Unraveling the RNA Modification Code with Mass Spectrometry

Abstract

The discovery and analysis of modifications on proteins and nucleic acids has provided functional information that has rapidly accelerated the field of epigenetics. While protein post-translational modifications (PTMs), especially on histones, have been highlighted as critical components of epigenetics, the post-transcriptional modification of RNA has been a subject of more recently emergent interest. Multiple RNA modifications have been known to be present in tRNA and rRNA since the 1960s, but the exploration of mRNA, small RNA, and inducible tRNA modifications remains nascent. Sequencing-based methods have been essential to the field by creating the first epitranscriptome maps of m⁶A, m⁵C, hm⁵C, pseudouridine, and inosine; however, these methods possess significant limitations. Here, we discuss the past, present, and future of the application of mass spectrometry (MS) to the study of RNA modifications.

Introduction

The central dogma of biology describes DNA as being the blueprint for RNA and eventually the translation of proteins. While this accurately represents much of the function of nucleic acids, RNA can have several other distinct functions ranging from translation inhibition to the editing of its own transcripts through self-splicing mechanisms. In addition to this deviation from the central dogma, the canonical RNA moieties adenosine, guanosine, cytidine, and uridine were first recognized as incomplete with the discovery of pseudouridine in 1956, long considered to be the fifth base of RNA^1-3. This 5-base paradigm persisted for decades despite the discovery of additional modifications before the end of the 1950s⁴. Mass spectrometry (MS) was first applied to nucleic acids by McCloskey and Biemann in 1962, when they began studying multiple tRNA modifications simultaneously⁵. This would eventually lead to the understanding of the structural importance of tRNA modifications; however, this was considered the only role of these modifications for many years. This perspective, along with technological limitations, led to stagnation of the biologically relevant work in the field until the discovery of m6A in mRNA led to an RNA modification renaissance. m6A was shown to be critical in multiple cellular processes across species, and it is necessary for life in many eukaryotes^6,7. Recently, it has been shown to be required for proper splicing and is inducible in various conditions to generate the correct spliceoform of certain proteins⁸. The broad resurgence of interest in RNA modifications led to additional discoveries of functional importance, including RNA processing mechanisms, tRNA structure and stability, host recognition through the m7G 5’ mRNA cap, and translation fidelity, among others⁹. Most recently, additional modifications have been found internally in mRNA, including m6Am, m1A, and m5C^10-12. These discoveries clearly demonstrate that more mRNA modifications could be lurking at low abundances, and the functions of many of these modifications have not been characterized. Additionally, interest has returned to tRNA modifications due to disabilities associated with mutations in tRNA and/or their modifying proteins, variable wobble position modifications, and inducible changes in modifications¹³. The functions of RNA modifications have led to their writers and erasers being targeted for therapy, and the rapidly expanding knowledge on these pathways has high potential for medical impact. As such, the discovery of novel RNA modifications and the expansion of methods to identify them is paramount to enabling these translational studies.

The advancement of epitranscriptomics remains full of challenges, and MS presents one of the most powerful tools in overcoming them. MS is an analytical technique that has long been considered the gold standard for both identification and quantification of proteins and small molecules, such as metabolites. The technology has been applied to nucleic acids to less prominence, but it has proven critical, if not necessary, to the broad analysis of nucleosides, nucleotides, oligonucleotides, and interacting proteins. Indeed, MS is the most effective technique at identifying previously undiscovered nucleic acid modifications and remains unsurpassed in detecting proteins interacting with an immobilized oligonucleotide¹⁴. These factors demonstrate the unique utility of applying mass spectrometry to epitranscriptomics. In this review, we discuss the past, present, and future of MS techniques and technologies that allow for interrogation of RNA modification abundance, localization, structure, function, and interaction.

The biological impact of RNA modifications is most strongly related to of the RNA type bearing the mark. Modifications of mRNA, especially m⁶A, have been most extensively studied since the advent of modification sequencing^9,15. Modifications are much less common in mRNA than other RNA types, but due to their low stoichiometry, understanding their functions without localization was insurmountable. Sequencing technologies yielded a surge in excitement and investigation into the effects of these modifications, but many of their functions remain enigmatic. However, broad associations have been drawn for some of the modifications in mRNA, which include m⁷G, inosine, pseudouridine, m⁵C, m⁶A, and ac4C (Fig. 1). The best studied, m⁶A, has been implicated in structure, splicing, and viral response^8,16-20. Another well-understood modification is m7G, which is used to identify host RNA when incorporated into the 5’ cap²¹. Unlike poly(A) tails, which are often used to pull down mRNA, the m⁷G cap is present in all translatable mRNA²². Another modification, known for being deposited in DNA as well as RNA, is m5C. While its role is still being fully developed, its known writer protein NSUN2 is required for HIV genomic replication and for HIV transcription^23,24. As it is studied by bisulfite sequencing, it can be localized to sites of other cytidine modifications, such as hm⁵C, m³C, and ac⁴C²⁵. The latter modification, ac4C, is the mark most recently discovered to be present in mammalian mRNA and like other mRNA modifications, it has been shown to increase stability and translation fidelity²⁵. Previous studies using bisulfite sequencing misidentified ac⁴C as m⁵C, highlighting the low accuracy of the approach²⁶. Importantly, while mass spectrometry has not been able to analyze a full profile of mRNA in a single run, it does not suffer from the same misidentification issues as sequencing. Sequencing of m⁶A had similar issues in its infancy, with many of the hits being false positives^27,28. The technology has improved dramatically in its accuracy since, and as the field has grown, more awareness of technological limitations has been acknowledged^29-31. For example, m⁶A identifications were not always noted as bases that could contain other modifications as well, such as m⁶Am^32,33. In fact, m6Am recently has been identified to mark alternative transcription start sites without altering translation¹⁰.

Figure 1:

Functions of modifications found on mRNA, tRNA, and rRNA.

Like many of the modifications of mRNA, marks on tRNA can be critical for translation fidelity. The wobble position, 34, and a nearby stabilizing position, 37, often bear modifications that are found nowhere else^34,35. Queuosine, 5-carbamoylhydroxymethyluridine, wybutosine, and other bulky bases can facilitate accurate codon recognition³⁶. Heavily modified uridines are common, often derivatized off the 5 position (Fig. 2), forming a stable carbon-carbon bond. Other marks decorate the rest of the 76-90mer tRNAs, many important in proper folding, recognition, and aminoacylation^37,38. Of greater recent interest is the discovery of functional tRNA fragments (tRFs), which can alter gene expression and induce paternally-derived epigenetic changes in offspring^39-41. In particular, m5C has been implicated in the function of some tRFs, however, the mechanism by which is operates is unknown³⁹. Further, derivations of m5C such as hm5C, hm5Cm, and further oxidations may play roles in tRF generation and function^42,43. These modifications can be challenging to study as the diverse modification profile of tRNA can create issues for sequencing and the marks of interest cannot trivially be distinguished in sequencing²⁶. On the other hand, tRNA is one of the simplest analytes for mass spectrometric analysis. As the tRNAs are small, they can be digested into uniform fragments of easily analyzable size, the sequence degeneracy is low allowing for small oligonucleotides to be mapped to specific tRNAs, and the many of the modifications can be accurately identified. While sequencing has many useful applications for investigating RNA modifications, mass spectrometry presents multiple context dependent advantages that should not be overlooked.

Figure 2:

Example numbering schemes for pyrimidines (uracil, top) and purines (guanosine, bottom).

Beyond just the identification of RNA modifications is the study of their relationships with the writer, reader, and eraser proteins which regulate them. Recent discoveries with m⁶A sequencing have shown direct splicing regulation of HNRNPs and Serine and Arginine rich Splicing Factors (SRSFs) reading and interacting with the modification directly or interacting with known m⁶A reader YTHDC1 respectively, for exon definition (Fig. 3) ^44-47. During heat shock transcripts are regulated by RNA modifications, with the heat shock protein 70 (HSP70) transcripts bound by the m⁶A reader protein YTHDF2 blocking the eraser protein FTO and allowing for the selective translation of the HSP transcript^48-51. Protein and RNA interactions themselves can also change the epitranscriptomic landscape by modulating the expression of specific transcripts, examples of this being targeted mRNA degradation mediated by Argonaut 2/RISC complex, and context-dependent inhibition of the histone methyltransferase activity of the PRC2 complex by RNA binding^52,53. Methods to identify both the key interactors of RNA modifications and the context of RNA modifications within their greater role in biology, but current work on the RNA modification and protein interactions is limited due to the merger of these two fields.

Figure 3:

Known writers, readers, and erasers of common nucleotide modifications.

RNA Nucleotides and Nucleosides

Analysis of RNA at the monomer level has typically been performed by electrospray ionization liquid chromatography-tandem mass spectrometry (ESI LC-MS/MS) of nucleosides (dephosphorylated) or nucleotides (phosphorylated). While nucleotides can provide slightly more information due to uncommon modifications such as cyclizations, methylations, and the presence of multiple phosphate groups between sugars, analyzing nucleosides is simpler in terms of both LC and MS. When studying these molecules, the phosphorylation state, LC, and MS methodology must be considered interdependently.

For nucleotide analysis, LC is typically performed with the use of ion-pairing agent chromatography or of hydrophilic interaction chromatography (HILIC)^54,55. Using ion-pairing agents allows these phosphorylated molecules to be bound to commonly used C18 resin in a reverse-phase chromatographic approach; however, ion-pairing agents can linger in LCs lines, suppress ionization at the MS source, and contaminate the interior of the MS^56-58. A common alternative approach, HILIC, uses hydrophilic stationary phases and ammonium acetate as opposed to C18 and ion-pairing agents, which is less disruptive to ionization, fragmentation, and instrumentation⁵⁹. Because HILIC chromatography depends on hydrophilicity, which is not highly variable between nucleotides largely due to the presence of the phosphate group, separation of analytes is not as robust as it is with reverse-phase approaches⁶⁰. However, because the analytes elute at high concentrations of acetonitrile, ionization is more efficient and MS sensitivity can be superior.

Analyzing RNA at the nucleoside level is a more popular approach that typically employs reverse-phase chromatography. This is most often performed with ammonium acetate buffers and C18 stationary phase but the use of formic acid buffers with alternative or additional stationary phases is gaining in popularity⁶¹. This approach was presented by the McCloskey lab in 1990 and prominently used by McCloskey’s and Limbach’s groups from the 1990s onward, and remains popular due its simplicity, reproducibility, and efficacy of analyte separation^62,63. While this methodology is effective, there remains room for improvement. For example, development of advanced techniques with improved analyte separation using formic acid buffers, which increase sensitivity and are the same as those used for conventional proteomics, would increase adoption of nucleoside analysis in MS labs.

Although C18 would be the most attractive stationary phase to use with formic acid buffers, it has significant limitations in retaining pyrimidines⁶⁴. An ideal technique would rapidly separate all isobaric species while maintaining high reproducibility and sensitivity, however, the currently available methodology is useful for the vast majority of applications. This type of approach is ideal for labs using chromatographic systems; however, chromatography is not required for analysis. Although the technique does not yet have widespread use, the Fabris lab has demonstrated reliable quantitation of modified nucleosides through direct infusion⁶⁵. This approach allows for much more rapid quantitation but requires a very clean sample and a high-resolution instrument capable of MSⁿ to differentiate base-modified isobars. In particular, isobaric species such as m3C and m5C must be differentiated using unique fragment ions and high enough fragmentation energy to generate the unique fragments^66,67: m3C can be identified by the unique fragment ion corresponding to loss of ribose and CH₃NCO; fragment ion at 69 m/z, and m5C can be identified by the unique fragment ion corresponding to loss of ribose and NH2CHO; fragment ion at 81 m/z (Fig 4.).

Figure 4.

Fragmentation spectra of two isobaric species methyl cytidine bases.

Chromatographic methods are relatively approachable; however, MS methods can be more complex due to their dependence on desired information and chromatography. Chromatographic methods will indicate which polarity the MS must be run in, which affects multiple aspects of data acquisition. For example, attempting to detect HILIC-resolved nucleotides in positive mode will yield significantly lower sensitivity than detection in negative mode, as the phosphate group must be protonated in addition to the protonation of a nitrogen in the base⁶⁰. Analyzing nucleotides in negative mode decreases efficacy of MS/MS scans, as fragmentation primarily occurs at the glycosidic bond and the phosphate group is the negative charge carrier. Thus, after fragmentation, the sugar will be observed rather than the base, preventing differentiation of base-modified isobars through unique fragment ions. Indeed, detecting nucleosides is substantially easier than detecting nucleotides, and it is recommended to dephosphorylate nucleotides unless there are specific modifications of interest involving the phosphate group. This will mean that the MS can be run with higher sensitivity in positive mode and that fragmentation will allow for detection of product ions of the base.

The two other considerations, instrumentation and desired data, are also intertwined. While there is a plethora of instrument options, the context of nucleoside analysis has different methodologies and instrument requirements for identification and high precision quantification. The minimalist instrumentation would include a low-resolution MS lacking the ability to fragment, such as a single quadrupole. A single quadrupole MS can identify modified species at the monoisotopic mass level; however, it requires chromatography capable of separating all isobaric species to yield accurate quantitation. High-resolution instruments must be capable of MS/MS and are useful for accurate identification of species of very similar mass (e.g. 5-carbamoylmethyl-2-thiouridine and 5-carbamoylhydroxymethyluridine, which differ by 0.018 Da), quantitation, and for discovery of new RNA modifications⁶⁸. Targeted fragmentation allows for quantitation at the MS² level, which has lower signal to noise than the MS¹ level but is incompatible with discovery methodology. Instruments capable of MS³ also provide the benefit of being able to identify isobaric modifications added to the nucleobase as they do not always show different fragmentation profiles in relatively low energy collision induced dissociation (CID) MS/MS. Further, as the data will often include full scans and data-dependent acquisition (DDA) MS/MS, there is the potential to mine old data if novel modifications are discovered after analysis. The most quantitative MS is the triple quadrupole (QQQ), which is a very rapid instrument ideally suited for targeted analysis. The QQQ is most sensitive due to its rapid targeting capabilities but will only analyze modifications that are specified a priori⁶⁹. Because of these properties, it is best to have both a high-resolution instrument and a QQQ to use differentially based on experimental design.

Fragmentation of nucleosides is dependent on instrumentation; however, it is relatively trivial. As noted previously, the weakest bond in a nucleoside is the glycosidic bond, so the primary fragmentation product in positive mode will most often be the nucleobase. Fragmentation is most commonly performed using beam-type CID in QQQ instruments and higher-energy collisional dissociation (HCD) in high-resolution instruments. Product ions are most commonly used for accurate identification, however, higher-energy CID, higher-energy HCD, and stepped collision energies can be used to generate additional fragment ions, which may be required for isobar identification of overlapping chromatographic peaks or direct infusions approaches⁷⁰. The resultant fragmentation spectra appear similar to the MS³ spectra (Fig. 4).

Collectively, these methods demonstrate high applicability and versatility. Indeed, there is no ideal method and method selection can be cleanly separated into two categories, quantitation or discovery, or, more simply, analysis of either knowns or unknowns. A future method may optimally converge the two approaches to some extent; however, a targeted method will always have greater sensitivity than untargeted due to noise reduction. Thus, though the field is open to novel techniques, the methodology is at adequate maturity for the analysis of nucleotides and nucleosides to be relatively simple.

Oligoribonucleotide Identification and Quantitation by Mass Spectrometry

While the analysis of nucleosides and nucleotides can provide rich information on the fundamental pieces of RNA, it does not provide information about the location of modifications within transcripts. Current NGS sequencing methods are limited to single modifications per sequencing technique and multiple modification sequencing results must be overlaid to expand coverage, increasing the possibility of experimental bias. The use of LC-MS/MS allows for an unbiased approach to the sequencing of RNA oligonucleotides which can identify any number of modifications, given specific criteria are met. The information provided by LC-MS/MS RNA sequencing is rich, but the difficulties surrounding the retention, fragmentation, and sequencing can be quite difficult to the novice. As well, depth of coverage can be an issue for any MS method and with cellular RNA consisting mainly of rRNA (80%) followed by tRNA (15%) and mRNA (5%), RNA sequencing by tandem MS can be challenging without enrichment to sequence a single RNA species⁷¹. Even with enrichment, extremely diverse samples are plagued with isomeric coelution which limits identification RNA sequences. This being said, the information gained contextualizes modifications within the identified RNA and is recently beginning to grow into a well-established field of its own.

While the RNA extraction methods are similar for NGS and MS, the two workflows diverge at the point of sample prep for their respective instruments. Many MS workflows require oligonucleotides of length less than 100 bases due to the difficulty of ionizing these large biomolecules, this limitation is compounded by lengths of many relevant RNA biomolecules, such as microRNAs and long non coding RNA (lncRNAs), ranging from 20 bases to kilobases respectively^72,73. Cleaving the RNA reproducibly into suitable sizes for MS sequencing is the first step. Most commonly used for enzymatic cleavage of RNA is the endonuclease RNAse T1, which targets specifically guanines, followed by RNase A, which cleaves primarily at purines^74,75. Due to the lack of complexity in some RNA sequences, the Limbach group designed new endonucleases to nonspecifically cleave long RNA transcripts to increase the heterogeneity and could thereby increase the confidence in localizing the sequenced RNA to a specific transcript⁷⁶. The choice of which nuclease is highly dependent on the region of interest: over-digest and the region of interest is lost, under-digest and the oligonucleotides do not enter the gas phase easily or fragment too poorly to sequence. For example, if the region of interest is G rich, it would be advised to not use RNase T1 as the RNA will be over digested. After the proper size has been achieved, the next step is ionizing the analyte of interest.

A common means of ionizing RNA oligonucleotides is matrix-assisted laser desorption ionization (MALDI). MALDI is performed by laser irradiation of a dry sample on a plate and is high-throughput, however the major drawback is the required simplicity of each sample⁷⁷. Ionization by MALDI provides easier gas-phase entry for oligonucleotides by inducing very soft ionization in a stable matrix^78,79. MALDI sources are commonly paired with Time of Flight (TOF) analyzers, which provide the high-resolution mass accuracy necessary for distinguishing oligonucleotides, but cannot identify oligonucleotides without prior knowledge of the sample⁸⁰. Ribooligonucleotide identification by MALDI-TOF has decreased in use due to challenges in simultaneous analysis of multiple oligonucleotides and lack of fragmentation; however, ESI LC-MS/MS or direct infusion is still routinely performed today^75,81,82. ESI is commonly used for RNA sequencing due to its compatibility with chromatographic separation, despite the low sensitivity of negative mode MS and the relative fragility of phosphodiester bonds^83-86. Chromatography poses an even greater challenge as oligoribonucleotides do not bind to C18 columns without the use of ion-pairing reagents⁸⁷. Ion-pairing reagents act as a bridge between the negative charges of the phosphate backbone and the hydrophobic alkanes of the C18 resin ^88,89. Ion-pairing reagents disassociate from the analyte during ionization, but can lead to ion suppression, ion burn, and other difficulties with MS as they begin to accumulate^57,90. In the recently-published paper by the Limbach lab, the use of HILIC without the use of ion-pairing was reported to retain and release synthetic oligonucleotides⁸⁷. Since RNA is sensitive to both high and low pH, it is crucial to not use extreme (less than 2 or greater than 9 pH) mobile phase buffers for the binding of RNA without on column hydrolysis or fragmentation⁹¹. Due to these complications it can be advised for larger sequences to simply remove LC completely and move to direct infusion^70,92,93. This removes many of the previously mentioned complications, but with the caveat of limiting the complexity without some means of separation. One means of circumventing the separation problem is the use of ion mobility to separate analytes by size, charge, and base content^70,94,95. With the RNA of interest properly retained, ionized, and contained within the gas phase, the next step is to then identify the analyte by fragmentation.

Fragmentation in any MS/MS workflow is the standard means of fingerprinting and identifying the analyte of interest, with RNA MS workflows as no exception. Fragmentation of the ribooligonucleotide, as seen in Fig. 5, can consist of cleavage along the phosphate backbone (a/w, b/x, c/y, d/z ions), fragmentation at the glyosidic bond (base/a-base or m-base), and the loss of a single phosphate, all of which can be useful for analysis^96,97. Common ribonucleotide fragmentation techniques consist of previously described “beam-type” fragmentation such as HCD and trapped fragmentation techniques such as CID. Fragmentation using HCD/CID creates primarily w/m-B, c/y and w/a ions (Fig. 6), assuming a properly tuned fragmentation energy⁹⁸. Even normalized HCD fragmentation energies can be difficult to properly tune for highly charged oligonucleotides and Fig. 6 demonstrates the effect of normalized fragmentation on increasing length of oligonucleotides and charge, illustrating the need to control charge along with length for fragmentation. Similarly to peptides, the use of electron transfer to fragment negatively charged RNA described as electron-detachment dissociation (EDD) and the light-based fragmentation such ultraviolet photodissociation (UVPD) and infrared multiphoton dissociation (IRMPD) for highly charged, longer oligonucleotides^96,99,100. These two less prevalent fragmentation types have shown parity in fragmentation sites for RNA oligonucleotides, creating the possibility of specialized fragmentation; EDD fragments along the phosphate backbone creating w/d and a/z ions and in the order of bases by G>U/T>A>C, while EPD fragments in the order of G>A>C>U/T^101,102. Radical-based fragmentation techniques for positively charged species such as ECD and ETD have been used primarily for DNA and been adapted for RNA, but are of course limited due to the acidic nature of the RNA where neutral losses would be abundant^99,103-105. Beyond these are the combination of the aforementioned techniques as hybrid activation techniques such as ET-IRMPD and ET-UVPD which retain the modification and generally yield better sequence coverage with a greater distribution of fragments for ease of site localization¹⁰⁶. These diverse fragmentations are distinguished from HCD/CID by providing a wide range of fragments, including w/m-b, c/y, d/z ions, allowing for greater sequence coverage and more confident modification site localization. The fragmentation types CID/HCD are best for smaller-ranging oligonucleotides between 10-20 bases which overlaps with radical or photon-based fragmentation for the same length, but the major difference being the ability to fragment higher charge state species efficiently. Oligonucleotide charge control is paramount for proper fragmentation, with charge states of −4 to −9 being the most common charge states for 20mers. Looking forward beyond current techniques of fragmentation we believe the advent of new technologies to allow for more complex and better fragmentation of longer RNA.

Figure 5:

An example RNA oligonucleotide with possible fragmentation sites in blue and possible modifications (mostly not co-occurrent) in red.

Figure 6.

Complications with oligonucleotide fragmentation by HCD with highly charged species. HCD fragmentation of two human rRNA oligonucleotides at 10% HCD. Sequence coverage by NASE algorithm.

The de novo sequencing of the RNA can be difficult to master due to the complex spectra. As seen in Fig 4 and Fig. 6, the fragmentation points are numerous starting from in and around the phosphate to the loss of the base and pentose. This of course can lead to some troublesome mapping of RNA as many fragments may have isobaric species or convoluted spectra creating difficulty in localizing the modification. For instance, a methyl mass shift may be a 2’-o-methyl mark or a methyl base, illustrating the need for rich spectra and diverse fragments. An example is illustrated in Fig. 7 where the modification is not present on the base (as seem where there is no mass shift) and the presence of the 2’-o-methyl mark can be seen or another example being the uridine isomer pseudo-uracil lacking a labile glyosidic bond which allows for the identification of the base and sugar but not base alone¹⁰⁷. To localize a base methylation however, the method must include a higher energy targeted MS³ to properly identify each isobaric base modification, as referenced previously in this manuscript. Manually assigning spectra is quite difficult without the use of new sequencing software, especially keeping in mind 150 individual modified and isobaric modifications on only 4 bases²⁷. With the rise of newly discovered RNA modifications, it has become paramount to design new sequencing software to localize modifications on RNA transcripts. Since the mid 2000’s, sequencing algorithms for RNA mass spectra have been introduced, but with increasing interest in RNA modifications their population has begun to grow^108,109. More recently, progress is being made by the Limbach and Isobe groups on combining modified RNA sequencing by MS to provide a more comprehensive analysis with new software akin to the proteomic sequencing software^110-113. As the field grows however, the field should be consider a point to rectify these differences in RNA-seq statistical analysis and those commonly performed by proteomics workflows to better improve the overlap of the two fields^114,115.

Figure 7.

A)An example of mapping data RNA oligonucleotides to specific sites using HCD fragmentation. Using 10% nHCD to fragment a small RNA oligomer with a 2’-o-methyl uracil modification using NASE software for support. B) The zoomed in spectra identifying bases and the lack of the methyl uracil base indicating a 2’-o-methyl modification.

With the ease of new computing software, it becomes easier to perform label free quantification (LFQ), relative and absolute quantification of oligoribonucleotides to answer more complex biological questions. Although limited in comparison to proteomics, oligonucleotide LFQ has been performed with the use of direct infusion to localize modification on RNA isomers¹¹⁶. Relative quantification has been used to identify the change in sequence abundance (labeling the nucleobase) or the changes of modifications (labeling the modification) with focused results. Previous use of the Stable Isotope Labeled ribonucleic Acid as an internal Standard (SILNAS) approach, akin to stable-isotope labeling by amino acids in cell culture (SILAC), identified the stoichiometry of post-transcriptional modifications within rRNA by using labeled the uracil and cytidine bases in media -- this approach being counter the use of N¹⁵ or O¹⁸ as labeling techniques, which are less targeted incorporation and quite expensive if used in media^117-121. Absolute quantification of RNA modifications by stable isotope labeled internal standard (SILIS) was performed by the Helm group to identify the changes in bacterial rRNA and has been used with both HPLC and direct infusion for MS identification^122,123. To the future, we envision a simple chemical reaction can be performed to create a labeling technique for multiplexed quantitation of RNA oligonucleotides, analogous to TMT peptide labeling^124-126. Attaching a 5’ or 3’ terminal label may provide information about the enzymatic cleavage, revealing any incidental fragmentation cleavage, and would also enable the ability to multiplex samples. Further use and development of each of these quantitative techniques can provide an ever-expanding map of the epitranscriptome, with the ability to monitor individual transcripts increasing the depth of the field dramatically.

Conclusions and Future

As shown, there is a multitude of mass spectrometry-based methods for the study of RNA modifications. These methods can provide critical data for advancing the field of epitranscriptomics and some of the methods have great promise, but has yet to rise to its full potential within the greater RNA modification field. There is ample room for improvement of technology, some of which can be accomplished through effort and expertise rather than creative innovation. For example, search algorithms that can map oligonucleotides and their modifications to known sequences are likely to be produced within five years, and iterative improvements on chromatography are published annually. These stepwise efforts in combination with creative innovation has the capability of leading to an optimized, high throughput method for shotgun RNA sequencing akin to bottom-up proteomics or the use of labeled quantitation such as TMT or SILAC. With the newfound excitement in the field of RNA modifications, the potential for realizing this goal is higher than ever.