Methods 156 (2019) 91–101

Contents lists available at ScienceDirect

Methods
journal homepage: [Link]/locate/ymeth

Surpassing limits of static RNA modification analysis with dynamic NAIL-MS T

Valentin F. Reichle, Steffen Kaiser, Matthias Heiss, Felix Hagelskamp, Kayla Borland,
⁎
Stefanie Kellner
LMU Munich, Faculty of Chemistry and Pharmacy, Department of Organic Chemistry, Butenandtstr. 5, 81377 Munich, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: Ribonucleic acids (RNA) are extensively modified. These modifications are quantified by mass spectrometry (LC-
Epitranscriptome MS/MS) to determine the abundance of a modification under certain conditions or in various genetic back-
tRNA modification grounds. With LC-MS/MS the steady state of modifications is determined, and thus we only have a static view of
LC-MS/MS the dynamics of RNA modifications. With nucleic acid isotope labeling coupled mass spectrometry (NAIL-MS) we
Nucleic acid isotope labeling coupled mass
overcome this limitation and get access to the dynamics of RNA modifications. We describe labeling techniques
spectrometry (NAIL-MS)
tRNA damage repair
for E. coli, S. cerevisiae and human cell culture and the current instrumental limitations. We present the power of
Demethylation NAIL-MS but we also outline validation experiments, which are necessary for correct data interpretation.
As an example, we apply NAIL-MS to study the demethylation of adenine and cytidine, which are methylated
by the damaging agent methyl-methanesulfonate in E. coli. With NAIL-MS we exclude the concurrent processes
for removal of RNA methylation, namely RNA degradation, turnover and dilution. We use our tool to study the
speed and efficiency of 1-methyladenosine and 3-methylcytidine demethylation.
We further outline current limitations of NAIL-MS but also potential future uses for e.g. relative quantification
of tRNA isoacceptor abundances.

1. Introduction Studies of modified RNA rely on sensitive detection of modified

nucleosides, which is commonly done with triple quadrupole mass
1.1. State-of-the-art quantification of modified nucleosides spectrometry (described in various reviews [2–4]). The principle and
workflow of such analyses is shown in Fig. 1a. After purification of the
Ribonucleic acids (RNA) are key players in the central dogma of RNA of interest (here tRNA), the RNA is enzymatically digested into
molecular biology. Messenger RNA (mRNA), ribosomal RNA (rRNA) nucleosides and ideally an internal standard is added. The sample is
and transfer RNA (tRNA) participate in protein synthesis, while the injected on the LC-MS/MS system where the nucleosides are separated
group of non-coding RNAs (ncRNAs) are crucial for many processes, chromatographically and their mass transitions are monitored in the
including gene regulation as interfering RNAs (miRNAs) and guide mass spectrometer. Using calibration curves, the absolute quantity of
RNAs. As RNAs fulfill many important and diverse functions, more than each nucleoside can be calculated and the number of modification per
4 building blocks are needed. Therefore, a large chemical diversity of tRNA can be plotted.
ribonucleoside modifications can be found [1]. The chemical altera-
tions of the canonical RNA nucleosides are comprised of simple me- 1.2. Nucleic acid isotope labeling (NAIL)
thylations, isomerization, thiolation or even addition of complex groups
like amino acids. Modifications that occur on the base are usually in- Key to the quantification of nucleosides by mass spectrometry is the
dicated by a small letter (e.g. m- for methylation or s- for thiolation) availability of the synthetic nucleoside in weighable quantities (or a
before calling the base (C, U, G or A). The position on the base is in- known extinction coefficient) and an internal standard. The internal
serted as superscript between the short abbreviation and the base; e.g. standard is ideally an isotopomer of the nucleoside of interest. While
5-methylcytidine is abbreviated as m5C. Some modifications, like the the natural nucleoside contains regular isotopes like hydrogen (H)
hypermodification queuosine, are abbreviated with their own capital carbon-12 (12C), nitrogen-14 (14N) its isotopomer will contain one or
letter, e.g. Q. Specific enzymes insert RNA modifications post-tran- more heavy isotopes like deuterium (D), carbon-13 (13C) or nitrogen-15
scriptionally. (15N). The isotopomer is therefore heavier than the natural nucleoside

⁎
Corresponding author.
E-mail address: [Link]@[Link] (S. Kellner).

[Link]
Received 1 August 2018; Received in revised form 25 October 2018; Accepted 31 October 2018
Available online 03 November 2018
1046-2023/ © 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
([Link]
V.F. Reichle et al. Methods 156 (2019) 91–101

Fig. 1. a General workflow for quantification of modified nucleosides. The RNA is isolated and digested to the nucleoside building blocks (here, 5-methylcytidine).
The stable isotope label internal standard (SILIS) is added and the sample is subjected to LC-MS/MS analysis. After chromatographic separation, the nucleosides are
detected in the mass spectrometer and the abundance can be calculated and plotted. b Current and potential uses of NAIL-MS (nucleic acid isotope labeling coupled
mass spectrometry).

and it can be used as an internal standard in mass spectrometry. Such functional analyses by, e.g. multiplexing. Multiplexing allows the direct
internal standards are referred to as stable isotope labeled internal comparison of samples by mixing them within a single tube and per-
standards (SILIS) and the technique as isotope dilution mass spectro- forming a single analysis. They can be distinguished by mass spectro-
metry. The generation of these isotopomers can be done either syn- metry either by a chemical labeling step prior to mixing or by metabolic
thetically [5–7] or biosynthetically by metabolic labeling in bacteria, isotope labeling during sample generation. Especially metabolic isotope
yeast or algae [8–10]. labeling has the advantage of overcoming purification biases. Ad-
Metabolic labeling relies on the incorporation of heavy isotopes ditionally, mass spectrometric detection fluctuations are of no con-
from heavy-labeled growth media into the RNA. The heavy labeled RNA sequence. Metabolic isotope labeled proteomics (SILAC-Proteomics
is isolated and processed for preparation of a SILIS for quantification. [17]) allows a direct comparison of proteomes within a single mea-
The usage of SILIS reduces the detection fluctuations of the mass surement with high accuracy. In principle, SILAC-like multiplexing is
spectrometer and thus reliable quantitative data becomes available to possible by NAIL, and we see a significant potential in such a SILAC-like
study RNA modifications. approach in RNA modification research. We refer to this technique as
However, nucleic acid isotope labeling (NAIL) should not be limited comparative NAIL, which is also a useful tool for validation of dy-
to quantification purposes. One can envision the use in a variety of namic NAIL-MS experiments.
experiments, which are or will become valuable tools in the research of
modified nucleosides. Fig. 1b reviews the current usage of NAIL and 1.2.2. Dynamic NAIL
potential future uses in RNA modification analysis. In addition to the While epigenetics is an intensively studied area, the analogue pro-
usage of NAIL in quantification, its use to facilitate the discovery of cess in RNA, termed epitranscriptomics, is far less studied. This is
novel modified nucleosides is widespread. Here, isotope labeling has mainly caused by our limited number of tools to study the dynamics of
become a key tool for sum formula generation and is highly helpful for RNA modifications and, in addition, the complex process of finding
structure prediction and verification [11–14]. biological consequences to RNA modifications. DNA is the storage of
the genetic code. Therefore, modifications of DNA must be removed by
1.2.1. Comparative NAIL an enzymatic process, which leaves the DNA sequence untouched and
One of the caveats in the area of RNA modification research is the the DNA intact. Otherwise, mutations would occur and harm the or-
lack of key technologies comparable to those boosting protein research ganism. While it should be possible for the cell to use similar removal
[15,16]. In the field of proteomics, a variety of mass spectrometry tools mechanisms in RNA, RNA has a second option for removal of an un-
have emerged and allowed scientists to study proteins and their net- wanted modification – the whole RNA itself is degraded and an un-
works in more detail. Many of these tools focus on the comparison of modified new RNA is transcribed. The potential competition between
proteomes (all proteins within a cell) in the context of stress studies or these two processes makes it difficult to study the dynamics of RNA

92
V.F. Reichle et al. Methods 156 (2019) 91–101

modifications. Although many studies described that mRNA is en- With LC-MS/MS the steady state of modifications is determined and
zymatically demodified, the mechanisms are still reviewed skeptically thus is limited to a static view on dynamic RNA modification processes.
[18]. These doubts arise from the used techniques for in vivo analysis of Here, we present nucleic acid isotope labeling coupled mass spectro-
the demodification process. In these studies, quantitative mass spec- metry (NAIL-MS) which overcomes these current limitations and allows
trometry of RNA modifications was used to observe an active demodi- dynamic analysis of RNA modifications. We describe labeling techni-
fication in vivo [19]. However, the absolute number of a modification ques for E. coli, S. cerevisiae and human cell culture and the current
within an RNA does not reflect the origin of the modification. A de- instrumental limitations and recommended validation experiments.
crease in modification density can be explained by enzymatic demo- We apply NAIL-MS to study the repair of adenine and cytidine,
dification processes but also by increased degradation of modified RNA which are methylated by the damaging agent MMS in E. coli. With
or even by increased transcription of unmodified RNA. Vice versa, an NAIL-MS, we exclude the concurrent processes for removal of RNA
increase in modification density can be explained by additional mod- methylation, namely RNA degradation, turnover and dilution. The
ification events in the original RNA or by increased degradation of non- power of NAIL-MS is demonstrated by showing the kinetics of de-
modified RNAs. methylation of damage-derived 1-methyladenosine and 3-methylcyti-
Current analyses of nucleic acid modifications are limited in respect dine in E. coli in vivo. We further outline current limitations of NAIL-MS
to their blindness towards the underlying mechanisms, which lead to but also potential future uses for e.g. relative quantification of RNA
changes of modification content. NAIL-MS expands our current analy- abundances.
tical toolbox and overcomes limitations by providing insight into the
dynamics of nucleic acid modifications. 2. Material and methods

1.3. Nucleic acid isotope labeling as a tool to observe nucleic acid 2.1. Salts, reagents, isotopes and nucleosides
modification dynamics
All salts were obtained from Sigma Aldrich (Munich, Germany) at
From the natural RNA modifications known to date (∼160), 70 molecular biology grade unless stated otherwise. Isotopically labeled
contain a methylation of the base or the ribose [1]. These methylations compounds: 15N-NH4Cl (≥98% atom) and L-methionine-methyl-D3
are enzymatically incorporated at defined positions of the RNA. Espe- (98% atom) from Sigma-Aldrich. 13C6-glucose (≥99% atom) and
cially m6A in mRNA has gained major interest since this is the first RNA Na234SO4 (≥99.1% atom) from Eurisotop (Saarbruecken, Germany).
modification, which is incorporated by a methyltransferase (writer) and 1,3-15N2-uracil (98% atom) from Cambridge Isotope Laboratories
can be removed by demethylation via FTO or AlkBH5 (erasers). (Tewksbury, MA, USA). All solutions and buffers were made with water
In yeast, it was found that tRNA modifications are highly dynamic from a Millipore device (Milli-Q, Merck). Nucleosides: adenosine (A),
during cellular stress and are used by the cell for efficient stress survival cytidine (C), guanosine (G), uridine (U) and N2-methylguanosine (m2G)
by changing the translational speed of stress response proteins [20]. from Sigma Aldrich. 1-Methyladenosine (m1A), 2-methyladenosine
Although the dynamic nature of tRNA modifications is crucial to cell (m2A), N3-methylcytidine (m3C), N6-methyladenosine (m6A), 7-me-
survival, it remains unclear how the cell achieves the adaptation of the thylguanosine (m7G), 5-methylcytidine (m5C), 5-methyluridine (m5U),
modification profile. Additionally, cells are facing various stressors that 2′-O-methylcytidine (Cm), 2′-O-methylguanosine (Gm), 1-methylgua-
can actively methylate the RNA such as methyl-methanesulfonate nosine (m1G) and 3-methyluridine (m3U) from Carbosynth (Newbury,
(MMS). Such methylations are randomly distributed across the acces- UK).
sible sites in RNA. In vitro tests showed that methylation of positions 1
of adenine and 3 of cytosine are quickly repaired by AlkB in E. coli [21]. 2.2. Specific laboratory equipment
Only one study showed the repair of m1A by AlkB in vivo. This early
pulse-chase study, based on radioactively labeled adenine, showed re- Injection vial for HPLC and LC-MS: 0.3 mL PP Snap Ring Micro Vial,
moval of m1A from MMS damaged RNA by AlkB in vivo. Other studies 32 × 11.6 mm, transparent, VWR (Radnor, PA, USA), Cat. No. 548-
that report demethylation processes in RNA did not present clear in vivo 0120.
evidence of demethylation due to the limitations of static LC-MS/MS Injection vial cap: 11 mm Snap Ring Cap, tr., natural rubber/TEF,
quantification. 60°, 1.0 mm, VWR (Radnor, PA, USA), Cat. No. 548-0014.
Our first steps to overcome this limitation was in 2017, when we Fraction Collector glass vial: 1.5 mL Screw vial, 32 × 11.6 mm
utilized non-radioactive isotope labeling of DNA to observe damage and clear, VWR (Radnor, PA, USA), Cat. No. VWRI548-0018.
repair of a DNA modification in bacteria. The combination of different Glass vial cap: 8 mm PP-Screw cap black hole, VWR (Radnor, PA,
labeling media allowed the creation of a pulse-chase experiment, which USA), Cat. No. VWRI548-3322.
was used to observe repair of phosphorothioates in bacterial DNA [22]. Culture tube: Centrifuge tube 50, TPP (Trasadingen, Switzerland),
The principle is shown in Fig. 2a. The bacteria were pre-cultured in Product No.91050
heavy labeled media resulting in complete heavy labeling of the DNA
nucleosides. In the heavy media, the bacteria were exposed to hypo- 2.3. Experimental settings
chlorous acid and the DNA present at this moment was damaged. After
the exposure the bacteria were placed into unlabeled media and re- 2.3.1. Metabolic isotope labeling
plication produced unlabeled DNA. By this approach it was possible to [Link]. Bacteria. For each experiment a single colony from an LB agar
distinguish DNA present during the exposure from DNA synthesized plate with E. coli BW25113 was picked and used for culture inoculation.
after the exposure by mass spectrometry. Thus, we could see the loss of We used minimal media M9 with and without the indicated isotopes
original phosphorothioates upon reaction with the chemical in vivo and for all bacterial cultures.
the subsequent repair by the phosphorothiolating dnd enzymes. In the Unlabeled 10× M9 stock solution: mix 68 g/L Na2HPO4, 30 g/L
same year, we adapted the approach to RNA modification analysis in KH2PO4, 2.5 g/L NaCl and 10 g/L NH4Cl and autoclave. Store at room
yeast [9]. As shown in Fig. 2b, we followed the modification density of temperature.
tRNA in dependence of the growth phase and we identified the un- Nitrogen-15-labeled 10x M9 stock solution: mix 68 g/L Na2HPO4,
derlying mechanisms for several modified nucleosides (here 7-methyl- 30 g/L KH2PO4, 2.5 g/L NaCl and (!) 10 g/L 15N-NH4Cl (!) and auto-
guanosine, m7G, is shown). With NAIL-MS we can assess RNA turnover, clave. Store at room temperature.
RNA biosynthesis and dilution effects and determine the modification Other stock solutions: MgCl2 (0.1 M), CaCl2 (0.1 M), Na2SO4 (0.1 M)
content of RNAs in response to e.g. growth or stress. and 20 w/w% glucose (or 13C6-glucose) are prepared by sterile

93
V.F. Reichle et al. Methods 156 (2019) 91–101

Fig. 2. a Principle of a NAIL-MS assay, which allows the observation of DNA modification damage repair in vivo. The bacteria are cultured in heavy isotope
containing media, and therefore the DNA is heavy labeled, as observed by mass spectrometry. After exposure to the damage, the media is exchanged to a light media
and newly replicated DNA is light labeled. The repair of original, but damaged DNA is observed by the formation of an intermediately labeled species in the mass
spectrometer. (Adapted from [22].) b Principle of a dynamic NAIL-MS assay of tRNA in S. cerevisiae and the dynamics of 7-methylguanosine (m7G) in tRNAs after
experiment initiation. The drop in pre-existing m7G is masked by an increase of post-methylated m7G added to the pre-existing tRNAs (from [9]).

filtration. Depending on the desired labeling either 13C6-glucose and L-

For an unlabeled 5 mL pre-culture (or 50 mL exposure culture, re- methionine-methyl-D3 or their unlabeled isotopomers were used. The
spectively) mix 500 µL (5 mL) 10x M9 stock solution with 100 µL (1 mL) cells were incubated at 10% CO2 atmosphere and cultivated in the
glucose, 100 µL (1 mL) MgCl2, 100 µL (1 mL) Na2SO4 and 5 µL (50 µL) labeled media for at least 2 days (5 days for complete labeling). For
CaCl2. splitting, the cells were treated with TrypLE Express (Gibco, Carlsbad,
For 13C-labeled cultures, 13C6-glucose was used. CA, USA).
For 15N-labeled cultures, the 15N-10x M9 stock solution was used.
For CD3-labeling, 200 µL of L-methionine-methyl-D3 (stock 5 g/L)
2.3.2. RNA isolation and tRNA purification
were added to 5 mL of culture volume.
Bacteria cultures were centrifuged at 1200×g for 5 min, yeast cul-
tures at 3500×g for 5 min. The supernatant was discarded and the cell
[Link]. Yeast. A single colony of a S. cerevisiae BY4741 YPD agar plate pellet was suspended in 1 mL TRI-Reagent® (Sigma-Aldrich) per 5 mL
was picked and used for inoculation of 5 mL YNB media. 10x YNB (Carl culture, respectively. Yeast cells were additionally vortexed for 5 min
Roth GmbH, Karlsruhe, Germany) was prepared according to using acid-washed Glass beads (Sigma-Aldrich) equivalent to ∼200 μL.
manufacturer’s manual. 1x YNB media was supplemented with the HEK cells were harvested directly in cell culture dishes/flasks using
following metabolites to a final concentration of: 10 g/L glucose, 1 mL TRI-Reagent® per 25 cm2. After transfer into an Eppendorf tube
0.02 g/L uracil, 0.02 g/L methionine, 0.02 g/L arginine, 0.1 g/L and incubation at room temperature for 5 min, 200 µL of chloroform
aspartic acid, 0.1 g/L glutamine, 0.02 g/L histidine, 0.06 g/L leucine, (≥99% purity, Roth) were added to 1 mL of the TRI-Reagent® solution
0.03 g/L lysine, 0.05 g/L phenylalanine, 0.4 g/L serine, 0.2 g/L of each organism and the mixture was vortexed for at least 10 s. The
threonine, 0.04 g/L tryptophan, 0.03 g/L tyrosine and 0.15 g/L valine biphasic solution was allowed to settle at room temperature for 5 min
(only L-amino acids were used). Depending on the desired labeling, and centrifuged at room temperature for 20 min at 10,000×g. The clear
13
C6-glucose and L-methionine-methyl-D3 were used instead of the upper phase (∼500 µL) was transferred into a new vial, 500 µL of iso-
unlabeled compounds. propanol (Roth, Karlsruhe, Germany) were added and the solution was
mixed thoroughly. The mixture was stored at -20 °C overnight. The
[Link]. Mammalian cells. HEK 293 T cells were cultured in Dulbeccós precipitated total RNA was pelleted by centrifugation for 40 min at
Modified Eagle Media (DMEM) or RPMI 1640 media (Gibco, Carlsbad, 12,000×g and 4 °C. The RNA pellet was washed two times with
CA, USA). DMEM media was prepared by dissolving 8.4 g DMEM 100–200 µL 70% EtOH and finally dissolved in 30 µL water.
powder D5030 (Sigma Aldrich) in 1 L milli-Q water. Before sterile For purification of tRNA from total RNA size exclusion chromato-
filtration, carbonate and phenol red were added to a final concentration graphy (SEC) [23] was used on an Agilent 1100 HPLC system (Degasser,
of 3.7 g/L NaHCO3 and 0.0159 g/L phenol red. Stocks of glucose G1279A; Quat Pump, G1311A; ALS, G1313A; COLCOM, G1316A; VWD,
(200 g/L) and L-glutamine (15 g/L) were prepared and sterile filtered. G1314A; Analyt FC, G1364C) with an AdvanceBio column, 300 Å pore
These were added to the DMEM media before usage to a final size, 2.7 µm particle size, 7.8 × 300 mm (Agilent, Waldbronn, Ger-
concentration of 2 g/L glucose, 0.584 g/L L-glutamine and 10% many). For elution, a 1 mL/min isocratic flow of 0.1 M ammonium
dialyzed fetal calf serum (Sigma Aldrich, Product No. F0392-500ML). acetate was used. Eluting RNA was detected at 254 nm with a diode
The methionine concentration was 0.15 g/L in the final media. array detector. Under these conditions, tRNA elutes at a retention time

94
V.F. Reichle et al. Methods 156 (2019) 91–101

between 7 and 8 min. The 1 mL tRNA fraction was collected and eva- MS grade, purity ≥99.95). A Kinetex EVO column (Phenomenex®,
porated (GeneVac, EZ-2 PLUS, Ipswich, UK) to a volume of ∼100 µL Torrance, California, USA; Kinetex® 1.7 µm EVO C18 100 Å,
before precipitation by addition of 0.1 vol of 5 M ammonium acetate 150 × 2.1 mm) at a temperature of 35 °C with an eluent flow rate of
and 2.5 vol of ice-cold ethanol (100%). (!) Ammonium acetate is the 0.35 mL/min was used. The gradient started at 100% solvent A, fol-
precipitation method of choice if mass spectrometric analysis is desired. lowed by an increase to 10% solvent B over 10 min. From 10 to 15 min
Sodium ions interfere with mass spectrometric detection of nucleosides solvent B was increased to 45% and maintained for 3 min before re-
and should be avoided if possible (!) After rigorous mixing, the tRNA turning to 100% solvent A and a 3 min re-equilibration period. Alter-
was allowed to precipitate at −20 °C overnight. The tRNA was pelleted natively, a Synergi Fusion-RP column (Phenomenex®, Torrance, Cali-
by centrifugation (12,000g, 40 min, 4 °C), washed with 70% ethanol fornia, USA; Synergi® 2.5 µm Fusion-RP 100 Å, 150 × 2.0 mm) at 35 °C
and resuspended in 30 µL water. and a flow rate of 0.35 mL/min was used for yeast and mammalian cell
analysis. Gradient elution started with 100% A for 1 min, increased to
2.3.3. Pulse-Chase NAIL-MS 10% B after 5 min, and to 40% after 7 min. The column was flushed
A single E. coli colony was picked and grown in 5 mL unlabeled M9 with 40% B for 1 min. After regeneration of starting condition for
media overnight (37 °C, shaking at 250 rpm). From this pre-culture, a 0.5 min the column was re-equilibrated at 100% A for 3 additional
second overnight culture was prepared by inoculating 50 mL unlabeled minutes.
media with the complete 5 mL pre-culture. On the next day, these
bacteria were added to 120 mL of unlabeled M9 media to a final OD of 2.4.2. High-resolution mass spectrometry
1.0. After 60 min of growth at 37 °C and 250 rpm, the first 7 mL aliquot The ribonucleosides were separated using a Dionex Ultimate 3000
was taken for RNA isolation. The remaining culture was equally split in HPLC system on an Interchim Uptisphere120-3HDO C18 or an RP-18
2 Erlenmeyer flasks (100 mL glass size) of 56.5 mL each. One was ex- column (Synergi, 2.5 µm Fusion-RP C18 100 Å, 100 × 2 mm;
posed to 95.7 µL MMS (99% purity), the other to the same amount of Phenomenex®, Torrance, California, USA). Mobile phase A was 2 mM
water (MOCK) and gently stirred before both cultures were left to grow ammonium acetate and mobile phase B was 80% acetonitrile con-
for 60 min at 37 °C with shaking at 250 rpm. After 60 min exposure, an taining 2 mM ammonium acetate. Gradient elution started with 0% B
aliquot of 7 mL was drawn from each culture and the RNA was isolated. and increased to 12% B after 10 min and to 80% after 12 min. After
The remaining bacteria were centrifuged (1200×g, 5 min) and the 4 min elution at 80% B and subsequently regeneration of starting
MMS/MOCK containing supernatants were discarded. The bacteria conditions to 100% A after 5 min, the column was equilibrated at 100%
pellets were washed with 5 mL 15N/CD3-methionine labeled M9 media A for 8 min. The flow rate was 0.2 mL/min and the column temperature
and each suspended in 50 mL fresh 15N/CD3-methionine M9 media. For 30 °C. High-resolution mass spectra of precursor and product ions were
recovery, the bacteria were grown at 37 °C, 250 rpm and after 1, 2, 3, 4 recorded by a ThermoFinnigan LTQ Orbitrap XL. The parameters of the
and 10 h 7 mL aliquots were drawn for RNA isolation. mass spectrometer were tuned with a freshly mixed solution of ade-
nosine (5 μM). The parameters were sheath gas flow rate, 5 arb; aux-
2.3.4. tRNA digestion for mass spectrometry iliary gas flow rate, 35 arb; sweep gas flow rate, 0 arb; spray voltage,
100 ng tRNA in 30 µL aqueous digestion mix were digested to single 5.0 kV; capillary temperature, 200 °C; capillary voltage, 20 V, tube lens
nucleosides by using Alkaline Phosphatase (0.2 U, Sigma-Aldrich, 65 V.
Munich, Germany), Phosphodiesterase I (0.02 U, VWR, Radnor,
Pennsylvania, USA) and Benzonase (0.2 U Sigma-Aldrich, Munich, 2.5. Calibration and equations
Germany) in Tris (5 mM, pH 8.0) and MgCl2 (1 mM) containing buffer.
Furthermore, tetrahydrouridine (THU, 0.5 µg from Merck), butylated For calibration, synthetic nucleosides were weighed and dissolved
hydroxytoluene (BHT, 1 µM, Sigma-Aldrich, Munich, Germany) and to a stock concentration of 1–10 mM. Calibration solutions ranging
Pentostatin (0.1 µg, Sigma-Aldrich, Munich, Germany) were added to from 0.15 pmol to 500 pmol for each canonical nucleoside and from
avoid deamination and oxidation of nucleosides [2]. All mentioned 0.15 fmol to 500 fmol for each modified nucleoside were prepared in
concentrations/amounts are final concentrations/amounts used in a water. The calibration solutions were mixed with the E. coli SILIS and
30 µL final digestion volume. The mixture was incubated with the RNA analyzed with the appropriate method. The value of each integrated
for 2 h at 37 °C and filtered through 96 well filterplates (AcroPrep peak area of the nucleoside was divided through the respective SILIS
™Advance 350 10 K Omega™, PALL Corporation, New York, USA) at area. The linear regression for each nucleoside’s normalized signal/
4 °C for 30 min at 3000×g, or through single tubes (VWR, 10 kDa concentration plot gives the relative response factor for nucleosides
MWCO) at room temperature for 7 min at 5000×g. The filtrate was (rRFN) [8]. The sample data were analyzed by the Quantitative and
mixed with E. coli SILIS 10:1 (stable isotope labeled internal standard Qualitative MassHunter Software from Agilent. The areas of the nu-
[8]) and measured with the triple quadrupole mass spectrometer. cleoside signals were integrated for each modification and their isotope
derivatives. The area was divided through the respective SILIS area and
2.4. LC-MS instruments and methods divided through the rRFN value from the respective calibration to re-
ceive the absolute amount of the modification or canonical. Finally, the
2.4.1. Triple quadrupole instrument absolute amounts of the modifications were referenced to the absolute
For quantification an Agilent 1290 Infinity II equipped with a DAD amounts of the precursor canonical. In case of the validation and dy-
combined with an Agilent Technologies G6470A Triple Quad system namic NAIL-MS experiment the different isotopomers were referenced
and electro-spray ionization (ESI-MS, Agilent Jetstream) was used. to their respective labeled canonicals, so that original modifications are
Optimized operating parameters: positive ion mode, skimmer voltage referenced to original canonicals and new modifications were refer-
15 V, Cell Accelerator Voltage 5 V, N2 gas temperature 230 °C and N2 enced to new canonicals. See the following equations for m1A as an
gas flow 6 L/min, sheath gas (N2) temperature 400 °C with a flow of example:
12 L/min, Capillary Voltage of 2500 V, Nozzle Voltage of 0 V and the
Nebulizer at 40 psi. The instrument was operated in dynamic MRM
m1A (fmol) A (fmol) Normalization
mode and the individual mass spectrometric parameters for the nu-
cleosides are given in Supplementary Tables S2 and S3. Mobile phase A Original area m1A (unlabeled) area A (unlabeled) m1A (original)
Here: unla- RFN m1A × area m1A (SILIS ) RFN A × area A (SILIS ) A (original)
was 5 mM NH4OAc (≥99%, HiPerSolv CHROMANORM®, VWR),
beled
brought to pH = 5.3 with glacial acetic acid (≥99%, HiPerSolv CHR-
OMANORM®, VWR). Mobile phase B was pure acetonitrile (Roth, LC-

95
V.F. Reichle et al. Methods 156 (2019) 91–101

New area m1A (labeled) area A (labeled) m1A (new ) efficiency. Fig. 3c shows the guanosine signal from tRNA of a 5 day-
Here: labeled RFN m1A × area m1A (SILIS ) RFN A × area A (SILIS ) A (new )
labeled HEK culture in the presence of 2 g/l13C6-glucose. Here, a mass
shift of +5, +6, +7 and +8 is observed for G. The +5 reflects a
guanosine with a 13C5-labeled ribose but unlabeled base. The +6 to +8
2.6. Other instruments
labeled peaks reflect guanosine isotopomers with 13C5-labeled ribose
and 13CX-labeled base. Such a labeling technique is not suitable for
Orbital Shaker-Incubator ES-20 (BioSan, Riga, Latvia); Rotina 380 R
comparative NAIL-MS studies (but potentially for dynamic NAIL-MS).
centrifuge (Hettich, Tuttlingen, Germany); Centrifuge 5427 R 13
C6-glucose appears not to be the metabolite of choice for successful
(Eppendorf, Hamburg, Germany); Perfect Spin 24 R refrigerated micro
NAIL-MS experiments in cell culture systems. The determination of a
centrifuge (PeQlab/VWR, Erlangen, Germany); Vortexer (Heathrow
metabolite or even a mixture of metabolites, which can be used to
Scientific, IL, USA), Product Code: 120212; Nanophotometer N60
achieve a defined labeling in cell culture, is the major bottleneck in
Touch (Implen, Munich, Germany), T60966; Speedvac EZ-2PLUS
establishing NAIL-MS in cell culture.
(GeneVac, Ipswich, UK).
When choosing a labeling technique for comparative or pulse-chase
NAIL-MS, the abundance of natural isotopes, especially 13C (1.1% rel.
3. Important considerations for NAIL-MS studies abundance) has to be considered. Pyrimidines and purines have 9 and
10 Carbon atoms, respectively. Statistically, ∼10% of all nucleosides
As the term NAIL-MS states, there are two key features for setting up carry one 13C atom (m/z +1), 1% carry two 13C atoms (m/z +2) and
such studies. The first is defined isotope labeling of the nucleic acid 0.1% carry three 13C atoms (m/z +3). We recommend labeling tech-
(NAIL) and the second is the availability of a mass spectrometer (MS). niques, which increase the mass by at least 3 Dalton to avoid the de-
In addition, we recommend rigorous validation of NAIL-MS studies, tection of the natural 13C-isotopomers of the nucleosides.
which will be discussed in detail. The preparation of isotopically labeled media requires the acquisi-
tion of isotopically labeled compounds, which are more expensive than
3.1. Labeling techniques the unlabeled compound. We have summarized the cost of 1 Liter NAIL-
MS media in an overview in Table 1. Per NAIL-MS experiment around
Before starting a NAIL-MS experiment, a labeling technique must be 20–50 mL heavy labeled media is necessary. The least expensive isotope
established, which leads to a defined labeling of the nucleic acid. labeling is achieved in bacteria (76 €/L) and cell culture (290 €/L), but
Microorganisms like E. coli [8] and S. cerevisiae [9] are effortlessly la- here the labeling efficiency is quite low. Yeast minimal media (945 €/L)
beled in minimal media or complete media (e.g. from Silantes, Munich, labeling is affordable, but yeast complete media (2025 €/L) is the most
Germany). In addition, algae like Chlamydomonas reinhardtii or worms expensive and therefore solely recommended for the production of in-
such as Caenorhabditis elegans can be labeled [10,24]. Fig. 3a shows the ternal standards.
mass spectra of digested tRNA from an E. coli culture in minimal media
M9 using 13C6-glucose as the carbon source (left) or using CD3-me- 3.2. Mass spectrometry
thionine (right). By feeding 13C6-glucose overnight, 83% of carbon
atoms in the tRNA are 13C labeled in the exemplary nucleoside gua- For the detection of modified nucleosides, sensitive triple quadru-
nosine (G). Thus a mass increase of +10 is observed for G compared to pole instruments are used. The first quadrupole filters for the nucleo-
unlabeled G (and ∼17% of a +9 species). Most enzymatic methylation sides’ m/z values (e.g. m/z 298 for m7G). In the collision cell (histori-
reactions require S-adenosyl methionine (SAM) as the methyl donor. cally second quadrupole), the nucleoside is fragmented into nucleobase
After the methylation, the resulting S-adenosyl homocysteine is re- and ribose and the charge remains on the nucleobase. The third
charged with a methyl group by methionine. Feeding organisms CD3- quadrupole filters for the nucleobases’ m/z values (e.g. m/z 166 for
methionine leads to the formation of CD3-SAM and transfer of heavy m7G). The detection of product ions from defined precursor ions is
methyl marks onto nucleic acids [25,26]. This is also observed for 7- called a mass transition. For m7G, the mass transition is 298 → 166.
methylguanosine (m7G) from CD3-methionine supplemented bacterial Commonly, 20–30 modified nucleosides are analyzed in a single run.
cultures, which has a mass increase of +3 compared to unlabeled m7G This is achieved by fast switching from one nucleoside’s mass transition
(note: the +4 is the natural 13C isotope peak of m7G). We recently to the next (around every 5–10 ms). If ions from the previous nucleoside
shared this labeling technique, which helped Dal Magro et al. to de- remain in the mass spectrometer, although the instrument selects for
termine the structure of a novel RNA modification in bacteria, namely the next nucleoside, false positive signals can be observed. Especially
msms2i6A [11]. for co-eluting isotopomers with small differences in mass transitions,
For purines in S. cerevisiae, we mainly observe a +6 mass increase carry-over in the collision cell falsifies the detected quantities of the
upon culturing in 13C6-glucose media overnight and a +3 for m7G in compounds. Linear collision cells with hexapoles and octopoles have
CD3-methionine media (Fig. 3b). It is noteworthy that both labeling high carry-over tendencies and should be validated for usability for
techniques do not lead to a single defined isotopomer. Further valida- NAIL-MS methods. The desired method can be tested by injection of
tion is necessary to assess the impact of multiple isotopomer formation single-labeled samples, which should only produce signals for the iso-
(see chapter on validation). In addition to the presented 13C6-glucose topomers from the used label. Signals from other isotopomer mass
and CD3-methionine labeling, it is possible to use 15NH4Cl or 34Na2SO4 transitions are considered artefacts and the method is not usable for
[22] in E. coli or 15N2-uracil in yeast for further labeling options. It is NAIL-MS studies. The carry-over error can be reduced by programming
also possible to combine the carbon, nitrogen, sulfur and methyl only 1–2 nucleosides into the method or by using time-gated selection
sources into a single media. of mass transitions. The new generation of QQQ instruments use faster
While yeast is already less efficiently labeled in minimal media collision cells (curved or T-wave), which have less to no carry-over.
compared to E. coli (compare Fig. 3a and b), labeling in human cell Note: Carry-over is rarely a problem in common nucleoside quantifi-
culture is even more difficult. We have tested HEK 293 T and Hela cell cation as most nucleosides are chromatographically separated and thus
lines in both DMEM and RPMI media using 13C6-glucose and CD3-me- do not disturb each other in the mass spectrometer.
thionine. While labeling of methyl groups with CD3-methionine is si-
milarly successful in HEK cells (Fig. 3c) as it is in yeast cells, 13C6- 3.3. Validation of NAIL-MS experiments
glucose labeling is less promising. We observed for all nucleosides a
variety of formed isotopomers, which were found to be independent of Insufficient labeling and a slow mass spectrometry can result in false
the used serum. Culturing for a longer time did not increase labeling positive results or misinterpretation. Thus, NAIL-MS experiments must

96
V.F. Reichle et al. Methods 156 (2019) 91–101

Fig. 3. High resolution mass spectra from 13C labeled guanosine and CD3-labeled 7-methylguanosine in E. coli (a), S. cerevisiae (b) and HEK 293 T (c). The grey bars in
the spectra point out the expected m/z value of the unlabeled compounds. *Coeluting compound.

Table 1 variant.
Costs of NAIL-MS suitable media per Liter media in €uro. At this step it is crucial that both media are identical except for the
Used isotope CD3 13
C 15
N 34
S
isotope composition. Small differences in media composition already
influence the abundance of modified nucleosides and the validation
Bacteria (E. coli) 17 360 76 510 result. Thus, we recommend parallel preparation of the media using
Yeast minimal 45 945 / / labeled and unlabeled stock solutions (!).
Yeast complete 1125 2025 / /
The two media are inoculated with the same number of cells and
Human cell culture 123 290 / /
grown in parallel under identical conditions. After the appropriate
amount of time (e.g. overnight for E. coli, 24 h for yeast and 5 days for
be carefully validated before they can be used for biological questions. HEK cells), the cells are harvested in e.g. TRI® reagent and mixed in a
The first step in validation should be to determine the labeling effi- single container. From now on all processing steps are performed in
ciency. As previously mentioned, it is crucial that the labeling leads to parallel to avoid purification biases. From the isolated total RNA, the
only one labeled isotopomer (e.g. 95% m/z of +5) and not several (e.g. RNA of interest (e.g. tRNA) is isolated and prepared for LC-MS/MS by
37% m/z +5, 31% m/z +6 and 18% m/z +7 as seen for guanosine in enzymatic digestion. The sample contains labeled and unlabeled nu-
Fig. 3c). We recommend testing each labeling strategy by scanning the cleosides, and the amount of each modified nucleoside can be de-
nucleoside mass range (e.g. m/z 240–400) and validate the abundance termined, normalized and plotted. Fig. 4b shows a successful validation
of all nucleoside isotopomers. If more than one isotopomer is formed of an E. coli 13C6-glucose labeling procedure. Here, the quantities of
under one labeling condition, we recommend optimization of the la- most modified nucleosides from total tRNA are identical in labeled and
beling technique. Sometimes low purity of the isotopically labeled unlabeled media.
metabolites/salts can also result in labeling inefficiency. In these cases, The abundance of D (dihydrouridine) is in general higher in the 12C
materials from other suppliers with higher isotope purity should be samples compared to the 13C samples. This is explained by con-
tested. taminating 12C-dihydrouridine from the used deaminase inhibitor tet-
For validation of multiplexing and dynamic NAIL-MS we further rahydrouridine. Other uridine and cytidine derivatives showed a com-
recommend a control multiplexing experiment as shown in Fig. 4a. The parable modification profile in the 12C and 13C tRNA. A more than 1.1
media of choice is prepared in the unlabeled and isotopically labeled fold difference in modification density is observed for m2A. The

97
V.F. Reichle et al. Methods 156 (2019) 91–101

Fig. 4. a The principle of comparative NAIL-MS for validation of labeling. b and c E. coli comparative NAIL validation from a 13C labeled culture mixed with an
unlabeled culture (b) and a 15N-labeled culture mixed with an unlabeled culture (c). y-axis labeling: Mod. per 1000 nts (Modification per 1000 nucleotides). Results
from labeled RNA are marked in red, from unlabeled in black. (data represents 3 biol. replicates and error bars reflect the standard deviation.) (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

observed fold difference between labeled and unlabeled tRNA is defined 4. Results
as the limit of precision in comparative NAIL-MS experiments and
should be given in a table for any NAIL-MS study (Supplementary Table 4.1. Observing the repair of methylation damage in E. coli tRNA by pulse-
S1). The 13C6-glucose labeling can be used for comparative NAIL studies chase NAIL-MS
of organisms with varying genetic background or under various growth
conditions. It is also very useful if samples from complex purification We recently applied NAIL-MS to discriminate the origin of methy-
procedures are compared as co-purification eliminates potential pur- lated RNA nucleosides in E. coli treated with the methylating agent
ification biases. It is also suitable for pulse-chase NAIL-MS studies, MMS. We observed direct methylation of all canonical nucleosides and
which focus on both the original RNA modification content and the new 7-methylguanosine, 1-methyladenosine, 6-methyladenosine, 3-methyl-
transcript modification content. cytidine and 3-methyluridine as the main damage products. In a dy-
Fig. 4c shows the validation of a 15N labeling technique in E. coli, namic NAIL-MS assay we followed the fate of these damage products
which is less successful. Here, we observe lower modification content in and we observed demethylation of 1-methyladenosin and 3-methylcy-
the 15N labeled bacterial tRNA compared to the unlabeled tRNA. The tidine after 24 h [27]. Here, we want to present the assay in more detail
limit of precision of this validation experiment is larger as shown in and we performed a NAIL-MS time-course experiment, which allows the
Supplementary Table S1. 15N labeling is more error prone compared to determination of the demethylation kinetics. An assay to observe the
13
C labeling in E. coli due to the media preparation. M9 media requires repair of methylated RNA nucleosides is possible by NAIL-MS under the
the preparation of a M9 salt stock, which also contains the 15NH4Cl salt following conditions: A) A “backbone” labeling technique is needed,
for 15N labeling. This M9 stock mix has a limited shelf life, and upon which allows discrimination of the damaged RNA from newly tran-
aging of the M9 stock mix the modification content in bacteria changes. scribed RNA. B) Availability of a labeling technique, which additionally
Thus, 15N labeling can be only used for comparative NAIL when both distinguishes the damaged modification from the natural modification
unlabeled and labeled M9 stock mix are freshly prepared and ideally in (here CD3-methyl groups) [27]. C) The “damage” label and “backbone”
parallel. For pulse-chase NAIL-MS 15N labeling is acceptable if, for ex- label must be sufficiently resolvable in the mass spectrometer, i.e. if the
ample, the labeling is only used to distinguish original RNA from newly “damage” label is +3, the “backbone label” cannot be +3 but should
transcribed RNA. In these studies, modification profiles cannot be be ideally larger than +5. D) An internal standard must be available
compared, but changes in the RNA modification profile of e.g. original which is clearly distinguishable from all potential heavy isotope com-
RNA can be studied and effects of transcription can be excluded. binations from the biological assay.
In this study, we used 15-Nitrogen for the “backbone” labeling
which results in a labeling of +5 for purines, +3 for cytidine and +2

98
V.F. Reichle et al. Methods 156 (2019) 91–101

Fig. 5. a Concept of a pulse-chase NAIL-MS assay to distinguish the modified nucleosides from damaged tRNAs and newly transcribed tRNAs. (Adapted from [27].) b
and c Pulse-chase NAIL-MS assay for 1-methyladenosine (b, m1A) and 3-methylcytidine (c, m3C). Color code: Black: damaged tRNAs. Red: newly trtranscribed tRNAs
(Error bars represent the standard error of 3 biological replicates.)

Table 2 In E. coli, the quantities of the usually unnatural nucleosides m1A

Mass transitions of 1-methyladenosine isotopomers in pulse-chase NAIL-MS for and m3C peak after 2 h with around 1% damaged adenosines (Fig. 5b)
RNA repair observation. and only 0.08% damaged cytidines (Fig. 5c) in tRNA. The abundance of
Compound Compound Precursor Product Ret Time m1A is slowly decreasing in the damaged tRNA over the 10 h recovery
Group Name Ion Ion (min) period. Two potential scenarios can explain this: The first scenario is
degradation of m1A-damaged tRNAs. The sequence of each tRNA con-
Unlabeled (original) A 268 136 5.7
tains at least 10 adenines in E. coli [1]. 1% of all adenosine methylation
m1A 282 150 1.7
15
N and 13C labeled A SILIS 283 146 5.7 damage thus translates to a statistical abundance of m1A in 1 out of 10
(internal standard) m1A SILIS 298 161 1.7 tRNAs. Thus 10% of the total tRNA pool needs to be degraded for re-
CD3 labeled m1A CD3 285 153 1.7 moval of m1A. If m1A-targeted tRNA degradation was the cause, a faster
(post-methylated) tRNA dilution rate in comparison to unstressed E. coli would be ob-
15
N labeled (new) A 15N 273 141 5.7
15 servable. However, we observe the opposite (Supplementary Fig. S2b).
N and CD3 labeled m1A 15N_CD3 290 158 1.7
(new) The original tRNA pool from MMS stressed bacteria is less quickly di-
luted by newly transcribed tRNA in comparison to the unstressed bac-
teria. Thus, we consider the scenario of m1A-targeted tRNA degradation
for uridine. As the “damage/methylation” label we used CD3-methio- as unlikely.
nine (+3). The assay was set up as outlined in Fig. 5a and samples were The second scenario is the proposed repair by demethylation. We
drawn every hour after removal of the MMS. The bacteria are grown in see this hypothesis as proven, since we did not observe increased de-
unlabeled M9 media and exposed to the LD50 dose of MMS gradation of the original, damaged tRNA pool. Therefore, the only
(Supplementary Fig. S1). The original RNA as well as all methylated cause for the decrease in m1A abundance is active demethylation of
nucleosides were unlabeled (m/z ± 0). After 60 min, the MMS is re- m1A (potentially by AlkB [28]).
moved by media exchange. The new media contains only 15N as the With our NAIL-MS assay, we can also follow the speed of m3C repair
nitrogen source and CD3-methionine. Newly transcribed RNA is now by demethylation in vivo. The repair is even quicker (Fig. 5c), which is
labeled with 15N (m/z +5, +3 or +2) and methylated nucleosides most likely caused by the comparably low abundance of m3C sites in the
have an additional +3 label (m/z is thus +8, +6 or +5). Original tRNAs.
tRNA, which is enzymatically methylated after media exchange re-
ceives a +3 label from the CD3-methionine (m/z is thus +3). Using 5. Discussion and outlook
mass spectrometry, the tRNA exposed to the MMS and the newly
transcribed tRNA can be clearly distinguished and the abundance of LC-MS/MS is the method of choice for quantification of modified
modified nucleosides in the original tRNA can be quantified. An ex- nucleosides. However, it is currently limited and only provides data on
emplary list for the mass transitions of all m1A isotopomers is given in the quantities of modified nucleosides in an RNA but not about the
Table 2 and for the other nucleosides in Supplementary Table S3. Here, dynamic changes and the underlying mechanisms. These limitations
and in the recently published study [27], we solely focused on the fate can be overcome by using nucleic acid isotope labeling coupled mass
of the damaged tRNA from unlabeled bacteria culture. In such a case, spectrometry (NAIL-MS). We present currently used approaches of
the validation of NAIL-MS can be omitted, as the modification profile of isotope labeling in microorganisms such as E. coli and S. cerevisiae.
the new transcripts is not of interest. Furthermore, we demonstrate the difficulties of adapting these methods

99
V.F. Reichle et al. Methods 156 (2019) 91–101

to human cell culture. In our hands, cell culture labeling by addition of to determine relative abundances of RNAs.
13
C6-glucose is not sufficient to allow successful application of com- As an example, we show the repair kinetics of 1-methyladenosine
parative NAIL-MS experiments. However, for dynamic pulse-chase ex- (m1A) and 3-methylcytidine (m3C) in bacterial tRNA after exposure to
periments where e.g. only the original RNA is the focus of the analysis, MMS in vivo. We find that m1A is slowly removed from original tRNA
the labeling should allow sufficient resolution to distinguish the ori- over the timeframe of observation. This is in accordance with previous
ginal RNA from new transcripts. Thus, we foresee a wide usage of cell work done with radioisotope labeling [28]. In contrast to the published
culture labeling by addition of 13C6-glucose in dynamic NAIL-MS ex- work, we used E. coli without previous induction of AlkB which is re-
periments to answer current questions regarding open questions in the flected in the rather slow repair of m1A observed in our NAIL-MS study.
field such as: Is the change in the epitranscriptome due to a complete Similarly, we can observe the removal of m3C from the bacterial tRNA
renewal of the transcriptome? Are modifications added to the existing which appears to happen a lot faster. The reason can be the relatively
transcriptome to confront e.g. stress? And are modifications removed low abundance of m3C damage compared to m1A damage. It is also
actively in vivo? Due to the many isotopomers formed for each nu- possible that m3C is the better substrate for the tRNA demethylase.
cleoside and its respective modifications, we also foresee that massive In this study, we found around 1% of all adenines in tRNA are
validation is necessary to answer these questions by 13C6-glucose in methylated by the methylating agent MMS after one hour of exposure
dynamic NAIL-MS. Ideally, other metabolites, e.g. precursors of nu- to the LD50 dose. From a chemical point of view, the N1 position of
cleoside biosynthesis are found which lead to single isotopomer for- adenine can be considered a good nucleophile which is easily methy-
mation in cell culture. Thus, the usage of comparative NAIL and more lated by an electrophile such as MMS. Therefore, we wonder how
elegant pulse-chase experiments would become possible in cell culture. strongly the N1 position of adenine interacts with natural electrophiles
Regarding comparative NAIL-MS, we also foresee a unique oppor- such as S-adenosyl methionine (SAM). SAM is the natural methylating
tunity towards determination of changes in RNA abundance as a con- agent of the cell, and the methylation reaction of nucleic acids usually
sequence of stress or the loss of an RNA modification. Instead of depends on an enzyme which activates the nucleoside first. However,
quantifying the abundance of modified nucleosides, it should be pos- chemically, the N1 position of adenosine is already a good nucleophile,
sible to focus on the quantities of canonical nucleosides instead. As and it should be possible, that some methylation occurs by reaction
Fig. 6 shows, the total RNA from a comparative NAIL experiment is with SAM inside the cell, especially in those environments, which are
analyzed and the ratio of canonical nucleosides from labeled and un- rich in SAM and those RNAs that have one or more non base-paired
labeled total RNA is determined. The ratio from the total RNA is later adenosines. Considering the current dispute on the distribution of m1A
used for normalization and set to 100%. From the total RNA mixture, in mRNA [30–32], we wonder if some of the 0.02% m1A (per A) [33] in
the RNA of interest, here isoacceptor tRNAs, can be purified [29]. The mRNA is due to non-enzymatic methylation by SAM. In eukaryotic
quantities of canonical nucleosides from an isoacceptor tRNA can be mRNA, the 5′ end is methylated (m7G of the mRNA cap) and thus the 5′
determined and subsequently the ratio of labeled and unlabeled is end and its adenosines are always exposed to high amounts of SAM.
formed. In a control experiment, the abundance of all tRNA iso- Statistically, a sub-stoichiometric methylation of these adenosines is
acceptors is expected to be identical in the unlabeled and labeled possible and should be taken into account during data interpretation. Of
samples. The ratio of the pure isoacceptor tRNA can be compared to the course, the origin of the 5′ UTR methylation, enzymatic or chemical,
total RNA ratio and plotted in %. Indeed, we observe in our 13C6-glu- does not play a major role upon determining the function of the me-
cose validation experiments from E. coli (Fig. 6) the same ratio of la- thylated adenosine.
beled and unlabeled canonical nucleosides for total RNA and several With nucleic acid isotope labeling coupled mass spectrometry
purified tRNA isoacceptors. (NAIL-MS) we overcome current limitations and assess the dynamics of
In a comparative NAIL experiment, it should be possible to detect RNA modifications. We show the repair of m1A and m3C in vivo by
changes in e.g. tRNA isoacceptor abundances because of stress or ge- discriminating the concurrent processes for removal of RNA methyla-
netic manipulation of an RNA writer. By mixing an unlabeled control tion, namely RNA degradation, turnover and dilution. Dynamic NAIL-
sample with a labeled, but e.g. stressed sample, the potential changes of MS and comparative NAIL-MS are powerful tools which finally allow
tRNA isoacceptor abundance should be detectable. We expect that the the observation of dynamic processes of RNA and its modifications.
ratio of canonical nucleosides would be different in total RNA and the
purified tRNA isoacceptor and thus the impact of the stress on the re- Acknowledgements
lative abundance of the tRNA isoacceptor should be revealed. If our
assumption is correct, we see a broad usability of comparative NAIL-MS The Kellner lab thanks Thomas Carell and his group for instrument

Fig. 6. Concept of relative tRNA isoacceptor abundance determination by NAIL-MS. The ratio of co-purified total RNA from unlabeled (ul) and 13C labeled cells is 1:1
in this example. From the same RNA sample, a tRNA isoacceptor is purified by oligonucleotide hybridization. After digestion the ratio of the canonical nucleosides is
determined and compared to the ratio from total RNA. The resulting data of each isoacceptor tRNA is plotted for all isoacceptors (n = 3).

100
V.F. Reichle et al. Methods 156 (2019) 91–101

time (high-resolution mass spectrometer) and valuable discussion. The derivatives, Proc. Natl. Acad. Sci. U.S.A. 113 (11) (2016) E1452–E1459.
bacteria projects are funded by the Fonds der Chemischen Industrie, [13] C.E. Dumelin, et al., Discovery and biological characterization of geranylated RNA
in bacteria, Nat. Chem. Biol. 8 (11) (2012) 913–919.
Frankfurt, Germany (FCI) and the DFG (CIPSM and SFB 1309). Yeast [14] S. Kellner, et al., Profiling of RNA modifications by multiplexed stable isotope la-
work was funded by the SPP 1784 of the DFG and the cell culture work belling, Chem. Commun. (Camb.) 50 (26) (2014) 3516–3518.
by the Emmy Noether program of the DFG. VFR is funded by the FCI. [15] N.L. Anderson, N.G. Anderson, Proteome and proteomics: new technologies, new
concepts, and new words, Electrophoresis 19 (11) (1998) 1853–1861.
[16] R. Apweiler, et al., Approaching clinical proteomics: current state and future fields
Appendix A. Supplementary data of application in cellular proteomics, Cytometry A 75 (10) (2009) 816–832.
[17] S.E. Ong, et al., Stable isotope labeling by amino acids in cell culture, SILAC, as a
simple and accurate approach to expression proteomics, Mol. Cell. Proteomics 1 (5)
Supplementary data to this article can be found online at https:// (2002) 376–386.
[Link]/10.1016/[Link].2018.10.025. [18] N.A. Rosa-Mercado, J.B. Withers, J.A. Steitz, Settling the m(6)A debate: methyla-
tion of mature mRNA is not dynamic but accelerates turnover, Genes Dev. 31 (10)
(2017) 957–958.
References
[19] K.D. Meyer, S.R. Jaffrey, The dynamic epitranscriptome: N6-methyladenosine and
gene expression control, Nat. Rev. Mol. Cell Biol. 15 (5) (2014) 313–326.
[1] P. Boccaletto, et al., MODOMICS: a database of RNA modification pathways. 2017 [20] C.T. Chan, et al., A quantitative systems approach reveals dynamic control of tRNA
update, Nucleic Acids Res. 46 (D1) (2018) D303–D307. modifications during cellular stress, PLoS Genet. 6 (12) (2010) e1001247.
[2] W.M. Cai, et al., A platform for discovery and quantification of modified ribonu- [21] P.A. Aas, et al., Human and bacterial oxidative demethylases repair alkylation da-
cleosides in RNA: application to stress-induced reprogramming of tRNA modifica- mage in both RNA and DNA, Nature 421 (6925) (2003) 859–863.
tions, Methods Enzymol. 560 (2015) 29–71. [22] S. Kellner, et al., Oxidation of phosphorothioate DNA modifications leads to lethal
[3] K. Thuring, et al., Analysis of RNA modifications by liquid chromatography-tandem genomic instability, Nat. Chem. Biol. 13 (8) (2017) 888–894.
mass spectrometry, Methods 107 (2016) 48–56. [23] Y.H. Chionh, et al., A multidimensional platform for the purification of non-coding
[4] C. Wetzel, P.A. Limbach, Mass spectrometry of modified RNAs: recent develop- RNA species, Nucleic Acids Res. 41 (17) (2013) e168.
ments, Analyst 141 (1) (2016) 16–23. [24] P. van Delft, et al., The profile and dynamics of RNA modifications in animals,
[5] C. Brandmayr, et al., Isotope-based analysis of modified tRNA nucleosides corre- ChemBioChem 18 (11) (2017) 979–984.
lates modification density with translational efficiency, Angew. Chem. Int. Ed. Engl. [25] T. Pfaffeneder, et al., Tet oxidizes thymine to 5-hydroxymethyluracil in mouse
51 (44) (2012) 11162–11165. embryonic stem cell DNA, Nat. Chem. Biol. 10 (7) (2014) 574–581.
[6] T. Bruckl, et al., Parallel isotope-based quantification of modified tRNA nucleosides, [26] M. Bachman, et al., 5-Hydroxymethylcytosine is a predominantly stable DNA
Angew. Chem. Int. Ed. Engl. 48 (42) (2009) 7932–7934. modification, Nat. Chem. 6 (12) (2014) 1049–1055.
[7] D. Pearson, et al., LC-MS based quantification of 2'-ribosylated nucleosides Ar(p) [27] V.F. Reichle, V. Weber, S. Kellner, NAIL-MS in E. coli determines the source and fate
and Gr(p) in tRNA, Chem. Commun. (Camb.) 47 (18) (2011) 5196–5198. of methylation in tRNA, ChemBioChem (2018).
[8] S. Kellner, et al., Absolute and relative quantification of RNA modifications via [28] R. Ougland, et al., AlkB restores the biological function of mRNA and tRNA in-
biosynthetic isotopomers, Nucleic Acids Res. 42 (18) (2014) e142. activated by chemical methylation, Mol. Cell 16 (1) (2004) 107–116.
[9] M. Heiss, V.F. Reichle, S. Kellner, Observing the fate of tRNA and its modifications [29] R. Hauenschild, et al., The reverse transcription signature of N-1-methyladenosine
by nucleic acid isotope labeling mass spectrometry: NAIL-MS, RNA Biol. 14 (9) in RNA-Seq is sequence dependent, Nucleic Acids Res. 43 (20) (2015) 9950–9964.
(2017) 1260–1268. [30] S. Schwartz, m(1)A within cytoplasmic mRNAs at single nucleotide resolution: a
[10] L.P. Sarin, et al., Nano LC-MS using capillary columns enables accurate quantifi- reconciled transcriptome-wide map, RNA 24 (11) (2018) 1427–1436.
cation of modified ribonucleosides at low femtomol levels, RNA 24 (10) (2018) [31] X. Xiong, X. Li, C. Yi, N(1)-methyladenosine methylome in messenger RNA and non-
1403–1417. coding RNA, Curr. Opin. Chem. Biol. 45 (2018) 179–186.
[11] C. Dal Magro, et al., A vastly increased chemical variety of RNA modifications [32] X. Li, et al., Base-resolution mapping reveals distinct m(1)A methylome in nuclear-
containing a thioacetal structure, Angew. Chem. Int. Ed. Engl. 57 (26) (2018) and mitochondrial-encoded transcripts, Mol. Cell 68 (5) (2017) 993–1005.e9.
7893–7897. [33] X. Li, et al., Transcriptome-wide mapping reveals reversible and dynamic N(1)-
[12] J.J. Thiaville, et al., Novel genomic island modifies DNA with 7-deazaguanine methyladenosine methylome, Nat. Chem. Biol. 12 (5) (2016) 311–316.

101