Impact of 3-deazapurine nucleobases on RNA properties

Article abstract of DOI:10.1093/nar/gkab256

Deazapurine nucleosides such as 3-deazaadenosine (c3A) are crucial for atomic mutagenesis studies of functional RNAs. They were the key for our current mechanistic understanding of ribosomal peptide bond formation and of phosphodiester cleavage in recentl

Full text of DOI:10.1093/nar/gkab256

Published online 15 April 2021
Nucleic Acids Research, 2021, Vol. 49, No. 8 4281–4293
doi: 10.1093/nar/gkab256
Impact of 3-deazapurine nucleobases on RNA
properties
Raphael Bereiter1,†, Maximilian Himmelstoß1,†, Eva Renard2, Elisabeth Mairhofer1,
Michaela Egger1, Kathrin Breuker 1, Christoph Kreutz 1, Eric Ennifar2 and
1,*
Ronald Micura
1Institute of Organic Chemistry, Center for Molecular Biosciences Innsbruck, University of Innsbruck,
Innsbruck, Austria and 2Architecture et Re´activite´ de l’ARN - CNRS UPR 9002, Universite´ de Strasbourg, Strasbourg,
France
Received March 05, 2021; Editorial Decision March 26, 2021; Accepted March 29, 2021
ABSTRACT
INTRODUCTION
Deazapurine nucleosides have been intensively applied in
atomic mutagenesis studies of biologically relevant RNAs
(1–5). The replacement of a single nitrogen atom by a car-
bon atom can be critical because the acid-base properties
of the nucleobase are dramatically changed and hydrogen
acceptor/donor properties are erased at the specific posi-
tion (6–8). This can be determining for RNA base pairing
(8,9), for RNA recognition of other nucleic acids (e.g. DNA,
2ꢀ-OCH3 RNA) (6), proteins (3), small molecules (8), and
ions (9), and this can also be crucial with respect to RNA-
catalyzed reactions (4,10–13). Concerning the latter, atomic
mutagenesis lead to our current in-depth understanding of
the chemical mechanism of phosphodiester cleavage of the
recently discovered twister and pistol ribozymes (14–16).
Similarly, atomic mutagenesis of the peptidyl transferase
center in the ribosome center allowed to critically evaluate
the mechanistic proposals for peptide bond formation that
arose from interpretations of static crystal structures (17–
20). These functional assays contributed significantly to a
profound comprehension of ribosomal protein synthesis.
Suitable deazanucleosides for informative RNA atomic
mutagenesis experiments are 7-deazaadenosine (c7A)
(4,11–13,15,16), 3-deazaadenosine (c3A) (14,20), 1-
deazaadenosine (c1A) (11,12,14,20), 7-deazaguanosine
(c7G) (13,15) and 3-deazacytidine (c3C) (13,21). Moreover,
3-deazaguanosine (c3G) and 1-deazaguanosine (c1G)
would be highly useful for RNA atomic mutagenesis,
however, such studies are very rare (22) because syn-
thetic access to appropriate phosphoramidite building
blocks and to the corresponding RNAs is challenging
(22,23). In general, 3-deazapurine modified RNAs appear
underinvestigated when compared to the 7-deazapurine
modified counterparts (24–28). Besides, more is known
about deaza-modified DNA compared to deaza-modified
RNA (29–38).
Deazapurine nucleosides such as 3-deazaadenosine
(c3A) are crucial for atomic mutagenesis studies of
functional RNAs. They were the key for our cur-
rent mechanistic understanding of ribosomal pep-
tide bond formation and of phosphodiester cleav-
age in recently discovered small ribozymes, such as
twister and pistol RNAs. Here, we present a compre-
hensive study on the impact of c3A and the thus far
underinvestigated 3-deazaguanosine (c3G) on RNA
properties. We found that these nucleosides can de-
crease thermodynamic stability of base pairing to a
significant extent. The effects are much more pro-
nounced for 3-deazapurine nucleosides compared
to their constitutional isomers of 7-deazapurine nu-
cleosides (c7G, c7A). We furthermore investigated
base pair opening dynamics by solution NMR spec-
troscopy and revealed significantly enhanced imino
proton exchange rates. Additionally, we solved the
X-ray structure of a c3A-modified RNA and visual-
ized the hydration pattern of the minor groove. Im-
portantly, the characteristic water molecule that is
hydrogen-bonded to the purine N3 atom and always
observed in a natural double helix is lacking in the
3-deazapurine-modified counterpart. Both, the find-
ings by NMR and X-ray crystallographic methods
hence provide a rationale for the reduced pairing
strength. Taken together, our comparative study is
a first major step towards a comprehensive under-
standing of this important class of nucleoside modi-
fications.
*To whom correspondence should be addressed. Tel: +43 512 507 57710; Fax: +43 512 507 57799; Email: ronald.micura@uibk.ac.at
†The authors wish it to be known that, in their opinion, the first two authors should be regarded as Joint First Authors.
ꢁ
C
The Author(s) 2021. Published by Oxford University Press on behalf of Nucleic Acids Research.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

4282 Nucleic Acids Research, 2021, Vol. 49, No. 8
nucleoside phosphoramidite building blocks and 2ꢀ-O-
Therefore, we set out to accomplish a thorough chemical
and biophysical analysis of 3-deazapurine containing RNA,
along with the improvement of the synthesis of c3G contain-
ing RNA beforehand. We here describe the impact of c3G
and c3A on RNA properties. Based on UV-spectroscopic
melting experiments, a detailed thermodynamic analysis of
duplex and hairpin stabilities is provided and the effects on
base pairing are discussed in the light of the sequence con-
text. Furthermore, solution NMR spectroscopy sheds light
on base pair opening dynamics. In addition, we have solved
the X-ray structure of a c3A containing RNA at atomic res-
olution to disclose crucial structural features, such as ri-
bose puckers, hydrogen-bonding networks, and hydration
patterns of the deazanucleoside, and to correlate them to
base pairing properties.
˚
Tbs 1000 A CPG solid support (>15nt) were purchased
from ChemGenes, Primer supportTM 5G (<15nt) was
purchased from GE Healthcare. All oligonucleotides were
synthesized on an ABI 391 or ABI 392 Nucleic Acid
Synthesizer following standard methods: detritylation (90
sec) with dichloroacetic acid/1,2-dichloroethane (4/96);
coupling (5.0 min) with phosphoramidites/acetonitrile
(100 mM, 200 ␮l) and benzylthiotetrazole / acetonitrile
(300 mM, 500 ␮l); capping (2 × 25 s) with Cap A/Cap
B (1/1) for c3A, c7G & c7A modified RNA, Cap A:
4-(dimethylamino)pyridine/acetonitrile (500 mM), Cap
B: acetic anhydride/sym-collidine/acetonitrile (2/3/5)
and Cap A mild/Cap B mild (1/1) for c3G modified
RNA, Cap A mild: phenoxyacetic anhydride/acetonitrile
(100 mM), Cap
B
mild: N-methylimidazole/sym-
collidine/tetrahydrofuran (0.160/0.265/10); oxidation (60
sec) with iodine (20 mM) in tetrahydrofuran/pyridine/H2O
(35/10/5). Solutions of phosphoramidites, tetrazole and
MATERIAL AND METHODS
Synthesis and characterization of organic compounds
˚
Cap were dried over activated molecular sieves (3 A)
Reagents were purchased in the highest available quality
from commercial suppliers (Merck/Sigma-Aldrich, ABCR,
VWR, ChemGenes, CarboSynth) and used without fur-
ther purification. All reactions were carried out under ar-
gon atmosphere, unless otherwise noted. Analytical thin-
layer chromatography (TLC) was performed on Macherey-
Nagel Polygram® SIL G/UV254 plates. Silica gel 60 (mesh
size 0.04 – 0.063 mm) for column chromatography was pur-
chased from Macherey-Nagel. The procedures for chem-
ical synthesis of phosphoramidites 8 and 9, and the 6-
chloro-3-deazapurine 14 and their characterization data are
available in the Supporting Information (phosphoramidite
8: eight steps, eight chromatographic purifications, 13%
overall yield; total amount synthesized: 0.90 g; phospho-
ramidite 9: ten steps, 10 chromatographic purifications, 9%
overall yield; total amount synthesized: 0.80 g; 6-amino-3-
deazapurine 15: 4 steps, four chromatographic purifications,
54% overall yield; total amount synthesized: 5 g). 1H, 13C,
and 31P NMR spectra were recorded on a Bruker Ultra-
shield™ 400 Plus spectrometer. Chemical shifts (␦) are re-
ported relative to tetramethylsilane (TMS), referenced to
the residual solvent signal (DMSO-d6: 2.50 ppm for 1H and
39.52 ppm for 13C NMR spectra; CDCl3: 7.26 ppm for 1H
and 77.16 ppm for 13C NMR spectra). The following ab-
breviations were used to denote multiplicities: s = singulet,
d = doublet, t = triplet, q = quadruplet, m = multiplet, b
overnight.
Deprotection, purification and quantification of natural and
modified RNA
For basic deprotection of natural, c3G, c3A, c7G and c7A
modified RNA, the solid support was mixed with aqueous
methylamine (40%, 0.65 ml) and aqueous ammonia (28%,
0.65 ml) for 15 min at 65◦C or 3 h at 37◦C. The supernatant
was removed and the solid support was washed twice with
0.5 ml tetrahydrofuran/H2O (1/1). Combined supernatant
and washings were evaporated to dryness and the residue
was dissolved in a solution of tetrabutylammonium fluo-
ride in tetrahydrofuran (1.0 M, 1.5 ml) and incubated for
14 h at 37◦C for removal of 2ꢀ-O-silyl protecting groups.
The reaction was quenched by addition of triethylammo-
nium acetate/H2O (1.0 M, 1.5 ml, pH 7.4). Tetrahydro-
furan was removed under reduced pressure and the sam-
ple was desalted with size-exclusion column chromatogra-
phy (GE Healthcare, HiPrep™ 26/10 Desalting; Sephadex
G25) eluting with H2O; collected fractions were evapo-
rated and the RNA dissolved in H2O (1 ml). The crude
RNA was purified by anion exchange chromatography on
a semipreparative Dionex DNAPac® PA-100 column (9
mm × 250 mm) at 80◦C with a flow rate of 1 ml/min (elu-
ent A: 6 M urea, 25 mM Tris·HCl, pH 8.0; eluent B: 500
mM NaClO4, 6 M urea, 25 mM Tris·HCl, pH 8.0). Frac-
tions containing RNA were diluted with 0.1 M triethy-
lammonium bicarbonate solution, loaded on a C18 Sep-
Pak Plus® cartridge (Waters/Millipore), washed with H2O
and eluted with acetonitrile/H2O (1/1). Crude and purified
RNA were analyzed by anion exchange chromatography on
1
= broad. Signal assignments are based on H–1H-COSY,
1H–13C-HSQC and 1H–13C-HMBC experiments. High res-
olution mass spectra were recorded in positive ion mode on
a Thermo Scientific Q Exactive Orbitrap, ionized via elec-
trospray at 3.7 kV spray voltage.
RNA solid-phase synthesis
¨
a GE Healthcare Akta Explorer HPLC System containing
Standard phosphoramidite chemistry was applied for RNA
strand elongation and incorporation of 3-deazaguanosine
(N2-Tfa, 2ꢀ-O-Tbs and N2-Tfa, 2ꢀ-O-CEM; >98% coupling
yield), 3-deazadenosine (N6-Bz, 2ꢀ-O-TIPS; >92% cou-
pling yield), 7-deazaguanosine (N2-iBu, 2ꢀ-O-Tbs; >98%
coupling yield) and 7-deazaadenosine (N6-Bz, 2ꢀ-O-Tbs;
>98% coupling yield). 2ꢀ-O-TOM and acetyl protected
a Dionex DNAPac® PA-100 column (4 mm × 250 mm)
at 80◦C with a flow rate of 1 ml/min. For RNA shorter or
equal to 15 nucleotides, a gradient of 0–40% B in 30 minutes
and for RNA longer than 15 nucleotides a gradient of 0–
60% B was used; Eluent A: 6 M urea, 25 mM Tris·HCl, pH
8.0; Eluent B: 500 mM NaClO4, 6 M urea, 25 mM Tris·HCl,
pH 8.0. HPLC traces were recorded at UV absorption by

Nucleic Acids Research, 2021, Vol. 49, No. 8 4283
260 nm. RNA quantification was performed on an Implen
P300 Nanophotometer.
proton––bulk water exchange rates due to differences in the
buffer composition. The CLEANEX-PM pulse sequence is
available from the Bruker standard experiment collection
(zgcxesgp) with an excitation sculpting water suppression
element. A standard excitation sculpting water suppression
experiment (zgesgp) was used as the reference experiment.
For the determination of the water T1 relaxation times un-
der the experimental conditions a saturation recovery exper-
iment was used and the longitudinal water relaxation time
ranged between 3.03 s (c3G RNA) and 3.13 s (unmodified
RNA). The following experimental parameters were used:
spectral width 24 ppm, o1p: 4.7 ppm, number of scans 2048,
dummy scans 32, interscan delay 1.5 s. Shaped pulse param-
eters were set via the getprosol command and the hard 90◦
1H pulse. The mixing times were set to 5, 25, 50, 50, 100,
100, 150, 200, 300, 400, 400 and 500 ms for both duplexes.
The NMR spectra were processed in TOPSPIN 4.0.9 us-
ing a line broadening factor of 10 Hz (lb value set to 10).
The absolute peak intensities of the CLEANEX-PM exper-
iments and the reference experiments were determined and
used to obtain the relative intensities. The data was then ex-
ported to MATLAB and the build-up curves were fitted to
the following equation:
Mass spectrometry of oligoribonucleotides
RNA samples (3 ␮l) were diluted with 40 mM
Na2H2(EDTA)/H2O (5/4) for a total volume of 30
␮l, injected onto a C18 XBridge column (2.5 ␮m, 2.1 mm
× 50 mm) at a flow rate of 0.1 ml/min and eluted with 0–
100% B gradient at 30◦C (eluent A: 8.6 mM triethylamine,
100 mM 1,1,1,3,3,3-hexafluoroisopropanol in H2O; eluent
B: methanol). RNA HPLC runs were analyzed on a Finni-
gan LCQ Advantage Max electrospray ionization mass
spectrometer with 4.0 kV spray voltage in negative mode.
Characterization of the c3G modified 47mer was performed
on a 7 T FT-ICR-MS (Bruker Daltonics – Apex Ultra 70).
Sample preparation: 2 ␮M RNA in 1:1 H2O/CH3OH with
20 mM imidazole and 20 mM piperidine (pH ∼ 10). The
sample was measured in negative ion mode at a flow rate of
90 ␮l/h and a skimmer potential of –50 V for ESI and –60
V for CAD (spectra averaged CAD: 400, spectra averaged
ESI: 50, capillary voltage 3.1 kV, spray shield 2.7 kV, dry
temperature 200◦C. A collision energy of 12 V was applied
in the CAD experiments.
ꢀ
k
I I0 =
(R1 A+ k − R1,water
)
Melting curve measurements of oligoribonucleotides
ꢁ
ꢂ
× e−R
− e(−R
(1,water)τmix
)
1,A+k)
τ
mix
RNA samples were lyophilized as triethylammonium salts,
dissolved in 800 or 330 ␮l buffer (10 mM Na2HPO4 (pH
7.0) containing 150 mM NaCl) and transferred into UV per-
meable high precision cells made of quartz SUPRASIL®
with a light path of 10 mm or 1 mm. UV melting profiles
were recorded at 250 and 260 nm on a Varian Cary-100
spectrophotometer equipped with a multiple cell holder and
a peltier temperature control device. Each RNA was mea-
sured at five different concentrations ranging from 1 to 100
␮M. Tm values were determined by calculating the first
derivative. Thermodynamic parameters were determined
according to (49,50). The error limits for ꢀG, ꢀH and ꢀS
reflect the standard deviation of three (duplex) or five (hair-
pin) independent measurements for a confidence interval of
95%.
with I/I0 relative peak intensity, k imino proton–bulk wa-
ter exchange rate (s−1), R1,water longitudinal water exchange
rate, τmix CLEANEX-PM mixing times (s) and R1,A is a
combination of the longitudinal and transverse relaxation
rate of the imino proton resonance (floating parameter dur-
ing fitting). Errors in exchange rates were estimated from
replicate experiments and obtained from 1000 Monte-Carlo
runs.
Determination of the pKa value of 3-deazaguanosine by NMR
spectroscopy. For the pH dependent H and 13C NMR
1
experiments, 22 mg of 3-deazaguanosine were dissolved in
600 ␮l buffer (15 mM sodium phosphate, 25 mM NaCl, 0.1
mM EDTA, pH 6.9) and transferred into a 5 mm NMR
tube. The chemical shifts of C2 and C6 were recorded on a
Bruker 600 MHz Avance II+ equipped with a Prodigy TCI
probe. The pH-value was determined using a Sigma-Aldrich
micro pH combination glass electrode. At each pH-value,
1H NMR spectrum and 13C NMR spectra at natural abun-
dance with power gated decoupling were measured.
The pKa values were obtained by fitting the data to the
following equation:
NMR experiments
RNA samples were lyophilized as triethylammo-
nium salts, dissolved in 500 ␮l NMR buffer (15 mM
Na[AsO2(CH3)2]·3H2O, 25 mM NaCl, 3 mM NaN3, in
H2O/D2O 9:1, pH 6.5) and transferred into 5 mm NMR
tubes. Sample concentrations varied between 0.05 and 0.3
mM and experiments were run at 298 K unless otherwise
stated. All NMR experiments were conducted on a Bruker
600 MHz Avance II+ NMR or a 700 MHz Avance Neo
NM both equipped with a Prodigy TCI probe.
δdeprotonated x 10(pH−pk ) + δprotonated
a
δobs
=
1 + 10(pH−pk )
a
Determination of imino proton water exchange rates by
CLEANEX-PM experiment. The CLEANEX-PM NMR
experiments were carried out in the following buffer: 15
mM Na[AsO2(CH3)2]·3H2O, 25 mM NaCl, 3 mM NaN3,
in H2O/D2O 9:1, pH 6.5. The same buffer stock solu-
tion was used for all samples to rule out changes in imino
With δobs observed 13C chemical shift at the respective
pH value, δdeprotonated 13C chemical shift of deprotonated
species and δdeprotonated 13C chemical shift of the protonated
species.

4284 Nucleic Acids Research, 2021, Vol. 49, No. 8
Table 1. Thermodynamic data of deazapurine modified RNAa
c
c
c
No.
RNA sequences 5ꢀ to 3ꢀ
Tm (◦C) b
ꢀTm (◦C)
ꢀG◦298 (kcal mol−1
)
ꢀH◦298 (kcal mol−1
)
ꢀS◦ (cal mol−1 K−1
)
I
Ia
Ib
Ic
Id
II
IIa
IIb
IIc
IId
III
IIIa
IIIb
IIIc
IIId
III’
III’a
III’b
III’c
III’d
GGCAGAGGC / GCCUCUGCC
GGCAc3GAGGC / GCCUCUGCC
GGCAGc3AGGC / GCCUCUGCC
GGCAc7GAGGC / GCCUCUGCC
GGCAGc7AGGC / GCCUCUGCC
GAAGGGCAACCUUCG
GAAc3GGGCAACCUUCG
GAc3AGGGCAACCUUCG
GAAc7GGGCAACCUUCG
GAc7AGGGCAACCUUCG
GGUCGACC
66.7
62.8
63.8
67.2
66.5
73.3
64.2
65.4
68.5
69.8
58.3
44.8
48.5
55.9
55.6
60.7
47.7
47.7
58.6
59.7
-
–16.5
–15.2
–16.0
–16.7
–16.1
–7.0
–6.2
–6.4
–6.5
–6.6
–13.2
–11.4
–11.7
–13.4
–13.0
–14.5
–11.8
–10.7
–13.7
–14.2
0.4
0.5
0.3
0.6
0.6
0.2
0.2
0.3
0.3
0.4
0.9
0.5
0.9
1.0
0.6
1.1
0.8
0.6
0.7
0.6
–79.7
–74.8
–80.0
–80.1
–76.6
–52.8
–53.5
–53.8
–51.5
–52.3
–64.6
–74.7
–68.1
–70.6
–67.0
–72.3
–70.9
–56.5
–68.8
–71.8
4.6
4.7
1.8
5.4
4.2
1.3
1.6
2.0
1.9
3.3
8.6
6.4
8.8
8.5
4.6
9.5
8.3
6.2
7.3
6.9
–212
–199
–215
–213
–203
–153
–159
–159
–151
–153
–172
–213
–189
–192
–181
–194
–199
–154
–185
–193
14
14
5
16
12
4
5
6
5
10
26
20
26
25
14
28
25
19
22
21
–3.9
–2.9
+0.5
–0.2
-
–9.1
–7.9
–4.8
–3.5
-
–13.5
–9.8
–2.4
–2.7
-
–13.0
–13.0
–2.1
–1.0
GGUCc3GACC
GGUCGc3ACC
GGUCc7GACC
GGUCGc7ACC
GGCUAGCC
GGCUAc3GCC
GGCUc3AGCC
GGCUAc7GCC
GGCUc7AGCC
aBuffer: 10 mM Na2HPO4, 150 mM NaCl, pH 7.0. ꢀH and ꢀS values were obtained by van’t Hoff analysis or based on RNA concentration dependent measurements according
to references 49 and 50.
bThe estimated errors of UV-spectroscopically determined Tm values are 0.3◦C.
cErrors for ꢀH and ꢀS were determined from three (RNAs I, III, and III’) and five (RNAs II) independent measurements; in general, errors arising from noninfinite cooperativity
of two-state transitions and from the assumption of a temperature-independent enthalpy, are typically 10−15%. Additional error is introduced when free energies are extrapolated
far from melting transitions; errors for ꢀG are typically 3−5%.
Crystallization and structure solution of a c3A modified RNA
RESULTS AND DISCUSSION
Improved c3G building block with N2-Tfa and 2ꢀ-OTbs pro-
Oligoribonucleotides of 27 nucleotide length correspond-
ing to the sequence of Escherichia coli 23 S rRNA sarcin-
ricin loop (SRL) and containing a 3-deazaadenosine at po-
sition 2670 were used for crystallization. The RNA was dis-
solved at a concentration of ∼350 ␮M in a buffer made
of Tris–HCl (10 mM), Na2H2EDTA (1 mM), pH 8.0. The
RNA sample was then heated to 55◦C and cooled down
to 10◦C using a temperature-controlled device equipped
with a Peltier element. Only one unique cubic-shaped crys-
tal of c3A2670-modified RNA could be obtained. It grew
after one month at 20◦C using vapor diffusion method
by mixing two volumes of RNA sample with one vol-
ume of a crystallization buffer made of ammonium sul-
fate (2.5 M), magnesium acetate (10 mM), and 2-(N-
morpholino)ethanesulfonic acid (MES) (50 mM), pH 5.6
(the other drops made in identical conditions led to spherul-
lites). Prior data collection, the crystal was cryoprotected
for about 5 min in a reservoir solution containing 15% of
glycerol and 3.0 M of ammonium sulfate, flash-frozen in
liquid ethane and then transferred into liquid nitrogen. The
collection of X-ray diffraction data has been done on the
X06DA beamline at the SLS synchrotron, Villigen, Switzer-
land. Processing of the data was done with the XDS Pack-
age (71) and the structure was solved by molecular replace-
ment with MOLREP (72) using the related PDB ID 3DVZ
unmodified SRL RNA model. The structure was refined
with the PHENIX package (73). Models were built us-
ing Coot (74). Alternative conformations were visible for
residues 2647–2650. Coordinates have been deposited with
the PDB database (PDB ID 7L3R for c3A2670-modified
SRL).
tection
For the solid-phase synthesis of c3A and c3G modified
RNA, appropriate phosphoramidite building blocks are
needed. Their syntheses have been a bottleneck for a long
time (18,22), with new and more powerful routes coming
up only recently (23,28).
A practical synthesis for c3G phosphoramidites from in-
expensive starting materials has been introduced by our re-
search group (23). Therein, the phenoxyacetyl (Pac) and
2-cyanoethoxymethyl (Cem) groups block the N2- and 2ꢀ-
hydroxyl functionalities, while the O6 atom remains unpro-
tected. This building block was successfully incorporated
into a short 5 nt RNA strand. Deprotection of the N2-
phenoxyacetyl group, however, requires rather long reac-
tion times and high pH values which causes degradation
when longer oligoribonucleotides are envisaged. To solve
this problem, we here demonstrate the installation of a more
base-labile protection group for the exocyclic amine of c3G,
resulting in N2-trifluoroacetyl protected 3-deazaguanosine
phosphoramidite 8. Distinct to our previously introduced
c3G building block, we used 2ꢀ-O-tert.-butyldimethylsilyl
(Tbs) protection (instead of the 2ꢀ-O-Cem group) to guaran-
tee widest possible compatibility with standard RNA solid-
phase synthesis.
The synthesis started from the previously reported 3-
deazaguanosine key intermediate 1 (23) (Scheme 1). The ap-
plication of the temporary O-tert.-butyldimethylsilyl (Tbs)
instead O-acetyl protection (compound 2) was necessary
because of otherwise partial acetyl transfer to the N2 amino

Nucleic Acids Research, 2021, Vol. 49, No. 8 4285
Scheme 1. Synthesis of the novel c3G phosphoramidite building block 8. Reagents and conditions: (a) MeNH2 in ethanol, room temperature, 1 h, 99%; (b) 6
equiv tert-butyldimethylsilyl chloride (TbsCl), 7 equiv imidazole in DMF, room temperature, 72 h, 90%; (c) 1.2 equiv trifluoroacetamide, 1.4 equiv Cs2CO3,
0.05 equiv Pd2(dba)3, 0.15 equiv Xantphos in 1,4-dioxane, 110◦C, 5 h, 90%; (d) 4.1 equiv tetra-n-butylammonium fluoride trihydrate in tetrahydrofuran,
room temperature, 2 h, 87%; (e) 1.15 equiv 4,4ꢀ-dimethoxytriphenylmethyl chloride, 0.05 equiv 4-(N,N-dimethylamino)pyridine in pyridine, room temper-
ature, 16 h, 74%; (f) 1.9 equiv TbsCl, 1.9 equiv AgNO3, in tetrahydrofuran, room temperature, 16 h, 39% (after two rounds of 2ꢀ,3ꢀ equilibration); (g) 0.33
equiv Pd/C 10%, H2 in tetrahydrofuran/ethanol 4:1, room temperature, 4 h, 79%; (h) 3.0 equiv 2-cyanoethyl-N,N,Nꢀ,Nꢀ-tetraisopropylphosphorodiamidite,
0.5 equiv 5-(benzylthio)-1H-tetrazole in dichloromethane, room temperature, 18 h, 79%.
tial attempts to take advantage of the Beigelman 5ꢀ,3ꢀ-
silyl clamp and 2ꢀ-O-Tbs protection concept (40) for com-
pound 1 because of low yields. The O6-benzyl group of
nucleoside 6 was cleaved by Pd/C-catalyzed hydrogena-
tion to provide the free lactam moiety in derivative 7.
Finally, phosphitylation was executed with 2-cyanoethyl-
N,N,Nꢀ,Nꢀ-tetraisopropylphosphorodiamidite in the pres-
ence of 5-(benzylthio)-1H-tetrazole. Starting from nucle-
oside 1, our route provides phosphoramidite 8 with 13%
overall yield in eight steps involving eight chromatographic
purifications; in total, 0.90 g of compound 8 was obtained
during the course of this study.
group under the conditions required for the next step,
namely the installation of the N2-trifluoroacetylamino (Tfa)
functionality: Pd catalyzed cross coupling between trifluo-
roacetamide and bromo compound 2 by using Buchwald-
Hartwig conditions gave derivative 3 in excellent yields.
Deprotection of the Tbs groups proceeded smoothly with
tetra-n-butylammonium fluoride (TBAF) in tetrahydrofu-
ran to furnish compound 4. Then, the functionalization
of 4 into the desired c3G phosphoramidite 8 required
four more transformations. First, nucleoside 4 was con-
verted into the dimethoxytritylated compound 5, using
4,4ꢀ-dimethoxytriphenylmethyl chloride in pyridine. Then,
Tbs-protection according to Ogilvie (39) with silver ni-
trate proceeded with rather low stereoselectivity (about
3:2), but in combination with base-induced equilibration
the overall yield was increased to 39% for compound 6.
At this point, we note that we did not follow up our ini-
Additionally, we synthesized the N2-Tfa protected build-
ing block 9 to take advantage of the less bulky 2ꢀ-O-
(2-cyanoethoxy)methyl protecting group if short coupling
times during solid-phase synthesis would be needed (see the
Supporting Information and Scheme 2).

4286 Nucleic Acids Research, 2021, Vol. 49, No. 8
Scheme 2. Synthesis of the novel c3G phosphoramidite building block 9. For individual reaction steps, reagents and conditions see the Supporting Infor-
mation.
deprotecting of the base labile groups were accomplished
by treatment with methylamine/ammonia in water (1:1 mix-
ture of 40% aqueous methylamine and 30% aqueous ammo-
nia (AMA) for 3 h at 37◦C room temperature or 15 min at
65◦C). Subsequently, deprotection of the 2ꢀ-O-silyl groups
(and the 2ꢀ-O-Cem group in case of 9) was carried out with
tetra-n-butylammonium fluoride trihydrate in tetrahydro-
furan for 14 hours at 37◦C. The reaction was quenched
by the addition of triethylammonium acetate buffer at pH
7.4. Salts were removed by size-exclusion chromatography.
Analysis by anion exchange chromatography under strong
denaturing conditions usually gave a major peak for the de-
sired RNA which was further purified by anion exchange
chromatography on a semipreparative column. The molec-
ular weights of the purified RNAs were confirmed by LC–
ESI-MS (Supplementary Table S1) and the sequences were
confirmed by top-down mass spectrometry (MS) using a
Fourier-transform ion cyclotron resonance (FT-ICR) mass
spectrometer (42) (Supplementary Table S2).
Synthesis of c3A containing RNA
The synthetic path of the c3A nucleoside building block
Figure 1. Characterization of c3G-modified RNA. (A) Anion-exchange
has been merged from two routes that we published ear-
HPLC traces of 8 nt RNA (top) and LC–ESI iontrap mass spectrum (bot-
lier (Scheme 3) (18,28). The key reaction is a silyl-Hilbert–
tom); (B) Anion-exchange HPLC traces of 47 nt RNA (top) and ESI FT-
Johnson nucleosidation using unprotected 6-amino-3-
ICR mass spectrum (bottom). HPLC conditions: Dionex DNAPac col-
umn (4 × 250 mm), 80◦C, 1 ml min–1, 0–40% buffer B (for 8 nt RNA) in
deazapurine 15 and benzoyl-protected 1-O-acetylribose
(28). We then proceeded with N6-benzoyl protection
buffer A within 30 min; 0–60% buffer B (for 47 nt RNA) in buffer A within
(18) instead of N6-(di-n-butyl)amidine protection (28)
30 min; buffer A: Tris–HCl (25 mM), urea (6 M), pH 8.0; buffer B: Tris-
HCl (25 mM), urea (6 M), NaClO4 (0.5 M), pH 8.0. See the experimental
for the following reason: The introduction of the 2ꢀ-O-
for LC–ESI MS conditions.
triisopropylsilyl group is not regioselective and results in
a mixture of 2ꢀ- and 3ꢀ-O-regioisomers; the required chro-
matographic purification of this mixture is more straight-
Synthesis of c3G containing RNA
forward for the N6-benzoyl protected regioisomers (com-
pared to the N6-(di-n-butyl)amidine counterparts) because
of the significantly larger gap in retention times. Further-
more, in this work we also optimized the access to 6-amino-
3-deazapurine 15 and its precursor 6-chloro-3-deazapurine
14 on large scale and starting from cheap materials (Scheme
3) (43,44). Solid-phase synthesis of c3A-modified building
block 17 into RNA was performed using standard proto-
cols as described earlier (15).
The incorporation of building blocks 8 and 9 into olig-
oribonucleotides proceeds with high coupling rates and
excellent yields using standard RNA solid-phase synthe-
sis protocols and, as intended, significantly milder condi-
tions can be used to reach complete deprotection of the
c3G containing oligonucleotides. Figure 1 exemplarily il-
lustrates the synthesis of 8 and 47 nt long RNA oligonu-
cleotides with a single c3G modification (for a complete
list of synthesized oligos see Supplementary Table S1). In
short, the c3G phosphoramidite building blocks 8 or 9
were applied in combination with standard N-acetylated 2ꢀ-
O-[(triisopropylsilyl)oxy]methyl (TOM) phosphoramidites
and the oligomers were assembled on controlled pore glass
(CPG) supports (41). Cleavage from the solid support and
Base pairing properties of c3G and c3A containing RNA
To the best of our knowledge, hardly anything is known
about the impact of 3-deazapurine nucleosides on base pair-
ing in regular RNA. We therefore set out to analyze the

Nucleic Acids Research, 2021, Vol. 49, No. 8 4287
Scheme 3. Synthesis of the novel c3A phosphoramidite building block 16. Reagents and conditions: (a) in formic acid, 140◦C, 48 h, 97%; (b) 2.0 equiv
3-chloroperbenzoic acid in 2:1 (v/v) CH2Cl2/CH3OH, room temperature, 16 h, 96%; (c) in POCl3, 110◦C, 4 h, 67%; (d) in 1:1 (v/v) 30% NH3 in H2O/1,4-
dioxane, 190◦C, autoclave, 120 h, 87%.
two and three regular Watson–Crick base pairs can form
next to them: we remind that nucleation of a bimolecular
double helix of oligonucleotides becomes thermodynami-
cally favorable only when at least (three to) four continuous
Watson–Crick base pairs can form (45,46). Therefore, these
palindromic RNAs are expected to markedly respond to a
nucleobase modification expressed in significant alterations
of their thermal stabilities (Tm) and thermodynamic param-
eters (ꢀG, ꢀH, ꢀS).
Table 1 summarizes the thermal and thermodynamic data
we obtained for the four RNA systems by UV-spectroscopic
melting profile measurements (for melting profiles see the
Supplementary Figures S1 to S16). The native type I RNA
melts at 66.7◦C. Both 3-deazapurines exhibit a destabiliza-
tion, by 3.9 and 2.9◦C, for c3G (in Ia) and c3A (in Ib) re-
spectively. Interestingly, in this sequence context, the corre-
sponding 7-deazapurines (c7G in Ic and c7A Id) have negli-
gible effect on the Tm and thermodynamic parameters. For
the hairpin RNA (type II), the destabilization of both 3-
Figure 2. Sequence design for thermodynamic analysis of base pairing
deazapurines is significant, reflected in a decrease of the Tm-
of deazapurine modified RNAs. Cartoon presentation to highlight inter-
value by 9.1 and 7.9◦C, for c3G (in IIa) and c3A (in IIb) re-
strand stacking interactions (in orange).
spectively, when compared to the native hairpin II. In con-
trast to type I RNA, the monomolecular system (type II)
reveals a slight destabilizing effect of 7-deazapurines as well
influence of c3G and c3A on the stability of RNA double
helices, and additionally, to compare them to the c7G and
c7A modified counterparts. Figure 2 illustrates the sequence
design of the double helices investigated. The first motif
constitutes an asymmetric bimolecular duplex of nine base
pairs with a single deazapurine modification in the center
(Type I). The second motif forms a hairpin with an extra-
stable GNRA loop (GCAA), a five base pair stem with
the modification again in the center, and additionally a 3ꢀ-
dangling guanosine to reduce fraying of the terminal base
pair (Type II). The third RNA motif covers two palindromic
RNAs of eight base pairs and identical pyrimidine–purine
stacking patterns (Type III and IIIꢀ) with the purine mod-
ification either side-by-side in the center (thereby enabling
direct inter-strand stacking), or separated by two base pairs.
The latter sequence design is highly sensitive for the thermo-
dynamic impact arising from a modification because only
(c7G in IIc and c7A IId).
To further elucidate the impact of 3- and 7-deazapurines
on base pairing, we investigated the thermodynamic base
pairing properties of 3- and 7-deazapurines in short palin-
dromic RNAs. As mentioned above, the effects become in-
tensified because the self-pairing results in two modified
base pairs with regularly Watson-Crick–paired regions next
to them that are below the required number of base pairs
needed for successful double helix nucleation (45,46). In-
deed, for c3G the reduction in both palindromic RNAs
was dramatic, reflected in –13.5/–13.0◦C reduced Tm val-
ues (IIIa and III’a), accounting for –6.8/–6.5◦C destabiliza-
tion per single modification which is more than twice of
the destabilization that we observed for a single c3G-C base
pair in the bimolecular 9 bp duplex Ia. The destabilizing na-
ture was also very pronounced for c3A in both palindromic
RNAs (–9.8◦C in IIIb and –13.0◦C in III’b). In contrast,

4288 Nucleic Acids Research, 2021, Vol. 49, No. 8
only little destabilization was found for both 7-deazapurines
in both palindromic RNAs (c7G: –2.4◦C in IIIc and –2.1◦C
in III’c; c7A: –2.7◦C in IIId and –1.0◦C in III’d).
To the best of our knowledge, our data provides the first
insights into thermal and thermodynamic stabilities of 3-
deazaadenine and 3-deazaguanine containing RNA. Thus
far, c3G was investigated only within the context of 2ꢀ-
OCH3–RNA; c3G caused destabilization when paired to
C (–6◦C with complementary RNA; –9◦C with comple-
mentary DNA in a Type I like sequence design) (47). In-
terestingly, when an acetyl group resided at the exocyclic
amino group (N2-acetyl-3-deazaguanine) the native pairing
strength was restored (48).
Furthermore, a single paper points at the strong destabi-
lizing effect of 3-deazaguanine on DNA base pairing (36).
In contrast, in several early studies, the slight destabiliz-
ing effect of 3-deazaadenine within DNA double helices
has been recognized (31,32,35,36). 7-Deazaguanine and 7-
deazaadenine modified DNAs have been investigated most
intensively with respect to base pairing stabilities; gener-
ally, they act slightly destabilizing compared to their native
counterparts (29,30,34–38).
NMR spectroscopy of 3-deazapurine modified RNA
The NMR resonances for the hydrogen-bonded protons of
Watson–Crick base-pairs (‘imino protons’) directly reflect
the double helical segments within folded RNA. The chem-
ical shifts of these signals are characteristic for A-U (>14
ppm) and C-G base pairs (∼12–13 ppm), and the linewidths
reflect proton exchange with the solvent. Clearly, they are
very sensitive to modifications, in particular, if they concern
the nucleobases. Figure 3A depicts 1H NMR spectra of the
palindromic duplex 5ꢀ-GGUCGACC III that were collected
as a function of temperature, over the range 5–25◦C. The
1
signal assignment was based on 2D H/1H-NOESY spec-
troscopy (Supplementary Figure S17). With increasing tem-
perature, the signal of the base pair G1-C becomes broader
and loses intensity, consistent with base pair fraying of the
duplex termini; the internal base pairs remain and reflect
stable duplex formation at ambient temperatures.
Figure 3. Comparative 1H NMR imino proton spectra of (A) unmodified,
(B) c3G-, and (C) c3A-modified oligoribonucleotides; conditions: cRNA
0.1 mM, 15 mM Na[AsO2(CH3)2]•3H2O, 25 mM NaCl, 3 mM NaN3, in
H2O/D2O 9/1, pH 6.5; (D) the graph illustrates the imino proton exchange
rates determined by 1D CLEANEX-PM experiments (see Supporting In-
formation for details).
=
In comparison, the c3G imino proton resonance of the
modified duplex IIIa was shifted highfield (Figure 3B), con-
sistent with the increased pKa of c3G (compared to G) of
almost 3 pKa units (Table 2). Interestingly, the neighboring
U3-A base pair senses the perturbation caused by c3G, re-
flected in a decrease in intensity of the U3 imino proton sig-
nal at higher temperature. The enhanced exchange with the
solvent indicates a dynamic hot spot of base pair opening
at this particular position.
Table 2. pKa values of nucleobases and deaza derivativesa
A•H+ (N1)
c3A•H+ (N1)
c7A•H+ (N1)
U (N3)
pKa
pKa
3.7
G (N1)
9.5
6.8
c3G (N1)b
12.3
5.3
c7G (N1)
10.3
9.2
C•H+ (N3)
4.1
Concerning the c3A-modified duplex IIIb, the thermal
1
destabilization is reflected in the imino proton H NMR
aValues are summarized from references 7 and 51. Atom position of
protonation/deprotonation are provided in brackets.
bFor an independent NMR-based pKa determination of c3G see Supple-
mentary Figure S22.
spectra as well (Figure 3C), with a tiny signal barrow at
5◦C in the typical ppm region of A–U base pairs; no signal
was observed at higher temperature. This stands for inten-
sive proton exchange with the solvent of the uridine N3-H
in the c3A–U base pair and is reminiscent to 1H NMR spec-
troscopy of 5F–U–A base pairs (52).
To investigate these effects in more detail, we quanti-
fied the water imino proton exchange kinetics of a c3G–
C base pair containing duplex (IIIa) by the application of
1D Phase-Modulated CLEAN Chemical EXchange Spec-
troscopy (CLEANEX-PM) NMR experiments (Figure 3D,
and Supplementary Figure S18) (53–55). Interestingly, the
largest exchange rate k(H2O ex.) of 1.68 s−1 was observed for

Nucleic Acids Research, 2021, Vol. 49, No. 8 4289
the A-U base pair next to the c3G–C modification. Also for
the c3G–C base pair itself the rate k(H2O ex.) was increased
compared to the unmodified duplex (1.11 s−1 versus 0.76
s
−1), while the G–C base pair two positions away almost
the same exchange rates k(H2O ex.) for modified and unmod-
ified duplexes were observed (0.72 s−1 versus 0.62 s−1). This
is consistent with a local structural perturbation induced by
the 3-deazapurine hotspot.
Crystal structures of a c3A containing RNA
To further shed light on 3-deazapurine containing base
pairs in RNA we put efforts into a high resolution X-
ray crystallographic analysis. We utilized the 27 nt frag-
ment of the E. coli 23S rRNA sarcin−ricin loop (SRL)
which is a robust and well-behaved crystallization scaf-
fold (56,57). For the incorporation of c3A, we deemed nu-
cleotide A2670 appropriate, which forms a Watson−Crick
base pair with C2650 in the regular A-form double helical
region. Crystallization trials with c3A-modified SRL RNA
indeed provided crystals that diffracted to subatomic res-
olution (Supplementary Table S3). X-ray structure deter-
mination showed that the c3A nucleobase is well-defined in
the electron density maps for the c3A-modified RNA (Fig-
ure 4). Superimpositions of the c3A-modified RNA struc-
ture with the unmodified RNA showed a root-mean-square
˚
deviation (rmsd) of 0.12 A (within the errors on coordi-
˚
nates of 0.13 A). Overall, the structure revealed that the 3-
deazanucleobase does not significantly affect the base pair
geometry and the SRL RNA fold. However, what is dif-
ferent is the local hydration network in the minor groove
of the c3A–U base pair (Figure 4A). While three ordered
water molecules are found in the minor groove for the un-
modified base pair, four are observed for the modified base
pair. An obvious distinction is that the N3 atom of A2670
Figure 4. X-ray structure of 3-deazaadenosine modified RNA at atomic
resolution. (A) Secondary structure of the E. coli Sarcin-ricin stem-loop
(SRL) RNA used for crystallization. The c3A nucleotide is labeled in red.
(B) Side view on the U2650/A2670 base pair in the unmodified duplex with
ordered water molecules in the minor groove highlighted in beige color
(PDB ID 3DVZ). (C) Side view on the U2650/c3A2670 base pair with or-
dered water molecules in the minor groove highlighted in red color (PDB
ID 7L3R). (D) 2Fobs – Fcalc electron density map contoured at 1.5 ␴ level
showing the U2650/c3A2670 base pair. Water molecules are shown as red
spheres (PDB ID 7L3R) and are superposed to the water molecules ob-
served in the unmodified RNA (PDB ID 3DVZ). Numbers are distances
˚
is hydrogen-bonded to one of the water molecules (2.8 A
distance) while for the c3A2670/U2650 base pair, a contin-
uous hydrogen-bonded water chain spans from the 2ꢀ-OH
of c3A2670 to the 2ꢀ-OH of U2650 without involving the
3
˚
c A nucleobase; the closest water is in 3.3 A-distance from
˚
in Angstro¨m (A). We note that the alternative conformations of U2650
its C3 atom (Figure 4D, Supplementary Figure S19). Al-
though these structural differences in the hydration pattern
appear minor in the static X-ray structure, they may con-
tribute to increased base pair opening dynamics as reflected
in the imino proton NMR spectrum of the c3A modified
SRL RNA by significant signal broadening (Supplementary
Figure S20). The destabilizing effect of the c3A modifica-
tion in the SRL RNA hairpin is further reflected in the UV-
melting profiles measured under diverse conditions (Sup-
plementary Figure S21).
(and C2649, G2648) are also observed in the unmodified SRL RNA scaf-
fold (PDB ID 3DVZ).
deazapurine) are of potential origin for these differences in
thermal stabilities and thermodynamic parameters (30).
It has been proposed that the pKa of a nucleobase is one
significant factor that contributes to base-pairing strength
(7,58–60). Table 2 summarizes the pKa values of the nucle-
obases investigated in this study, and the following trend be-
comes obvious: The pKa decreases in the order of A•H+ <
c7A•H+ < c3A•H+, and the same is true for G < c7G < c3G.
This trend is consistent with the notion that 7-deazapurines
are less destabilizing than 3-deazapurines compared to A
and G in standard Watson-Crick pairs in RNA double he-
lices (Table 1). We furthermore note that c3A with a pKa
of 6.8 is partially protonated at physiological pH values
around 7.0 which may affect base pairing in addition.
Another aspect that was documented earlier is that natu-
ral Watson–Crick base pairs A–U and C–G are character-
ized by a pKa gap between the H-donor (U, G; N3-H, N1-
Towards a rationale for the destabilization by c3G and c3A
For all four modifications––c3G, c7G and c3A, c7A––the
base pair geometry and hydrogen bonding pattern to C and
A respectively, are the same as encountered for the native
Watson–Crick G–C and A–U base pairs. Therefore, other
factors such as altered nucleobase dipole moments that can
affect nucleobase stacking (58), or altered pKa values of the
modified nucleobases (7,59–61), as well as altered hydra-
tion of minor groove (3-deazapurines) and major groove (7-

4290 Nucleic Acids Research, 2021, Vol. 49, No. 8
of hydrated Mg2+ ions that critically participate in proton
transfer to release the 5ꢀ-O leaving group in the course of
phosphodiester cleavage (15,16,68,69). Recent prominent
examples refer to the NAD+ class-I riboswitch (68) and the
pistol ribozyme (15,16). In pistol RNA, a highly conserved
guanine (G33) interacts with a Mg2+ cation through inner-
sphere coordination with its N7 atom. Replacement of this
guanine by c7G renders the ribozyme inactive (15). Together
with distance analysis of pistol ribozyme crystal structures,
this suggests that a water molecule of the hydrated Mg2+ ion
assists in stabilizing the 5ꢀ-O leaving group during the course
of the reaction (15,16). In the NAD+ class-I riboswitch, a
highly conserved adenine (A10) exhibits N7-innersphere co-
ordination to a Mg2+ cation that additionally coordinates to
the oxygen atom of a sequence-distant phosphate backbone
unit and thus molds a critical part of the binding pocket.
Replacement of this adenine by c7A abolishes binding to
NAD+ (68).
H) and the H-acceptor (A, C; N1, N3) of ꢀpKa 5.5 and 5.4,
respectively, which appears optimal for efficient base pair-
ing (7,58). This gap becomes significantly altered for isos-
teric c7A–U (ꢀpKa 3.9), c3A–U (ꢀpKa 2.4), c7G–C (ꢀpKa
6.2) and c3G–C (ꢀpKa 8.2) deazapurine base pairs.
Clearly, caution has to be taken because many more fac-
tors in addition to the pKa of the heterocycles (as listed
above) contribute to macroscopically observed double he-
lix stabilities. This might be the reason why concepts for
predicting hydrogen-bond strengths from acid−base molec-
ular properties (e.g. proton affinity/pKa equalization, pKa
slide rule, etc.) cannot be directly applied to nucleic acid
base pairing strengths, although being promising for an in-
tegrated theoretical and experimental approach to solve this
long-standing problem in the future (62).
Finally, we point at the hydration of RNA grooves which
is distinct between 3-deazapurines (c3G and c3A) affecting
the minor groove, and 7-deazapurines (c7G and c7A) affect-
ing the major groove. In particular, the first solved crystal
structure of a c3A modified RNA and the comparison to the
unmodified counterpart reveal that the characteristic water
molecule that always docks on the purine N3 atom and of-
ten spans further to the ribose 2ꢀ-OH in native RNA double
helices (63,64), is lacking in the 3-deaza-modified counter-
parts. The altered hydration pattern is a clear structural dis-
tinction that should be taken into account when searching
for the concurrent causes of the decrease in stability.
With respect to the ribosome, atomic mutagenesis involv-
ing 3-deazaadenine in the ribosomal peptidyl transferase
center (PTC) demonstrated that an initial suspect (N3 of
A2451) did not actively participate in proton transfer to cat-
alyze peptide bond formation, but instead, the 2ꢀ-OH of the
very adenosine assists in positioning of the substrate (19;
see also reference 17,18,20,70).
For atomic mutagenesis experiments using deazanucle-
obases, a profound knowledge of the intrinsic properties
of these heterocycles in an RNA context is imperative to
avoid misinterpretations. Functional assays to evaluate ri-
bozyme or riboswitch activity in response to a specific
purine-deazapurine replacement are usually designed to im-
pair a potential H-bond or innersphere coordination to a
divalent metal ion, as suggested by the observed distances
in the three-dimensional structure. Provided such an inter-
action is crucial for activity, it can thus be easily identi-
fied and the mechanistic interpretation becomes straight-
forward. Importantly, in all of the examples mentioned
above, the purine/deazapurine finds itself in an exposed
and unpaired position. Clearly, in the light of our present
study, functional assays that would involve atomic muta-
genesis with 3-deazanucleobases in double helical regions
would encounter severe limitations because of thermody-
namic destabilization of the stem containing the modifica-
tion. This aspect is, however, less critical for atomic mutage-
nesis using 7-deazanucleobases which do not or only mini-
mally destabilize double helices.
Role of deazapurines in RNA atomic mutagenesis studies
Deazapurine modified RNAs have been frequently applied
in biochemical mutagenesis studies of ribozymes to verify
or falsify mechanistic proposals for the chemical reactions
they catalyze (65,66). Several examples are found in the lit-
erature where atomic mutagenesis experiments led to an in-
depth understanding of the chemical mechanism, revealing
the functionally crucial imino groups of purines that partic-
ipate in general acid-base catalysis of small nucleolytic ri-
bozymes for phosphodiester cleavage (14–16). For instance,
this concerns the twister ribozyme where proton transfer
from the (protonated) N3 of the conserved adenine (A6) at
the cleavage site to the 5ꢀ-O leaving group had been postu-
lated to contribute to reaction catalysis (i.e. ␦-catalysis; for
explanation of ␣, ß, ␥, ␦-catalysis see reference 67). Indeed,
replacement of this adenine by c3A or c1c3A rendered the
twister ribozyme inactive (14,67). A follow-up NMR spec-
troscopic analysis revealed a pKa shift of the very adenine
N3 from 3.7 to 5.1 (14). As a second example, the pistol
ribozyme is pointed out. At first, distance analysis and in-
terpretation of the three-dimensional structure of this ri-
bozyme implicated the N3 of a conserved purine (A32 or
G32, respectively) as a possible candidate for general acid–
base catalysis, being involved in leaving group stabilization
(␦-catalysis). However, upon replacement of this purine by
c3A, the ribozyme retained full activity. Later, it was found
that the 2ꢀ-OH of this purine nucleoside plays a crucial role,
and not the N3 (15).
CONCLUDING REMARKS
In this study, we synthesized and comprehensively analyzed
the properties of 3-deazaguanine and 3-deazaadenine mod-
ified RNA. When these nucleobases are embedded in a dou-
ble helical environment we found that they can decrease the
stability of base pairing. Interestingly, the effects were much
more pronounced for 3-deazapurines compared to the iso-
meric 7-deazapurines. Moreover, destabilization is more se-
vere in short stems interfering with double helix nucleation,
while a more modest decrease is observed if located in the
center of an extended double helix. We furthermore found
that the imino proton exchange for deazapurine-modified
Watson Crick base pairs is significantly higher compared
Additionally, deaza-nucleobase substitutions have identi-
fied crucial metal binding sites that are responsible for struc-
turing of riboswitch aptamers and ribozyme active sites
(13,67), as well as have disclosed key coordination sites

Nucleic Acids Research, 2021, Vol. 49, No. 8 4291
7. Krishnamurthy,R. (2012) Role of pKa of nucleobases in the origins
of chemical evolution. Acc. Chem. Res., 45, 2035–2044.
8. Acharya,P., Cheruku,P., Chatterjee,S., Acharya,S. and
Chattopadhyaya,J. (2004) Measurement of nucleobase pKa values in
model mononucleotides shows RNA−RNA duplexes to be more
stable than DNA−DNA duplexes. J. Am. Chem. Soc., 126,
2862–2869.
9. Bande,O., Braddick,D., Agnello,S., Jang,M., Pezo,V., Schepers,G.,
Rozenski,J., Lescrinier,E., Marlie`re,P. and Herdewijn,P. (2016) Base
pairing involving artificial bases in vitro and in vivo. Chem. Sci., 7,
995–1010.
10. Fuchs,E., Falschlunger,C., Micura,R. and Breuker,K. (2019) The
effect of adenine protonation on RNA phosphodiester backbone
bond cleavage elucidated by deaza-nucleobase modifications and
mass spectrometry. Nucleic Acids Res., 47, 7223–7234.
11. Spitale,R.C., Volpini,R., Heller,M.G., Krucinska,J., Cristalli,G. and
Wedekind,J.E. (2009) Identification of an imino group indispensable
for cleavage by a small ribozyme. J. Am. Chem. Soc., 131, 6093–6095.
12. Spitale,R.C., Volpini,R., Mungillo,M.V., Krucinska,J., Cristalli,G.
and Wedekind,J.E. (2009) Single-atom imino substitutions at A9 and
A10 reveal distinct effects on the fold and function of the hairpin
ribozyme catalytic core. Biochemistry, 48, 7777–7779.
13. Zheng,L., Falschlunger,C., Huang,K., Mairhofer,E., Yuan,S.,
Wang,J., Patel,D.J., Micura,R. and Ren,A. (2019) Hatchet ribozyme
structure and implications for cleavage mechanism. Proc. Natl. Acad.
Sci. U.S.A., 116, 10783–10791.
14. Kosˇutic´,M., Neuner,S., Ren,A., Flu¨r,S., Wunderlich,C.,
Mairhofer,E., Vusˇurovic´,N., Seikowski,J., Breuker,K., Ho¨bartner,C.
et al. (2015) A mini-twister variant and impact of residues/cations on
the phosphodiester cleavage of this ribozyme class. Angew. Chem. Int.
Ed., 54, 15128–15133.
15. Neuner,S., Falschlunger,C., Fuchs,E., Himmelstoss,M., Ren,A.,
Patel,D.J. and Micura,R. (2017) Atom-specific mutagenesis reveals
structural and catalytic roles for an active-site adenosine and hydrated
Mg2+ in pistol ribozymes. Angew. Chem. Int. Ed., 56, 15954–15958.
16. Teplova,M., Falschlunger,C., Krasheninina,O., Egger,M., Ren,A.,
Patel,D.J. and Micura,R. (2020) Crucial roles of two hydrated Mg2+
ions in reaction catalysis of the pistol ribozyme. Angew. Chem. Int.
Ed., 59, 2837–2843.
to their native counterparts and simultaneously induces en-
hanced opening of the very neighbor base pair. These per-
turbations in local structural dynamics provide a rationale
for the observed reduced base pairing strength. Addition-
ally, we solved the X-ray structure of a c3A modified RNA
to visualize the hydration pattern of the minor groove. The
structure points out that the hydrophobic minor grove face
of c3A impairs hydration because the water molecule that
is usually hydrogen bonded to the N3 atom becomes ex-
truded from the 3-deaza-modified base pair. Taken together,
our comparative study sheds light on Watson–Crick base
pairing of 3- and 7-deazapurines and thus represents a first
important step towards a comprehensive understanding of
the intrinsic properties of these RNA modifications that are
critical for their proper application in atomic mutagenesis
experiments of biologically relevant RNA.
DATA AVAILABILITY
Atomic coordinates and structure factors for the reported
crystal structures have been deposited with the Protein Data
bank under accession numbers 7L3R.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
ACKNOWLEDGEMENTS
We thank Jonas Kottersteger for synthetic contributions,
Sarah Heel for assistance in FT-ICR mass spectrometry,
and Daniel Fellner (Innsbruck) for technical support.
17. Erlacher,M.D., Lang,K., Shankaran,N., Wotzel,B., Hu¨ttenhofer,A.,
Micura,R., Mankin,A.S. and Polacek,N. (2005) Chemical engineering
of the peptidyl transferase center reveals an important role of the
2ꢀ-hydroxyl group of A2451. Nucleic. Acids. Res., 33, 1618–1627.
18. Erlacher,M.D., Lang,K., Wotzel,B., Rieder,R., Micura,R. and
Polacek,N. (2006) Efficient ribosomal peptidyl transfer critically relies
on the presence of the ribose 2ꢀ-OH at A2451 of 23S rRNA. J. Am.
Chem. Soc., 128, 4453–4459.
19. Lang,K., Erlacher,M., Wilson,D.N., Micura,R. and Polacek,N.
(2008) The role of 23S ribosomal RNA residue A2451 in peptide bond
synthesis revealed by atomic mutagenesis. Chem. Biol., 15, 485–492.
20. Polikanov,Y.S., Steitz,T.A. and Innis,C.A. (2014) A proton wire to
couple aminoacyl-tRNA accommodation and peptide-bond
formation on the ribosome. Nat. Struct. Mol. Biol., 21, 787–793.
21. Das,S.R. and Piccirilli,J.A. (2005) General acid catalysis by the
hepatitis delta virus ribozyme. Nat. Chem. Biol., 1, 45–52.
22. Seela,F., Debalak,H., Andrews,L. and Beigelman,L. (2003) Synthesis
and enzymic hydrolysis of oligoribonucleotides incorporating
3-deazaguanosine: the importance of the nitrogen-3 atom of single
conserved guanosine residues on the catalytic activity of the
hammerhead ribozyme. Helv. Chim. Acta, 86, 2726–2740.
23. Mairhofer,E., Flemmich,L., Kreutz,C. and Micura,R. (2019) Access
to 3-deazaguanosine building blocks for RNA solid-phase synthesis
involving Hartwig-Buchwald C-N cross-coupling. Org. Lett., 21,
3900–3903.
FUNDING
Austrian Science Fund FWF [P31691, F8011-B to R.M.,
P32773 to C.K., P30087 to K.B.]; Austrian Research Pro-
motion Agency FFG [West Austrian BioNMR 858017].
Funding for open access charge: Austrian Science Fund
FWF.
Conflict of interest statement. None declared.
REFERENCES
1. Polacek,N. (2013) Atomic mutagenesis of the ribosome: towards a
molecular understanding of translation. Chimia – Int. J. Chem., 67,
322–326.
2. Hoernes,T.P, Clementi,N., Juen,M.A., Shi,X., Faserl,K., Willi,J.,
Gasser,C., Kreutz,C., Joseph,S., Lindner,H. et al. (2018) Atomic
mutagenesis of stop codon nucleotides reveals the chemical
prerequisites for release factor-mediated peptide release. Proc. Natl.
Acad. Sci. U.S.A., 115, E382–E389.
3. Hu,T., Suter,S.R., Mumbleau,M.M. and Beal,P.A. (2018) TLR8
activation and inhibition by guanosine analogs in RNA: Importance
of functional groups and chain length. Bioorg. Med. Chem., 26, 77–83.
4. Forconi,M., Benz-Moy,T., Gleitsman,K.R., Ruben,E., Metz,C. and
Herschlag,D. (2012) Exploring purine N7 interactions via atomic
mutagenesis: the group I ribozyme as a case study. RNA, 18,
1222–1229.
5. Seela,F., Debelak,H., Usman,N., Burgin,A. and Beigelman,L. (1998)
1-Deazaadenosine: synthesis and activity of base-modified
hammerhead ribozymes. Nucleic Acids Res., 26, 1010–1018.
6. Kapinos,L.E., Operschall,B.P., Larsen,E. and Sigel,H. (2011)
Understanding the acid-base properties of adenosine: the intrinsic
basicities of N1, N3 and N7. Chem. Eur. J., 17, 8156–8164.
24. Perl´ıkova´,P. and Hocek,M. (2017) Pyrrolo[2, 3-d]pyrimidine
(7-deazapurine) as a privileged scaffold in design of antitumor and
antiviral nucleosides. Medicinal Res. Rev., 37, 1429–1460.
25. Richardson,K.E. and Znosko,B.M. (2016) Nearest-neighbor
parameters for 7-deaza-adenosine·uridine base pairs in RNA
duplexes. RNA, 22, 934–942.
26. Ingale,S.A., Leonard,P. and Seela,F. (2018) Glycosylation of
pyrrolo[2,3-d]pyrimidines with 1-O-acetyl-2, 3,
5-tri-O-benzoyl-␤-d-ribofuranose: substituents and protecting groups

4292 Nucleic Acids Research, 2021, Vol. 49, No. 8
effecting the synthesis of 7-deazapurine ribonucleosides. J. Org.
Chem., 83, 8589–8595.
27. Seela,F., Mersmann,K., Grasby,J.A. and Gait,M.J. (1993)
7-Deazaadenosine – oligoribonucleotide building-block synthesis and
autocatalytic hydrolysis of base-modified hammerhead ribozymes.
Helv. Chim. Acta, 76, 1809–1820.
28. Mairhofer,E., Fuchs,E. and Micura,R. (2016) Facile synthesis of a
3-deazaadenosine phosphoramidite for RNA solid-phase synthesis.
Beilstein J. Org. Chem., 12, 2556–2562.
45. Saenger,W. (1984) In: Principles of Nucleic Acid Structure, Springer,
Berlin.
46. Majlessi,M. and Becker,M.M. (2008) Formation of the double helix:
a mutational study. Nucleic Acids Res., 36, 2981–2989.
47. Seio,K., Sasami,T., Ohkubo,A., Ando,K. and Sekine,M. (2007)
Highly selective recognition of cytosine over uracil and adenine by a
guanine analogue, 2-N-acetyl-3-deazaguanine, in
2ꢀ-O-methyl-RNA/RNA and DNA duplexes. J. Am. Chem. Soc., 129,
1026–1027.
29. Gold,B., Stone,M.P. and Marky,L.A. (2014) Looking for Waldo: a
potential thermodynamic signature to DNA damage. Acc. Chem.
Res., 47, 1446–1454.
30. Kowal,E.A., Ganguly,M., Pallan,P.S., Marky,L.A., Gold,B., Egli,M.
and Stone,M.P. (2011) Altering the electrostatic potential in the
major groove: thermodynamic and structural characterization of
7-deaza-2ꢀ-deoxyadenosine:dT base pairing in DNA. J. Phys. Chem.
B., 115, 13925–13934.
31. Ganguly,M., Wang,R.W., Marky,L.A. and Gold,B. (2010)
Thermodynamic characterization of DNA with 3-deazaadenine and
3-methyl-3-deazaadenine substitutions: the effect of placing a
hydrophobic group in the minor groove of DNA. J. Phys. Chem. B.,
114, 7656–7661.
32. Lever,C., Li,X., Cosstick,R., Edel,S. and Brown,T. (1993)
Thermodynamic stability and drug-binding properties of
oligodeoxyribonucleotide duplexes containing
48. Seio,K., Sasami,T., Tawarada,R. and Sekine,M. (2006) Synthesis of
2ꢀ-O-methyl-RNAs incorporating a 3-deazaguanine, and UV melting
and computational studies on its hybridization properties. Nucleic
Acids Res., 34, 4324–4334.
49. Marky,L.A. and Breslauer,K.J. (1987) Calculating thermodynamic
data for transitions of any molecularity from equilibrium melting
curves. Biopolymers, 26, 1601−1620.
50. Petersheim,M. and Turner,D.H. (1983) Base-stacking and
base-pairing contributions to helix stability: thermodynamics of
double-helix formation with CCGG, CCGGp, CCGGAp, ACCGGp,
CCGGUp, and ACCGGUp. Biochemistry, 22, 256−263.
51. Siraiwa,S., Suzuki,A., Katoh,R. and Saito,Y. (2016) Design and
synthesis of a novel fluorescent benzo[g]imidazo[4, 5-c]quinoline
nucleoside for monitoring base-pair-induced protonation with
cytosine: distinguishing cytosine via changes in the intensity and
wavelength of fluorescence. Org. Biomol. Chem., 14, 3934–3942.
52. Puffer,B., Kreutz,C., Rieder,U., Ebert,M.O., Konrat,R. and
Micura,R. (2009) 5-Fluoro pyrimidines: labels to probe DNA and
RNA secondary structures by 1D 19F NMR spectroscopy. Nucleic.
Acids. Res., 37, 7728–7740.
3-deazaadenine:thymine base pairs. Nucleic Acids Res., 21,
1743–1746.
33. Cosstick,R., Li,X., Tuli,D.K., Williams,D.M., Connolly,B.A. and
Newman,P.C. (1990) Molecular recognition in the minor groove of
the DNA helix. Studies on the synthesis of oligonucleotides and
polynucleotides containing 3-deaza-2ꢀ-deoxyadenosine. Interaction of
the oligonucleotides with the restriction endonuclease EcoRV. Nucleic
Acids Res., 18, 4771–4778.
34. Minakawa,N., Kawano,Y., Murata,S., Inoue,N. and Matsuda,A.
(2008) Oligodeoxynucleotides containing 3-bromo-3-deazaadenine
and 7-bromo-7-deazaadenine 2ꢀ-deoxynucleosides as chemical probes
to investigate DNA-protein interactions. ChemBioChem, 9, 464–470.
35. Seela,F. and Grein,T. (1992) 7-Deaza-2ꢀ-deoxyadenosine and
3-deaza-2ꢀ-deoxyadenosine replacing dA within d(A6)-tracts:
differential bending at 3ꢀ- and 5ꢀ-junctions of d(A6).d(T6) and
B-DNA. Nucleic Acids Res., 20, 2297–2306.
36. Grein,T., Lampe,S., Mersmann,K., Rosemeyer,H., Thomas,H. and
Seela,F. (1994) 3-Deazapurines and 7-deazapurines - duplex stability
of oligonucleotides containing modified adenine or guanine bases.
Bioorg. Med. Chem. Lett., 4, 971–976.
37. Ganguly,M., Wang,F., Kaushik,M., Stone,M.P., Marky,L.A. and
Gold,B. (2007) A study of 7-deaza-2ꢀ-deoxyguanosine
2ꢀ-deoxycytidine base pairing in DNA. Nucleic Acids Res., 35,
6181–6195.
53. Strebitzer,E., Rangadurai,A., Plangger,R., Kremser,J., Juen,M.A.,
Tollinger,M., Al-Hashimi,H.M. and Kreutz,C. (2018) 5-Oxyacetic
acid modification destabilizes double helical stem structures and
favors anionic Watson-Crick like cmo5U-G base pairs. Chem. Eur. J.,
24, 18903–18906.
54. Hwang,T.-L., van Zijl,P.C.M. and Mori,S. (1998) Accurate
Quantitation of Water-amide Proton Exchange Rates Using the
Phase-Modulated CLEAN Chemical EXchange (CLEANEX-PM)
Approach with a Fast-HSQC (FHSQC) Detection Scheme. J. Biomol.
NMR, 11, 221–226.
55. Gue´ron,M. and Leroy,J.L. (1995) Studies of base pair kinetics by
NMR measurement of proton exchange. Methods Enzymol., 261,
383–413.
56. Olieric,V., Rieder,U., Lang,K., Serganov,A., Schulze-Briese,C.,
Micura,R., Dumas,P. and Ennifar,E.,2009, A fast selenium
derivatization strategy for crystallization and phasing of RNA
structures. RNA, 15, 707–715.
57. Correll,C.C., Wool,I.G. and Munishkin,A. (1999) The two faces of
the Escherichia coli 23 S rRNA sarcin/ricin domain: the structure at
1.11 A resolution. J. Mol. Biol., 292, 275–287.
38. Ganguly,M., Wang,R.-W., Marky,L.A. and Gold,B. (2009)
Introduction of cationic charge into DNA near the major groove edge
of a guanine cytosine base pair: characterization of
58. Sponer,J., Leszczynski,J. and Hobza,P. (2001) Electronic properties,
hydrogen bonding, stacking, and cation binding of DNA and RNA
bases. Biopolymers, 61, 3–31.
59. Krishnamurthy,R. (2009) A search for structural alternatives of
RNA. J. Mex. Chem. Soc., 53, 22–33.
60. Thaplyal,P. and Bevilacqua,P.C. (2014) Experimental approaches for
measuring pKa’s in RNA and DNA. Methods Enzymol., 549,
189–219.
61. Acharya,P., Cheruku,P., Chatterjee,S., Acharya,S. and
Chattopadhyaya,J. (2014) The measurement of nucleobase pKa of the
model mononucleotides shows why RNA-RNA duplex is more stable
than DNA-DNA duplex. J. Am. Chem. Soc., 126, 2862–2869.
62. Gilli,P., Pretto,L., Bertolasi,V. and Gilli,G. (2009) Predicting
hydrogen-bond strengths from acid−base molecular properties. The
pKa slide rule: toward the solution of a long-lasting problem. Acc.
Chem. Res., 42, 33–44.
oligodeoxynucleotides substituted with
7-aminomethyl-7-deaza-2ꢀ-deoxyguanosine. J. Am. Chem. Soc., 131,
12068–12069.
39. Hakimelahi,G.H., Proba,Z.A. and Ogilvie,K.K. (1981) Nitrate ion as
catalyst for selective silylations of nucleosides. Tetrahedron Lett., 22,
4775–4778.
40. Serebryany,V. and Beigelman,L. (2002) An efficient preparation of
protected ribonucleosides for phosphoramidite RNA synthesis.
Tetrahedron Lett., 43, 1983–1985.
41. Micura,R. (2002) Small interfering RNAs and their chemical
synthesis. Angew. Chem. Int. Ed., 41, 2265–2269.
42. Glasner,H., Riml,C., Micura,R. and Breuker,K. (2017) Label-free,
direct localization and relative quantitation of the RNA nucleobase
methylations m6A, m5C, m3U, and m5U by top-down mass
spectrometry. Nucleic Acids Res., 45, 8014–8025.
43. Crey-Desbiolles,C., Lhomme,J., Dumy,P. and Kotera,M. (2004)
3-Nitro-3-deaza-2ꢀ-deoxyadenosine as a versatile photocleavable
2ꢀ-deoxyadenosine mimic. J. Am. Chem. Soc., 126, 9532–9533.
44. Maiti,M., Persoons,L., Andrei,G., Snoeck,R., Balzarini,J. and
Herdewijn,P. (2013) Synthesis and anti-herpetic activity of
phosphoramidate ProTides. Chem. Med. Chem., 8, 985–993.
63. Wahl,M.C., Rao,S.T. and Sundaralingam,M. (1996) The structure of
r(UUCGCG) has a 5ꢀ-UU-overhang exhibiting Hoogsteen-like trans
U • U base pairs. Nat. Struct. Biol., 3, 24–31.
64. Adamiak,D.A., Milecki,J., Adamiak,R.W. and Rypniewski,W. (2010)
The hydration and unusual hydrogen bonding in the crystal structure
of an RNA duplex containing alternating CG base pairs. New J.
Chem., 34, 903–909.

Nucleic Acids Research, 2021, Vol. 49, No. 8 4293
65. Micura,R. and Ho¨bartner,C. (2020) Fundamental studies of
functional nucleic acids: aptamers, riboswitches, ribozymes and
DNAzymes. Chem. Soc. Rev., 49, 7331–7353.
66. Scheitl,C.P.M, Ghaem Maghami,M., Lenz,A.K. and Ho¨bartner,C.
(2020) Site-specific RNA methylation by a methyltransferase
ribozyme. Nature, 587, 663–667.
67. Gebetsberger,J. and Micura,R. (2017) Unwinding the twister
ribozyme: from structure to mechanism. Wiley Interdiscip. Rev. RNA,
8, e1402.
68. Chen,H., Egger,M., Xu,X., Flemmich,L., Krasheninina,O., Sun,A.,
Micura,R. and Ren,A. (2020) Structural distinctions between NAD+
riboswitch domains 1 and 2 determine differential folding and ligand
binding. Nucleic Acids Res., 48, 12394–12406.
70. Erlacher,M.D., Chirkova,A., Voegele,P. and Polacek,N. (2011)
Generation of chemically engineered ribosomes for atomic
mutagenesis studies on protein biosynthesis. Nat. Protoc., 6, 580–592.
71. Kabsch,W. (2010) Integration, scaling, space-group assignment and
post-refinement. Acta Cryst., D66, 133–144.
72. Vagin,A. and Teplyakov,A. (2010) Molecular replacement with
MOLREP. Acta Cryst., D66, 22–25.
73. Afonine,P.V., Grosse-Kunstleve,R.W., Echols,N., Head,J.J.,
Moriarty,N.W., Mustyakimov,M., Terwilliger,T.C., Urzhumtsev,A.,
Zwart,P.H and Adams,P.D. (2012) Towards automated
crystallographic structure refinement with phenix.refine. Acta Cryst.,
D68, 352–367.
74. Emsley,P., Lohkamp,B., Scott,W.G. and Cowtan,K. (2010) Features
and development of Coot. Acta Cryst., D66, 486–501.
69. Fuchs,E., Falschlunger,C., Micura,R. and Breuker,K. (2019) The
effect of adenine protonation on RNA phosphodiester backbone
bond cleavage elucidated by deaza-nucleobase modifications and
mass spectrometry. Nucleic Acids Res., 47, 7223–7234.