The authors acknowledge the R21 grant from National Institutes of Health&#160;(1R21HG009576) to S. Z. and W. L. and New York Institute of Technology (NYIT) Institutional Support for Research and Creativity grants to S. Z., which supported this work. The authors would like to thank PhD student Xuanting Wang (Columbia University) for assisting in figure-making, and thank Prof. Michael Hadjiargyrou (NYIT), Prof. Jingyue Ju (Columbia University), Drs. James Russo, Shiv Kumar, Xiaoxu Li, Steffen Jockusch, and other members of the Ju lab (Columbia University), Dr. Yongdong Wang (Cerno Bioscience), Meina Aziz (NYIT), and Wenhao Ni (NYIT) for helpful discussions and suggestions for our manuscript.

1. Design RNA oligonucleotides
<ol>
	<li>Design synthetic RNA oligonucleotides with different lengths (19 nt, 20 nt and 21 nt), including one (RNA #6) with both canonical and modified nucleotides. &#968; is employed as a model for non-mass-altering modifications, which is challenging for MS sequencing because it has an identical mass to U. m5C is chosen as a model for mass-altering modifications to demonstrate the robustness of the approach. 
	 
	RNA #1: 5&#180;-HO-CGCAUCUGACUGACCAAAA-OH-3&#180; 
	RNA #2: 5&#180;-HO-AUAGCCCAGUCAGUCUACGC-OH-3&#180; 
	RNA #3: 5&#180;-HO-AAACCGUUACCAUUACUGAG-OH-3&#180; 
	RNA #4: 5&#180;-HO-GCGUACAUCUUCCCCUUUAU-OH-3&#180; 
	RNA #5: 5&#180;-HO-GCGGAUUUAGCUCAGUUGGGA-OH-3&#180; 
	RNA #6: 5&#180;-HO-AAACCGU&#968;ACCAUUAm5CUGAG-OH-3&#180; 
	&#160;</li>
	<li>Dissolve each synthetic RNA in nuclease-free diethyl pyrocarbonate (DEPC)-treated water (expressed as DEPC-treated H2O unless otherwise indicated) to obtain a 100 mM RNA stock solution. Stock solutions are stored long-term at -20 &#176;C.</li>
	<li>To avoid possible RNA sample degradation, use RNase-free experimental supplies including DEPC-treated water, microcentrifuge tubes, and pipette tips. Frequently wipe down surfaces of lab supplies using&#160;RNase elimination wipes.</li>
</ol>
2. Label the 3&#180;-end of RNAs with biotin
<ol>
	<li>Two-step reaction protocol (adenylation and ligation)
	<ol>
		<li>Add 1 &#181;L of 10x adenylation reaction buffer containing 50 mM sodium acetate, pH 6.0, 10 mM MgCl2, 5 mM dichlorodiphenyltrichloroethane (DTT), 0.1 mM ethylenediaminetetraacetic acid (EDTA), 1 &#181;L of 1 mM ATP, 1 &#181;L of 100 &#181;M biotinylated cytidine bisphosphate (pCp-biotin), 1 &#181;L of 50 &#181;M Mth RNA ligase, and 6 &#181;L of DEPC-treated H2O (a total volume of 10 &#181;L) into an RNase-free thin-walled 0.2 mL PCR tube. 
		NOTE: Store the reagents at -20 &#176;C before the two-step reaction. Thaw the reagents at room temperature and mix well by vortexing and centrifuging before adding to the reaction.</li>
		<li>Incubate the reaction in a PCR machine at 65 &#176;C for 1 h and inactivate the reaction at 85 &#176;C for 5 min.</li>
		<li>Conduct the ligation step in an RNase-free, thin walled 0.2 mL PCR tube containing 10 &#181;L of reaction solution from the previous step by adding 3 &#181;L of 10x T4 RNA ligase reaction buffer containing 50 mM tris(hydroxymethyl)aminomethane (Tris)-HCl, pH 7.8, 10 mM MgCl2, 1 mM DTT, 1.5 &#181;L of the 100 mM sample stock of the RNA to be sequenced, 3 &#181;L of anhydrous dimethyl sulfoxide (DMSO) to reach 10% (v/v), 1 &#181;L of T4 RNA ligase (10 units/&#181;L), and 11.5 &#181;L of DEPC-treated H2O (for a total volume of 30 mL). Incubate the reaction overnight at 16 &#176;C in a PCR machine. 
		NOTE: Combine reaction components at room temperature due to the high freezing point of DMSO (18.45 &#176;C).</li>
		<li>Incubate the reaction overnight at 16 &#176;C.</li>
		<li>Quench and purify the reaction by column purification to remove enzymes and free pCp-biotin using Oligo Clean &#38; Concentrator (Zymo Research, Irvine, CA, USA). Oligo Binding Buffer, DNA Wash Buffer, spin columns and collection tubes are provided in the kit. Add 20 mL of DEPC-treated H2O to the reaction solution to reach a 50 mL sample volume prior to adding the Binding Buffer.</li>
		<li>Add 100 mL of binding buffer to each reaction solution. Add 400 &#181;L of ethanol, mix by pipetting, and transfer the mixture to the column. Centrifuge at 10,000 x g for 30 s. Discard the flow-through.</li>
		<li>Add 750 &#181;L of DNA Wash Buffer to the column. Centrifuge at 10,000 x g and maximum speed for 30 s and 1 minute, respectively.</li>
		<li>Transfer the column to a 1.5 mL microcentrifuge tube. Add 15 &#181;L of DEPC-treated H2O to the column and centrifuge at 10,000 x g for 30 s to elute the RNA product. 
		NOTE: Samples can be stored at -20 &#176;C at this stage until the next step is performed.</li>
	</ol>
	</li>
	<li>One-step reaction protocol
	<ol>
		<li>Perform a one-step labeling reaction by combining 2 &#181;L of 150 &#956;M adenosine-5&#180;-5&#180;-diphosphate-{5&#180;-(cytidine-2&#180;-O-methyl-3&#180;-phosphate-TEG}C-biotin (AppCp-biotin), 3 &#181;L of 10x ligase reaction buffer, 1.5 &#181;L of the 100 mM sample stock of the RNA to be sequenced, 3 &#181;L of anhydrous DMSO to reach 10% (v/v), 1 &#181;L of T4 RNA ligase (10 units/&#181;L), and 19.5 &#181;L of DEPC-treated H2O (for a total volume of 30 mL) in a 1.5 mL RNase-free microcentrifuge tube.</li>
		<li>Incubate the reaction overnight at 16 &#176;C in a PCR machine.</li>
		<li>Perform column purification as described above in steps 2.1.5-2.1.8. 
		NOTE: Prepare a separate/exclusive reaction tube for each RNA sample (150 pmol scale of RNA). Labeling of the 5&#180;-end of the RNA(s) with sulfo-Cyanine3 (Cy3) or Cy3 may be needed (e.g., for bidirectional sequencing verification). The method is different than that of 3&#180;-biotinylation and is described in a previous publication9.</li>
	</ol>
	</li>
</ol>
3. Capture biotinylated RNA sample on streptavidin beads
<ol>
	<li>Activate 200 &#181;L of streptavidin C1 magnet beads by adding 200 &#181;L of 1x B&#38;W buffer (5 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 1 M NaCl) in a 1.5 mL RNase-free microcentrifuge tube. Vortex this solution and place it on a magnet stand for 2 min. Then discard the supernatant by carefully pipetting out the solution.</li>
	<li>Wash the beads twice with 200 &#181;L of Solution A (DEPC-treated 0.1 M NaOH and DEPC-treated 0.05 M NaCl) and once in 200 &#181;L of Solution B (DEPC-treated 0.1 M NaCl). For each wash step, vortex the solution and place it on a magnet stand for 2 min, followed by discarding of the supernatant. Then add 100 &#181;L of 2x B&#38;W buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 2 M NaCl).</li>
	<li>Add 1x B&#38;W buffer to the biotinylated RNA sample until the volume is 100 &#181;L. Then add this solution to the washed beads stored in 100 &#181;L of 2x B&#38;W buffer. Incubate for 30 min at room temperature on a rocking platform shaker at 100 rpm. Place the tube on a magnet stand for 2 min and discard the supernatant.</li>
	<li>Wash the coated beads 3 times in 1x B&#38;W buffer and measure the final concentration of supernatant in each wash step by Nanodrop for recovery analysis, to confirm that the target RNA molecules remain on the beads.</li>
	<li>Incubate the beads in 10 mM EDTA, pH 8.2 with 95% formamide at 65 &#176;C for 5 min in a PCR machine. Keep the tube on the magnet stand for 2 min and collect the supernatant (containing the biotinylated RNAs released from the streptavidin beads) by pipet. 
	NOTE: This physical separation step prior to acid degradation is only used for sequencing of RNA#1 in Figure 1c, and is not mandatory for the 2D-HELS MS Seq since the hydrophobic biotin label can cause the 3&#180;-labeled ladder fragments to have a significantly delayed tR during LC-MS measurement, which can clearly distinguish the labeled 3&#180;-ladder fragments from the unlabeled 5&#180;-ladder fragments in the 2D mass-tR plot.</li>
</ol>
4. Acid hydrolysis of RNA to generate MS ladders for sequencing
<ol>
	<li>Divide each RNA sample into three equal aliquots. For instance, divide an RNA sample with a volume of 15 &#181;L RNA sample into three aliquots of 5 &#181;L.</li>
	<li>Add an equal volume of formic acid to achieve 50% (v/v) formic acid in the reaction mixture8,9.</li>
	<li>Incubate the reaction at 40 &#176;C in a PCR machine, with one reaction running for 2 min, one for 5 min, and one for 15 min, respectively.</li>
	<li>Quench the acid degradation by immediately freezing the sample on dry ice after each reaction finishes.</li>
	<li>Use a centrifugal vacuum concentrator to dry the sample. The sample is typically completely dried within 30 min, and formic acid is removed together with H2O during the drying process because formic acid has a boiling point (100.8 &#176;C) similar to that of H2O (100 &#176;C).</li>
	<li>Suspend and combine a total of three dried samples in 20 &#181;L of DEPC-treated H2O for LC-MS measurement. 
	NOTE: Samples can be stored at -20 &#176;C at this stage while waiting for LC-MS measurement.</li>
</ol>
5. Convert &#968; to CMC-&#968; adduct
<ol>
	<li>Add 80 &#181;L of DEPC-treated H2O into a 1.5 mL RNase-free microcentrifuge tube containing 0.0141 g of N-cyclohexyl-N&#697;-(2-morpholinoethyl)-carbodiimide metho-p-toluenesulfonate (CMC) and 0.07 g of urea. Add 10 &#181;L of the 100 &#181;M sample stock of the RNA to be sequenced, 8 &#181;L of 1 M bicine buffer (pH 8.3), and 1.28 &#181;L of 0.5 M EDTA. Add DEPC-treated H2O to reach a total volume of 160 &#181;L. Final concentrations are 0.17 M CMC, 7 M urea, and 4 mM EDTA in 50 mM bicine (pH 8.3)11. 
	NOTE: This protocol is applicable to either a single synthetic RNA sequence or RNA mixtures.</li>
	<li>Divide the 160 &#181;L reaction solution into four equal aliquots in RNase-free, thin walled 0.2 mL PCR tubes and incubate at 37 &#176;C for 20 min in a PCR machine. 
	NOTE: 50 &#181;L per tube is the maximum reaction volume that can be used in a PCR machine.</li>
	<li>Quench each reaction with 10 &#181;L of 1.5 M sodium acetate and 0.5 mM EDTA (pH 5.6).</li>
	<li>Perform column purification with four parallel spin columns to remove excessive reactants according to the procedure as described in steps 2.1.5-2.1.8. Dissolve the purified product in 15 &#181;L of DEPC-treated H2O in each 1.5 mL RNase-free microcentrifuge tube.</li>
	<li>Transfer the purified product to four RNase-free, thin walled 0.2 mL PCR tubes. Add 20 &#181;L of 0.1 M Na2CO3&#160;buffer (pH 10.4) into each 15 &#181;L of purified product and add DEPC-treated H2O to make a final volume of 40 &#181;L for each reaction tube (in total four tubes). Incubate the reaction at 37 &#176;C for 2 h in a PCR machine.</li>
	<li>Quench and purify the reaction by column purification with four parallel spin columns as described in step 2.1.5. Elute the CMC-&#968; converted product to a 1.5 mL RNase-free microcentrifuge tube each with 15 &#181;L of DEPC-treated H2O.</li>
	<li>Combine the purified CMC-&#968; converted sample from four collection tubes into one tube. Perform formic acid degradation 50% (v/v) according to the procedures as described in steps 4.1-4.6 to generate MS ladders for sequencing.</li>
</ol>
6. LC-MS measurement
<ol>
	<li>Prepare mobile phases for LC-MS measurement. Mobile phase A is 25 mM hexafluoro-2-propanol with 10 mM diisopropylamine in LC-MS grade water; mobile phase B is methanol.</li>
	<li>Transfer the sample to LC-MS sample vial for analysis. Each sample injection volume is 20 &#181;L containing 100-400 pmol of RNA.</li>
	<li>Use the following LC conditions: column temperature of 35 &#176;C, flow rate of 0.3 mL/min; a linear gradient from 2&#8211;20% mobile phase B over 15 min followed by a 2 min wash step with 90% mobile phase B. 
	NOTE: For more hydrophobic end-labels such as Cy3 and sulfo-Cy3 as mentioned in Section 2, a higher percentage of organic solvent may be necessary for sample elution (i.e., a similar gradient can be used but with an increased percentage range of mobile phase B). For instance, 2&#8211;38% mobile phase B over 30 min with a 2 min wash step with 90% mobile phase B.</li>
	<li>Separate and analyze samples on an Agilent Q-TOF (Quadrupole Time-of-Flight) mass spectrometer coupled to an LC system equipped with an autosampler and an MS HPLC (High Performance Liquid Chromatography) system. The LC column is a 50 mm x 2.1 mm C18 column with a particle size of 1.7 &#956;m. Use the following MS settings: negative ion mode; range, 350 m/z to 3200 m/z; scan rate, 2 spectra/s; drying gas flow, 17 L/min; drying gas temperature, 250 &#176;C; nebulizer pressure, 30 psig; capillary voltage, 3500 V; and fragmentor voltage, 365 V. Please note that these parameters are specific to the type or model of mass spectrometer being used.</li>
	<li>Acquire data with Agilent MassHunter acquisition software. Use Agilent molecular feature extraction (MFE) workflow to extract compound information including mass, retention time, volume (the MFE abundance for the respective ion species), and quality score, etc. Use the following MFE settings: &#8220;centroid data format, small molecules (chromatographic), peak with height &#8805; 100, up to a maximum of 1000, quality score &#8805; 50&#8221;. 
	NOTE: Optimize MFE settings to extract as many potential compounds as possible, up to a maximum of 1000, with quality scores of &#8805; 50.</li>
</ol>
7. Automate RNA sequence generation by a computational algorithm
NOTE: This procedure is shown only for RNA #1 in Figure 1c.
<ol>
	<li>Sort MFE extracted compounds in order of decreasing volume (peak intensity) and tR. Perform data pre-selection via 1) setting tR from 4 to 10 min to select the RNA fragments labeled by the biotin, since the tRs of the biotin-labeled mass ladder components are shifted to this tR window (4 min to 10 min), and 2) using an order-of-magnitude higher of input compounds than the number of ladder fragments for algorithm computation to reduce data amount based on volume. For instance, for a 20 nt RNA, 20 labeled mass-tR ladder components will be required for sequencing of the 20 nt RNA, and thus, 200 compounds from MFE data file will be selected based on volume. Please note that the tR window may be different when a different type or model of mass spectrometer is used.</li>
	<li>Perform data processing and sequence generation of RNA #1 using a revised version of a published algorithm8. The source codes of the revised algorithm are described previously (https://academic-oup-com.remotexs.ntu.edu.sg/nar/article/47/20/e125/5558343#supplementary-data)9.</li>
	<li>In addition to automating sequence generation using the algorithm, manually calculate the mass differences between two adjacent ladder components for base calling. All bases in the RNA can be called manually and matched with the theoretical ones in the RNA nucleotide and modification database8; thus, the complete sequence of the RNA strand can be accurately read out manually, which is used to confirm the accuracy of the algorithm-reported sequence read. More structures of RNA modifications can be found in RNA modification databases12, and their corresponding theoretical masses are obtained by ChemBioDraw. In Tables S1&#8211;S2, the ppm (parts-per-million) mass difference is shown when comparing the observed mass to its theoretical mass for a specific ladder component, and a value less than 10 ppm is considered a good match for each base calling.</li>
</ol>
8. Sequencing RNA mixtures
<ol>
	<li>Label a mixture of five RNA strands (RNA #1 to #5) at their 3&#180;-ends with A(5&#180;)pp(5&#180;)Cp-TEG-biotin using a one-step protocol described in step 2.2. In a total volume of 150 &#181;L reaction solution, add 15 &#181;L of 10x T4 RNA ligase reaction buffer, 1.5 &#181;L of each RNA strand (100 &#181;M stock of RNA #1 to #5, respectively, for a total volume of 7.5 &#181;L), 10 &#181;L of 150 &#181;M A(5&#180;)pp(5&#180;)Cp-TEG-biotin, 15 &#181;L of anhydrous DMSO, 5 &#181;L of T4 RNA ligase (10 units/&#181;L), and 97.5 &#181;L of DEPC-treated H2O. Equally distribute the reaction solution into five aliquots. Each RNase-free microcentrifuge tube contains 30 &#181;L of reaction solution.</li>
	<li>Incubate the reaction overnight at 16 &#176;C in a PCR machine.</li>
	<li>Perform column purification according to the procedure as described in steps 2.1.5-2.1.8&#160;with five parallel spin columns. Elute a mixture sample of 3&#180;-biotinylated 5 RNA strands (mixture of RNA #1 to #5) to a 1.5 mL RNase-free microcentrifuge tube each with 15 &#181;L of DEPC-treated H2O.</li>
	<li>Combine the purified mixture samples from the five collection tubes into one tube. Perform formic acid degradation according to the procedure described in Section 4.</li>
	<li>Measure samples by LC-MS as described in Section 6, and analyze the data using the data analysis software with optimized MFE settings to extract data containing mass, tR, and volume as described in step 6.5. The typical processing and base-calling algorithm is not applied due to the significantly increased data complexity resulting from the mixture. All bases in the RNA of the mixed sample are called manually in a method similar to Section 7.3 and match well with the theoretical ones in the RNA nucleotide and modification database8, thus the complete sequences of all five RNA strands in the mixed sample are accurately read out. In Tables S7&#8211;S11, all information is listed including observed mass, tR, volume, quality score and ppm mass difference.</li>
</ol>

<ol>
	<li>Addepalli, B., Venus, S., Thakur, P., Limbach, P. A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Novel+ribonuclease+activity+of+cusativin+from+Cucumis+sativus+for+mapping+nucleoside+modifications+in+RNA.">Novel ribonuclease activity of cusativin from Cucumis sativus for mapping nucleoside modifications in RNA.</a> Analytical and Bioanalytical Chemistry. 409 (24), 5645-5654 (2017).</li><li>Gao, H., Liu, Y., Rumley, M., Yuan, H., Mao, B. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Sequence+confirmation+of+chemically+modified+RNAs+using+exonuclease+digestion+and+matrix-assisted+laser+desorption/ionization+time-of-flight+mass+spectrometry.">Sequence confirmation of chemically modified RNAs using exonuclease digestion and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry.</a> Rapid Communications in Mass Spectrometry. 23 (21), 3423-3430 (2009).</li><li>McLuckey, S. A., Van Berkel, G. J., Glish, G. L. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Tandem+mass+spectrometry+of+small,+multiply+charged+oligonucleotides.">Tandem mass spectrometry of small, multiply charged oligonucleotides.</a> Journal of The American Society for Mass Spectrometry. 3 (1), 60-70 (1992).</li><li>Fountain, K. J., Gilar, M., Gebler, J. C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Analysis+of+native+and+chemically+modified+oligonucleotides+by+tandem+ion-pair+reversed-phase+high-performance+liquid+chromatography/electrospray+ionization+mass+spectrometry.">Analysis of native and chemically modified oligonucleotides by tandem ion-pair reversed-phase high-performance liquid chromatography/electrospray ionization mass spectrometry.</a> Rapid Communications in Mass Spectrometry. 17 (7), 646-653 (2003).</li><li>Taucher, M., Breuker, K. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Characterization+of+modified+RNA+by+top-down+mass+spectrometry.">Characterization of modified RNA by top-down mass spectrometry.</a> Angewandte Chemie International Edition in English. 51 (45), 11289-11292 (2012).</li><li>Kellner, S., Burhenne, J., Helm, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Detection+of+RNA+modifications.">Detection of RNA modifications.</a> RNA Biology. 7 (2), 237-247 (2010).</li><li>Thomas, B., Akoulitchev, A. V. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Mass+spectrometry+of+RNA.">Mass spectrometry of RNA.</a> Trends in Biochemical Sciences. 31 (3), 173-181 (2006).</li><li>Bjorkbom, A., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Bidirectional+direct+sequencing+of+noncanonical+RNA+by+two-dimensional+analysis+of+mass+chromatograms.">Bidirectional direct sequencing of noncanonical RNA by two-dimensional analysis of mass chromatograms.</a> Journal of the American Chemical Society. 137 (45), 14430-14438 (2015).</li><li>Zhang, N., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+general+LC-MS-based+RNA+sequencing+method+for+direct+analysis+of+multiple-base+modifications+in+RNA+mixtures.">A general LC-MS-based RNA sequencing method for direct analysis of multiple-base modifications in RNA mixtures.</a> Nucleic Acids Research. 47 (20), 125 (2019).</li><li>Zhang, N., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Direct+sequencing+of+tRNA+by+2D-HELS-AA+MS+Seq+reveals+its+different+isoforms+and+dynamic+base+modifications.">Direct sequencing of tRNA by 2D-HELS-AA MS Seq reveals its different isoforms and dynamic base modifications.</a> ACS Chemical Biology. 15 (6), 1464-1472 (2020).</li><li>Bakin, A., Ofengand, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Four+newly+located+pseudouridylate+residues+in+Escherichia+coli+23S+ribosomal+RNA+are+all+at+the+peptidyltransferase+center:+analysis+by+the+application+of+a+new+sequencing+technique.">Four newly located pseudouridylate residues in Escherichia coli 23S ribosomal RNA are all at the peptidyltransferase center: analysis by the application of a new sequencing technique.</a> Biochemistry. 32 (37), 9754-9762 (1993).</li><li>Cantara, W. A., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+RNA+Modification+Database,+RNAMDB:+2011+update.">The RNA Modification Database, RNAMDB: 2011 update.</a> Nucleic Acids Research. 39 (Database issue), D195-D201 (2011).</li></ol>

The authors have filed a provisional patent related to the technology discussed in this manuscript.

Unlike tandem-based MS fragmentation, highly controlled acidic hydrolysis is used in the 2D-HELS MS Seq approach to fragment the RNA before analysis with a mass spectrometer9,10. As a result, each acid-degraded fragment can be detected by the instrument, forming the equivalent of a sequencing ladder. Under optimal conditions, this method creates an &#8220;ideal&#8221; sequence ladder from RNA via, on average, one-per-molecule site-specific RNA cleavage exclusively at a phosphodiester bond8,9,10. After each degraded fragment is measured by the mass spectrometer in a single run, the mass difference between two adjacent ladder fragments corresponds to the exact mass of the RNA nucleotide or modification at that position. Each RNA modification either has an intrinsic unique mass that can help to identify and locate it in the RNA, or can be converted to one with a unique mass. Thus, in theory, this method can report the identity and location of both canonical and modified nucleotides for de novo and direct sequencing of any RNA. However, different sequence ladders may overlap with each other, complicating MS data analysis and making it difficult for RNA sequencing by MS in practice.&#160;
One of the benefits of the 3&#180;-hydrophobic tag is that it overcomes a major challenge in any fragmentation method, i.e., that every RNA molecule must be cleaved into exactly two fragments (and ideally no more): one fragment containing the original 5&#180;-end, and the other containing the original 3&#180;-end of the RNA. Therefore, each cleavage event produces two fragments, producing two ladders&#8212;one measured from the 5&#180;-end, and the other from the 3&#180;-end. There is always ambiguity in determining which MS peak belongs to which ladder. This becomes more problematic in a mixture of several different RNAs, due to generation of a large number of overlapping sequence ladders. However, since all ladder fragments from the 3&#180;-ends are labeled with a hydrophobic tag, they exhibit much longer tRs (Figure 1a). As a result, we can obtain clear and unambiguous ladders in the 2D mass-tR data exclusively derived from just the 3&#180;-labeled RNA. Notably, we are optimizing approaches to selectively tag either the 5&#180;- or 3&#180;-end of any RNA using different chemical conjugation methods9. We can also perform bidirectional sequencing, which is not used to determine the terminal base(s) here, but is used to provide identical sequence information twice when reading from both 5&#180;- and 3&#180;- directions (i.e., bidirectional sequencing verification), and thus further improving the accuracy of sequencing.&#160;
For de novo sequencing of unknown RNA samples, especially for complex biological samples, a general and robust algorithm is required to process a large amount of LC-MS data for sequence generation in an accurate and efficient manner, which has recently become available via other published work10. Although these algorithms have been used for sequencing of more complicated samples10, in this study, we performed manual base calling for sequence generation unless indicated otherwise. We aim to cover all key steps in the 2D-HELS MS Seq, and would like to illustrate the process during which even without using additional sequencing algorithms, we can still manually read out sequences of the RNA to be sequenced. For ease of visualization and to more quickly identify ladder fragments needed for sequencing in the 2D mass-tR plot, the MFE files of each LC-MS run are processed by a revised version of a published algorithm8 before reading their sequences, unless indicated otherwise. The published algorithm cannot be used directly to read out the sequences from the LC-MS data, but part of its function can still be used to process the data&#8212;hierarchically clustering mass adducts through this algorithm will augment&#160;the intensity of each ladder component, which in turn reduces the data complexity, especially in the crucial region where sequence reads are generated8,9.&#160;
One of the crucial steps during sample preparation for 2D-HELS MS Seq results in the improvement of RNA hydrophobic tag end-labeling efficiency. A high labeling efficiency can help to reduce the amount of RNA sample needed for generating MS signals that sequence data rely on. In order to increase the labeling efficiency, we employ new labeling strategies, including using activated AppCp-biotin to avoid the adenylation step when labeling the 3&#180;-end of the RNA. The yield of the reaction for labeling the 3&#180;-end of a 19 nt RNA with biotin (see step 2.2) can be improved from 60% to ~95%9 using this one-step method. With the efficient labeling, we are able to sequence a mixed sample containing up to 12 distinct RNAs as previously described9. In this study, we use a mixture of five RNAs as a representative example to illustrate the sequencing process.&#160;We also detect all ladder fragments needed for accurate sequencing and read out the complete sequences of each of the five RNA sequences in the mixture. Higher labeling efficiency not only assists in minimizing the sample loading amount, but it also assists in significant reduction of data complexity during downstream data analysis for sequence generation. Novel reactions are currently under development to achieve quantitative yield in labeling RNAs on both 5&#180;- and 3&#180;-ends.&#160;
When sequencing RNA #1 as shown in Figure 1c, streptavidin capture and release steps are used to physically separate biotinylated RNA #1 prior to acid degradation (see Section 3). This removes a small portion of unlabeled RNA, and subsequently results in greater ease of visual identification of the labeled mass ladders in the 2D mass-tR plot. However, the physical separation step is not mandatory because the biotinylated RNA ladder fragments have delayed/longer tRs due to the hydrophobicity from the biotin tag&#160;when compared to their unlabeled counterparts. In addition, base calling does not rely on physical separation, but relies on the mass differences of adjacent mass ladder components, thus, the correct base call can be achieved as long as the mass differences of two adjacent ladder components match well with the corresponding masses of a particular nucleotide or modification in the RNA nucleotide and modification datebase8. A computational algorithm is currently under development to automate base-calling and sequence generation.&#160;
The MFE settings during original LC-MS data export (in the file type of .d) into spreadsheet files are highly crucial to the data processing and subsequent sequence generation (see Section 6.5). For instance, we tested the MFE setting &#8220;peak with height&#8221; in a range from 100 to 1000 and noticed that setting of 100 can provide us with 2-fold more compounds than those of setting 1000. In order to avoid missing any ladder components, we can adjust the MFE setting during the sequencing workflow. This setting is likely dependent on instrument mass resolution, the amount of mass ladder fragments, and data complexity. In addition, it is important to use the centroid dataset and chromatographic type setting for small molecules. The quality score can be varied from 50 % to 100% based on the data quality.&#160;
The LC-MS instrument we use in the study has an upper mass resolution of ~40K, limiting the method to only sequencing RNA less than 35 bases long. However, the exact read length of this method is instrument-dependent; more advanced instruments with higher resolving power may lead to longer read length. Similarly, the throughput, i.e., how many RNA sequences can be simultaneously sequenced in a single LC-MS run, remains to be explored, although we manually sequenced a mixture of RNA sample up to 12 distinct RNA strands even without use of any algorithm9. With the current workflow, ~100 pmol short RNA (&#60;35 nt) is required for each LC-MS run. The loading amount increases when additional experiments are needed: for differentiating isomeric nucleotide modifications, typically up to 400 pmol of RNA is required. For sequencing specific tRNA like tRNAPhe, ~1000 pmol of sample may be needed for sequencing and modification analysis. However, we expect required sample loading amounts will be decreased on LC-MS instruments with greater sensitivity. With improvements in sample labeling efficiency, sequencing algorithm, and instrument sensitivity and resolution, we expect our method to be applicable to a wider range of RNA samples, especially those with various RNA modifications.&#160;

Mass spectrometry (MS)-based sequencing methods, including top-down MS and tandem MS1,2,3,4, have been developed for direct sequencing of RNA. However, in situ fragmentation techniques for effectively generating high-quality RNA ladders in mass spectrometers currently can not be applied to de novo sequencing5,6. Furthermore, it is not very trivial to analyze the traditional one-dimensional (1D) MS data for de novo sequencing of even one purified RNA sequence, and it would be even more challenging for MS sequencing of mixed RNA samples7,8. Therefore, a two-dimensional (2D) liquid chromatography (LC)-MS-based RNA sequencing method has been developed, incorporating production of 2D mass-retention time (tR) ladders to replace 1D mass ladders, making it much easier to identify ladder components needed for de novo sequencing of RNAs8. However, the 2D LC-MS-based RNA sequencing method is mainly limited to purified synthetic short RNA, as it cannot read a complete sequence solely based on one single ladder, but must rely on two co-existing adjacent ladders (5&#180;- and 3&#180;-ladders)8. More specifically, this approach requires bidirectional paired-end reads for reading terminal nucleobases in the low-mass region8. The added complexity of the paired-end reading results in this method being untenable for sequencing of RNA mixtures because confusion is raised on which ladder fragment belongs to which ladder for the unknown samples.&#160;
To overcome the abovementioned barriers in MS-based RNA sequencing approaches and to broaden such applications in direct RNA sequencing, two issues must be addressed: 1) how to generate a high-quality mass ladder that can be used to read a complete sequence, from the first nucleotide to the last in an RNA strand,&#160;and 2) how to effectively identify each RNA/mass ladder in a complex MS dataset. Together with well-controlled acid degradation, we have developed a new sequencing method by introducing a hydrophobic end labeling strategy (HELS) into the MS-based sequencing technique, and successfully addressed these two issues by adding a hydrophobic tag at either 5&#180;- and/or 3&#180;-end of the RNAs to be sequenced9. This method creates an &#8220;ideal&#8221; sequence ladder from RNA&#8212;each ladder fragment derives from site-specific RNA cleavage exclusively at each phosphodiester bond, and the mass difference between two adjacent ladder fragments is the exact mass of either the nucleotide or nucleotide modification at that position 8,9,10. This is possible because we include a highly controlled acidic hydrolysis step, which fragments the RNA, on average, once per molecule, before it is injected into the instrument. As a result, each degradation fragment product is detected on the mass spectrometer and all fragments together form a sequencing ladder8,9,10. This new strategy enables complete reading of an RNA sequence from one single ladder of an RNA strand without paired-end reading from the other ladder of the RNA, and additionally allows MS sequencing of RNA mixtures with multiple different strands that contain combinatorial nucleotide modifications9. By adding a tag at the 5&#180;- and/or 3&#180;-end of the RNA, the labeled ladder fragments display a significant delay of tR, which can help to distinguish the two mass ladders from each other and also from the noisy low-mass region. The mass-tR shift caused by adding the hydrophobic tag facilitates mass ladder identification and simplifies data analysis for sequence generation. Furthermore, the addition of the hydrophobic tag can help to identify the terminal base in the strand by preventing its corresponding ladder fragment from being in the noisy low-mass-tR region due to the mass and hydrophobicity increase caused by the tag, thus allowing identification of the complete sequence of an RNA from a single ladder; no paired-end reads are required. As a result, we have previously demonstrated the successful sequencing of a complex mixture of up to 12 RNA distinct strands without the use of any advanced sequencing algorithm9, which opens the door for de novo MS sequencing of RNA containing both canonical and modified nucleotides and makes it more feasible for the sequencing of mixed and more complex RNA samples. In fact, using 2D-HELS MS Seq, we have even successfully sequenced a mixed population of tRNA samples10 and are actively expanding its application to other complex RNA samples.&#160;
To facilitate 2D-HELS MS Seq to directly sequence a broader range of RNA samples, here we will focus on the technical aspects of this sequencing approach and will cover all of the essential steps needed when applying the technique towards direct sequencing of RNA samples. Specific examples will be used to illustrate the sequencing technique, including synthetic single RNA sequences, mixtures of multiple distinct RNA sequences, and modified RNAs containing both canonical and modified nucleotides such as pseudouridine (&#968;) and 5-methylcytosine (m5C). Since RNAs all contain phosphodiester bonds, any type of RNA can be acid-hydrolyzed to generate an ideal sequence ladder for 2D-HELS MS Seq under optimal conditions8,9. However, detection of all ladder fragments of a given RNA is instrument dependent. On a standard high-resolution LC-MS (40K), the minimal loading amount for sequencing a purified short RNA sample (&#60;35 nt) is 100 pmol per run. However, more material is required (up to 400 pmol per RNA sample) when additional experiments must be conducted (e.g., to distinguish isomeric base modifications that share identical masses). The protocol used in sequencing the model synthetic modified RNAs will also be applicable to sequencing broader RNA samples, including biological RNA samples with unknown base modifications. However, an even larger sample amount, such as 1000 pmol for sequencing tRNA (~76 nt) using a standard LC-MS instrument, is required to sequence the complete tRNA with all the modifications, and an advanced algorithm must be developed for its de novo sequencing10.

<table><tbody><tr><td>5&#039; DNA Adenylation kit</td><td>New England Biolabs</td><td>E2610S</td><td>50uM concentration</td></tr><tr><td>6550 Q-TOF mass spectrometer</td><td>Agilent Technologies</td><td>5991-2116EN</td><td>Coupled to a 1290 Infinity LC system</td></tr><tr><td>A(5&amp;acute;)pp(5&amp;acute;)Cp-TEG-biotin-3&amp;acute;</td><td>ChemGenes</td><td>91718</td><td>HPLC purified</td></tr><tr><td>ATP&amp;gamma;S</td><td>Sigma-Aldrich</td><td>11162306001</td><td>Lithium salt</td></tr><tr><td>Bicine</td><td>Sigma-Aldrich</td><td>B8660</td><td>BioXtra, &amp;ge;99% (titration)</td></tr><tr><td>Biotin maleimide</td><td>Vector Laboratories</td><td>SP-1501</td><td>Long arm</td></tr><tr><td>C18 column</td><td>Waters</td><td>186003532</td><td>50 mm &amp;times; 2.1 mm Xbridge C18 column with a particle size of 1.7 &amp;mu;m</td></tr><tr><td>Centrifugal Vacuum Concentrator</td><td>Labconco</td><td>Refrig 115v/60hz 7310022</td><td>Labconco CentriVap</td></tr><tr><td>ChemBioDraw</td><td>PerkinElmer</td><td>ChemDraw Prime</td><td>Generate a chemical structure and property data of structures &amp;amp; fragments</td></tr><tr><td>CMC (N-cyclohexyl-N?-(2-morpholinoethyl)-carbodiimide metho-p-toluenesulfonate)</td><td>Sigma-Aldrich</td><td>2491-17-0</td><td>95% Purifiy</td></tr><tr><td>Cyanine3 maleimide (Cy3)</td><td>Lumiprobe</td><td>11080</td><td>Water insoluble</td></tr><tr><td>DEPC-treated water</td><td>Thermo Fisher Scientific</td><td>AM9906</td><td>Autoclaved, certified nuclease-free</td></tr><tr><td>Diisopropylamine (DIPA)</td><td>Thermo Fisher Scientific</td><td>108-18-9</td><td>99% Alfa Aesar</td></tr><tr><td>DMSO</td><td>Sigma-Aldrich</td><td>276855</td><td>Anhydrous dimethyl sulfoxide, 99.9%</td></tr><tr><td>EDTA</td><td>Sigma-Aldrich</td><td>E6758</td><td>Anhydrous, crystalline, BioReagent, suitable for cell culture</td></tr><tr><td>Formic acid</td><td>Merck</td><td>64-18-6</td><td>98-100%, ACS reag, Ph Eur</td></tr><tr><td>Hexafluoro-2-propanol (HFIP)</td><td>Thermo Fisher Scientific</td><td>920-66-1</td><td>99% Acros Organics</td></tr><tr><td>LC-MS sample vials</td><td>Thermo Fisher Scientific</td><td>C4000-11</td><td>Plastic screw thread vials</td></tr><tr><td>LC-MS vial caps</td><td>Thermo Fisher Scientific</td><td>C5000-54A</td><td>Autosampler vial screw thread caps</td></tr><tr><td>Na&lt;sub&gt;2&lt;/sub&gt;CO&lt;sub&gt;3&lt;/sub&gt; buffer</td><td>Sigma-Aldrich</td><td>88975</td><td>BioUltra, &gt;0.1 M Na&lt;sub&gt;2&lt;/sub&gt;CO&lt;sub&gt;3&lt;/sub&gt;, &gt;0.2 M NaHCO&lt;sub&gt;3&lt;/sub&gt;</td></tr><tr><td>Oligo Clean &amp;amp; Concentrator</td><td>Zymo Research</td><td>D4060</td><td>Spin column</td></tr><tr><td>OriginLab</td><td>OriginLab</td><td>OriginPro</td><td>Data analysis and graphing software</td></tr><tr><td>pCp-biotin</td><td>TriLink BioTechnologies</td><td>NU-1706-BIO</td><td>20 ul (1 mM)</td></tr><tr><td>RNA #1--#6</td><td>Integrated DNA Technologies</td><td>Custom RNA oligos</td><td>19nt-21nt single-stranded RNAs, used without further purification</td></tr><tr><td>Rocking platform shaker</td><td>VWR</td><td>Orbital Shaker Standard 1000</td><td>Speed Range 40 to 300 rpm</td></tr><tr><td>Streptavidin magnetic beads</td><td>Thermo Fisher Scientific</td><td>88816</td><td>Binding approx. 55ug biotinylated rabbit lgG per mg of beads</td></tr><tr><td>Sulfonated Cyanine3 maleimide</td><td>Lumiprobe</td><td>11380</td><td>Water soluble</td></tr><tr><td>T4 DNA ligase 1</td><td>New England Biolabs</td><td>M0202S</td><td>400 units/uL</td></tr><tr><td>T4 polynucleotide kinase</td><td>Sigma-Aldrich</td><td>T4PNK-RO</td><td>From phage T4 am N81 pse T1 infected Escherichia coli BB</td></tr><tr><td>Tris-HCl buffer</td><td>Sigma-Aldrich</td><td>T6455</td><td>Tris-HCl Buffer, pH 10, 10&amp;times;, Antigen Retriever</td></tr><tr><td>Urea</td><td>Sigma-Aldrich</td><td>81871</td><td>Urea for synthesis. CAS No. 57-13-6, EC Number 200-315-5.</td></tr></tbody></table>

2d-hels ms seq: a general lc-ms-based method for direct and de novo sequencing of rna mixtures with different nucleotide modifications

Mass spectrometry (MS)-based sequencing approaches have been shown to be useful in&#160;direct&#160;sequencing&#160;RNA without the need for a complementary DNA (cDNA) intermediate. However, such approaches are rarely applied as a de novo RNA sequencing method, but used mainly as a tool that can assist in quality assurance for confirming known sequences of purified single-stranded RNA samples. Recently, we developed&#160;a direct RNA sequencing method by integrating a 2-dimensional mass-retention time hydrophobic end-labeling strategy into MS-based sequencing (2D-HELS MS Seq). This method is capable of accurately sequencing single RNA sequences as well as mixtures containing up to 12 distinct RNA sequences. In addition to the four canonical ribonucleotides (A, C, G, and U), the method has the capacity to sequence RNA oligonucleotides containing modified nucleotides. This is possible because the modified nucleobase either has an intrinsically unique mass that can help in its identification and its location in the RNA sequence, or can be converted into a product with a unique mass. In this study, we have used RNA, incorporating two representative modified nucleotides&#160;(pseudouridine (&#936;) and 5-methylcytosine (m5C)), to illustrate the application of the method for the de novo sequencing of a single RNA oligonucleotide&#160;as well as a mixture of RNA oligonucleotides, each with a different sequence and/or modified nucleotides. The procedures and protocols described here to sequence these model RNAs will be applicable to other short RNA samples (&#60;35 nt) when using a standard high-resolution LC-MS system, and can also be used for&#160;sequence verification of modified therapeutic RNA oligonucleotides.&#160;In the future, with the development of more robust algorithms and with better instruments, this method could allow sequencing of more complex biological samples.&#160;

Introducing a biotin tag to the 3&#180;-end of RNA to produce easily-identifiable mass-tR ladders. The workflow of the 2D-HELS MS Seq approach is demonstrated in Figure 1a. The hydrophobic biotin label introduced to the 3&#180;-end of the RNA (see Section 2) increases the masses and tRs of the 3&#180;-labeled ladder components when compared to those of their unlabeled counterparts. Thus, the 3&#180;-ladder curve is shifted to greater y-axis values (due to the increase in the tRs) and shifted to greater x-axis values (due to the increase in masses) in the 2D mass-tR plot. Figure 1b shows the sample preparation protocol including introducing a biotin tag to the 3&#180;-end of RNA for 2D-HELS MS Seq. Figure 1c demonstrates separation of the 3&#180;-ladder from the 5&#180;-ladder and other undesired fragments on a 2D mass-tR plot based on systematic changes in tRs of the 3&#180;-biotin-labeled mass-tR ladder fragments of RNA #1. The 3&#180;-ladder curve alone gives a complete sequence of RNA #1, and the 5&#180;-ladder curve that does not show a tR shift provides the reverse sequence, but it requires end-pairing for reading the terminal base8. With this strategy of 2D-HELS, end-pairing is not&#160;required as reported before and the entire RNA sequence can be read out completely from only one labeled ladder curve8. As such, it is possible to sequence mixed samples containing multiple RNAs, e.g., two RNA strands of different lengths (RNA #1 and RNA #2, 19 nt and 20 nt, respectively) with a 5&#180;-biotin label at each RNA (Figure 1d).&#160;
Converting &#968; to its CMC-&#968; adduct for 2D-HELS MS Seq. &#968; is a difficult nucleotide modification for MS-based sequencing because it has the same mass as uridine (U). To differentiate these two bases from each other, we treat the RNA with CMC, which converts a &#968; to a CMC-&#968; adduct (see Section 5). The adduct has a different mass than U and can be differentiated in the 2D-HELS MS Seq. Figure 2a shows the HPLC profile of the crude product of the reaction converting &#968; to its CMC-adduct in RNA #6. By integrating their UV peaks, we calculated the percent conversion and 42% &#968; is converted to its CMC-&#968; adduct after the process illustrated in Section 5. After acid degradation and LC-MS measurement, we manually acquired the sequence based on both non-CMC-converted ladders and CMC-converted ladders identified from the algorithm-processed data8,9. A red curve branches up off of the grey curve starting from &#968; at position 8 in RNA #6 (Figure 2b) due to partial conversion of &#968; to the CMC-&#968; adduct. Because of the mass and hydrophobicity of the CMC, this conversion results in a 252.2076 Dalton increase in mass and a significant increase in tR for each CMC-&#968; adduct-containing ladder component when compared to its unconverted counterpart. Thus, a dramatic shift starting at position 8 in RNA #6 can be observed in the 2D mass-tR plot, indicating that position 8 is indeed a &#968; in RNA #6.
Sequencing RNA mixtures. A mixture of five different RNA strands is sequenced by the 2D-HELS MS Seq approach with 3&#180;-end labeling (see Section 8). The concern for sequencing mixed RNAs is that multiple ladder curves in the 2D mass-tR plot may overlap with each other when they all share the same starting points (the hydrophobic tag in the 2D mass-tR plot). However, base calling is made one by one, each based on a mass difference between two adjacent ladder fragments in the MFE data.&#160;The correct base-call can be made as long as each mass difference matches well (a PPM MS difference &#60; 10) with one of the theoretical masses of canonical or modified nucleotides in the data pool8,9. In the analysis of the multiplexed RNA samples, the typical processing and base-calling algorithm used in Figure 1 and 2 is not used mainly due to the significantly increased data complexity resulting from the mixture. These sequences are base-called manually via calculating the mass difference between two adjacent mass ladder fragments, and comparing it to the theoretical mass of the nucleotide in the data pool9. Any matched base with a mass PPM &#60;10 is chosen as the base identity at this position. With this base-by-base manual calculation for base-calling, all sequences in the mixture are accurately sequenced. OriginLab software is used to re-construct a 2D mass-tR plot, in which the starting tR for each sequence is normalized systematically for better visualizing five different RNA sequences (Figure 3). Without such normalization, the letter codes (i.e., A, C, G, and U) for the sequences of all five RNA would be crowded together on the plot (Figure S1), resulting in less ease of visualization compared to that reported in Figure 3. The sequencing results demonstrate that 2D-HELS MS Seq approach is not just limited to sequencing of&#160;purified single-stranded RNAs, but also, more importantly, RNA mixtures with multiple RNA strands. Algorithms are currently under development to automate the process of base-calling and sequence generation.
<img alt="Figure 1" class="xfigimg" src="/files/ftp_upload/61281/61281fig1v2.jpg" /> 
Figure 1. 2D-HELS MS Seq of representative RNA samples. (a) Workflow for 2D-HELS MS Seq. The major steps include 1) hydrophobic tag-labeling of RNA to be sequenced, 2) acid hydrolysis, 3) LC-MS measurement, 4) extraction and analysis of MFE data, and 5) sequence generation via algorithms or manual calculation. (b) Sample preparation protocol including introducing a biotin tag to the 3&#180;-end of RNA for 2D-HELS MS Seq. (c) Separation of the 3&#180;-ladder from the 5&#180;-ladder and other undesired fragments in a 2D mass-retention time (tR) plot based on systematic changes in tRs of 3&#180;-biotin-labeled mass-tR ladder fragments of RNA #1 (19 nt). The sequences are de novo and automatically read out directly by a base-calling algorithm9. (d) Simultaneous sequencing of 5&#180;-biotin labeled RNA #1 and RNA #2, 19 nt and 20 nt, respectively. Methods for introducing a biotin tag to the 5&#180;-end of RNA are different than that of 3&#180;-biotinylation, and can be found in the previous published protocol9. The 5&#180;-end of two RNAs (RNA #1 and RNA #2) are biotinylated and their 5&#180;-biotinylated ladders can be easily identified; both 5&#180;-biotinylated ladders are easily separated from their unlabeled 3&#180;-ladders in the 2D mass-tR plot after LC-MS, because the biotinylated ladder components have the larger tR shifts due to the hydrophobicity of the biotin, while unlabeled ladder components are in the lower tR region. Although the 5&#180;-ladders and 3&#180;-ladders co-exist, they do not interfere with the sequence interpretation of two mixed RNA strands. Each sequence of these two RNAs are manually acquired from 5&#180;-biotinylated ladders based on the computational algorithm-processed data8,9. This figure has been modified from Zhang et al.9. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/61281/61281fig1v2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 2" class="xfigimg" src="/files/ftp_upload/61281/61281fig2v2.jpg" /> 
Figure 2. Converting pseudouridine (&#968;) to its adduct for 2D-HELS MS Seq. (a) HPLC profile of the crude product of the reaction converting &#968; to its CMC adduct in a 20 nt RNA (RNA #6) that contains one &#968;. (b) Sequencing of a &#968;-containing RNA #6. The conversion of the &#968; to the CMC-&#968; adducts (&#968;*) results in a 252.2076 Dalton increase in mass and a significant increase in tR because of its mass and hydrophobicity of the CMC. Thus, a dramatic shift starting at the position of 8 can be observed in the mass-tR plot, indicating that this is a &#968; at the position of 8 in the RNA sequence. The sequences are manually acquired based on the computational algorithm-processed data8,9. This figure has been modified from Zhang et al.9. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/61281/61281fig2v2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 3" class="xfigimg" src="/files/ftp_upload/61281/61281fig3.jpg" /> 
Figure 3. Sequencing RNA mixtures containing five&#160;distinct RNAs. A biotin is used to label each RNA at their 3&#180;-end before 2D-HELS MS Seq. For each sequence, the starting tR values are normalized systematically to start at 7 min intervals for ease of visualization. The absolute differences between the starting tR value and subsequent tRs remain unchanged for each of the five RNAs, and thus it is easier to visualize each of them in the same plot. All bases are identified by manually calculating the mass differences of two adjacent ladder components and matching them with the theoretical mass differences in the RNA nucleotide and modification database8; plots for Figure 3 are re-constructed using OriginLab based on manual base-calling and sequencing data (see Section of Sequencing RNA mixtures in Representative Results). The 2D mass-tR figure of the five mixed RNAs without tR normalization is shown in Figure S1. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/61281/61281fig3large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
Figure S1.&#160; <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/61281/SI_6-22.docx" target="_blank">Please click here to download this file.</a>

Watch this Scientific Journal Video about 2D-HELS MS Seq: A General LC-MS-Based Method for Direct and de novo Sequencing of RNA Mixtures with Different Nucleotide Modifications at JoVE.com

2D-HELS MS Seq: A General LC-MS-Based Method for Direct and de novo Sequencing of RNA Mixtures with Different Nucleotide Modifications

Here, we describe a detailed protocol for an LC-MS-based sequencing method that can be used as a direct method to sequence short RNA (&#60;35 nt per run) without a cDNA intermediate, and as a general method to sequence different nucleotide modifications in a single study at single-base precision.

Mass spectrometry (MS)-based sequencing approaches have been shown to be useful in&#160;direct&#160;sequencing&#160;RNA without the need for a ...

2d-hels-ms-seq-general-lc-ms-based-method-for-direct-de-novo

Nanyang Technological University

Research

JoVE Journal

Genetics

1.9K Views.  New York Institute of Technology. Hier beschrijven we een gedetailleerd protocol voor een LC-MS-gebaseerde sequencingmethode die kan worden gebruikt als een directe methode om kort RNA (

Video: 2D-HELS MS Seq: Een algemene LC-MS-gebaseerde methode voor directe en de novo sequencing van RNA-mengsels met verschillende nucleotidemodificaties

A Method for Measuring RNA N6-methyladenosine Modifications in Cells and Tissues

N6-Methyladenosine (m6A) modifications of RNA are diverse and ubiquitous amongst eukaryotes. They occur in mRNA, rRNA, tRNA, and microRNA. Recent studies have revealed that these reversible RNA modifications affect RNA splicing, translation, degradation, and localization. Multiple physiological processes, like circadian rhythms, stem cell pluripotency, fibrosis, triglyceride metabolism, and obesity are also controlled by m6A modifications. Immunoprecipitation/sequencing, mass spectrometry, and modified northern blotting are some of the methods commonly employed to measure m6A modifications. Herein, we present a northeastern blotting technique for measuring m6A modifications. The current protocol provides good size separation of RNA, better accommodation and standardization for various experimental designs, and clear delineation of m6A modifications in various sources of RNA. While m6A modifications are known to have a crucial impact on human physiology relating to circadian rhythms and obesity, their roles in other (patho)physiological states are unclear. Therefore, investigations on m6A modifications have immense possibility to provide key insights into molecular physiology.

A Method for Measuring RNA N6-methyladenosine Modifications in Cells and Tissues

A modified northern blotting method for measuring N6-methyladenosine (m6A) modifications in RNA is described. The current method can detect modifications in diverse RNAs and controls under various experimental designs.

N6-Methyladenosine (m6A) modifications of RNA are diverse and ubiquitous amongst eukaryotes. They occur in mRNA, rRNA, tRNA, and microRNA. Recent studies ...

Biochemistry

Sample Preparation for Mass Spectrometry-based Identification of RNA-binding Regions

Noncoding RNAs play important roles in several nuclear processes, including regulating gene expression, chromatin structure, and DNA repair. In most cases, the action of noncoding RNAs is mediated by proteins whose functions are in turn regulated by these interactions with noncoding RNAs. Consistent with this, a growing number of proteins involved in nuclear functions have been reported to bind RNA and in a few cases the RNA-binding regions of these proteins have been mapped, often through laborious, candidate-based methods.
Here, we report a detailed protocol to perform a high-throughput, proteome-wide unbiased identification of RNA-binding proteins and their RNA-binding regions. The methodology relies on the incorporation of a photoreactive uridine analog in the cellular RNA, followed by UV-mediated protein-RNA crosslinking, and mass spectrometry analyses to reveal RNA-crosslinked peptides within the proteome. Although we describe the procedure for mouse embryonic stem cells, the protocol should be easily adapted to a variety of cultured cells.

We describe a protocol to identify RNA-binding proteins and map their RNA-binding regions in live cells using UV-mediated photocrosslinking and mass spectrometry.

Noncoding RNAs play important roles in several nuclear processes, including regulating gene expression, chromatin structure, and DNA repair. In most ...

Improving Small RNA-seq: Less Bias and Better Detection of 2'-O-Methyl RNAs

The study of small RNAs (sRNAs) by next-generation sequencing (NGS) is challenged by bias issues during library preparation. Several types of sRNA such as plant microRNAs (miRNAs) carry a 2&#39;-O-methyl (2&#39;-OMe) modification at their 3&#39;&#160;terminal nucleotide. This modification adds another difficulty as it inhibits 3&#39;&#160;adapter ligation. We previously demonstrated that modified versions of the &#39;TruSeq (TS)&#39;&#160;protocol have less bias and an improved detection of 2&#39;-OMe RNAs. Here we describe in detail protocol &#39;TS5&#39;, which showed the best overall performance. TS5 can be followed either using homemade reagents or reagents from the TS kit, with equal performance.

We present a detailed small RNA library reparation protocol with less bias than standard methods and an increased sensitivity for 2&#39;-O-methyl RNAs. This protocol can be followed using homemade reagents to save cost or using kits for convenience.

The study of small RNAs (sRNAs) by next-generation sequencing (NGS) is challenged by bias issues during library preparation. Several types of sRNA such as ...

Analyzing tRNA Dynamics Using AQRNA-seq

AQRNA-seq for Quantifying Small RNAs

AQRNA-seq provides a direct linear relationship between sequencing read counts and small RNA copy numbers in a biological sample, thus enabling accurate quantification of the pool of small RNAs. The AQRNA-seq library preparation procedure described here involves the use of custom-designed sequencing linkers and a step for reducing methylation RNA modifications that block reverse transcription processivity, which results in an increased yield of full-length cDNAs. In addition, a detailed implementation of the accompanying bioinformatics pipeline is presented. This demonstration of AQRNA-seq was conducted through a quantitative analysis of the 45 tRNAs in Mycobacterium bovis BCG harvested on 5 selected days across a 20-day time course of nutrient deprivation and 6 days of resuscitation. Ongoing efforts to improve the efficiency and rigor of AQRNA-seq will also be discussed here. This includes exploring methods to obviate gel purification for mitigating primer dimer issues after PCR amplification and to increase the proportion of full-length reads to enable more accurate read mapping. Future enhancements to AQRNA-seq will be focused on facilitating automation and high-throughput implementation of this technology for quantifying all small RNA species in cell and tissue samples from diverse organisms.

Author Spotlight: AQRNA-seq Role in Mapping Small RNAs and Unraveling Protein Translation Mechanisms

Absolute quantification RNA sequencing (AQRNA-seq) is a technology developed to quantify the landscape of all small RNAs in biological mixtures. Here, both the library preparation and data processing steps of AQRNA-seq are demonstrated, quantifying changes in the transfer RNA (tRNA) pool in Mycobacterium bovis BCG during starvation-induced dormancy.

AQRNA-seq provides a direct linear relationship between sequencing read counts and small RNA copy numbers in a biological sample, thus enabling accurate ...

A Rapid High-throughput Method for Mapping Ribonucleoproteins (RNPs) on Human pre-mRNA

Sequencing RNAs that co-immunoprecipitate (co-IP) with RNA binding proteins has increased our understanding of splicing by demonstrating that binding location often influences function of a splicing factor.  However, as with any sampling strategy the chance of identifying an RNA bound to a splicing factor is proportional to its cellular abundance.  We have developed a novel in vitro approach for surveying binding specificity on otherwise transient pre-mRNA. This approach utilizes a specifically designed oligonucleotide pool that tiles across introns, exons, splice junctions, or other pre-mRNA.  The pool is subjected to some kind of molecular selection. Here, we demonstrate the method by separating the oligonucleotide into a bound and unbound fraction and utilize a two color array strategy to record the enrichment of each oligonucleotide in the bound fraction. The array data generates high-resolution maps with the ability to identify sequence-specific and structural determinates of ribonucleoprotein (RNP) binding on pre-mRNA.  A unique advantage to this method is its ability to avoid the sampling bias towards mRNA associated with current IP and SELEX techniques, as the pool is specifically designed and synthesized from pre-mRNA sequence. The flexibility of the oligonucleotide pool is another advantage since the experimenter chooses which regions to study and tile across, tailoring the pool to their individual needs.  Using this technique, one can assay the effects of polymorphisms or mutations on binding on a large scale or clone the library into a functional splicing reporter and identify oligonucleotides that are enriched in the included fraction.  This novel in vitro high-resolution mapping scheme provides a unique way to study RNP interactions with transient pre-mRNA species, whose low abundance makes them difficult to study with current in vivo techniques.  



A Rapid High-throughput Method for Mapping Ribonucleoproteins RNPs on Human pre-mRNA

Due to the transient nature of pre-mRNA, it can be difficult to isolate and study in vivo. Here, we present a novel in vitro approach to investigate RNA-protein interactions using a synthetic oligo pool that tiles across selected regions of pre-mRNA.

Sequencing RNAs that co-immunoprecipitate (co-IP) with RNA binding proteins has increased our understanding of splicing by demonstrating that binding ...

Biology

Characterizing RNA Modifications in Single Neurons Using Mass Spectrometry

Post-transcriptional modifications (PTMs) of RNA represent an understudied mechanism involved in the regulation of translation in the central nervous system (CNS). Recent evidence has linked specific neuronal RNA modifications to learning and memory paradigms. Unfortunately, conventional methods for the detection of these epitranscriptomic features are only capable of characterizing highly abundant RNA modifications in bulk tissues, precluding the assessment of unique PTM profiles that may exist for individual neurons within the activated behavioral circuits. In this protocol, an approach is described&#8212;single-neuron RNA modification analysis by mass spectrometry (SNRMA-MS)&#8212;to simultaneously detect and quantify numerous modified ribonucleosides in single neurons. The approach is validated using individual neurons of the marine mollusk, Aplysia californica, beginning with surgical isolation and enzymatic treatment of major CNS ganglia to expose neuron cell bodies, followed by manual single-neuron isolation using sharp needles and a micropipette. Next, mechanical and thermal treatment of the sample in a small volume of buffer is done to liberate RNA from an individual cell for subsequent RNA digestion. Modified nucleosides are then identified and quantified using an optimized liquid chromatography-mass spectrometry method. SNRMA-MS is employed to establish RNA modification patterns for single, identified neurons from A. californica that have known morphologies and functions. Examples of qualitative and quantitative SNRMA-MS are presented that highlight the heterogeneous distribution of RNA modifications across individual neurons in neuronal networks.

Post-transcriptional modifications of RNA represent an understudied layer of translation regulation that has recently been linked to central nervous system plasticity. Here, sample preparation and liquid chromatography-tandem mass spectrometry approach is described for simultaneous characterization of numerous RNA modifications in single neurons.

Post-transcriptional modifications (PTMs) of RNA represent an understudied mechanism involved in the regulation of translation in the central nervous ...

Chemistry

Probing RNA Structure with Dimethyl Sulfate Mutational Profiling with Sequencing In Vitro and in Cells

The role of RNA structure in virtually any biological process has become increasingly evident, especially in the past decade. However, classical approaches to solving RNA structure, such as RNA crystallography or cryo-EM, have failed to keep up with the rapidly evolving field and the need for high-throughput solutions. Mutational profiling with sequencing using dimethyl sulfate (DMS) MaPseq is a sequencing-based approach to infer the RNA structure from a base&#39;s reactivity with DMS. DMS methylates the N1 nitrogen in adenosines and the N3 in cytosines at their Watson-Crick face when the base is unpaired. Reverse-transcribing the modified RNA with the thermostable group II intron reverse transcriptase (TGIRT-III) leads to the methylated bases being incorporated as mutations into the cDNA. When sequencing the resulting cDNA and mapping it back to a reference transcript, the relative mutation rates for each base are indicative of the base&#39;s &#34;status&#34; as paired or unpaired. Even though DMS reactivities have a high signal-to-noise ratio both in vitro and in cells, this method is sensitive to bias in the handling procedures. To reduce this bias, this paper provides a protocol for RNA treatment with DMS in cells and with in vitro transcribed RNA.

Probing RNA Structure with Dimethyl Sulfate Mutational Profiling with Sequencing In Vitro and in Cells

The protocol provides instruction for modifying RNA with dimethyl sulfate for mutational profiling experiments. It includes in vitro and in vivo probing with two alternative library preparation methods.

The role of RNA structure in virtually any biological process has become increasingly evident, especially in the past decade. However, classical ...

A Nonsequencing Approach for the Rapid Detection of RNA Editing

RNA editing is a process that leads to posttranscriptional sequence alterations in RNAs. Detection and quantification of RNA editing rely mainly on Sanger sequencing and RNA sequencing techniques. However, these methods can be costly and time-consuming. In this protocol, a portable microtemperature gradient gel electrophoresis (&#181;TGGE) system is used as a nonsequencing approach for the rapid detection of RNA editing. The process is based on the principle of electrophoresis, which uses high temperatures to denature nucleic acid samples as they move across a polyacrylamide gel. Across a range of temperatures, a DNA fragment forms a gradient of fully double-stranded DNA to partially separated strands and then to entirely separated single-stranded DNA. RNA-edited sites with distinct nucleotide bases produce different melting profiles in &#181;TGGE analyses. We used the &#181;TGGE-based approach to characterize the differences between the melting profiles of four edited RNA fragments and their corresponding nonedited (wild-type) fragments. Pattern Similarity Scores (PaSSs) were calculated by comparing the band patterns produced by the edited and nonedited RNAs and were used to assess the reproducibility of the method. Overall, the platform described here enables the detection of even single base mutations in RNAs in a straightforward, simple, and cost-effective manner. It is anticipated that this analysis tool will aid new molecular biology findings.

Rapid detection and reliable quantification of RNA editing events at a genomic scale remain challenging and currently rely on direct RNA sequencing methods. The protocol described here uses microtemperature gradient gel electrophoresis (&#181;TGGE) as a simple, quick, and portable method of detecting RNA editing.

RNA editing is a process that leads to posttranscriptional sequence alterations in RNAs. Detection and quantification of RNA editing rely mainly on Sanger ...

Bioengineering

Highly Efficient Ligation of Small RNA Molecules for MicroRNA Quantitation by High-Throughput Sequencing

MiRNA cloning and high-throughput sequencing, termed miR-Seq, stands alone as a transcriptome-wide approach to quantify miRNAs with single nucleotide resolution. This technique captures miRNAs by attaching 3&#8217; and 5&#8217; oligonucleotide adapters to miRNA molecules and allows de novo miRNA discovery. Coupling with powerful next-generation sequencing platforms, miR-Seq has been instrumental in the study of miRNA biology. However, significant biases introduced by oligonucleotide ligation steps have prevented miR-Seq from being employed as an accurate quantitation tool. Previous studies demonstrate that biases in current miR-Seq methods often lead to inaccurate miRNA quantification with errors up to 1,000-fold for some miRNAs1,2. To resolve these biases imparted by RNA ligation, we have developed a small RNA ligation method that results in ligation efficiencies of over 95% for both 3&#8217; and 5&#8242; ligation steps. Benchmarking this improved library construction method using equimolar or differentially mixed synthetic miRNAs, consistently yields reads numbers with less than two-fold deviation from the expected value. Furthermore, this high-efficiency miR-Seq method permits accurate genome-wide miRNA profiling from in vivo total RNA samples2.

MicroRNAs (miRNAs) are a widely conserved class of regulatory molecules. Here we describe a miRNA cloning method that relies upon two potent ligation steps followed by high-throughput sequencing. Our method permits accurate genome-wide quantitation of miRNAs.

MiRNA cloning and high-throughput sequencing, termed miR-Seq, stands alone as a transcriptome-wide approach to quantify miRNAs with single nucleotide ...

Targeted RNA Sequencing Assay to Characterize Gene Expression and Genomic Alterations

RNA sequencing (RNAseq) is a versatile method that can be utilized to detect and characterize gene expression, mutations, gene fusions, and noncoding RNAs. Standard RNAseq requires 30 - 100 million sequencing reads and can include multiple RNA products such as mRNA and noncoding RNAs. We demonstrate how targeted RNAseq (capture) permits a focused study on selected RNA products using a desktop sequencer. RNAseq capture can characterize unannotated, low, or transiently expressed transcripts that may otherwise be missed using traditional RNAseq methods. Here we describe the extraction of RNA from cell lines, ribosomal RNA depletion, cDNA synthesis, preparation of barcoded libraries, hybridization and capture of targeted transcripts and multiplex sequencing on a desktop sequencer. We also outline the computational analysis pipeline, which includes quality control assessment, alignment, fusion detection, gene expression quantification and identification of single nucleotide variants. This assay allows for targeted transcript sequencing to characterize gene expression, gene fusions, and mutations.

We describe a targeted RNA sequencing-based method that includes preparation of indexed cDNA libraries, hybridization and capture with custom probes and data analysis to interrogate selected transcripts for gene expression, mutations, and gene fusions. Targeted RNAseq permits cost-effective, rapid evaluation of selected transcripts on a desktop sequencer.

RNA sequencing (RNAseq) is a versatile method that can be utilized to detect and characterize gene expression, mutations, gene fusions, and noncoding ...

2D-HELS MS Seq: A General LC-MS-Based Method for Direct and de novo Sequencing of RNA Mixtures with Different Nucleotide Modifications

Samenvatting

Meer video's verkennen

Hoofdstukken in deze video