Structure-fluorescence activation relationships of a large Stokes shift fluorogenic RNA aptamer

Article abstract of DOI:10.1093/nar/gkz1084

The Chili RNA aptamer is a 52 nt long fluorogen-activating RNA aptamer (FLAP) that confers fluorescence to structurally diverse derivatives of fluorescent protein chromophores. A key feature of Chili is the formation of highly stable complexes with different ligands, which exhibit bright, highly Stokes-shifted fluorescence emission. In this work, we have analyzed the interactions between the Chili RNA and a family of conditionally fluorescent ligands using a variety of spectroscopic, calorimetric and biochemical techniques to reveal key structure-fluorescence activation relationships (SFARs). The ligands under investigation form two categories with emission maxima of ～540 or ～590 nm, respectively, and bind with affinities in the nanomolar to low-micromolar range. Isothermal titration calorimetry was used to elucidate the enthalpic and entropic contributions to binding affinity for a cationic ligand that is unique to the Chili aptamer. In addition to fluorescence activation, ligand binding was also observed by NMR spectroscopy, revealing characteristic signals for the formation of a G-quadruplex only upon ligand binding. These data shed light on the molecular features required and responsible for the large Stokes shift and the strong fluorescence enhancement of red and green emitting RNA-chromophore complexes.

Full text of DOI:10.1093/nar/gkz1084

Nucleic Acids Research, 2019
1
doi: 10.1093/nar/gkz1084
Structure–fluorescence activation relationships of a
large Stokes shift fluorogenic RNA aptamer
*
Christian Steinmetzger, Irene Bessi, Ann-Kathrin Lenz and Claudia H o¨ bartner
Institute of Organic Chemistry, Julius-Maximilians-University W u¨ rzburg, Am Hubland, 97074 W u¨ rzburg, Germany
Received September 13, 2019; Revised October 13, 2019; Editorial Decision October 30, 2019; Accepted November 01, 2019
ABSTRACT
tween the nucleic acid and its ligand. The different classes
of known FLAP chromophores include derivatives of 4-
hydroxybenzylidene imidazolon (HBI) (1,10,11), triphenyl-
methane dyes (12), cyanine dyes (9,13–15) and rhodamines
The Chili RNA aptamer is a 52 nt long fluorogen-
activating RNA aptamer (FLAP) that confers fluores-
cence to structurally diverse derivatives of fluores-
cent protein chromophores. A key feature of Chili is
the formation of highly stable complexes with dif-
ferent ligands, which exhibit bright, highly Stokes-
shifted fluorescence emission. In this work, we have
analyzed the interactions between the Chili RNA and
a family of conditionally fluorescent ligands using a
variety of spectroscopic, calorimetric and biochemi-
cal techniques to reveal key structure–fluorescence
activation relationships (SFARs). The ligands under
investigation form two categories with emission max-
ima of ∼540 or ∼590 nm, respectively, and bind with
affinities in the nanomolar to low-micromolar range.
Isothermal titration calorimetry was used to eluci-
date the enthalpic and entropic contributions to bind-
ing affinity for a cationic ligand that is unique to
the Chili aptamer. In addition to fluorescence acti-
vation, ligand binding was also observed by NMR
spectroscopy, revealing characteristic signals for the
formation of a G-quadruplex only upon ligand bind-
ing. These data shed light on the molecular features
required and responsible for the large Stokes shift
and the strong fluorescence enhancement of red and
green emitting RNA–chromophore complexes.
(
16,17) including silicon rhodamine (SiR) (18). Despite the
large variety of reported FLAPs, only few have been thor-
oughly characterized with respect to structure–fluorescence
activation relationships and ligand diversity (3,17,19).
The structural features responsible for fluorescence acti-
vation by several HBI-binding RNA aptamers are well doc-
umented on the basis of X-ray co-crystal structures with
their respective ligands (3,20–22). Spinach and its variants
(
Spinach2 (23), Baby Spinach (24), iSpinach (25), Broc-
coli (26)) that activate the fluorescence of 3,5-difluoro-
-hydroxybenzylidene imidazolone (DFHBI), as well as
4
Corn, a dimeric RNA aptamer for the yellow-emitting
dye 3,5-difluoro-4-hydroxybenzylidene imidazolone oxime
(DFHO) (11), adopt quadruplex folds composed of two G-
quartets and one or more mixed-sequence quartets. Spinach
contains a U–A–U base triple on top of a G-quartet
with the ligand intercalated in between the two layers
(
24,27,28). Such a base triple is absent in Corn, which forms
a homodimer that encloses the ligand at the interface of the
two individual quadruplexes (29). In both aptamers a con-
fined environment is created that decreases the nonradiative
deactivation rate of bound photoexcited HBI and results in
a significant fluorescence turn-on.
Despite
a recently reported exception (30), G-
quadruplexes are suggested to be a privileged tertiary
structure motif for fluorogen-activating aptamers (20–22)
and were also found in the family of Mango RNAs, which
activate the Thiazole orange conjugate TO1-biotin (13–15).
In Mango aptamers, three tiers of G-quartets are ar-
ranged in an unusual mixed parallel–antiparallel topology
and undergo van der Waals interactions with the entire
fluorophore-linker-biotin unit (31,32). Even though the
nature of the binding site and the mode of interaction are
unlike those found in Spinach–DFHBI and Corn–DFHO,
conceptionally the same effect of decreased nonradiative
fluorophore deactivation is achieved.
INTRODUCTION
In the wake of the seminal report of the Spinach RNA
aptamer (1), fluorogen-activating aptamers (FLAPs) have
found their footing as biochemical tools for studying RNA
in vitro and in cells. As RNA labeling tools, FLAPs sup-
plement the multitude of available fluorescent proteins
in live-cell imaging applications (2–9). While fluorescent
proteins exist as monolithic units of a fluorophore cova-
lently embedded in the protein, fluorogenic aptamers offer
the opportunity for dual engineering of chromophore and
RNA due to the non-covalent nature of the interaction be-
Apart from the core structure of the binding site and
the chromophore, ancillary nucleobases as well as ligand
side chains can also play important roles in determining the
*To whom correspondence should be addressed. Tel: +49 931 31 89693; Fax: +49 931 31 84755; Email: claudia.hoebartner@uni-wuerzburg.de
ꢀ
C
The Author(s) 2019. 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.

2
Nucleic Acids Research, 2019
fluorescence properties of the RNA–ligand complexes. For
example, quantum yield and binding affinity of Spinach2
were tuned by the introduction of trifluoromethyl groups to
the parent ligand 3,5-difluoro-4-hydroxybenzylidene imida-
zolone (DFHBI) to generate DFHBI-1T (10). In DFHO-
binding variants of Broccoli, mutation of an adenosine in
close proximity to the ligand was shown to shift the emis-
sion maximum over a range of 20 nm. This effect was pro-
posed to be caused by specific hydrogen bonding interac-
tions between the oxime moiety of DFHO and the adjacent
nucleobase. DFHBI, lacking the oxime substituent, was un-
affected by any mutation at that position (33).
Recently, the 52-nt RNA Chili has been reported to ex-
hibit Stokes shifts on the order of 130 nm when inducing
fluorescence in HBI-derived dyes. Chili is an engineered ver-
sion of the 13–2 RNA aptamer, which was in vitro-selected
to bind 4-hydroxy-3,5-dimethoxybenzylidene imidazolone
instrument settings. Background spectra for blank subtrac-
tion were obtained from samples containing no RNA. The
resulting emission spectra were integrated over the whole
peak width and the resulting intensities were normalized to
the Chili complex of DMHBI.
Fluorescence titration
A dilution series (15 steps) was prepared from a 2× solu-
tion of the Chili aptamer and the RNA was annealed as de-
scribed above. Titration samples in binding buffer were ob-
2
tained by adding a 4× solution of the dye and MgCl and
◦
were incubated at 4 C for 16 h prior to measurement. Emis-
sion spectra were collected at the optimum wavelength for
each RNA-dye combination. Background spectra for blank
subtraction were obtained from pure binding buffer. The
resulting emission spectra were integrated over the whole
peak width and then fitted to a single site binding model.
(
DMHBI, 1) (1). Compared to 13–2, Chili shows brighter
fluorescence and tighter binding with low-nanomolar affini-
ties, as well as greatly reduced Mg2+ requirements for fluo-
rescence enhancement of DMHBI and extensively modified
HBI ligands (34). Here, we present a comprehensive survey
of the performance of the Chili aptamer with various ligand
modifications and RNA point mutations. The ligand struc-
ture was altered by introduction of additional H-bonding
sites, expansion of the conjugated ␲ electron system and
addition of a cationic side chain. NMR and variable tem-
perature optical spectroscopy revealed the strictly ligand-
dependent folding of a G-quadruplex domain in the Chili
RNA. Several green and red emitting RNA–chromophore
complexes were discovered that exhibit bright fluorescence
enhancement and high binding affinity and revealed novel
structure-fluorescence activation relationships.
Isothermal titration calorimetry
A 150 ␮M stock solution of RNA was dialyzed against ul-
tra pure H O according to standard protocols. This solu-
2
tion was diluted and annealed as described above to obtain
a 300 ␮l sample containing 10 or 15 ␮M RNA in binding
buffer, which was loaded into the measurement cell of the
calorimeter. In parallel, a 100 or 150 ␮M dye sample in bind-
ing buffer was prepared in an identical manner and loaded
into the syringe of the calorimeter. The dye was titrated into
◦
the RNA at 25 C using the default protocol for 13 or 17 in-
jections. Background heat measurements for baseline cor-
rection were obtained by titrating the dye into an identically
prepared buffer containing no RNA. The data points were
fitted to a single site binding model as implemented in the
MicroCal PEAQ-ITC analysis software.
MATERIALS AND METHODS
RNA synthesis and folding
Binding kinetics
The Chili RNA aptamer and its derivatives were prepared
by in vitro transcription with T7 RNA polymerase using
synthetic DNA templates (see Supplementary Table S4 for
sequence information). RNAs were purified by denaturing
polyacrylamide gel electrophoresis, extraction and precip-
itation. The purity of the transcripts was checked by an-
ion exchange HPLC and concentrations were determined
by measuring UV absorbance at 260 nm. For RNA folding,
The RNA was annealed as described above to make a 26 nM
solution in 1.05× binding buffer. In parallel, solutions of the
dye (16, 21, 32, 42 and 53 ␮M) in water were prepared. Un-
der constant stirring, 20 ␮l of the respective dye solution
were rapidly injected into 400 ␮l of the RNA solution at
◦
2
5 C. Fluorescence time courses were monitored for 10–30
min at the optimum wavelengths for each RNA–dye com-
bination. The data points were fitted to a biexponential as-
sociation model.
◦
the aqueous RNA solution was heated to 95 C for 3 min in
◦
a buffer containing KCl and HEPES and then kept at 20 C
for 20 min before MgCl2 was added. The final composition
of the binding buffer was 125 mM KCl, 40 or 80 mM HEPES
pH 7.5 and 5 mM MgCl2.
Thermal melting analysis
◦
The RNA was heated to 95 C for 3 min in a buffer contain-
◦
ing KCl and HEPES and then kept at 20 C for 20 min be-
Fluorescence screening
+
fore adding DMHBI to make a solution containing 2 ␮M
The RNA was annealed as described above to make a 1 ␮M
solution in binding buffer. In parallel, a 1 ␮M dye solution
in binding buffer was prepared. 7.5 ␮l of each solution were
RNA, 2 ␮M ligand 125 mM KCl and 40 mM HEPES pH 7.5.
The absorbance of the solution at 260 and 295 nm was mea-
sured in a 1 cm cuvette, collecting four temperature ramps
between 10 and 95 C with a heating rate of 0.5 C/min. Af-
terwards, the fluorescence intensity (Ex/Em: 413/542 nm)
of the sample was measured in a 10 × 2 mm cuvette using
identical temperature ramps. A sample containing no ligand
served as a control.
◦
◦
◦
mixed in a 3 × 3 mm cuvette and incubated at 20 C for 3 min
prior to measurement. Measurements were repeated after
2
4 h to check for time-dependent changes. Excitation and
emission spectra were collected at the optimum wavelengths
for each RNA–dye combination using otherwise identical

Nucleic Acids Research, 2019 3
Full details of the synthesis and characterization of HBI
derivatives, NMR measurements and computational meth-
ods are given in the Supporting Information.
response for the respective Chili RNA complex (Supple-
mentary Figures S2 and S3). The fluorescence enhancement
upon RNA binding for each chromophore is depicted in
Figure 1, relative to the fluorescence of the Chili–DMHBI
1
complex, which was set to 100. The excitation and emis-
RESULTS AND DISCUSSION
sion maxima as well as relative fluorescence values are sum-
marized in Table 1. Several dyes showed strongly enhanced
fluorescence, while others were only weakly activated upon
binding to Chili. A low emission intensity may be caused
by one or more of the following reasons: low affinity (under
Synthesis of fluorogenic HBI derivatives
The 36 derivatives of DMHBI investigated in this study
were synthesized by an imine-imidate [3+2] cycloaddition
strategy (34,35). Briefly, an aryl aldehyde (A) was first re-
acted with an alkyl or aryl amine to form the respective
imine (B). The cycloaddition partner, an N-alkyl imidate
−1
the experimental conditions, a KD of 0.5 ␮mol l or higher
would result in less than 40% complex formation), slow up-
take kinetics, slow RNA folding (restructuring of the bind-
ing site) or low quantum yield of the formed complex.
A first set of chromophores was used to examine the re-
lationship between fluorescence activation and steric de-
mand of the ligand side chain, irrespective of additional
electrostatic and hydrogen bonding interactions. Therefore,
(
C), was generated from methyl glycinate and an ethyl im-
idate (D). Ethyl imidates were prepared from their par-
ent nitriles (E) by reaction with ethanolic HCl in case
they were not commercially available. Cycloaddition be-
tween the imines (B) and N-alkyl imidates (C) afforded
the series of HBI chromophores of the general structure F
3
the R substituent at the imidazolone-N3 of DMHBI (1)
(
boxed in Scheme 1). A first series is represented by com-
was varied from methyl to ethyl (2), isopropyl (3) and tert-
butyl (4), respectively. A 40% loss of fluorescence inten-
sity was observed when a linear substituent (compounds 1,
pounds 1–20 containing alkyl residues at C2 of the imi-
4
dazolone ring (R = methyl or phenylethyl, Route A) and
3
3
variable residues R . In some cases with bulky R groups,
the cycloaddition required extended reaction times and re-
sulted in the competing formation of a side product 16 in
which the imidate-N-methylglycinyl substituent instead of
the imine-N substituent was incorporated into the HBI
scaffold. The chromophores were isolated in pure form by
chromatography and obtained as colored solids in medium
2
) was replaced by a branched N-alkyl substituent (com-
pounds 3, 4). Interestingly, compound 5 with the steri-
cally even more demanding trans-4-methylcyclohexyl sub-
stituent showed comparable fluorescence intensity to com-
plexes with 1 and 2, suggesting that steric bulk is only a sec-
ondary factor. In contrast, the benzyl and p-methoxybenzyl
substituents in compounds 6 and 7 resulted in complexes
with 5-fold lower fluorescence intensity compared to the
parent Chili–DMHBI complex. These effects are likely to be
caused by an increased number of rotational degrees of free-
dom for larger substituents, which may increase the rates
of internal conversion, thus resulting in lower fluorescence
quantum yields (36).
4
to good yields. Selected examples with R = CH3, were
oxidized with SeO2 to access the formylated intermediates
G. In a second series, the ␲ electron system was expanded
to obtain chromophores 21–36 via one of three methods:
1
(
) Sc(OTf)3-catalzyed aldol condensation of 1 or 9 with
hetero)aromatic aldehydes (Route B), 2) Wittig reaction
of formylated HBI derivatives G with phosphorous ylides
Route C), or 3) condensation of the formylated HBI deriva-
A more direct interplay between the RNA and a bound
ligand could arise from enhanced ␲–␲ stacking interac-
tions. We therefore attached a phenyl group directly at the
imidazolone-N3 without any alkyl linker resulting in com-
pound 8, and modulated the electron density of the phenyl
group by means of the para substituent, resulting in a se-
ries of dyes 9–15. Besides its strongly electron-withdrawing
properties, the trifluoromethyl group was chosen because
it is known to enhance the lipophilic character (37), which
could have an impact on RNA binding of compounds 11
and 12, in comparison to their non-fluorinated analogs 9
and 10. Geometry optimization (DFT: B3LYP-D3/def2-
TZVP) shows that these derivatives possess largely simi-
lar three-dimensional structures, i.e. the HBI core is pla-
(
tive obtained from 13 with hydroxylamine (Route D). The
cationic HBI derivatives 14, 34 and 36 were prepared by
quaternization of the dimethylamino group of 13, 33 and 35
with MeI as the final step of the reaction sequence of routes
A, C and D.
For all DMHBI derivatives carrying an arylvinyl sub-
1
stituent at imidazolone-C2, H NMR showed that the vinyl
3
group was exclusively in the E configuration with JHH cou-
pling constants of 15–16 Hz. Additionally, H- H NOESY
spectra show through-space correlations between the prox-
imal vinyl proton and protons of the R residue, suggesting
a predominant s-cis configuration of the exocyclic bond at
1
1
3
C2 (see e.g. Supplementary Figure S1 for compound (32).
nar and the N3-substituent is twisted out of plane by 50–
◦
59 (Supplementary Figure S4). One key difference is found
in the calculated dipole moment of these chromophores,
which is substantially larger for the heteroatom-containing
derivatives 10–14 compared to 9 and 15. Most significant
Fluorescence screening
A simple screening protocol was used to examine the fluo-
rescence turn-on of all the newly synthesized HBI deriva-
tives by the Chili RNA aptamer. Briefly, ligands and pre-
folded RNA were combined at a concentration of 0.5 ␮M
each in a binding buffer containing 125 mM KCl and 5 mM
MgCl2 at pH 7.5 and incubated at 25 C for 3 min; fluores-
cence emission spectra were collected by exciting the sample
at the wavelength that elicited the maximum fluorescence
+
is the large value of 16.7 D for DMHBI (14) due to its
positively charged sidechain (Supplementary Table S1). In
the fluorescence activation assay several interesting results
were observed for this family of chromophores. First, we
note that compound 9 (DMHBTI) in complex with Chili
gave a 20-fold higher fluorescence intensity than its con-
stitutional isomer 6 (DMHBI-Bn). Thus, the direct attach-
◦

4
Nucleic Acids Research, 2019
3
Scheme 1. Synthesis of functionalized HBI scaffolds (Route A) and their ␲-extended derivatives (Route B–D). Reagents and conditions: (a) R NH2,
3
◦
MgSO4, CH2Cl2, r.t., 24 h; or R NH2, toluene, reflux, 16 h. (b) AcCl, EtOH, 0 C, 7 h. (c) Glycine methyl ester hydrochloride, NEt3, CH2Cl2, r.t., 3 h.
◦
5
◦
(
d) EtOH, r.t., 16 h; or toluene, 120 C, 16 h. (e) MeI, DMF, r.t., 24 h. (f) R CHO, cat. Sc(OTf)3, dioxane, 110 C, 24 h. (g) SeO2, dioxane, reflux, 2 h. (h)
R CH2PPh3Br, nBuLi, 0 C, 30 min; r.t., 16 h. (i) NH2OH·HCl, K2CO3, MeOH, r.t., 24 h.
5
◦
Figure 1. Fluorescence screening results for various HBI derivatives with the Chili aptamer (0.5 ␮M RNA, 0.5 ␮M dye, 125 mM KCl, 5 mM MgCl2, 80 mM
HEPES pH 7.5). The blank-corrected intensity is given relative to the Chili-DMHBI 1 complex. Spectral data and intensity values are listed in Table 1 and
spectra are shown in Supplementary Figures S2 and S3.

Nucleic Acids Research, 2019 5
Table 1. Substitution patterns and spectral data for HBI derivatives 1–36. Fluorescence intensity is given relative to the Chili-DMHBI 1 complex
Rel.
Fl.Int.
R1
R2
R3
R4
R5
route
λEx/λEmnm
[a]
Nr.
Name
1
2
3
4
5
6
7
8
9
DMHBI
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
H
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
Me
Et
iPr
tBu
4MeCy
PhCH2
4MeOPhCH2
Ph
4MePh
4MeOPh
4CF3Ph
4CF3OPh
4Me2NPh
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Ph(CH2)2
−
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
400/537
400/537
400/537
400/534
400/537
400/537
400/535
410/539
410/539
410/538
413/540
413/540
413/540
413/542
410/539
n.m.
100
122
61
DMHBI-Et
DMHBI-iPr
DMHBI-tBu
DMHBI-MeCy
DMHBI-Bn
DMHBI-PMBn
DMHBPI
61
113
17.9
16.3
144
400
531
429
487
10.2
735
71.0
n.m.
280
n.d.
142
4.3
9.9
5.9
12.3
n.d.
74.3
45.6
2.3
[
*]
DMHBTI
DMHBAI[
#]
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
F
DMHBTI
DMHBAI
F
DMHBI-DMA
+
+
DMHBI
4Me3N Ph
C
DMHBI
4tBuPh
MeO(CO)CH2
4MeOPh
4MeOPh
Me
Me
Me
Me
Me
Me
Me
Me
4MePh
4MePh
4MePh
4MePh
4MePh
Me
DMHBI-spdt
MHBAI
DMBAI
-
-
-
-
397/513
n.d.
OMe
Br
[
b]
BMHBI
386/520
400/539
462/601
467/616
465/611
n.d.
463/545,594
469/539
464/618
467/613
n.d.
416,461/542,592
478/539
460/573
n.m.
465/603
n.m.
DMHBI-PhEt
DMHBI-Styr
DMHBI-2Py
DMHBI-3Py
DMHBI-4Py
DMHBI-Imi
DMHBI-Ind
DMHBTI-2Py
DMHBTI-3Py
DMHBTI-4Py
DMHBTI-Imi
DMHBTI-Ind
DMHBI-Fc[
DMHBI-Styr-DMA
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
OMe
A
B
Ph
2Py
3Py
4Py
4Imi
3Ind
2Py
3Py
4Py
4Imi
3Ind
Fc
Ph
Ph
−
[
c]
−
B
B
B
B
B
B
B
B
B
B
C
C
C
D
D
−
−
−
−
−
−
3.2
−
n.d.
92.0
50.1
2.3
n.m.
85.2
n.m.
147
[
d]
−
−
†]
−
4Me2NPh
4Me3N Ph
4Me2NPh
4Me3N Ph
−
+
+
DMHBI-Styr
DMHBO-DMA
DMHBO
−
−
+
+
−
−
456/592
aMeasurement conditions: 0.5 ␮M RNA, 0.5 ␮M dye, 125 mM KCl, 5 mM MgCl2, 80 mM HEPES pH 7.5, λEx was set to the value indicated in the table.
Samples without added RNA were used for the blank correction.
b
Erlenmeyer azlactone synthesis.
c
ZnCl2 was used as the catalyst instead of Sc(OTf)3 in the aldol condensation.
d
DMHBTI-Imi shows a dual emission profile (Supplementary Figure S3j). n.d. no fluorescence detected. n.m. not measured. [*] T = p-tolyl. [#] A =
p-anisyl. [†] Fc = ferrocenyl.
ment of an aryl group to N3 proved beneficial in contrast
to the alkylation with a benzyl group. While the Chili com-
chromophore 73-fold. This strong fluorescence enhance-
ment of 14 is a result of two factors: a higher affinity of
3
+
plex of DMHBPI (8) with R = phenyl showed similar flu-
Chili for DMHBI (leading to a more complete complex
orescence to Chili–DMHBI, a 4- to 5-fold enhanced fluo-
rescence intensity was observed with ligands 9–12 that con-
tained electron-donating methyl and methoxy groups, or
electron-withdrawing CF3 and OCF3 substituents, respec-
formation under the measurement conditions) and a higher
fluorescence quantum yield of the resulting complex (34).
To help understand how these two factors depend on the
positive charge in the sidechain of 14, a neutral isomor-
3
+
C
tively, at the para-position of R . In stark contrast, a tenfold
phic analog of DMHBI was synthesized: DMHBI (15)
contained a tert-butyl group instead of the trimethylammo-
nium substituent. The geometry-optimized structures of 14
and 15 are superimposable in the gas phase (Supplemen-
tary Figure S5). Surprisingly, however, the simple N-to-C
substitution resulted in a tenfold loss of fluorescence en-
hancement under our screening conditions (Figure 2A). Al-
decreased fluorescence intensity compared to DMHBI was
observed when this substituent was changed to a dimethyl-
amino group in compound 13. This quenching effect was
expected, considering that the dimethylaniline (DMA) moi-
ety is a known donor for intramolecular photoinduced elec-
tron transfer (PET) which leads to non-radiative decay of
the excited state (38). N-methylation of 13 disrupts the PET
C
though the Chili–DMHBI complex retained 70% of the
+
and the resulting cationic ligand DMHBI (14) fluoresces
intensity of the parent DMHBI, the distinct benefits of 14
compared to 15 warranted a more detailed analysis of the
binding affinity and the influence of electrostatic effects (see
below).
more than seven times brighter than DMHBI when bound
to the Chili aptamer. In other words, methylation of the
DMA group enhances the fluorescence of the RNA-bound

6
Nucleic Acids Research, 2019
Red-emitting HBI chromophores (i.e. λEm > 570 nm)
differ from their green-emitting counterparts by an ex-
panded ␲ electron system at position C2 of the imi-
dazolone ring. In nature, this is implemented in differ-
ent classes of red fluorescent proteins through a small
palette of functional groups such as acylimine (DsRed),
2
-hydroxy-dihydrooxazolyl (mOrange) or vinylimidazolyl
(
Kaede) moieties, which are formed from specific tripep-
tides in the protein sequence during maturation (39). Their
exact spectral characteristics are determined by a delicate
interplay between the HBI chromophore and its surround-
ing protein environment (40–42). In designing new fluoro-
genic HBI ligands that interact with biomacromolecules,
more tractable and chemically stable entities such as oximes
(
11,34) and simple benzenoid aromatics (43,44) have been
favored so far.
DMHBI-Imi (25) and DMHBO+ (36) bind to Chili
with good to excellent affinities, forming yellow to red-
fluorescent complexes with emission maxima above 590 nm
and quantum yields of 8% and 10%, respectively (34). In
order to elucidate key structural features responsible for
fluorophore binding and activation of C2-extended chro-
mophores, we investigated several related derivatives. First,
two model compounds 20 and 21 were synthesized in which
the imidazolone-C2 methyl substituent of DMHBI was re-
placed by a saturated ethylphenyl (20) or an unsaturated
vinylphenyl group (21), respectively. Compound 20 exhib-
ited nearly the same excitation and emission maxima as
DMHBI when bound to Chili, albeit with a strongly re-
duced fluorescence intensity (Supplementary Figure S2).
In contrast, both the excitation and emission maxima of
the Chili-bound derivative 21 were redshifted by 62 nm, at
about twice the fluorescence intensity of the complex with
20. Exchanging the imidazolone-N3 methyl substituent of
21 for the cationic trimethylammoniumphenyl sidechain in
compound 34 led to a complete recovery of the fluorescence
intensity with respect to DMHBI. This beneficial effect of a
large dipole moment is consistent with the seven- to tenfold
enhancement observed for the same positively charged side
Figure 2. (A) The emission spectrum of Chili–DMHBI+ compared to
C
Chili–DMHBI . (B) Emission spectra of Chili–DMHBTI-Ind and Chili–
+
DMHBI-Styr . (0.5 ␮M RNA, 0.5 ␮M dye, 125 mM KCl, 5 mM MgCl2,
8
0 mM HEPES pH 7.5). Samples without added RNA were used for the
blank correction.
Next, we analyzed the spectral tuning of the N3-modified
DMHBI chromophores and their complexes with Chili. In-
3
troduction of R = aryl groups led to a negligible shift of the
free ligands’ 8–15 excitation and emission maxima (3 nm or
less) compared to the alkyl derivatives 1–7 (Supplementary
Table S2). However, when bound to the RNA, the differ-
ence in excitation maxima between the two groups of alkyl
and aryl substituents consistently was in the range of 10–
+
+
chain in DMHBI and DMHBO . Therefore, DMHBI-
+
1
3 nm, whereas it remained unchanged for the emission
Styr (34) represents a promising new ligand for Chili with
maxima (Table 1). It follows that the aryl groups undergo
a direct interaction with a specific part of the binding site
irrespective of their electron density, which destabilizes the
electronic ground state of the neutral ligand but not the ex-
cited state of its anion. A somewhat related effect was ob-
served for the Spinach aptamer complex with DFHBI-1T,
an imidazolone-N3-modified derivative of the parent lig-
and DFHBI (10). While the excitation maximum of the un-
bound ligand was only altered by 3 nm, its Spinach com-
plex exhibited a more pronounced redshift of 35 nm. The
emission maxima were shifted by 6 and 5 nm, respectively.
Incidentally, the N3 substituent of DFHBI(-1T) is in close
proximity to a specific nucleotide that was recently shown to
govern the spectral tuning of Spinach-related aptamers (33).
However, a significant difference between Spinach and Chili
FLAPs remains the pronounced preference for Chili to bind
exclusively the neutral phenol form of the chromophores
an emission maximum beyond 600 nm (Figure 2B).
Next, additional hydrogen bond acceptor sites were intro-
5
duced via R , by replacing the vinylphenyl group in 21 with
three different vinylpyridyl side chains. Compounds 22–24
were very weakly autofluorescent in aqueous buffer at pH
7.5, and binding to Chili resulted only in a minor increase
in fluorescence of DMHBI-2Py (22) and DMHBI-3Py (23),
reaching similar intensities as for compound 21 with emis-
sion maxima of 615 and 610 nm, respectively. However, no
fluorescence enhancement was observed for DMHBI-4Py
(24). DMHBI-4Py has recently been synthesized using a
method similar to ours and was screened as a potential lig-
and for the fluorogenic protein tag FAST, but also in this
case no increase in fluorescence intensity was detected (45).
In DMHBI-Imi 25, which resembles the chromophore of
the red fluorescent protein Kaede, both a hydrogen bond
donor and acceptor site are present. When bound to Chili,
it shows a remarkable, bimodal fluorescence spectrum upon
excitation at 463 nm. Two components with emission max-
ima of 542 nm (38% integrated intensity) and 592 nm (62%
(
and thereby shifts the phenol – phenolate equilibria even
at neutral pH), while Spinach can only bind and activate
the anionic phenolate of DFHBI and its analogs.

Nucleic Acids Research, 2019 7
integrated intensity) were extracted by fitting the experi-
mental spectrum with a Gaussian function (Supplementary
Figure S3o). Since the higher energy component closely re-
sembles the emission profile of Chili–DMHBI, it can be at-
tributed to the excited state of a fluorophore subpopula-
tion that has undergone proton transfer from the imidazole
moiety to a nearby acceptor and thereby lost its extended
sidering the high acidity of DFHBI compared to deriva-
tives of DMHBI, the required protonation step to gener-
ate the binding-competent phenol form is likely impeded
in this case. Additionally, because of significant differences
in size, electronegativity, and H-bonding potential, the flu-
orine substituents may be mismatched to the environment
that has been selected to accomodate methoxy substituents.
To address the possibilities of specific recognition of the
phenolic OH and the neighboring OMe groups, two analogs
of DMHBAI (10) were examined, in which either one of
the methoxy groups (MHBAI, 17) or the phenolic hydroxy
group (DMBAI, 18) were absent. The fluorescence emis-
sion of Chili–MHBAI (17) was half as intense as that of
Chili–DMHBAI (10), indicating that both OMe groups
contribute to the optimal occupancy of the binding site.
Interestingly, one OMe group of the original DMHBI lig-
and could be replaced by bromine, resulting in the lig-
and BMHBI (19), which gave rise to a chromophore–RNA
complex of comparable intensity. In stark contrast, how-
ever, no fluorescence emission was detected when DMBAI
(18) was combined with Chili under standard screening con-
ditions. This result highlights the importance of the phe-
nolic OH group. A competition experiment between DM-
BAI (18) and DMHBAI (10) was then performed to dis-
tinguish between non-binding and non-activation. Briefly,
one equivalent of the preformed Chili–DMHBAI complex
was treated with either buffer or a tenfold excess of DMBAI
in buffer. Since the observed decrease in fluorescence emis-
sion was identical in both cases, DMBAI could not have
displaced DMHBAI from the binding site (Figure 3A). In
the reverse direction, one or ten equivalents of DMHBAI
were added to a solution of Chili containing ten equivalents
of DMBAI. The identical increase in fluorescence intensity
shows that DMBAI could not have occupied the binding
site of Chili (Figure 3B). These results indicate that the bind-
ing site of Chili contains a hydrogen bond acceptor that
not only facilitates proton abstraction from the photoex-
cited ligand but that is also crucial for ligand binding in
the ground state. Secondary interactions with the methoxy
groups (or isosteric replacements) enhance the fluorescence
by constraining the bound ligand into a specific conforma-
tion.
␲
conjugation. To further corroborate this hypothesis, we
also synthesized DMHBI-Ind (26) containing a 3-indolyl
substituent in place of the imidazolyl group. Indole itself
is a known photoacid (46) and should therefore enable ef-
ficient NH proton transfer in Chili–DMHBI-Ind. Excita-
tion of the complex at 469 nm resulted in an emission band
with a prominent maximum at 537 nm and a broad shoul-
der on the red edge, which is consistent with the suggested
mechanism (Supplementary Figure S3). Similar results were
3
observed for a second series of chromophores, in which R
was a 4-methylphenyl (i.e. p-tolyl) group to obtain the re-
spective DMHBTI derivatives 27–31 (Figure 2B, Supple-
mentary Figure S3). DMHBTI-Imi (30) was obtained as a
10:1 mixture of E and Z isomers at the newly formed dou-
ble bond. Similar to DMHBI-Imi (25), its Chili complex
shows two prominent emission bands at 542 and 592 nm.
However, their relative intensity was dependent on the ex-
citation wavelength (Supplementary Figure S3j), which is
likely attributed the presence of electronically distinct ge-
ometric isomers. Finally, a ferrocenyl-expanded DMHBI
analog 32 was screened. The redox-active organometallic
side-chain could provide interesting properties for electro-
chemical sensor applications (47), but because only a very
low level of fluorescence activation upon addition of Chili
RNA was observed, this ligand was not investigated in fur-
ther detail.
Recognition of HBI by the Chili binding site
Molecular recognition of ligands in the RNA binding site
is governed by specific supramolecular interactions. For the
Spinach aptamer and its cognate DFHBI ligand, these in-
teractions have been revealed by X-ray crystallography, and
␲
stacking as well as direct and H2O-bridged hydrogen
bonding have been identified as important interactions be-
tween DFHBI and the RNA aptamer. The ligand is situ-
ated between the top face of a G-quartet and an U–A–U
base triple, where it acts as a hydrogen bond acceptor to
a nearby adenine (via imidazolone-N1) and two guanines
We also tested whether ligand recognition was influenced
by the presence or absence of specific divalent metal ions.
2+
Chili requires up to 5 mM Mg to effectively turn on the flu-
+
orescence of DMHBI, while strongly enhanced DMHBI
fluorescence was observed regardless of the Mg2 concen-
tration (34). When MgCl2 in the binding buffer was re-
placed by BaCl2, the fluorescence intensity for either lig-
and dropped by a factor of approx. 10 (Figure 3C). It re-
covered over the course of 24 h and remained stable at a
moderate level for several days. In contrast, in the presence
+
(
via benzylidene-O4 and imidazolone-O4) (24). A similarly
dense framework of interactions has been found in the crys-
tal structure of Corn–DFHO, where the oxime moiety of
the ligand additionally acts as an ambidentate hydrogen
bond donor and acceptor at the same time (29).
In the absence of a crystal structure of the Chili RNA,
the detailed mechanism of ligand recognition is unknown,
but biochemical experiments can reveal essential func-
tional groups for analogous supramolecular interactions.
We showed previously that DFHBI is not strongly acti-
vated by the Chili aptamer (34), indicating a different lig-
and binding mode in Chili compared to Spinach. The char-
acteristically large pseudo-Stokes shifts observed for Chili
complexes show that ligand binding and fluorescence acti-
vation involves a proton transfer cycle (Scheme S1). Con-
2+
of Mg , high fluorescence intensity was observed imme-
diately, which increased by <10% upon prolonged incuba-
tion (Figure 3C). Unlike Spinach–DFHBI, whose emission
maximum is strongly shifted depending on the nature of the
divalent metal ion (24), neither of the two Chili complexes
displayed any form of spectral tuning upon cation replace-
ment (Figure 3D). This suggests that Chili complexes lack
specific cation–ligand interactions; instead, only the global
stability of the active RNA conformation is influenced due

8
Nucleic Acids Research, 2019
Figure 4. ITC data for the formation of Chili–DMHBI (A, B) and Chili–
+
◦
Figure 3. (A) Integrated fluorescence emission intensities for Chili–
DMHBAI (0.5 ␮M RNA, 0.5 ␮M dye 10, 125 mM KCl, 5 mM MgCl2,
DMHBI (C, D) at 25 C. Background-corrected differential power and
offset-corrected integrated heat are shown for a single representative run
of three. The data was fitted with a one-site-binding model.
4
0 mM HEPES pH 7.5) before and after adding a tenfold excess of DM-
BAI (18) or an equivalent volume of water. (B) Integrated fluorescence
emission intensities for Chili–DMBAI (0.5 ␮M RNA, 5 ␮M dye, 125 mM
KCl, 5 mM MgCl2, 40 mM HEPES pH 7.5) before and after adding 0.5 or
Table 2. Thermodynamic data for the formation of Chili complexes from
ITC experiments
5
␮M DMHBAI. (C) Integrated fluorescence emission intensities for Chili–
+
DMHBI and Chili–DMHBI (0.5 ␮M RNA, 0.5 ␮M dye, 125 mM KCl,
mM MgCl2 or BaCl2, 80 mM HEPES pH 7.5) after 3 min and 24 h incuba-
tion time, respectively. (D) Fluorescence excitation (dashed) and emission
solid) spectra of Chili–DMHBI after 3 min incubation in the presence of
KDb,
5
a
ꢀ
mol
H, kJ
ꢀS, J
mol
ꢀG , kJ
KD, nmol nmol
−
1
−1
−1
−1
−1
−1
L
Ligand
K
mol
L
(
MgCl2 (black) or BaCl2 (red).
DMHBI −44
−276
−37
−44
371
24.1
570
63
+
DMHBI −42
+78
to differences in the coordination environment of the re-
spective metal ion.
a
ꢀ
G = ꢀH − T ꢀS at 298 K.
b
Determined by fluorescence titration (34).
Binding affinity
calorimetry (ITC) to determine the respective enthalpic
and entropic contributions of the binding process (Figure
4, Table 2). For both, the calorimetric dissociation con-
While association with the RNA aptamer is a requirement
for fluorescence turn-on of the ligand, affinity reflects in-
teractions in the electronic ground state of the complex,
whereas fluorogenic enhancement is an excited state prop-
erty. Both aspects are therefore subject to different, not
necessarily correlated, mechanisms. To examine the effects
of ligand substituents on binding affinities, the dissocia-
tion constants of Chili complexes with HBI derivatives 9–
+
stants (DMHBI: 371 ± 55 nM; DMHBI : 24.1 ± 5.6 nM)
are in good agreement with those obtained by fluores-
cence titration (34). While no substantial difference in bind-
−1
ing enthalpy was observed (−44 and −42 kJ mol , re-
spectively), the binding entropy is unfavorable for DMHBI
−
1
−1
+
(−276 J mol
K
), but slightly favorable for DMHBI
−
1
−1
1
2 (selected based on the strong fluorescence turn-on dur-
(78 J mol
K
) which results in an overall more negative
◦
ing the screening) were measured using fluorescence titra-
tion experiments, in which the ligand concentration was
held constant and the RNA concentration was increased
up to 40 ␮M. Both DMHBTI (9) and DMHBAI (10) are
strong binders with a KD of 151 ± 21 and 65 ± 7 nM, re-
spectively. In contrast, saturation binding was not achieved
ꢀG at 25 C for the latter.
In general, the entropy change of binding encompasses
a number of different factors such as the reduction of the
number of unbound molecules in solution, loss of confor-
mational freedom in the complex, and the release of water
molecules due to a loss of solvent-accessible surface area
(48). A positive ꢀS is characteristic for such a desolvation
effect and reflects the increased hydration of the cationic
F
F
for DMHBTI (11) and DMHBAI (12); their dissociation
constants are estimated to be greater than 1 ␮M. DMHBI
C
+
+
(
15), the neutral analog of DMHBI , was also found to be
side chain in DMHBI compared to DMHBI. This further
a weak binder with an estimated KD of >0.7 ␮M (Supple-
mentary Figure S6).
suggests that the lower affinity of HBI derivatives 11, 12 and
15 is the result of an entropic effect caused by the hydropho-
bic nature of their aromatic side chains. However, once a
complex with the Chili aptamer is formed, a higher degree
In addition, the binding affinities of DMHBI (1) and
+
DMHBI (14) were assessed by isothermal titration

Nucleic Acids Research, 2019 9
Table 3. Rate constants for the association and dissociation of Chili com-
plexes from kinetic experiments according to Figure 5D
a
−1
−1 −1
b
−1
Ligand
KD , nmolL
kon, L mol
s
koff , s
DMHBI+
DMHBO
63
12
8.2 × 10
4
5.1 × 10
7.8 × 10
−3
−4
+
4
6.5 × 10
a
Dissociation constants from (34).
koff = kon · KD.
b
+
Figure 5. (A) Fluorescence activation kinetics of Chili with DMHBI ,
+
F
DMHBO and DMHBAI under pseudo-first order conditions (0.025 ␮M
RNA, 2 ␮M dye, 125 mM KCl, 5 mM MgCl2, 40 mM HEPES pH 7.5). Data
points were collected in 2 s intervals, every 10th point is plotted. Solid lines
represent biexponential fits. (B) First 90 s of the data in panel a without
decimation. (C) kobs values for the biexponential fits in panel A). The inset
shows only the slow kobs values. (D) Plot of kobs against the dye concen-
tration (0.025 ␮M RNA, dye as indicated, 125 mM KCl, 5 mMM MgCl2,
4
0 mM HEPES pH 7.5). Solid lines represent linear fits to obtain kon.
of fluorescence turn-on compared to N3-alkylated deriva-
tives is achieved.
Binding kinetics
In contrast to other fluorogenic RNA systems like Spinach–
DFHBI, Chili was shown to exhibit biexponential kinetics
+
when activating the fluorescence of DMHBO 36, which
was attributed to a combination of ligand uptake and re-
folding of the Chili aptamer (34,49,50). We therefore ex-
amined the activation kinetics of two additional ligands
+
F
with different affinities, DMHBI (14) and DMHBAI (12)
to test whether this behavior is a general property of the
Chili RNA. Under pseudo-first order conditions, the fluo-
rescence signal was found to rise biexponentially for both
dyes (Figure 5A, B, Supplementary Figure S7). However,
Figure 6. Thermal melting profiles of Chili (black) and Chili–DMHBI+
(
red) (2 ␮M RNA, 2 ␮M dye, 125 mM KCl, 40 mM HEPES pH 7.5, tem-
◦
perature ramp 0.5 C/min). (A) Absorbance and (B) first derivative of the
absorbance at 260 nm. (C) Absorbance and (D) first derivative of the ab-
sorbance at 295 nm. (E) Fluorescence emission and (F) first derivative of
the fluorescence emission at 413/542 nm (Ex/Em). Curves are shown for
the second of four reversible ramps.
+
while the apparent rate constants (kobs) for DMHBI and
+
F
DMHBO were not significantly different, DMHBAI was
activated at a more than sixfold slower rate (Figure 5C).
Bimolecular association rates (kon) were obtained from
determining kobs over a range of different ligand concentra-
tions; the corresponding dissociation rates (koff) were then
estimated from the relation koff = kon · KD (Table 3). These
UV–Vis and fluorescence melting analysis
To study folding and unfolding of the Chili RNA and
its fluorescent complex with HBI ligands, temperature-
dependent changes in the absorbance of Chili and Chili–
+
+
results show that DMHBI and DMHBO bind to Chili
4
−1 −1
+
with similar rates on the order of 10 l mol
s
(Figure
DMHBI were monitored at different wavelengths between
◦
5D), which is typical for the complex formation between ap-
10 and 95 C. At 260 nm, where melting of canonical duplex
tamers or riboswitches and their cognate ligands (51). The
lower dissociation constant of Chili–DMHBO therefore
structures is dominant, both systems showed strong hyper-
chromicity with similar transitions at 73 C, which reflect
+
◦
reflects a slower unbinding of this fluorophore compared
to DMHBI , which may be explained among other reasons
by the larger number of molecular interactions due to the
extended oxime functional group.
the dissociation of the bottom stem and top stem loop of
Chili, respectively (Figure 6A, B). Additionally, a strongly
+
◦
hypochromic transition at 51 C was observed for Chili–
+
DMHBI at 295 nm, but not for Chili in the absence of

10 Nucleic Acids Research, 2019
slower reorganization of the mutant binding site. G9 and
C44 are predicted to form a base pair at the junction be-
tween the binding site and the bottom stem of Chili. Both
G9A and the double mutant G9A/C44U are completely in-
active. The mutation C44U, which introduces a G–U wob-
ble base pair, was tolerated, but resulted in a significant loss
of fluorescence enhancement to about one third of the par-
ent intensity (Supplementary Figure S8). The base pair at
the junction between the binding site and the top stem loop
is likely formed by C16 and G31. Here, a C–G Watson–
Crick base pair is strictly required for HBI activation, since
both single point mutants as well as the C16U/G31A dou-
ble mutant with a restored U–A base pair remained un-
+
able to bind DMHBI and DMHBI (Supplementary Fig-
Figure 7. Secondary structure of Chili and mutagenesis pattern of the
binding site. Color indicates fluorescence enhancement of DMHBI 1 rel-
ative to wt-Chili (0.5 ␮M RNA, 0.5 ␮M dye, 125 mM KCl, 5 mM MgCl2,
ure S8). Although these data are not sufficient to unam-
biguously identify the guanines involved in formation of
the quadruplex core structure, the formation of functional
G-quartets is further corroborated by the activation of two
quadruplex-binding fluorogenic dyes, thiazole orange (TO)
and thioflavin T (ThT) by wt-Chili and a number of tran-
sition mutants (Supplementary Figure S9), as well as by
NMR spectroscopy.
8
0 mM HEPES pH 7.5). See data in Supplementary Table S3.
+
DMHBI . This observation reflects the ligand-induced for-
mation of a G-quadruplex motif (Figure 6C, D) (52). At
+
the same time, Chili–DMHBI exhibited a linear decrease
in fluorescence emission at 542 nm; no fluorescence was de-
◦
tected above 63 C (Figure 6E). This decline correlates with
NMR spectroscopy of the Chili aptamer
the G-quadruplex unfolding, which further emphasizes the
interdependence of ligand binding, quadruplex formation
and fluorescence activation for Chili (Figure 6F). Parallels
can be drawn to the thermal properties of the Spinach and
Mango aptamers: Spinach does not gain stabilization from
ligand binding because, like in the present case, dissociation
of the complex precedes the melting of Watson–Crick sec-
ondary structure motifs. Mango, on the other hand, which
is essentially a minimal G-quadruplex domain, is strongly
stabilized by ligand binding, i.e. melting and fluorescence
loss occur simultaneously (31,19).
1
The imino region of the H NMR spectra of RNA provides
insightful information on the secondary structure because
the chemical shift of imino protons involved in hydrogen
bonding is highly sensitive to the nature of the hydrogen
bond: imino protons engaged in canonical Watson-Crick
base pairs resonate in the region of 12–15 ppm, while imino
protons involved in G–U wobbles or in the formation of
G-quartets via Hoogsteen base pairing typically resonate
between 10 and 12 ppm (53,54). The H NMR spectrum
of Chili in the apo state shows signals in the imino region
Figure 8A), which likely arise from partial hybridization
1
(
of the top and bottom stems, as revealed by comparison
with the spectra of synthetic stem loops that comprise the
nucleotides 1–9/44–52 and 16–31 of the Chili RNA (Sup-
plementary Figure S10). The imino pattern did not change
Mutagenesis of the Chili aptamer
Chili was designed as a folding-optimized variant of the
1
3–2 aptamer by modifying the stem loop regions adjacent
2+
+
to the central bulge, which itself was left unaltered (34).
We prepared single-nucleotide transition mutants for all nu-
cleotides of the central region (except A11, which was mu-
tated to U, in order to avoid seven consecutive Gs) and com-
pared their ability to activate DMHBI to that of the par-
ent Chili RNA. The majority of nucleotides could not be
mutated without rendering the aptamer virtually inactive,
which is reflected by the red colored boxes in Figure 7. Mu-
tants U34C, G35A, G36A and U38C retained their activity
partially or even fully, which is consistent with previously re-
ported structural probing data that demonstrated enhanced
susceptibility to hydrolysis of this region in the parent Chili
RNA (Figure 7) (34).
noticeably upon sequential addition of Mg and K , sug-
gesting that metal ions are not competent to induce fold-
+
ing of the aptamer core. Only after addition of DMHBI , a
new set of well-dispersed imino signals appeared, indicating
the ligand-induced formation of a binding pocket (Figure
8A). The chemical shift of the new signals suggests the for-
mation of canonical Watson-Crick hydrogen bonds as well
as non-canonical base pairs (e.g. Hoogsteen-type). Interest-
ingly, after transferring the sample into D2O a set of sig-
nals in the region 10.7–11.8 ppm as well as two signals at
12.2 and 12.8 ppm, respectively, were still observed (Sup-
plementary Figure S11). In particular, the signals around
11 ppm are still detectable at room temperature for sev-
eral hours, suggesting that these protons are highly pro-
tected from exchange with water and pointing at the for-
mation of a 2-tetrad G-quadruplex, as previously proposed
for the 13–2 aptamer in presence of the non-canonical lig-
and DFHBI (24). The formation of a G-quadruplex struc-
Based on these findings, several mutants were chosen for
a more detailed analysis, including a comparison between
activation of DMHBI and DMHBI . G35A and G36A,
representing the only positions in the binding site where mu-
tation of a G was tolerated, were found to activate DMHBI
to the same degree as the parent Chili; DMHBI fluores-
+
+
+
ture is also supported by the K dependence of the fluores-
cence, however, was weaker, especially with G35A. This dif-
ference lessened after overnight incubation, suggesting a
cence emission (34) and the thermal melting experiments
discussed above. This process is apparently decoupled from

Nucleic Acids Research, 2019 11
late their fluorescence profile, such as nucleotides stacking
onto their aromatic moieties. Therefore, the induction of a
G-quadruplex structure is necessary but not sufficient for
bright fluorescence activation. Additional capping struc-
tures must play an important role in defining the fluoro-
genic behavior of each dye. The imino protons detected in
the NMR spectrum report only on the formation of hydro-
gen bonds, but do not give any information on, e.g., a nu-
cleotide that is stacking but not base-paired. The slow re-
arrangement of such a capping structure could be also the
reason why a slow kinetic component of fluorescence acti-
vation was detected, but no corresponding time-dependent
changes in the NMR imino region were observed.
CONCLUSIONS
The fluorogenic RNA aptamer Chili was investigated for its
ability to activate the fluorescence of thirty-six differently
substituted HBI chromophores. The dye structures were in-
spired by chromophores found in fluorescent proteins, and
were synthetically obtained using a cycloaddition-based
strategy, which in most cases gave more reliable results than
the Erlenmeyer azlactone route. While alkyl substituents
were less favored at N3 of the imidazolone ring, a set of
strongly fluorogenic dyes was obtained that contained aryl
groups at this position. These chromophores showed overall
bright fluorescence emission, but the binding affinities var-
ied between less than 100 nM and above 1 ␮M. The tightest
binding was observed for ligands with cationic trimethylam-
moniumphenyl side chains. Combined with the expansion
of the ␲ system, this strategy resulted in new red-emitting
RNA complexes with emission maxima beyond 600 nm. A
large dipole moment of the chromophore was connected
to strong fluorescence enhancement, and importantly, the
specificity for the Chili aptamer was not impaired by en-
hanced electrostatic attraction. The thermodynamics of lig-
and binding was investigated by ITC, and the approximately
Figure 8. (A) Imino 1H NMR spectra of Chili before and after addition
+
of KCl followed by DMHBI (150 ␮M RNA, 1 mM MgCl2, 25 mM TRIS
1
pH 7.4, 10% D2O/90% H2O; + 50 mM KCl; + 150 ␮M dye). (B) Imino H
NMR spectra of Chili alone and in the presence of various ligands (20 ␮M
RNA, 20 ␮M dye, 50 mM KCl, 1 mM MgCl2, 25 mM KPi pH 7.4, 10%
D2O/90% H2O).
the slow kinetic component of fluorescence activation (dis-
cussed in Figure 5), as no time-dependent changes were ob-
served in the imino region of the NMR spectrum (Supple-
mentary Figure S12).
+
tenfold stronger binding of the cationic DMHBI was iden-
tified as an entropic effect. The differences in affinity com-
pared to uncharged analogs were also mirrored in the flu-
orescence activation kinetics, which showed a slower fluo-
rescence intensity increase for lower affinity ligands. Two
factors were found to contribute to fluorescence activation:
Fast formation of the aptamer-ligand complex, followed
Under similar experimental conditions, DMHBI gave
+
only slightly different NMR results to DMHBI (Figure
C
8
B). In case of DMHBI , despite 80% of the RNA being
bound to the ligand (estimated according to Supplemen-
tary Figure S6E) the imino resonances broadened, suggest-
1
by a slow structural reorganization of the binding site. H
C
ing that DMHBI binds to the Chili RNA in the interme-
NMR spectroscopy showed that the binding site of Chili
is not preorganized in the apo state, cannot be induced by
metal ions alone, but readily adopts its folded structure
only upon ligand binding. The newly formed imino pro-
ton signals as well as the results of temperature-dependent
UV spectroscopy are characteristic for the folding of a G-
quadruplex. Future NMR studies will employ site-specific
isotope labeling to identify individual nucleobases interact-
ing with the ligands. Overall, our study helps to understand
the unique features of the Chili aptamer on a structural ba-
sis to guide further optimization and application of the ap-
tamer in combination with suitable ligands as a bioanalyti-
cal tool.
diate exchange regime (Figure 8B). In fact, the residence
time of the ligand at the binding site is short compared to
+
that of DMHBI (∼200 s, calculated as 1/koff, see Table
3
4
). Because DMHBI-Styr (21) and its aza analog DMHBI-
Py (24) were found to be weakly or non-fluorogenic, re-
1
spectively, they were also tested for binding by H NMR.
Both ligands gave rise to an imino resonance pattern com-
parable to that observed for DMHBI+ and DMHBI in
terms of number of signals and resonance dispersion, but
1
with a larger degree of line broadening. Therefore, the H
NMR spectra suggest the formation of a ligand-induced
G-quadruplex structure in the binding pocket also in the
presence of DMHBI-4Py and DMHBI-Styr. The low flu-
orescence of their Chili complexes does not primarily re-
sult from weak binding to the G-quadruplex, but instead
from the lack of additional structural motifs able to modu-
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.

12 Nucleic Acids Research, 2019
ACKNOWLEDGEMENTS
16. Holeman,L.A., Robinson,S.L., Szostak,J.W. and Wilson,C. (1998)
Isolation and characterization of fluorophore-binding RNA
aptamers. Fold. Des., 3, 423–431.
17. Sunbul,M. and J a¨ schke,A. (2018) SRB-2: a promiscuous rainbow
aptamer for live-cell RNA imaging. A promiscuous rainbow aptamer
for live-cell RNA imaging. Nucleic Acids Res., 46, e110.
Access to a MicroCal PEAQ-ITC was kindly provided by
Malvern Panalytical GmbH Germany. Laura Garske and
Sebastian Mayer are acknowledged for technical assistance.
1
1
2
8. Wirth,R., Gao,P., Nienhaus,G.U., Sunbul,M. and J a¨ schke,A. (2019)
SiRA: a silicon rhodamine-binding aptamer for live-cell
super-resolution RNA imaging. J. Am. Chem. Soc., 141, 7562–7571.
9. Jeng,S.C.Y., Chan,H.H.Y., Booy,E.P., McKenna,S.A. and Unrau,P.J.
FUNDING
(
2016) Fluorophore ligand binding and complex stabilization of the
European Research Council [ERC consolidator grant
RNA Mango and RNA Spinach aptamers. RNA, 22, 1884–1892.
0. Trachman,R.J., Truong,L. and Ferr e´ -D’Amar e´ ,A.R. (2017)
Structural principles of fluorescent RNA aptamers. Trends
Pharmacol. Sci., 38, 928–939.
682586 to C.H.]; University of W u¨ rzburg. Funding for open
access charge: European Research Council.
Conflict of interest statement. None declared.
2
1. You,M. and Jaffrey,S.R. (2015) Structure and mechanism of RNA
mimics of green fluorescent protein. Annu. Rev. Biophys., 44, 187–206.
2. Trachman,R.J. and Ferr e´ -D’Amar e´ ,A.R. (2019) Tracking RNA with
light: selection, structure, and design of fluorescence turn-on RNA
aptamers. Q. Rev. Biophys., 52, e8.
2
REFERENCES
1
. Paige,J.S., Wu,K.Y. and Jaffrey,S.R. (2011) RNA mimics of green
fluorescent protein. Science, 333, 642–646.
. Bouhedda,F., Autour,A. and Ryckelynck,M. (2017) Light-Up RNA
aptamers and their cognate fluorogens: from their development to
their applications. from their development to their applications. Int. J.
Mol. Sci., 19, E44.
2
3. Strack,R.L., Disney,M.D. and Jaffrey,S.R. (2013) A superfolding
Spinach2 reveals the dynamic nature of trinucleotide
2
repeat-containing RNA. Nat. Methods, 10, 1219–1224.
4. Warner,K.D., Chen,M.C., Song,W., Strack,R.L., Thorn,A.,
Jaffrey,S.R. and Ferr e´ -D’Amar e´ ,A.R. (2014) Structural basis for
activity of highly efficient RNA mimics of green fluorescent protein,
Nat. Struct. Mol. Biol., 21, 658–663.
2
3
4
. Neubacher,S. and Hennig,S. (2019) RNA structure and cellular
applications of fluorescent light-up aptamers. Angew. Chem. Int. Ed.,
2
5. Autour,A., Westhof,E. and Ryckelynck,M. (2016) iSpinach. A
fluorogenic RNA aptamer optimized for in vitro applications. Nucleic
Acids Res., 44, 2491–2500.
6. Filonov,G.S., Moon,J.D., Svensen,N. and Jaffrey,S.R. (2014)
Broccoli. Rapid selection of an RNA mimic of green fluorescent
protein by fluorescence-based selection and directed evolution. J. Am.
Chem. Soc., 136, 16299–16308.
5
8, 1266–1279.
. Jepsen,M.D.E., Sparvath,S.M., Nielsen,T.B., Langvad,A.H.,
Grossi,G., Gothelf,K.V. and Andersen,E.S. (2018) Development of a
genetically encodable FRET system using fluorescent RNA aptamers.
Nat. Commun., 9, 18.
2
5
6
. Roszyk,L., Kollenda,S. and Hennig,S. (2017) Using a specific
RNA-protein interaction to quench the fluorescent RNA Spinach.
ACS Chem. Biol., 12, 2958–2964.
. Masuda,I., Igarashi,T., Sakaguchi,R., Nitharwal,R.G., Takase,R.,
Han,K.Y., Leslie,B.J., Liu,C., Gamper,H., Ha,T. et al. (2017) A
genetically encoded fluorescent tRNA is active in live-cell protein
synthesis. Nucleic Acids Res., 45, 4081–4093.
. Litke,J.L. and Jaffrey,S.R. (2019) Highly efficient expression of
circular RNA aptamers in cells using autocatalytic transcripts. Nat.
Biotechnol., 37, 667–675.
. You,M., Litke,J.L., Wu,R. and Jaffrey,S.R. (2019) Detection of
low-abundance metabolites in live cells using an RNA integrator. Cell
Chem. Biol., 26, 471–481.
2
7. Huang,H., Suslov,N.B., Li,N.-S., Shelke,S.A., Evans,M.E.,
Koldobskaya,Y., Rice,P.A. and Piccirilli,J.A. (2014) A
G-quadruplex-containing RNA activates fluorescence in a GFP-like
fluorophore. Nat. Chem. Biol., 10, 686–691.
2
8. Fernandez-Millan,P., Autour,A., Ennifar,E., Westhof,E. and
Ryckelynck,M. (2017) Crystal structure and fluorescence properties of
the iSpinach aptamer in complex with DFHBI. RNA, 23, 1788–1795.
9. Warner,K.D., Sjeklo c´ a,L., Song,W., Filonov,G.S., Jaffrey,S.R. and
Ferr e´ -D’Amar e´ ,A.R. (2017) A homodimer interface without base
pairs in an RNA mimic of red fluorescent protein. Nat. Chem. Biol.,
7
8
9
2
1
3, 1195–1201.
3
0. Shelke,S.A., Shao,Y., Laski,A., Koirala,D., Weissman,B.P.,
Fuller,J.R., Tan,X., Constantin,T.P., Waggoner,A.S., Bruchez,M.P.
et al. (2018) Structural basis for activation of fluorogenic dyes by an
RNA aptamer lacking a G-quadruplex motif. Nat. Commun., 9, 4542.
1. Trachman,R.J., Demeshkina,N.A., Lau,M.W.L.,
. Tan,X., Constantin,T.P., Sloane,K.L., Waggoner,A.S., Bruchez,M.P.
and Armitage,B.A. (2017) Fluoromodules consisting of a
promiscuous RNA aptamer and red or blue fluorogenic cyanine dyes.
selection, characterization, and bioimaging, J. Am. Chem. Soc., 139,
3
9
001–9009.
Panchapakesan,S.S.S., Jeng,S.C.Y., Unrau,P.J. and
1
1
0. Song,W., Strack,R.L., Svensen,N. and Jaffrey,S.R. (2014)
Plug-and-play fluorophores extend the spectral properties of Spinach.
J. Am. Chem. Soc., 136, 1198–1201.
1. Song,W., Filonov,G.S., Kim,H., Hirsch,M., Li,X., Moon,J.D. and
Jaffrey,S.R. (2017) Imaging RNA polymerase III transcription using
a photostable RNA-fluorophore complex. Nat. Chem. Biol., 13,
Ferr e´ -D’Amar e´ ,A.R. (2017) Structural basis for high-affinity
fluorophore binding and activation by RNA Mango. Nat. Chem.
Biol., 13, 807–813.
2. Trachman,R.J., Abdolahzadeh,A., Andreoni,A., Cojocaru,R.,
Knutson,J.R., Ryckelynck,M., Unrau,P.J. and Ferr e´ -D’Amar e´ ,A.R.
3
(
2018) Crystal structures of the mango-II RNA aptamer reveal
1
187–1194.
heterogeneous fluorophore binding and guide engineering of variants
with improved selectivity and brightness. Biochem., 57, 3544–3548.
3. Filonov,G.S., Song,W. and Jaffrey,S.R. (2019) Spectral tuning by a
single nucleotide controls the fluorescence properties of a fluorogenic
aptamer. Biochem., 58, 1560–1564.
4. Steinmetzger,C., Palanisamy,N., Gore,K.R. and H o¨ bartner,C. (2019)
A multicolor large stokes shift fluorogen-activating RNA aptamer
with cationic chromophores. Chem. Eur. J., 25, 1931–1935.
5. Lerestif,J.M., Perrocheau,J., Tonnard,F., Bazureau,J.P. and
Hamelin,J. (1995) 1,3-Dipolar cycloaddition of imidate ylides on
imino-alcohols. Synthesis of new imidazolones using solvent free
conditions. Tetrahedron, 51, 6757–6774.
1
1
2. Babendure,J.R., Adams,S.R. and Tsien,R.Y. (2003) Aptamers switch
on fluorescence of triphenylmethane dyes. J. Am. Chem. Soc., 125,
3
3
3
1
4716–14717.
3. Dolgosheina,E.V., Jeng,S.C.Y., Panchapakesan,S.S.S., Cojocaru,R.,
Chen,P.S.K., Wilson,P.D., Hawkins,N., Wiggins,P.A. and Unrau,P.J.
(
2014) RNA mango aptamer-fluorophore. A bright, high-affinity
complex for RNA labeling and tracking. ACS Chem. Biol., 9,
412–2420.
2
1
1
4. Autour,A., C Y Jeng,S., D Cawte,A., Abdolahzadeh,A., Galli,A.,
Panchapakesan,S.S.S., Rueda,D., Ryckelynck,M. and Unrau,P.J.
(
2018) Fluorogenic RNA Mango aptamers for imaging small
non-coding RNAs in mammalian cells. Nat. Commun., 9, 656.
5. Trachman,R.J., Autour,A., Jeng,S.C.Y., Abdolahzadeh,A.,
Andreoni,A., Cojocaru,R., Garipov,R., Dolgosheina,E.V.,
Knutson,J.R., Ryckelynck,M et al.2019) Structure and functional
reselection of the Mango-III fluorogenic RNA aptamer. Nat. Chem.
Biol., 15, 472–479.
3
6. Gepshtein,R., Huppert,D. and Agmon,N. (2006) Deactivation
mechanism of the green fluorescent chromophore. J. Phys. Chem. B,
1
10, 4434–4442.

Nucleic Acids Research, 2019 13
3
3
7. Leroux,F.R., Manteau,B., Vors,J.-P. and Pazenok,S. (2008)
Trifluoromethyl ethers–synthesis and properties of an unusual
substituent. Beilstein J. Org. Chem., 4, 13.
Gorbachev,D.A., Malyshevskaya,K.K., Gavrikov,A.S. et al. (2019)
Red-shifted substrates for FAST fluorogen-activating protein based
on the GFP-like chromophores. Chem. Eur. J., 25, 9592–9596.
46. Chen,Y., Rich,R.L., Gai,F. and Petrich,J.W. (1993) Fluorescent
species of 7-azaindole and 7-azatryptophan in water. J. Phys. Chem.,
97, 1770–1780.
47. van Staveren,D.R. and Metzler-Nolte,N. (2004) Bioorganometallic
chemistry of ferrocene. Chem. Rev., 104, 5931–5985.
48. Jelesarov,I. and Bosshard,H.R. (1999) Isothermal titration
calorimetry and differential scanning calorimetry as complementary
tools to investigate the energetics of biomolecular recognition. J. Mol.
Recognit., 12, 3–18.
49. Wang,P., Querard,J., Maurin,S., Nath,S.S., Le Saux,T., Gautier,A.
and Jullien,L. (2013) Photochemical properties of Spinach and its use
in selective imaging. Chem. Sci., 4, 2865–2873.
50. Han,K.Y., Leslie,B.J., Fei,J., Zhang,J. and Ha,T. (2013)
Understanding the photophysics of the spinach-DFHBI RNA
aptamer-fluorogen complex to improve live-cell RNA imaging. J. Am.
Chem. Soc., 135, 19033–19038.
51. Wickiser,J.K., Cheah,M.T., Breaker,R.R. and Crothers,D.M. (2005)
The kinetics of ligand binding by an adenine-sensing riboswitch.
Biochem., 44, 13404–13414.
52. Mergny,J.-L., Phan,A.-T. and Lacroix,L. (1998) Following G-quartet
formation by UV-spectroscopy. FEBS Lett., 435, 74–78.
53. F u¨ rtig,B., Richter,C., W o¨ hnert,J. and Schwalbe,H. (2003) NMR
spectroscopy of RNA, ChemBioChem, 4, 936–962.
8. Sessler,J.L., Sathiosatham,M., Brown,C.T., Rhodes,T.A. and
Wiederrecht,G. (2001) Hydrogen-bond-mediated photoinduced
electron-transfer: novel dimethylaniline-anthracene ensembles formed
via Watson-Crick base-pairing. J. Am. Chem. Soc., 123, 3655–3660.
9. Craggs,T.D. (2009) Green fluorescent protein: structure, folding and
chromophore maturation. Chem. Soc. Rev., 38, 2865–2875.
0. Pakhomov,A.A. and Martynov,V.I. (2008) GFP family: structural
insights into spectral tuning. Chem. Biol., 15, 755–764.
1. Amat,P. and Nifos `ı ,R. (2013) Spectral “fine” tuning in fluorescent
proteins: the case of the GFP-like chromophore in the anionic
protonation state. J. Chem. Theory Comput., 9, 497–508.
2. Drobizhev,M., Callis,P.R., Nifos `ı ,R., Wicks,G., Stoltzfus,C.,
Barnett,L., Hughes,T.E., Sullivan,P. and Rebane,A. (2015) Long- and
short-range electrostatic fields in GFP mutants: implications for
spectral tuning. Sci. Rep., 5, 13223.
3
4
4
4
4
4
3. Feng,G., Luo,C., Yi,H., Yuan,L., Lin,B., Luo,X., Hu,X., Wang,H.,
Lei,C., Nie,Z. et al. (2017) DNA mimics of red fluorescent proteins
(
RFP) based on G-quadruplex-confined synthetic RFP
chromophores. Nucleic Acids Res., 45, 10380–10392.
4. Liu,Y., Wolstenholme,C.H., Carter,G.C., Liu,H., Hu,H.,
Grainger,L.S., Miao,K., Fares,M., Hoelzel,C.A., Yennawar,H.P.
et al. (2018) Modulation of fluorescent protein chromophores to
detect protein aggregation with turn-on fluorescence. J. Am. Chem.
Soc., 140, 7381–7384.
54. Adrian,M., Heddi,B. and Phan,A.T. (2012) NMR spectroscopy of
G-quadruplexes. Methods, 57, 11–24.
4
5. Povarova,N.V., Zaitseva,S.O., Baleeva,N.S., Smirnov,A.Y.,
Myasnyanko,I.N., Zagudaylova,M.B., Bozhanova,N.G.,