successful incorporation of preQ 1 -L1 in HeLa human cells deprived of q (Supplementary Figure S7a). In analogy to the earlier presented analysis of E. coli polysomes, we also investigated the levels of Q-L1 in tRNA purified from F0 and F3 of accordingly treated HeLa cultures. Similar to our observations in E. coli, the amount of Q-L1 detected in the polysomal tRNA (F3 fraction) from preQ 1 -L1 fed HeLa cells was comparable to its level in the respective F0 fraction, indicating that Q-L1containing tRNAs actively engage in protein biosynthesis in cellulo (Supplementary Figure S7b). This result indicates relevance of our investigations with respect to biomedical considerations, e.g. potential therapeutic interventions.

DISCUSSION
Interest in concepts for the incorporation of modified and/or non-natural derivatives of metabolites into nucleic acids has been steadily increasing, boosted in part by a surge in RNA modification research, and, more recently, in mRNA-based vaccines. Post-synthetic derivatization of RNA in vitro, e.g. by methyltransferases has been exploited for labelling in conjunction with click chemistry (65)(66)(67)(68)(69). In the queuosine field, a number of q-derived compounds, including clickable tetrazine derivatives, have been incorporated into native RNA preparation in vitro using recombinant TGT, and applied to fluorescent labelling, affinity purification, and interactome research (42)(43)(44)70,71). In a previous study, Brooks et al. reported that azide congeners of preQ 1 lacking the methylene amine were not incorporated by the TGT which they traced to the necessity of this structural element for a successful binding to the enzyme forming hydrogen bonds between aminoacid residues Leu231 and Met260 of the enzyme (72,73). Although, as mentioned, strong indirect evidence (34) suggested that incorporation of nonnatural q derivatives should be feasible in principle, no in vivo labelling of Q-tRNAs with clickable q-derivatives has been demonstrated so far.
Overall, concepts and applications in the RNA field currently move from in vitro (74) to metabolic feeding approaches in cellulo and in vivo. Here, the use of noncanonical nucleoside structures has opened up new experimental avenues in the community. As an example, in RNA modification research feeding of methionine analogues featuring e.g. propargyl residues, has enabled their incorporation into RNA in lieu of methyl groups. Subsequent derivatization by click chemistry was exploited for determination of modification sites (75,76,66).
An important progress featured in our work is that we demonstrate low toxicity of the labelling compound and provide corresponding data at the molecular level. Elsewhere in the field, little attention is paid to the physiological impact of surrogate feeding. In most cases, a moderate survival rate in cell culture is sufficient to conduct e.g. -omics type analyses after incorporation (75)(76)(77). However, in the next steps of its development, the field might conceivably move to applications in model organisms. Here, by the latest, one will need  (1) and bio-conjugated in vitro with alkyne-functionalized biotin (2). Biotinlabelled RNAs were subsequently affinity-purified using streptavidin-coated magnetic beads (3), reverse-transcribed and subjected to high-throughput sequencing (5). As a control, metabolic labelling was performed in an S. pombe strain lacking TGT (qtr2Δ). (B) Log2 fold change of normalized read counts of RNAs from WT compared to qtr2Δ determined by exomePeak2. Red: tRNA Asp , tRNA His and tRNA Tyr ; (three independent replicates). (C) Q-RIP-Seq of tRNA Asp , tRNA His , tRNA Tyr and tRNA Asn after metabolic labelling with preQ 1 -L1 in S. pombe WT and qtr2Δ cells. Coverage of the tRNA sequences from modified (WT, black) and unmodified (qtr2Δ, grey) samples is shown. The transcript architecture is shown below with thin and thick parts representing introns and mature tRNA sequences. Replicate 1 of three independent experiments is shown. Plots were generated using IGV.
to adopt concepts from medicinal chemistry, such as cell permeability, and toxicity. In this respect, the work presented here pioneers the combination of metabolic feeding of clickable surrogates with investigations into their physiological molecular impact after cellular uptake and their usage for the enrichment and identification of RNA species that were labelled in vivo by endogenous TGT. Apart from the observation of growth inhibition of q derivatives in eukaryotic cell culture, which are somewhat suggestive (78), there is strong indirect evidence for the actual incorporation of a q-derivative by TGT in vivo in mouse (34), however without direct analysis of the tRNA. Significantly, said case features a background of medicinal chemistry, and the compound used is structurally related to our preQ1-L series used here. It does, however, feature a lipophilic phenylpropyl sidechain which is likely causative of, or enhancing the compound's cell permeability and biodistribution.
In the present work, we have developed the azido-propylderivative preQ 1 -L1 as a bioactive surrogate for preQ 1 in vivo. preQ 1 -L1 is taken up into unicellular prokaryotes as well as into eukaryotes, and incorporated into the known tRNA substrates of TGT. The resulting nucleoside is semi-synthetic in that its sugar moiety is native, while its nucleobase is synthetic. Its azide moiety can be employed to metabolically label and isolate Q-modified RNAs by affinity purification after conjugation by click chemistry. We used this feature to confirm similar data from human cells, obtained after UV-crosslinking (41). Taken together, this means that the single most important molecular interaction for a physiological impact of q (or preQ 1 -L1) is mediated through position 34 in the anticodons of the four known TGT substrate tRNAs.
Known molecular interactions issuing from this nucleobase are mostly restricted to tRNA aminoacylation and mRNA decoding, which we have interrogated by investigating the amount of Q-L1 carrying tRNAs on polysomes. While Q-L1 was less abundant there than was native Q, it was clearly present, featuring an equal distribution between actively translating tRNAs and the cytoplasmic pool in both bacterial and human cell preparations.
One other known effect of Q was also faithfully emulated by Q-L1, namely the stimulation of m 5 C38 formation by Dnmt2 in the anticodon stem of tRNA Asp , representing a so-called modification loop. Technically speaking, we report the first-ever manipulation of a modification loop by atomic mutagenesis in vivo.
In spite of numerous described Q-dependent implications in various diseases, starting from cancer (29)(30)(31)(32) to neurological and neuropsychiatric disorders, such as multiple sclerosis, schizophrenia and Parkinson (79,80,33,34), a defined mechanism explaining the role of Q in these pathologies is still missing (28). Recently, we discovered a direct connection between Q, accuracy and the speed of codon-biased translation (27,28), which promotes protein folding and prevents the accumulation of misfolded proteins. The fact that Q-L1 is functionally involved in the translational process in a 'minimally invasive' system, opens the possibility to study the roles of Q34 modifications in protein translation in normal and pathogenic human cell lines, directly combining click chemistry or LC-MS/MS with polysome profiling.
In summary, the combination of very few queuosinylation sites and the effective functional replacement of Q by Q-L1 on the molecular level, makes the q/Q system uniquely suited for a 'minimally invasive' placement of a non-natural nucleobase within the total cellular RNA.

DATA AVAILABILITY
HTS data for Q-RIP-Seq experiments are available in the NCBI GEO database, record GSE210404. All data needed to evaluate the conclusions in the paper are present in the paper and/or Supplementary Data. Additional data related to this paper may be requested from the authors.