Site-specific labeling of RNA at internal ribose hydroxyl groups: Terbium-assisted deoxyribozymes at work

Article abstract of DOI:10.1021/ja503864v

A general and efficient single-step method was established for site-specific post-transcriptional labeling of RNA. Using Tb³⁺ as accelerating cofactor for deoxyribozymes, various labeled guanosines were site-specifically attached to 2′-OH groups of internal adenosines in in vitro transcribed RNA. The DNA-catalyzed 2′,5′-phosphodiester bond formation proceeded efficiently with fluorescent, spin-labeled, biotinylated, or cross-linker-modified guanosine triphosphates. The sequence context of the labeling site was systematically analyzed by mutating the nucleotides flanking the targeted adenosine. Labeling of adenosines in a purine-rich environment showed the fastest reactions and highest yields. Overall, practically useful yields >70% were obtained for 13 out of 16 possible nucleotide (nt) combinations. Using this approach, we demonstrate preparative labeling under mild conditions for up to ～160-nt-long RNAs, including spliceosomal U6 small nuclear RNA and a cyclic-di-AMP binding riboswitch RNA.

Full text of DOI:10.1021/ja503864v

Article
pubs.acs.org/JACS
Site-Specific Labeling of RNA at Internal Ribose Hydroxyl Groups:
Terbium-Assisted Deoxyribozymes at Work
Lea Buttner, Fatemeh Javadi-Zarnaghi,† and Claudia Hobartner*
̈
̈
Research Group Nucleic Acid Chemistry, Max Planck Institute for Biophysical Chemistry, Am Fassberg 11, 37077 Gottingen,
̈
Germany
S
* Supporting Information
ABSTRACT: A general and efficient single-step method was
established for site-specific post-transcriptional labeling of
RNA. Using Tb3+ as accelerating cofactor for deoxyribozymes,
various labeled guanosines were site-specifically attached to 2′-
OH groups of internal adenosines in in vitro transcribed RNA.
The DNA-catalyzed 2′,5′-phosphodiester bond formation
proceeded efficiently with fluorescent, spin-labeled, biotiny-
lated, or cross-linker-modified guanosine triphosphates. The
sequence context of the labeling site was systematically
analyzed by mutating the nucleotides flanking the targeted
adenosine. Labeling of adenosines in a purine-rich environ-
ment showed the fastest reactions and highest yields. Overall, practically useful yields >70% were obtained for 13 out of 16
possible nucleotide (nt) combinations. Using this approach, we demonstrate preparative labeling under mild conditions for up to
∼160-nt-long RNAs, including spliceosomal U6 small nuclear RNA and a cyclic-di-AMP binding riboswitch RNA.
and Silverman prepared a labeled tagging RNA by in vitro
transcription using 5-aminoallyl-CTP and subsequent labeling
by N-hydroxysuccinimide chemistry. This tagging RNA was
then used as substrate for DNA-catalyzed ligation to the target
RNA, forming labeled 2′,5′-branched RNA. Despite the
elegance of this approach, it has not been widely used in the
RNA community. Limitations include the elaborate preparation
of labeled tagging RNAs, as well as their inconvenient lengths
of 8−17 nt, which could potentially interfere with RNA folding
or functions.
In this contribution, we report the direct post-transcriptional
installation of fluorescent, biotinylated, paramagnetic, and
photoreactive mononucleotides at specific ribose 2′-OH groups
within functional in vitro transcribed RNA. Key to this versatile
approach is the use of deoxyribozymes as functional single-
stranded DNA catalysts that have proven useful as tools for
various applications in nucleic acids research.16−18 A particular
deoxyribozyme, named 10DM24, originally developed by
Silverman and co-workers for the synthesis of 2′,5′-branched
RNA,19 was earlier shown to accept mononucleotides as
substrates for ligation to adenosine branch sites.20 We
harnessed this catalytic ability of 10DM24 and established a
mild and efficient labeling technique for RNA using chemically
modified guanosine triphosphates (Scheme 1). A strategically
positioned cytidine within the catalytic core of 10DM24 serves
as a recognition nucleotide for guanosine triphosphates by
Watson−Crick base-pairing at the center of a three-helix
INTRODUCTION
■
The site-specific attachment of labels onto RNA is of significant
value in many areas of RNA chemical biology. The study of
RNA folding and RNA−protein and RNA−ligand interactions
by spectroscopic methods requires installation of observable
reporter groups. Biochemical assays of RNA functions benefit
from position-specific placement of reactive functional groups
or probing agents.1 Emissive nucleotide (nt) analogues or
abiotic functional groups for bioorthogonal labeling can be site-
specifically installed by solid-phase synthesis.2−4 Synthesis of
RNAs longer than ∼40 nt often involves enzymatic
manipulation or ligation using T4 ligases or deoxyribo-
zymes.5−9 As alternatives to the fragment-based approach,
direct post-transcriptional labeling methods are gaining
increasing attention. Recent developments include the use of
specific methyltransferases together with functionalized S-
adenosylmethionine (SAM) analogues, as exemplified by the
alkylation of a specific guanosine in tRNAPhe with Trm1.10
More generally applicable may be the reconstitution of C/D
box small nucleolar ribonucleoprotein particles (snoRNPs),
using guide RNAs responsible for the selection of the labeling
site in the target RNA, as recently shown for the site-specific 2′-
alkynylation of tRNA and mRNA.11 A non-enzymatic example
of oligonucleotide-guided post-transcriptional RNA modifica-
tion was introduced with the functionality transfer reaction
from 6-thioguanosine in a donor DNA to the exocyclic amino
groups of cytosine and guanine residues in the target
RNA.12−14 Particularly attractive was the combination of a
guide and an enzyme in a single strand of DNA, introduced as
“DNA-catalyzed labeling of RNA”.15 In this approach, Baum
Received: April 17, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 1. DNA-Catalyzed Labeling of RNA
a
Installation of bioorthogonal functional groups (azides or primary
amines), spin labels, cross-linkers, affinity reagents, or fluorescent dyes
using indicated GTP derivatives. The binding arms of the DNA
enzyme select the target site in the RNA via hybridization (orange,
binding arms; green, catalytic core; gray, third stem of 3HJ with RΔ).
The recognition nucleotide for GTP binding is explicitly indicated. A
new phosphodiester bond is formed between the 2′-OH of the target
adenosine and the 5′-phosphate of the labeled guanosine nucleotide
upon incubation with Mg2+ and Tb3+ at pH 7.5. Various GTP
derivatives used in this study are shown. TEMPO = 2,2,6,6-
tetramethylpiperidinoxyl, MANT = methylanthraniloyl, BP =
benzophenone. Structures of fluorophores Cy3, Cy5, and F545 are
shown in Figure S1.
junction (3HJ) structure.20 The target nucleoside in the RNA
substrate is selected via hybridization of the DNA enzyme’s
binding arms (see Figure S2 for sequence information). In the
presence of metal ion cofactors, a 2′,5′-phosphodiester bond is
formed upon nucleophilic attack of the adenosine 2′-OH group
on the α phosphate of the labeled GTP and release of
pyrophosphate. In contrast to the earlier DNA-catalyzed
approach based on ligation of 17-nt-long fluorescently labeled
tagging RNAs,15 our current method benefits from the small
size of the labeled mononucleotide and the use of commercially
available labeled GTP analogues. Capitalizing on our recent
finding of lanthanides as accelerating cofactors for RNA-ligating
deoxyribozymes,21 we now report highly beneficial effects of
Tb3+ for DNA-catalyzed labeling of RNA. This finding was
essential in establishing a general protocol for preparative
labeling of RNA that works efficiently at pH 7.5 and at practical
concentrations of precious labeled GTP derivatives.
Figure 1. (a) Autoradiographs of PAGE analysis of DNA-catalyzed
ligation of GTP and its 2′-modified analogues to RNA R1 with 2 mM
GTPs and 80 mM Mg2+ at pH 9.0. Reaction products are indicated by
arrowheads. (b) PAGE analysis of CuAAC derivatization of N3-G-
labeled R1, using 5 equiv of BPA or fluorescent F545-alkyne, 0.4 mM
CuBr, 0.8 mM TBTA in DMSO/tBuOH/H2O. (c) Accelerating effect
of Tb3+: nucleotide concentration-dependent kinetics of DNA-
catalyzed GTP ligation to R1 at pH 7.5, in the absence (red, empty
symbols) or presence (green, filled symbols) of 100 μM Tb3+. Note
the faster reactions at lower GTP concentrations with Tb3+. (d)
Establishing generality of DNA-catalyzed labeling for 16 variants of R1.
DNA enzymes maintained Watson−Crick base-pairing with the target
RNA upstream and downstream of the labeling site. (e) Ligation yields
for 16 target RNAs after 10 min reaction time under indicated
conditions. PAGE = polyacrylamide gel electrophoresis, TBTA =
tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine, TP = triphosphate.
RESULTS AND DISCUSSION
■
Exploring the Substrate Scope and Establishing Mild
Reaction Conditions for DNA-Catalyzed Labeling of
RNA. Implementing deoxyribozymes as versatile tools for RNA
labeling requires knowledge about acceptable modifications in
the substrates to maintain high ligation efficiency. We tested
various functional group mutations in the GTP substrate and
found that ribose modifications, such as 2′-azido-, 2′-amino-,
and 2′/3′-aminoethylcarbamoyl (EDA) functional groups, were
well tolerated in the DNA-catalyzed reaction for labeling of
RNA. Figure 1a shows the time course of the reactions for RNA
R1 (for sequence see the Supporting Information) with 2 mM
GTP and 2′-modified analogues, resulting in quantitative
labeling within 1−2 h at 37 °C. This successful DNA-catalyzed
reaction of ribose-modified GTP derivatives allows the
installation of bioorthogonal functional groups that are
normally not present in RNA. Fluorophores, spin labels,
cross-linkers, and other structural probing agents can then be
attached via well-known bioconjugation reactions, such as
amide bond formation with active esters, or 3+2 cycloaddition
B
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reactions, such as Cu(I)-catalyzed azide−alkyne cycloaddition
(CuAAC). We have demonstrated the viability of this approach
with 2′-N3-G-modified RNA which was derivatized via CuAAC
using TEMPO- or benzophenone-modified alkynes (TA, BPA)
or the commercially available alkyne of fluorophore F545
(Figure 1b). The cycloadditions were performed with TBTA-
complexed Cu(I) instead of in situ generated Cu(I) which
requires, e.g., ascorbate as reducing agent. Thus, the integrity of
the nitroxide radical and other sensitive reporter groups was
maintained (see Supporting Information for details and
additional examples for CuAAC conjugation of 2′-N3-G-labeled
RNA, Figure S3). These results encouraged us to expand the
labeling approach to larger RNA targets. However, incubation
at pH 9.0 in the presence of Mg2+ raises the general concern of
RNA degradation, and using millimolar concentrations of
modified GTP analogues may be prohibitive for economic
reasons.22
To expand the scope and applicability of DNA-catalyzed
labeling of RNA, it is highly desirable to employ lower
concentrations of labeled GTP at neutral pH. Therefore, we
analyzed the influence of reaction conditions on rate and yield.
Decreasing the pH from 9.0 to 7.5 reduced the ligation rate 4−
6-fold, and lowering the GTP concentration 10-fold from 2
mM to 200 μM caused a further decrease of the ligation rate by
a factor of ∼5, resulting in rather slow reactions, which yielded
60−80% ligation product after 5 h incubation (Table 1).
than 10 min (Figure S5). Several nucleobase-modified GTP
analogues that were potentially suitable for further bioconju-
gation were also tested, including 8-oxo-GTP, 6-thio-GTP and
8-(6-aminohexylamino)-GTP (8-AHA-G), but all of those
derivatives reacted rather slowly and gave poorer yields (Figure
S6).
Within the original sequence context of RNA R1 for which
10DM24 was selected, the reaction is most efficient for ligation
at 2′-OH of adenosine.19 As the allowed sequence context
around the adenosine target site has not been thoroughly
investigated,23 we systematically explored adenosines flanked by
all 16 possible dinucleotide combinations and examined the
efficiency of GTP ligation at the target nucleosides (Figure 1d).
Watson−Crick base-pairing with the DNA binding arms was
maintained for each substrate (see Supporting Information for
sequences). Figure 1e depicts the labeling yields for all 16 RNA
substrates after 10 min reaction time, using either 3 mM GTP
at pH 9 with 100 mM Mg2+, or 200 μM GTP at pH 7.5 with 80
mM Mg2+ and 100 μM Tb3+ (kinetic data are given in Table
S2).
Several interesting conclusions were drawn from these
experiments. First, the ligation efficiency at high pH and high
GTP concentration was strongly sequence-dependent. Second,
the efficiency for all RNA substrates was strongly improved by
adding Tb3+ and reducing the GTP concentration. Under these
conditions, labeling was fast and high-yielding (80% in 10 min)
for adenosines that were followed by a purine in the 3′-
direction. For adenosines followed by a pyrimidine nucleotide,
ligation yields were very good when the 5′-neighbor was
guanosine but lower for other 5′-neighbors. Overall, 9 out of 16
substrates yielded >70% labeled RNA in 10 min, and four
additional RNAs yielded >70% labeled product in 2 h (Figure
S7). Summarizing these results, -NAR- and -GAY- are the
preferred sequence contexts for target adenosines for DNA-
catalyzed labeling with 10DM24 (N = any nucleotide, R =
purine, Y = pyrimidine). For preparative reactions, reliable
conditions for various target RNAs and 2′-modified GTPs were
found using RNA concentrations of 5−20 μM, with as little as
4−5 equiv of labeled GTP. The corresponding deoxyribozyme
in complex with RΔ was used in stoichiometric ratio with
respect to target RNA. Preparative incubation was performed at
37 °C for 1−2 h, or at 22 °C for 3−5 h.
Table 1. pH and Concentration Dependence of kobs for
DNA-Catalyzed Ligation of GTP, 2′-N3-GTP, 2′-NH2-GTP,
and 2′/3′-EDA-GTP
kobs (min−1
)
[NTP]
(mM)
[Tb3+]
(μM)
N3-
GTP
NH2-
GTP
EDA-
GTP
a
pH
GTP
9.0
7.5
7.5
7.5
7.5
7.5
2
0
0
0.11
0.016
0.003
0.001
0.67
1.2
0.13
0.025
0.004
0.002
0.053
0.009
0.002
2
0.034
0.007
0.2
0.05
0.2
0.05
0
0
b
b
b
100
100
∼1
∼1
∼1
a
pH 7.5, 50 mM HEPES; pH 9.0, 50 mM CHES. Constant conditions
for all experiments in this table: 150 mM NaCl, 2 mM KCl, 80 mM
b
Mg2+, 37 °C. >50% product after 30 s; >90% in 5 min.
Having established mild and economic reaction conditions,
we examined the reaction of GTP analogues with larger ribose
modifications that would allow fluorescent labels or affinity tags
to be directly installed on the RNA target in a single DNA-
catalyzed reaction step. We tested a series of commercially
available ribose-labeled GTP analogues and found that, e.g., 2′/
3′-MANT-GTP and 2′/3′-EDA-linked biotin- or Cy3-modified
GTP gave high yields of labeled RNA (Figures 2a and S8).
Encouraged by these results, we synthesized various labeled
GTP analogues by CuAAC, using 2′-N3-GTP and propargy-
lated derivatives of TEMPO (TA, serving as spin label) or
benzophenone (BPA, to be used as UV-inducible cross-linker),
as well as commercial alkynes of fluorescent dyes, including
C488, F545, Cy3, and Cy5 analogues. The labeled GTPs were
purified by RP-HPLC (Figure S9) and then used in the DNA-
catalyzed labeling reactions with Tb3+ as cofactor (Figure 2b).
All “click”-labeled GTPs were efficient substrates and produced
the site-specifically labeled RNAs in high yields with fast
reaction rates (kobs ≈ 1 min−1, Figure S10). Using 4 equiv of
labeled GTPs with 100 μM Tb3+ and 20 mM Mg2+ at pH 7.5
resulted in >80% labeled RNA within 30 min for most labeled
Inspired by our recent finding that lanthanide ions can
impressively enhance the ligation rates for the DNA-catalyzed
synthesis of 2′,5′-branched RNA,21 we explored the effect of
Tb3+ on the activity of 10DM24 for single-nucleotide ligation.
All reactions with Tb3+ were performed at pH 7.5, and it was
found that Tb3+ drastically accelerated the reaction. The
magnitude of the rate acceleration was dependent on the GTP
concentration (Figure 1c). While addition of 100 μM Tb3+
caused only 2-fold faster reaction with 5 mM GTP, the reaction
was 200-fold faster with 200 μM GTP, and 1000-fold faster
with 50 μM GTP (Table 1, see also Figure S4). At first sight,
this effect of faster rates at lower GTP concentrations may seem
counter-intuitive, but it can be rationalized by considering the
affinity of Tb3+ for phosphates (i.e., the effective concentration
of Tb3+ is likely reduced due to triphosphate coordination at
high GTP concentrations). Importantly, similar acceleration
effects with Tb3+ were observed for 2′-modified GTP
derivatives (Table 1); for example, 2′-N3-GTP, 2′-NH2-GTP,
and 2′/3′-EDA-GTP produced >90% labeled RNA R1 in less
C
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. DNA-catalyzed labeling of RNA with ribose-modified GTPs.
(a) Using commercial carbamoyl-linked MANT and biotin-labeled
GTP in comparison to unmodified GTP and EDA-GTP. RNA R1 was
used at 5 μM with 20 μM GTP (or labeled GTP), at pH 7.5, with 100
μM Tb3+, 37 °C, 30 min. Gel images compare labeling yields obtained
with 80 mM Mg2+ (left) and 20 mM Mg2+ (right). (b) Labeled GTP
derivatives synthesized by CuAAC (see Supporting Information for
details) were used under the same conditions as in panel a. Below the
autoradiograph is shown an overlay of fluorescence images (epi
illumination with blue, green, and red LEDs).
Figure 3. DNA-catalyzed labeling of RNA and characterization by
FRET. (a) Sequence of 46-bp duplex with label positions indicated.
Overlay of fluorescence images of PAGE for purification of labeled
RNAs: Cy3 is green, Cy5 is red. (b) UV−vis absorption spectra. (c)
UV melting curves. Conditions for panels b and c: 0.4 μM duplex in 10
mM potassium phosphate pH 7.0, 150 mM NaCl. (d) Normalized
fluorescence emission spectra upon donor excitation (at 545 nm, 0.1
μM RNA, 50 mM KMOPS pH 7.5, 100 mM KCl, 20 °C).
GTP analogues (Figure 2). Reactions were 2−4-fold faster with
80 mM Mg2+, but high metal ion concentrations might be
suboptimal for long RNA targets. Below 20 mM Mg2+ the
reactions were too slow to be of practical use (Table S3).
Labeled RNAs were isolated and characterized by ESI-MS,
thermal melting, and EPR spectroscopy (Table S4, Figure S11).
DNA-Catalyzed Labeling of RNA for FRET: Confirming
the Innocence of 2′-Attached Guanosine Labels. To
demonstrate the generality of this labeling strategy, we
examined the influence of 2′-labeling on stability, folding, and
function of various RNAs. As a first example we discuss a 46-bp
RNA duplex that was labeled with two fluorophores, Cy3-G
and Cy5-G, that form a well-known FRET pair. Cy3-G was
installed at two different positions, A16 or A24 of strand A, and
Cy5-G was installed at A10′ of the complementary strand B
(Figure 3a). Labeling efficiency was >85%, and isolated yields
were 50−62% (4 nmol of RNA). Labeled and unlabeled strands
A and B were used to prepare five duplex samples (I−V),
containing none (I), one (II, III), or two labels (IV, V). The
double-labeled samples contained the donor and acceptor
fluorophores in opposite strands, separated by 13 or 21 bp.
HPLC and UV−vis spectrophotometry confirmed an equi-
molar ratio of quantitatively labeled strands (Figures 3 and
S12). Thermal melting experiments revealed that the labeled
guanosines linked to the 2′-OH of adenosine branch sites
caused only minor destabilization of the duplex. Single-labeled
duplexes had the Tm lowered by 1.5−2 °C, and the Tm of
double-labeled duplexes dropped by less than 4 °C. As
expected, the FRET efficiency was higher for duplex IV (Cy3
and Cy5 separated by 13 bp) than for duplex V (Cy3 and Cy5
separated by 21 bp) (Figure 3d).
Figure 4. (a) DNA-catalyzed labeling of SAM-II RNA and analysis of
pseudoknot folding by FRET. (b) Images of PAGE analysis of
unlabeled, single-labeled, and double-labeled RNAs with C488-G at
A13 and F545-G at A41. Conditions: 10% PAGE, 35 W, 2.5 h. (c)
Mg2+-dependent folding monitored by FRET in the absence and
presence of 10 μM SAM. Conditions: 0.1 μM RNA, 50 mM KMOPS
pH 7.5, 100 mM KCl, 20 °C. IDA = fluorescence emission intensity at
520 nm of double-labeled RNA, ID = emission intensity at 520 nm of
donor-labeled RNA; excitation at 496 nm. [Mg2+]1/2 (+SAM) = 1.0
mM; [Mg2+]1/2 (−SAM) = 1.4 mM.
providing a solid basis for testing our RNA labeling method. To
examine the influence of 2′-attached labeled guanosines on
ligand binding, we used a previously described strategy based
on 2-aminopurine (2AP) fluorescence.26 Changes in 2AP
fluorescence upon titration with SAM provide a measure for
ligand affinity. Labeled SAM-II RNAs were prepared with 2AP
on position 14 and a guanosine label at the 2′-OH group of
A41. Alternatively, 2AP was introduced at position 41 and
As a second example we chose to study a ligand-binding
RNA. The 52-nt aptamer domain of the SAM-II riboswitch
folds into a pseudoknot structure upon binding SAM in the
presence of Mg2+ (Figure 4a).24,25 Structure and folding of
SAM-II have been well studied by various techniques,24−27
D
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
guanosine attached to A9 or A13 (Figure S13), and
fluorescence changes were monitored in SAM titration
experiments. The binding constants ranged from 0.4 to 2.1
μM (Figure S14) and were consistent with reported data,24,26
confirming that the guanosine labels do not interfere with
ligand binding. To demonstrate SAM-II RNA folding by FRET,
we installed C488-G at 2′-OH of A13 and F545-G at A41 by
two successive DNA-catalyzed labeling reactions. PAGE
analysis confirmed the presence of both labels, by the
detectable reduced mobility and by the overlapping fluo-
rescence signal of both chromophores (Figure 4b). The FRET
efficiency was monitored upon Mg2+-induced folding of the
pseudoknot structure in the presence and absence of SAM
(Figures 4c and S15). The observed left-shifted [Mg2+]1/2 value
in the presence of SAM was expected and consistent with
reported data,28 confirming the intact functionality of the
FRET-labeled RNA.
Direct DNA-Catalyzed Labeling of Large RNA Tran-
scripts. Many interesting functional RNAs are longer than the
∼50 nt RNA discussed in the previous section. We chose two
examples of longer non-coding RNAs and examined DNA-
catalyzed attachment of fluorescent reporter groups. The 120-
nt-long spliceosomal U6 snRNA29 was labeled with the
fluorescent reporter MANT-G at 10 positions (Figure 5a, b),
which were selected on the basis of the sequence preferences of
target adenosines. Analysis of labeling efficiencies becomes
more difficult for such larger targets since labeled and unlabeled
RNAs have equal electrophoretic mobilities. We demonstrate
that labeling yields can easily be determined upon site-specific
cleavage into shorter fragments using the well-known RNA-
cleaving deoxyribozymes 8−17 or 10−23 (Figures 5c and
S16).16 Labeling of U6 snRNA at any targeted position
between A40 and A56 was analyzed by cleavage of an analytical
aliquot of each labeled RNA at nts 35 and 62, resulting in
labeled 27-mers, which can easily be distinguished from the
unlabeled 27-mer upon denaturing PAGE separation (Figure
5d). Similarly, labeling between A62 and A94 was analyzed after
cleavage at positions 51 and 97, resulting in labeled 46-mers
which were separated and quantified (Figure 5e). The observed
labeling efficiencies correlated well with the predictions based
on the sequence context. For example, A51 which is embedded
between guanosines was almost quantitatively labeled, while
A41, residing in an unfavorable sequence context (AAC), was
poorly labeled. Overall, six tested sites were labeled with >75%,
two reached 40−50%, and three sites did not give preparatively
useful yields. The result for A76 (which was predicted to be a
good labeling site on the basis of its sequence context AAG)
showed only 15% MANT-G attached. This hints toward the
influence of strong secondary structure elements that limit the
accessibility for deoxyribozyme hybridization. Using strategi-
cally positioned disruptor oligonucleotides, double-stranded
secondary structures can be broken to improve the labeling
efficiency for difficult sites.
Figure 5. DNA-catalyzed labeling of large RNAs. (a) U6 snRNA
sequence; adenosines tested as labeling sites are marked in blue. (b)
PAGE image of purification of 10 MANT-G-labeled U6 snRNA
samples (10% PAGE). (c) Schematic of DNA-catalyzed RNA cleavage
(cut) for analysis of labeling efficiency. (d) Analysis after cut i at A35
and A62. (e) Analysis after cut ii at A51 and A97. For d and e, 15%
PAGE, full gel image in Figure S17). (f) ydaO riboswitch RNA
sequence. Cy3 is installed at A46 and Cy5 at A125. (g) Fluorescence
images of PAGE analysis of uncut and cut ydaO RNA samples; 15%
PAGE, corresponding autoradiography gel image in Figure S18.
Another structured RNA addressed in this work is the 156-
nt-long ydaO riboswitch RNA (Figure 5f), which was recently
reported to bind cyclic-di-AMP with sub-nanomolar affin-
ity.30,31 Details about folding and dynamics of this RNA upon
ligand binding are not yet known. Single-molecule FRET of
double-labeled RNAs has proven to be a highly valuable
method to study riboswitch RNAs.32−34 To prepare labeled
derivatives for spectroscopic studies, we targeted the DNA-
catalyzed installation of a FRET pair on the ydaO RNA. As an
example, we chose A46 and A125 as target sites for labeling
with Cy3- and Cy5-GTP, according to the established labeling
conditions described above. A DNA oligonucleotide comple-
E
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
by solid-phase synthesis or by in vitro transcription according to the
procedures previously described and commonly used in our
laboratory.36 The DNA template for transcription of ydaO RNA was
prepared by overlapping primer extension. A pUC18 plasmid
containing U6 snRNA sequence from S. cerevisiae was kindly provided
mentary to nts 57−89 was included as disruptor of secondary
structures to facilitate hybridization of the labeling deoxy-
ribozymes. Samples with only donor or only acceptor dye were
prepared by a single DNA-catalyzed labeling step. For
installation of the second dye, the single-labeled RNA was
first isolated by PAGE and then subjected to a second labeling
reaction with the other fluorophore. By splitting the single-
labeled RNAs for several reactions, this stepwise approach
allows parallel synthesis of a library of double-labeled RNAs for
screening several donor−acceptor distances. The fluorescence
scans of the polyacrylamide gel analysis showed overlapping
emission bands in the red and green channels, confirming the
attachment of both dyes in the double-labeled sample. To
determine the efficiency of labeling at each site, analytical
aliquots were cut by RNA-cleaving deoxyribozymes at positions
17 and 60 (cut iii), resulting in a Cy3-labeled 43-mer and a
Cy5-labeled 94-mer, or at positions 96 and 143 (cut iv), giving
a Cy3-labeled 98-mer and a Cy5-labeled 47-mer (Figure 5g).
Efficiency of Cy3-G labeling at A46 was 60%, and Cy5-G was
attached at A125 with 38% yield (Figure S18). In an approach
similar to the analysis by RNA-cleaving deoxyribozymes,
labeled transcripts can be enriched by digestion of the
unlabeled RNA due to the remaining accessibility of the 2′-
OH for transesterification.35
These results demonstrate that various fluorescent labels can
be conjugated to long RNAs with the DNA-catalyzed approach
developed in this work. With the ability to attach biotinylated
guanosine in an analogous fashion, this method will make
immobilizable FRET-labeled RNAs for single-molecule studies
accessible from long unmodified transcripts. Labeling reactions
addressing different target sites can easily be performed in
parallel, providing rapid access to a collection of fluorescent
RNAs for probing RNA folding landscapes. In addition, the
deoxyribozymes can be isolated upon PAGE purification of the
labeled RNAs and reused in subsequent reactions.
by Prof. Reinhard Luhrmann (MPI for Biophysical Chemistry).
̈
Modified GTP analogues used in this study were purchased from
TriLink Biotechnologies or Jena Bioscience.
Synthesis of Labeled GTPs by CuAAC. A solution of 2′-N3-GTP
(100 nmol, 1 equiv) in 24 μL of H2O and 10 μL of DMSO/tBuOH
3:1 was mixed with 3 μL of alkyne solution (100 mM; 3 equiv) and
combined with a freshly prepared mixture of CuBr (1 μL of 88 mM in
DMSO/tBuOH 3:1; 0.9 equiv) and TBTA (2 μL of 88 mM in
DMSO/tBuOH 3:1; 1.7 equiv) to give a final volume of 40 μL.
Alkynes were synthesized (TA, BPA, see Supporting Information) or
purchased from Jena Bioscience or Baseclick. The solution was
incubated at room temperature for 4 h and precipitated by addition of
1.2 mL of ice-cold 2% NaClO4 in acetone, followed by centrifugation
at 4 °C, 13 200 rpm for 30 min. The crude sample was dissolved in
water and purified by reversed-phase HPLC on a Nucleosil 100-5 C18
HD column at 40 °C (buffer A, 100 mM TEAA in H2O; buffer B, 20
mM TEAA in H2O:MeCN 1:4). The isolated yield ranged from 45 to
76%. The products were characterized by ESI-MS (see Table S1).
Kinetic Assays of DNA-Catalyzed RNA Labeling Reactions.
The 5′-32P-labeled target RNA (2 pmol), the respective deoxyribo-
zyme, and RΔ were mixed in a 1:5:10 ratio and annealed in 25 mM
HEPES pH 7.5, 15 mM NaCl, and 0.1 mM EDTA (95 °C, 2 min,
slowly cooling to room temperature, 15 min). The assays were
performed in 10 μL volume at 37 °C in 50 mM HEPES pH 7.5 or in
50 mM CHES pH 9.0, including 150 mM NaCl and 2 mM KCl. For
screening of reaction conditions, GTP was used at concentrations of
50 μM to 5 mM, and MgCl2 at 3−80 mM. Reactions in the presence
of Tb3+ contained 100 μM TbCl3. Kinetic assays with labeled GTP
analogues used 50−200 μM nucleotide and 100 μM TbCl3. Aliquots of
0.8−1.0 μL were removed at desired time points and quenched into
stop solution (80% formamide, 1× TBE, 50 mM EDTA, 0.025% each
bromophenol blue and xylene cyanol). The samples were analyzed by
PAGE (20% polyacrylamide), and band intensities were quantified by
PhosphorImaging. The yield versus time data were fit to yield (%) =
Y(1 − e−kt), where k = kobs and Y = final yield.
General Method for Labeling of RNA with GTP Analogues
and Terbium-Assisted 10DM24 Deoxyribozymes. The target
RNA (0.5−4 nmol), the respective deoxyribozyme, and RΔ were
mixed in a 1:1:1.5 ratio and annealed (as for kinetic assays). Next, 4
equiv of GTP analogue with respect to target RNA was added, and the
reaction was performed at room temperature or at 37 °C for 3−5 h in
50 mM HEPES pH 7.5, 150 mM NaCl, 2 mM KCl, 100 μM TbCl3,
and 20 or 80 mM MgCl2. The final concentration of target RNA was 5,
10, or 20 μM, and the final concentration of labeled GTP was
therefore 20, 40, or 80 μM. Nucleic acids were precipitated with
ethanol and separated by denaturing PAGE. Isolated yields of labeled
RNAs ranged from 25 to 62%. The deoxyribozymes were re-isolated
and used for subsequent reactions.
CONCLUSIONS
■
In summary, we have established a general RNA labeling
method using ribose-modified guanosine triphosphates as
labeling reagents and Tb3+-assisted DNA catalysts to install
fluorophores, spin labels, and cross-linkers at 2′-OH groups of
internal adenosines in in vitro transcribed RNA. Systematic
investigation of labeling efficiencies resulted in guidelines for
selecting labeling sites on the basis of sequence context and
secondary structure. This approach will be widely applicable for
labeling diverse RNAs for structural and functional studies. For
example, we envision labeling of mRNAs for single-molecule
FRET studies of RNA splicing and cross-linking experiments to
examine various non-coding RNAs and their interactions with
proteins. We point out that this approach holds the potential
for orthogonal DNA-catalyzed labeling of RNA, as engineering
of the recognition nucleotide will switch the specificity to other
labeled nucleotides, based on Watson−Crick base-pairing.20
For example, a donor-labeled ATP and an acceptor-labeled
GTP can then be used simultaneously with the corresponding
DNA enzymes in a one-pot FRET labeling reaction. In
addition, new deoxyribozymes for labeling of RNA target
sites other than adenosine will further broaden the generality of
the method.
Thermal Melting Analysis of RNA Duplexes. UV melting
curves were recorded and analyzed as previously described.8 Duplex
concentrations were 0.4 or 2 μM in 10 mM sodium phosphate pH 7
with 150 mM NaCl.
Analysis of SAM-II RNA: Ligand Binding and Pseudoknot
Folding Monitored by 2AP Fluorescence and FRET. Fluores-
cence titration experiments for measuring KD,app were performed
essentially as previously described.1,26 For 2AP fluorescence:
excitation, 308 nm; emission area integration, 325−500 nm. For
FRET with donor C488 and acceptor F545: excitation, 496 nm; donor
emission, 520 nm; acceptor emission, 585 nm. FRET efficiency was
calculated as E = 1 − IDA/ID, with IDA the donor emission intensity of
double-labeled RNA and ID the donor emission intensity of donor-
only-labeled RNA. Further details are given in the Supporting
Information.
METHODS
■
Materials and General Methods. Deoxyribozymes were
purchased as unmodified DNA oligonucleotides. RNAs were prepared
Analysis of Labeling Efficiency by RNA-Cleaving Deoxy-
ribozymes. Analytical aliquots of fluorescently labeled U6 or ydaO
F
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
RNA (containing traces of 32P body-labeled transcripts) isolated from
DNA-catalyzed labeling reactions were annealed with 2-fold excess of
the respective RNA-cleaving deoxyribozyme (10−23 or 8−17 DNAs,
see Supporting Information for sequences) and incubated with 20 mM
MgCl2 and 20 mM MnCl2 in 1× reaction buffer (40 mM Tris HCl pH
7.5, 150 mM NaCl) for 1.5 h at 37 °C. The samples were analyzed by
denaturing PAGE, and the RNA fragments were visualized by
PhosphorImaging and fluorescence imaging.
(16) Silverman, S. K.; Baum, D. A. Methods Enzymol. 2009, 469, 95.
(17) Willner, I.; Shlyahovsky, B.; Zayats, M.; Willner, B. Chem. Soc.
Rev. 2008, 37, 1153.
(18) Schlosser, K.; Li, Y. Chem. Biol. 2009, 16, 311.
(19) Zelin, E.; Wang, Y.; Silverman, S. K. Biochemistry 2006, 45,
2767.
(20) Hobartner, C.; Silverman, S. K. Angew. Chem., Int. Ed. 2007, 46,
̈
7420.
(21) Javadi-Zarnaghi, F.; Hobartner, C. J. Am. Chem. Soc. 2013, 135,
̈
12839.
ASSOCIATED CONTENT
■
(22) Most commercial labeled GTP analogues are quite expensive,
and often available only at low concentrations (e.g., at 1 mM).
(23) In an earlier study, Baum and Silverman used 10DM24 for
ligation of a 17-nt-long tagging RNA at 10 adenosine target sites
within the P4-P6 RNA of a group I intron.15
S
* Supporting Information
Additional information as noted in the text, including Figures
S1−S18 and Tables S1−S4; sequence information on all RNA
and DNA oligonucleotides used in this work; and methods for
chemical synthesis of TA and BPA. This material is available
free of charge via the Internet at http://pubs.acs.org.
(24) Gilbert, S. D.; Rambo, R. P.; Van Tyne, D.; Batey, R. T. Nat.
Struct. Mol. Biol. 2008, 15, 177.
(25) Corbino, K. A.; Barrick, J. E.; Lim, J.; Welz, R.; Tucker, B. J.;
Puskarz, I.; Mandal, M.; Rudnick, N. D.; Breaker, R. R. Genome Biol.
2005, 6, R70.
(26) Haller, A.; Rieder, U.; Aigner, M.; Blanchard, S. C.; Micura, R.
Nat. Chem. Biol. 2011, 7, 393.
AUTHOR INFORMATION
Corresponding Author
claudia.hoebartner@mpibpc.mpg.de
■
Present Address
(27) Lim, J.; Winkler, W. C.; Nakamura, S.; Scott, V.; Breaker, R. R.
Angew. Chem., Int. Ed. 2006, 45, 964.
(28) Chen, B.; Zuo, X.; Wang, Y. X.; Dayie, T. K. Nucleic Acids Res.
2012, 40, 3117.
†Department of Biology, Faculty of Sciences, University of
Isfahan, Iran
Notes
(29) Karaduman, R.; Dube, P.; Stark, H.; Fabrizio, P.; Kastner, B.;
The authors declare no competing financial interest.
Luhrmann, R. RNA 2008, 14, 2528.
̈
(30) Nelson, J. W.; Sudarsan, N.; Furukawa, K.; Weinberg, Z.; Wang,
ACKNOWLEDGMENTS
L.B. acknowledges funding by an excellence fellowship from the
Gottingen Graduate School for Neurosciences, Biophysics, and
J. X.; Breaker, R. R. Nat. Chem. Biol. 2013, 9, 834.
■
(31) The ydaO RNA was earlier shown to bind ATP with millimolar
affinity: Watson, P. Y.; Fedor, M. J. Nat. Chem. Biol. 2012, 8, 963.
(32) Haller, A.; Altman, R. B.; Souliere, M. F.; Blanchard, S. C.;
Micura, R. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 4188.
(33) Souliere, M. F.; Altman, R. B.; Schwarz, V.; Haller, A.;
Blanchard, S. C.; Micura, R. Proc. Natl. Acad. Sci. U.S.A. 2013, 110,
E3256.
(34) Suddala, K. C.; Rinaldi, A. J.; Feng, J.; Mustoe, A. M.; Eichhorn,
C. D.; Liberman, J. A.; Wedekind, J. E.; Al-Hashimi, H. M.; Brooks, C.
L., 3rd; Walter, N. G. Nucleic Acids Res. 2013, 41, 10462.
(35) Currently, the separation of double-labeled RNA from single-
labeled and unlabeled molecules remains challenging when labeling
positions need to be selected that cannot be efficiently cleaved by
deoxyribozymes. Using 8−17 and 10−23 deoxyribozymes, only A|G
and A|U labeling sites can be cleaved with high efficiency; in our
hands, alternative deoxyribozymes16,37 for A|C and A|A junctions were
not efficient enough for long RNAs.
̈
Molecular Biosciences (GGNB, DFG GSC 226/2). This work
was supported by the Cluster of Excellence and DFG Research
Center Nanoscale Microscopy and Molecular Physiology of the
Brain (CNMPB), the International Research Training Group
Metal Sites in Biomolecules (IRTG1422), and the Max Planck
Society. We thank Uwe Pleßmann and Jurgen Bienert for
measuring ESI-MS and NMR, and Giuseppe Sicoli for
recording CW-EPR spectra.
̈
REFERENCES
■
(1) RNA Structure and Folding; Klostermeier, D., Hammann, C., Eds.;
Walter de Gruyter GmbH: Berlin, Germany, 2013.
(2) Solomatin, S.; Herschlag, D. Methods Enzymol. 2009, 469, 47.
(3) Weisbrod, S. H.; Marx, A. Chem. Commun. 2008, 44, 5675.
(4) Fauster, K.; Hartl, M.; Santner, T.; Aigner, M.; Kreutz, C.; Bister,
K.; Ennifar, E.; Micura, R. ACS Chem. Biol. 2012, 7, 581.
(5) Lang, K.; Micura, R. Nat. Protoc. 2008, 3, 1457.
(36) Wachowius, F.; Javadi-Zarnaghi, F.; Hobartner, C. Angew.
̈
Chem., Int. Ed. 2010, 49, 8504.
(37) Cruz, R. P. G.; Withers, J. B.; Li, Y. Chem. Biol. 2004, 11, 57.
(6) Winz, M. L.; Samanta, A.; Benzinger, D.; Jaschke, A. Nucleic Acids
̈
Res. 2012, 40, e78.
(7) Paredes, E.; Evans, M.; Das, S. R. Methods 2011, 54, 251.
(8) Purtha, W. E.; Coppins, R. L.; Smalley, M. K.; Silverman, S. K. J.
Am. Chem. Soc. 2005, 127, 13124.
(9) Buttner, L.; Seikowski, J.; Wawrzyniak, K.; Ochmann, A.;
̈
Hobartner, C. Bioorg. Med. Chem. 2013, 21, 6171.
̈
(10) Motorin, Y.; Burhenne, J.; Teimer, R.; Koynov, K.; Willnow, S.;
Weinhold, E.; Helm, M. Nucleic Acids Res. 2011, 39, 1943.
(11) Tomkuviene, M.; Clouet-d’Orval, B.; Cerniauskas, I.; Weinhold,
E.; Klimasauskas, S. Nucleic Acids Res. 2012, 40, 6765.
(12) Onizuka, K.; Shibata, A.; Taniguchi, Y.; Sasaki, S. Chem.
Commun. 2011, 47, 5004.
(13) Onizuka, K.; Taniguchi, Y.; Sasaki, S. Bioconjugate Chem. 2009,
20, 799.
(14) Onizuka, K.; Taniguchi, Y.; Sasaki, S. Bioconjugate Chem. 2010,
21, 1508.
(15) Baum, D. A.; Silverman, S. K. Angew. Chem., Int. Ed. 2007, 46,
3502.
G
dx.doi.org/10.1021/ja503864v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX