Thioguanosine conversion enables mRNA life-time evaluation by RNA sequencing via double metabolic labeling.

Temporal information about cellular RNA populations is essential to understand the functional roles of RNA. We have developed hydrazine/NH 4 Cl/OsO 4 -based conversion of 6-thioguanosine (6sG)- into A'-containing RNA, where A' constitutes a 6-hydrazino purine derivative. The A' nucleoside retains the Watson-Crick base pair mode as shown by crystal structure analysis of a short palindromic duplex and thermodynamic analysis of UV-melting profiles. We further demonstrate that A' is efficiently decoded as adenosine in primer extension assays and in RNA sequencing experiments. Because 6sG is applicable to metabolic labeling of freshly synthesized RNA and because the novel conversion chemistry is fully compatible with the conversion of the frequently applied metabolic label 4-thiouridine (4sU) into C, the combination of both modified nucleosides in dual labeling set-ups enables high accuracy measurements of RNA decay. Our novel approach that we name TUC-seq DUAL uses the two modified nucleotides in subsequent pulses and their simultaneous detection. Thus, synthesis and decay of mRNA can be clearly distinguished based on the differential presence of G-to-A and U-to-C mutations. This enables mRNA life-time evaluation with unprecedented precision.


Supporting Figures 17
Supporting Figure S1. Optimization of 6sG conversion. 17 Supporting Figure S2. Melting profiles of 6sG, 6soG, and A'-containing RNA and unmodified counterparts 18 Supporting Figure S3. Crystal structure of 5'-CGCGA'AUUAGCG 19 Supporting Figure S4. Conversion of 6sG in structured RNAs 20 Supporting Figure S5. Characterization of suitability of 6sG for metabolic labeling 21 Supporting Figure S6. Ongoing incorporation of 4sU in a pulse-chase labeling experiment 22

Supporting Table 23
Supporting Table 1. X-ray data collection and crystallographic refinement statistics 23 Supporting Table 2. List of primer sequences 24

Conversion of 6sG containing RNA
OsO4 stock solution (100 mM, 1.5 mL; stored at 4 °C in a sealed glass vial) and ammonium chloride buffer (2 M, pH 8.88) were prepared according to references [1,2]. The hydrazine stock solution (1.5 M) was made by adding hydrazine monohydrate (7.51 g, 150 mmol) into 100 mL of an aqueous solution containing tris(hydroxymethyl)aminomethane (Tris base) (1.21 g, 10 mmol; 0.1 M) and ethylenediaminetetraacetic acid (EDTA) (146 mg, 0.5 mmol; 5 mM). The pH was adjusted with concentrated HCl to 8.98. For the conversion reaction, a 1 mM aqueous OsO4 solution was freshly prepared from the OsO4 stock solution.
Afterwards, the RNA was lyophilized, dissolved in water (15 µL), and hydrazine stock solution (5 µL, 1.5 M, pH 8.98) was added to give a final concentration of 375 mM hydrazine in a total volume of 20 µL. The solution was incubated for 2 hours at 40 °C. The RNA solution was transferred into centrifugal concentrators and washed four times with 400 µL of water as described before. Progress of the RNA conversion reactions was monitored by anion exchange HPLC; 6soG and A' modified RNA was stored at -20 °C in aqueous solution or lyophilized. As an alternative to ultrafiltration, the RNA can also be precipitated by adding 90 µL of precipitation solution (650 µL water, 150 µL 1 M NaOAc, pH 5.2, 10 µL glycogen (20 mg mL -1 )) and 250 µL of cold ethanol, stored at -20 °C for 30 minutes, followed by centrifugation (13000 rpm, 4 °C, 30 min). The supernatant was discarded and the RNA was lyophilized or dissolved in water (15 µL) and further processed as described above.
For isolated, metabolically labeled cellular RNA, a milder hydrazine stock solution was used. Briefly, after OsO4/NH4Cl treatment and precipitation, the RNA pellet was resuspended in water (15 µL) and hydrazine stock solution (5 µL; 0.5 M, pH 8.11, 5 mM EDTA, 0.5 M Tris) was added to give a final concentration of 125 mM hydrazine.

N 2 -Acetyl-2'-O-(tert-butyldimethylsilyl)-6-(2-cyanoethylthio)-guanosine (5)
Hydrogen fluoride in pyridine (Sigma Aldrich, 8 M, 174 µl, 6.88 mmol, 3.85 eq) was diluted under cooling with pyridine (1.08 mL) and added to a solution of compound 4 (1.16 g, 1.79 mmol) in dichloromethane (8.7 mL) at 0 °C. The solution was stirred for two hours at 0 °C, washed once with water and twice with saturated aqueous bicarbonate, dried over Na2SO4 and then evaporated. Product 5 as white foam was used in the next step without further purification. For NMR spectra the product was purified by column chromatography (SiO2

UV melting analysis
Solutions of duplex or hairpin RNAs were prepared containing 10 mM Na2HPO4 and 150 mM NaCl, at pH 7.0 and at RNA concentrations as indicated (usually from 2 µM to 30 µM). Thermal denaturation was monitored on a Cary 100 UV-Vis spectrophotometer equipped with a temperature control accessory. Heating and cooling cycles were performed between 10 to 90 °C three times at a rate of 0.7 °C/min and recorded at 260 nm.

Crystallography and X-ray analysis
Two synthetic RNA dodecamers, 5'-CGCGAA'UUAGCG-3' (RNA59H) and 5'-CGCGA'AUUAGCG-3' (RNA58H), were used for crystallization trials. RNA was dissolved in water at a concentration of 1 mM, heated at 80 °C for 10 min and cooled at 20 °C at a 1 °C min -1 rate. Crystals were grown at 20 °C by the vapor diffusion method using sitting drops by mixing 2 µL of RNA sample with 2 µL of a crystallization buffer made with 10% v/v 2-methyl-2,4-pentanediol (MPD), 40 mM sodium cacodylate pH 7.0, 12 mM spermine, 80 mM sodium chloride, 30 mM magnesium chloride against a reservoir made with 35% MPD. Prior to data collection, crystals were flash-frozen in liquid ethane. X-ray diffraction data were collected on the X10SA beamline at the SLS synchrotron, Villigen, Switzerland. Data were processed with the XDS Package 3 and the structure was solved by molecular replacement with MOLREP 4 using the related PDB 2Q1R RNA model. 5 The structure was refined with the PHENIX package 6 (Supporting Table 1). The model was built using Coot. 7 Coordinates have been deposited with the PDB database (entry numbers D_1292105554 and D_1292105478). Superscript III was used due to its known superior performance in primer extension analyses. 8,9 The primer extension reaction was stopped by addition of 1 μL 4 M NaOH and incubation at 95 °C for 5 min followed by cooling to 4 °C. The Alexa Fluor 647 labelled cDNA strands were precipitated by adding 90 µL of precipitation solution (650 µL water, 150 µL 1 M NaOAc pH 5.2, 10 µL of 20 mg/mL glycogen)) and 250 µL of cold ethanol and stored for 30 min at -20 °C. After centrifugation for 30 min at 4 °C at 13500 rpm, the samples were resuspended in 8 μL of gel loading buffer (97% formamide, 10 mM EDTA). Sequencing ladders were produced by adding 2 μL of 5 mM ddNTPs in addition to the 8 μL of reaction mixture to unmodified RNA samples, prior to incubation at 60°C. Samples were loaded next to a migration control dye (0.1% xylene cyanol, 95% formamide, 10 mM EDTA) on 10% polyacrylamide gels with 7 M urea and run for approximately 100 min at 45 W. The extension products were then analyzed by scanning the gel at 635 nm with a Typhoon FLA 9500 instrument (GE Healthcare).

PCR-mediated detection of 6sG-to-A' conversion
To determine OsO4-hydrazine-mediated 6sG-to-A' conversion by PCR-mediated detection, a 47 nt RNA oligo containing a single 6sG at position 16 was synthesized. Furthermore, a synthetic tRNA containing a single 6sG at position 29 was generated by synthesis of 37 nt and 39 nt half molecules and enzymatic ligation (S. cerevisiae tRNA Phe sequence and secondary structure shown in Supporting Figure 4A). The OsO4-treated and untreated RNA was purified, and 100 ng RNA was reverse transcribed using GoScript Reverse Transcriptase (Promega) according to the manufacturer's instructions with a specific stem loop primer complementary to the 3' end of the oligo or the tRNA Phe . The generated cDNA was PCR amplified employing primers specific for the universal stemloop sequence and the 3' ends of the tRNA and oligo, respectively (Primers sequences are listed in Supporting Table 2). PCR products were subcloned into a pGEM-T-Vector (Promega) and individual clones were picked for colony PCR with M13 primers. PCR products were gel purified according to the manufacturer's instructions (NEB) and subsequently subjected to Sanger sequencing.

Single and double labeling of HEK293T cells
For single labeling experiments, 1.5x10 7 HEK293T cells were seeded into 15 cm round cell culture dishes and grown overnight at 37°C and 5% CO2 in DMEM medium (Gibco). Medium was replaced with DMEM supplemented with 0.025, 0.05 (Supporting Figure S4) and 0.1 mM (Figure 3, Supporting Figure S4) 6thioguanosine (6sG; Figure 3) or with 0.05 mM 4-thiouridine (4sU; Jena Bioscience; Supporting Figure S5). Cells were incubated for 30 min, 1 h or 2 h and subsequently harvested. For pulse-chase experiments (Supporting Figure S5), 4sU-containing medium was replaced with DMEM supplemented with 1 mM uridine, and cells were further incubated for 4 h. Samples were collected at 2, 3 and 4 h of uridine chase. For double labeling experiments ( Figure 5), cells were labeled with 0.1 mM 6sG for 1 h before the medium was removed and replaced by DMEM containing 0.05 mM 4sU. Cells were collected right after 6sG incubation, and at 2 and 4 h of 4sU incubation.

Proliferation assay
Proliferation of cells treated with different 6sG concentrations (25, 50, and 100 µM) for 30 min or 1 h was monitored for a period of 72 hours by counting cells with a hemocytometer every 24 hours. The mean and standard deviation of four replicates was determined by GraphPad Prism 7 and plotted.

RNA isolation and RT-qPCR analysis
Total RNA was isolated using the innuPREP RNA Mini Kit (Analytik Jena) according to the manufacturer's instructions. To determine 6sG incorporation efficiency and for amplicon sequencing experiments, mRNA was prepared from total RNA using Magnetic mRNA isolation kit (NEB). For reverse-transcription real time PCR experiments (RT-qPCR), cDNA was generated by GoScript Reverse Transcriptase (Promega) and subjected to real time PCR using Lunaâ Universal qPCR Master Mix (NEB) with 1.25 ng/ml cDNA and 0.8 mM gene-specific primers in a QuantStudio 3 Real-Time PCR System (Applied Biosystems). Data were normalized against glycerinaldehyde-3-phosphate dehydrogenase (GAPDH) and are expressed as mean ± SEM 2 -ΔCT values (n=3). Statistical significance was calculated by unpaired t-test with a significance threshold of p<0.05 using GraphPad Prism 7.

6sG incorporation analysis by HPLC
Cells were labeled with 0.1 mM 6sG for 2 h before total RNA was isolated and mRNA was prepared as described above. mRNA (2 µg) was digested to mononucleosides as follows: RNA was denatured for 5 min at 95 °C and subsequent cooling on ice. Then RNA was digested with 178 U S1 nuclease (Promega) in a final volume of 50 µl for 4 h at 37 °C, followed by addition of 2 U shrimp alkaline phosphatase (NEB) and 0.2 U phosphodiesterase (Sigma) and further incubation at 37 °C for 1 h. Nucleosides were extracted twice with chloroform and the aqueous phase was analyzed by HPLC on a XBridge C18 5 µm column (4.6 mm x 150 mm) at 35 °C. Elution was in a gradient of 6 column values 0-10% eluent B (acetonitrile) in A (250 mM ammonium acetate, pH 6) at a flow rate of 1 ml min -1 . A solution of A, U, C, G, and 6sG ribonucleosides (0.5 mM each; 100 µl) was used as standard. Absorbance was measured at 260 nm and 320 nm (6sG).

Amplicon sequencing and data analysis
RNA from labeled and unlabeled cells was isolated and treated with OsO4-hydrazine as described in Materials and Methods. After purification and reverse transcription using GoScript Reverse Transcriptase (Promega) and random hexamer primers, selected targets were amplified with specific primers containing barcode overhangs using standard PCR conditions (primer sequences without barcodes are listed in Supporting Table 2). The products were separated on a 1.5% agarose gel, purified from the gel and pooled at equimolar ratio. Library preparation from the amplicon pool and sequencing using the Illumina HighSeq platform with 150 nt paired end reads was performed by Eurofins.
The multiplexed sequencing read data were split into single sample files according to the sample-specific barcodes using flexbar version 3.0.3. 10 The sample-specific sequencing reads were aligned to the respective reference sequences by running bowtie2 version 2.3.4.1 11 in a first round in "end-to-end" mode.
Reads that failed to align in "end-to-end" mode were then aligned in a second round by using the "local" mode. Amplicon positions with G-to-A and U-to-C conversions were called using Varscan2 version 2.4.3. 12 We set the maximum depth to 10 6 and the minimum base call quality score to 30. Only U and G positions with a minimum conversion frequency of 10 -4 were considered further. For identifying the background/baseline mutation/error frequency we analyzed all non U-to-C and non G-to-A changes according to the same criteria as used for U-to-C and G-to-A conversions. In order to minimize errors from potentially misaligned reads we only considered positions on the amplicons which were at maximum 146 bases distant from the amplicon ends. To quantify the number of reads with U-to-C and G-to-A conversions, we used sam2tsv 13 and a custom written perl script (available upon request) to analyze each aligned read and count the U-to-C and G-to-A conversions and the read-specific conversion frequency. Again, only sequence read bases with a minimum base call quality score of 30, and a maximum amplicon end distance of 146 were considered in our analyses.

Statistical analyses.
To determine the statistical significance of differences of G-to-A mutation frequencies of 6sG-labeled and OsO4-treated versus unlabeled and untreated, and 6sG labeled but not OsO4-treated samples, and of differences between G-to-A mutation frequencies versus A,C,U-to-N mutation frequencies, Chi-Square analyses with Yates' correction were performed using GraphPad Prism 7. Statistical significance was set to p<0.05.
Supporting Figure S1. Optimization of 6sG conversion. (A) Attempts for one-step conversion of 6sG into an A analog by simultaneous application of OsO4 and the nucleophiles NH4Cl (left), hydroxylamine (middle), and hydrazine (right). Oxidation only or incomplete conversions were observed. Assignments according to mass spectrometric analysis. (B) Two-step conversions: After isolation of 6soG RNA (step 1, see panel A, left), different nucleophiles for substitution were tested (step 2). While ammonia resulted in incomplete conversions (left) and degradation during long reaction times (not shown), hydroxylamine substitution required long reaction times (middle); complete conversion was observed for hydrazine in short reaction times. (C) Comparison of the optimized OsO4-hydrazine 6sG-to-A' conversion to OsO4-CF3CH2NH2 and NaIO4-CF3CH2NH2 treatments which resulted in incomplete or no conversion.
Supporting Figure S2. Melting profiles of 6sG, 6soG, and A'-containing RNA and unmodified counterparts. Supporting Figure S6. Ongoing incorporation of 4sU in a pulse-chase labeling experiment. Cells were labeled with 4sU for 1 h, before washout and addition of excess U. OsO4/NH4Cl treatment was performed, p21 RNA was analyzed by amplicon sequencing at the indicated time points and the fraction of newly synthesized transcripts was calculated.
Supporting Table 1. X-ray data collection and crystallographic refinement statistics.