Content uploaded by Edward A Curtis
Author content
All content in this area was uploaded by Edward A Curtis on Jun 14, 2024
Content may be subject to copyright.
Nucleic Acids Research , 2024, 1–13
https://doi.org/10.1093/nar/gkae467
Nucleic Acid Enzymes
Aurora: a uorescent deoxyribozyme for high-throughput
screening
Martin Volek
1 , 2
, Jaroslav Kurfürst
1 , 3
, Matúš Drexler
1
, Mic hal S v oboda
1
, Pa v el Srb
1
,
V áclav V everka
1 , 4 and Edward A. Curtis
1 ,
*
1
Institute of Organic Chemistry and Biochemistry of the Czech Academy of Sciences, Prague 166 10, Czech Republic
2
Department of Genetics and Microbiology, Faculty of Science, Charles University in Prague, Prague 128 44, Czech Republic
3
Department of Informatics and Chemistry, University of Chemistry and Technology, Prague 166 28, Czech Republic
4
Department of Cell Biology, Faculty of Science, Charles University in Prague, Prague 128 44, Czech Republic
*
To whom correspondence should be addressed. Te l : +420 733 169 654; Email: curtis@uochb.cas.cz
Abstract
Fluorescence facilitates the detection, visualization, and tracking of molecules with high sensitivity and specicity. A functional DNA molecule
that generates a robust uorescent signal would offer signicant advantages for many applications compared to intrinsically uorescent proteins,
which are e xpensiv e and labor intensiv e to synthesiz e, and uorescent RNA aptamers, which are unstable under most conditions. Here, we
describe a no v el deo xyribo yzme that rapidly and efciently generates a stable uorescent product using a readily a v ailable coumarin substrate.
An engineered version can detect picomolar concentrations of ribonucleases in a simple homogenous assay, and was used to rapidly identify
no v el inhibitors of the SARS-CoV-2 ribonuclease Nsp15 in a high-throughput screen. Our work adds an important new component to the toolkit
of functional DNA parts, and also demonstrates how catalytic DNA motifs can be used to solve real-world problems.
Gr aphical abstr act
Introduction
Fluorescence makes it possible to detect, visualize, and track
molecules with high sensitivity and specicity. It also facili-
tates analysis of dynamic interactions important for molecu-
lar function. Fluorescence-based techniques are widely used in
microscopy , immunology , cell sorting, DNA sequencing, diag-
nostics, and microarrays, and new applications continue to be
developed. Such techniques offer a number of advantages rel-
ative to those with other types of readouts: they are typically
more sensitive than colorimetric assays, offer greater exibil-
ity and control over experimental readouts that chemilumi-
nescent ones, and are safer than radioactive assays. One of the
most powerful uorescent tools is the uorescent protein GFP
( 1 ,2 ). Originally identied in the jellysh Aequorea victoria ,
uorescent proteins have now been discovered in a wide range
of organisms. The properties of these proteins have been en-
hanced by engineering, and variants have been developed that
fold more efciently, function over a wide range of conditions,
and generate uorescent signals with different colors. Such
proteins have greatly facilitated studies of protein expression
and localization. More recently, SELEX has been used to iden-
tify uorescent RNA aptamers with a wide range of functional
properties ( 3 ,4 ). These aptamers provide a way to investigate
the functions of cellular RNA molecules, and engineered vari-
ants can also be used to monitor metabolite concentrations in
real time. Although extremely useful for studies of biological
systems, such motifs are less suitable for in vitro applications
such as high-throughput screening. In the case of proteins this
is related to both time and money: proteins are generally ex-
pensive and time-consuming to produce, and more difcult
to evolve than nucleic acids. On the other hand, a signicant
limitation of uorescent RNA aptamers is that they are unsta-
ble under many conditions due to the ubiquitous presence of
ribonucleases ( 5 ).
DNA is another type of polymer capable of sophisticated
functions, and functional DNA molecules such as aptamers
Received: February 12, 2024. Revised: May 7, 2024. Editorial Decision: May 15, 2024. Accepted: May 23, 2024
©The Author(s) 2024. 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-NonCommercial License
(https: // creativecommons.org / licenses / by-nc / 4.0 / ), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the
original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
2 Nucleic Acids Research , 2024
and deoxyribozymes can be useful alternatives to functional
protein and RNA motifs ( 5–8 ). DNA can be chemically syn-
thesized at low cost, is stable over a wide range of conditions
(including in the presence of ribonucleases, which are often
present in samples), can typically be denatured and refolded
without losing activity, and can be readily engineered using
articial evolution. Motifs with new functions (such as de-
oxyribozymes allosterically regulated by ligands) can be con-
structed by rational design or selection ( 9 ). And powerful en-
zymatic methods such as the polymerase chain reaction (PCR)
( 10 ) and rolling circle amplication (RCA) ( 11 ) can be used to
copy and therefore amplify the signals generated by functional
DNA. Despite these advantages, few methods to generate uo-
rescent signals using functional DNA motifs have been devel-
oped ( 12–17 ). One approach uses DNA aptamers that bind
and enhance the uorescence of ligands ( 12–14 ). The signal
to noise ratios of these aptamers rarely exceed 100-fold, and
tend to be signicantly lower than those of their RNA coun-
terparts ( 14 ). In addition, because the uorophore must re-
main associated with the aptamer to generate a signal, this
approach provides a less permanent and robust readout than
a signal generated by a catalyst or enzyme. Another method
utilizes a nonspecic peroxidase reaction catalyzed by DNA
G-quadruplexes in the presence of hydrogen peroxide and a
hemin cofactor ( 15 ,16 ). Although typically used to generate a
colorimetric product, a uorogenic signal can also be gener-
ated when this reaction is performed using phenolic substrates
such as tyramine ( 17 ). Because the peroxidase reaction is also
promoted by hemin itself, this method suffers from high back-
ground. Hydrogen peroxide is also incompatible with some
types of assays, and high concentrations can inactivate the
hemin cofactor. These limitations highlight the need for new
and complementary methods to generate uorescent signals
using DNA.
In this study we used in vitro selection to identify a de-
oxyribozyme that generates a uorescent signal by converting
the coumarin substrate 4-MUP into the uorescent product 4-
MU ( 18 ). In a complementary study, a similar approach was
employed to identify a deoxyribozyme that generates a col-
orimetric signal by converting the colorless substrate pNPP
into the yellow product pNP ( 19 ). Our deoxyribozyme, which
we named Aurora, offers a number of advantages relative to
existing methods. Aurora works under mild conditions and
uses an inexpensive and commercially available substrate. It
is small, label free, and can be rapidly synthesized at low cost.
Aurora is a potent enhancer of uorescence, and generates a
signal in minutes with a signal to noise ratio of > 700. It is
highly specic for its substrate and orthogonal to a chemilu-
minescent deoxyribozyme previously discovered in our group
( 20 ). This means that it could potentially be useful for mul-
tiplex applications (e.g. by rst analyzing light production in
the absence of excitation and then, after the signal has de-
cayed, exciting the sample and analyzing uorescence). Au-
rora can be modied to only generate uorescence in the pres-
ence of an input of interest (such as a target molecule in a
sample). It is also useful for real-world applications: an en-
gineered variant can detect ribonuclease activity with a limit
of detection of ∼100 pM, and was used to identify small
molecule inhibitors of the S AR S-CoV-2 ribonuclease Nsp15 in
a high-throughput screen. Our results provide a new and im-
proved way to construct uorescent sensors using DNA. They
also show how such sensors can be used to solve real-world
problems.
Materials and methods
Oligonucleotides
Oligonucleotides were chemically synthesized by GENERI
BIOTECH s.r.o., Sigma-Aldrich or IDT and puried by PAGE
or HPLC. See Supplementary Tab l e S1 for the sequences of all
oligonucleotides used in this study.
Pool design
The library used in our initial selection (Pool 1 in
Supplementary Tabl e S1 ) was generated by randomly muta-
genizing the H1 variant of Supernova (a chemiluminescent
deoxyribozyme recently discovered in our group ( 20 )) at a
rate of 21% per position. A 20-nucleotide long primer-binding
site was also added to the 3
end. The library used for the re-
selection (Pool 2 in Supplementary Tabl e S1 ) was based on
the sequence of Hit10. This generated uorescence with the
highest signal to noise ratio of any of the deoxyribozymes
we tested from the initial selection. The 85 positions in Hit10
were randomly mutagenized at a rate of 21% per position and
a new 20-nucleotide long primer-binding site was added to the
3
end.
Initial selection
The single-stranded DNA pool (Pool1) and blocking oligonu-
cleotide (REV1) were mixed in Milli-Q water. After heating
at 65
◦C for 2 min and cooling at room temperature for 5
min, 5 ×selection buffer and then the disodium salt of the 4-
methylumbelliferyl phosphate substrate (4-MUP) were added.
Final concentrations were 1 μM Pool1, 1.5 μM REV1, 1 ×se-
lection buffer (200 mM KCl, 1 mM ZnCl
2
, 1 μM Ce(NO
3
)
4
,
0.1 μM PbCl
2
, 50 mM HEPES pH 7.4) and 1 mM 4-MUP.
After incubating for 2.4 h, DNA was concentrated by ethanol
precipitation. A short oligonucleotide (FWD1) was then lig-
ated to library members containing a 5
phosphate. To increase
the efciency of the ligation, the reaction was performed in
the presence of a splint oligo (Splint1) which was comple-
mentary to both FWD1 and the 5
end of Pool1. The liga-
tion reaction was incubated for 5 min at 37
◦C. Final con-
centrations were 2.5 μM Pool1, 2.5 μM FWD1, 2.5 μM
Splint1, 1 ×T4 DNA ligase buffer and 0.5 Wei ss units of
T4 DNA ligase per 1.0 μg of Pool1. DNA molecules were
then separated by 6% urea–PAGE and DNA molecules that
co-migrated with a 125-nucleotide marker were cut from the
gel, eluted and ethanol precipitated. They were then amplied
by PCR using Q5 HotStart DNA Polymerase and the FWD1r
and REV1p primers. Final concentrations were 500 ×di-
luted Pool1, 0.5 μM FWD1r, 0.5 μM REV1p, 1 ×Q5 re-
action buffer, 1 ×Q5 high GC enhancer, 0.2 mM dNTPs
and 0.02 U of Q5 HotStart DNA polymerase per 1 μl of the
PCR reaction mixture. Double-stranded PCR products were
isolated using a Macherey-Nagel PCR Clean-up kit. The re-
verse primer REV1p contained a 5
phosphate, and the strand
synthesized using this primer (which was complementary to
Pool1) was digested using λ-exonuclease. Final concentrations
were 5 μg of the double-stranded PCR product, 1 ×Lambda
Exonuclease reaction buffer and 1 μl (5 U) of Lambda Ex-
onuclease in a volume of 50 μl. The Lambda Exonuclease
mixture was incubated at 37
◦C for 1 h. The resulting single-
stranded DNA molecules (of length 125 nucleotides) were pu-
ried using a Macherey-Nagel PCR Clean-up kit. The FWD1r
primer used in the PCR contained a single RNA linkage at
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 3
its 3
end. This made it possible to regenerate the 5
end of
the Pool1 by base hydrolysis. To do this, DNA was heated at
65
◦C for 2 min, cooled at room temperature for 5 min, and
mixed with 10 ×hydrolysis buffer (1 ×hydrolysis buffer: 20
mM Tri zma base, 400 mM KOH , 4 mM EDTA). The RNA
linkage was then base hydrolyzed at 90
◦C for 10 min. The
resulting 105 nucleotide long DNA molecules (corresponding
to single-stranded Pool 1 molecules with a 5
hydroxyl group)
were then isolated by 6% urea–PAGE and ethanol precipita-
tion. After the fth round of selection the library was ampli-
ed by PCR, puried using a Macherey-Nagel PCR Clean-up
kit, and sequenced by Eurons Genomics using an amplicon
paired-end sequencing run.
Reselection
Reselection conditions were the same as those used in the ini-
tial selection except for the following differences. First, Pool2
was used instead of Pool1. Second, the library was incubated
with 4-MUP for 14.4 min rather than 2.4 h. Third, a new
blocking oligonucleotide and reverse primer (REV2 / REV2p)
was used ( Supplementary Tabl e S1 ). This library was se-
quenced after the sixth round by Eurons Genomics using an
amplicon paired-end sequencing run.
Analysis of uorescence production
Oligonucleotides corresponding to individual sequences from
evolved libraries were ordered from GENERI BIOTECH s.r.o.
Fluorescence production was measured as follows: oligonu-
cleotides were re-suspended in Milli-Q water, heated at 65
◦C
for 2 min, and cooled at room temperature for 5 min. After
adding 5 ×selection buffer or 5 ×Aurora buffer, samples
were transferred to a white half-area 96-well plate (Corn-
ing). 4-MUP was then added. In continuous assays, uo-
rescence was measured for 4 h using a Tecan Spark plate
reader (Tecan Group). In discontinuous assays, after incubat-
ing for a specic time, samples were quenched with 20 μl
of 1 M KOH and uorescence was measured using a plate
reader. In a typical experiment nal concentrations were 15
μM of the tested oligonucleotide and either 1 ×selection
buffer (200 mM KCl, 1 mM ZnCl
2
, 1 μM Ce(NO
3
)
4
, 0.1 μM
PbCl
2
, 50 mM HEPES pH 7.4) or 1 ×Aurora buffer (200
mM KCl, 1 mM ZnCl
2
, 50 mM HEPES pH 7.4, 5% (v / v)
DMSO) and 30 μM 4-MUP. Fluorescence was measured in a
white half-area 96-well plate (Corning) using a Tecan Spark
plate reader with the following settings: excitation 358 ( ±5)
nm, emission 455 ( ±5) nm, 97 nm wavelength gap, optimal
gain, 30 ashes, Z position calculated from one well in the
plate.
Analysis of phosphorylation
Oligonucleotides corresponding to individual sequences from
evolved libraries were ordered from GENERI BIOTECH
s.r.o., puried by 6% urea–PAGE or HPLC, and resuspended
in Milli-Q water. Self-phosphorylation reactions were per-
formed by rst heating deoxyribozymes at 65
◦C for 2 min
and cooling at room temperature for 5 min. After mixing with
5 ×selection buffer or 5 ×Aurora buffer, the 4-MUP substrate
was added. Final concentrations in a typical reaction were 1
μM deoxyribozyme, 1 ×selection buffer (200 mM KCl, 1 mM
ZnCl
2
, 1 μM Ce(NO
3
)
4
, 0.1 μM PbCl
2
, 50 mM HEPES pH
7.4) or 1 ×Aurora buffer (200 mM KCl, 1 mM ZnCl
2
, 50
mM HEPES pH 7.4, 5% (v / v) DMSO), and 1 mM 4-MUP
unless stated otherwise. Reactions were incubated for specic
times at room temperature and stopped by the addition of
EDTA to a nal concentration of 25 mM. Reactions were then
concentrated by ethanol precipitation, and reacted deoxyri-
bozymes (now containing a 5
phosphate) were ligated to a
short oligonucleotide as described in the section ‘Initial Selec-
tion.’ Reacted and unreacted molecules were separated by 6%
urea–PAGE. DNA was visualized by staining with GelRed us-
ing the protocol recommended by the manufacturer. Gels were
scanned using a Typhoon laser scanner and the percentage of
reacted and unreacted molecules was quantied using Image-
Quant TL software.
Calculation of signal to noise ratios
Signal to noise ratios were dened as the uorescence of a sam-
ple in the presence of deoxyribozyme divided by the uores-
cence of the sample in the absence of the deoxyribozyme. The
background signal was dened as the uorescence of 1 ×Au-
rora buffer (200 mM KCl, 50 mM HEPES, pH 7.4, 1 mM
ZnCl
2 and 5% (v / v) DMSO) and was subtracted before cal-
culating signal to noise ratios.
Optimization of reaction conditions
To maximize uorescence we searched for optimal reac-
tion conditions. The optimal DNA, 4-MUP, KCl, ZnCl
2 and
HEPES concentrations were determined by titration experi-
ments. We also tested the effects of different monovalent and
divalent metal ions, an organic solvent (DMSO), and a molec-
ular crowding agent (PEG 200) on activity. Titration experi-
ments to determine the optimal pH during and after the reac-
tion were also performed. Aurora 2 ( Supplementary Table S1 )
was used for these experiments if not stated otherwise. Ac-
tivity was measured by analysis of uorescence production
(using a plate reader assay) and self-phosphorylation (using
a ligation assay).
Kinetic measurements and analysis
Kinetic measurements were performed using a ligation assay.
Deoxyribozyme (either Aurora 1 or Aurora 2; Supplementary
Tabl e S1 ) was mixed with Milli-Q water, heated at 65
◦C for
2 min, and cooled at room temperature for 5 min. 5 ×Au-
rora buffer and 4-MUP were then added. Final concentra-
tions were 1 μM deoxyribozyme, 1 ×Aurora buffer (200
mM KCl, 1 mM ZnCl
2
, 50 mM HEPES pH 7.4, 5% (v / v)
DMSO) and 1 μM to 300 μM 4-MUP. Reactions were in-
cubated for specic times at room temperature and stopped
by the addition of EDTA to a nal concentration of 25 mM.
Reactions were stopped at time points that corresponded to
the linear phase of the reaction. After ethanol precipitation,
reacted deoxyribozyme (containing a 5
phosphate) was lig-
ated to a short oligonucleotide as described in the section
‘Initial Selection’. Reacted and unreacted molecules were sep-
arated by 6% urea–PAGE. DNA was visualized by staining
with GelRed using the protocol recommended by the manu-
facturer and gels were scanned using a Typhoon laser scanner.
The percentage of reacted and unreacted deoxyribozyme was
quantied using ImageQuant TL software. k
cat
and K
m
values
were obtained using Prism 9 software. Curves were t using
the equations: V
0
= V
max
×[S] / ( K
m
+ [S]) to obtain K
m
and
k
cat
= V
max
/ [E] to obtain k
cat
.
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
4 Nucleic Acids Research , 2024
Next generation sequencing and data analysis
All libraries were sequenced by Eurons Genomics using
amplicon paired-end sequencing runs. Raw reads were pro-
cessed using a pipeline consisting of adaptor trimming (cu-
tadapt v1.18), read merging (fastq-join v1.3.1), unifying of
read orientation (fastx barcode splitter), primer clipping (cu-
tadapt v1.18), length ltering (cutadapt v1.18) and counting
of unique sequences (bash). All further analysis was performed
using in house python scripts available at https://github.com/
Jardic/aurora _ selection _ analysis .
Oligonucleotide detection using an engineered
version of Aurora
The oligonucleotide sensor was mixed with the target oligonu-
cleotide in water, heated at 98
◦C for 2 min, and immediately
cooled on ice for 5 min. 5 ×Aurora buffer and DMSO were
then added. Samples were transferred to a white half-area 96-
well plate (Corning), 4-MUP was added, and the reaction mix-
ture was incubated for 4 h at room temperature. Final concen-
trations were 5 μM of the oligonucleotide sensor, 10 μM of the
target oligonucleotide, 1 ×Aurora buffer (200 mM KCl, 50
mM HEPES pH 7.4, 1 mM ZnCl
2
and 5% (v / v) DMSO) and
30 μM 4-MUP. After 4 h the reaction was stopped by adding
20 μl of 1 M KOH, and uorescence was then measured us-
ing a Tecan Spark plate reader. Analysis of uorescence pro-
duction was performed as described below in ‘Calculation of
signal to noise ratios’.
RNase A sensor based on Aurora
The RNase A sensor was heated at 65
◦C for 2 min, and cooled
at room temperature for 5 min. Then 5 ×Aurora buffer and
DMSO were added. Samples were transferred to a white half-
area 96-well plate (Corning), and 4-MUP and either RNase
A (Thermo Fisher Scientic) alone or RNase A and RiboLock
(Thermo Fisher Scientic) were added. The reaction mixture
was incubated for 4 h at room temperature. Final concentra-
tions were 5 μM of the RNase A sensor, 500 nM RNase A or
500 nM RNase A plus 500 nM RiboLock, 1 ×Aurora buffer
(200 mM KCl, 50 mM HEPES pH 7.4, 1 mM ZnCl
2
and 5%
(v / v) DMSO) and 30 μM 4-MUP if not stated otherwise. After
4 h the reaction was stopped by adding 20 μl of 1 M KOH to
the reaction mixture. Fluorescence was then measured using a
Tecan Spark plate reader. Analysis of uorescence production
was performed as described below in the section ‘Calculation
of signal to noise ratios’.
Plasmid construction, expression, and purication
of Nsp15 from SARS-CoV-2
Nsp15 cloning, expression and purication were performed
as described in Kim et al. ( 21 ) with minor modications. A
synthetic DNA sequence encoding an Esc heric hia coli codon
optimized version of Nsp15 was cloned into a pMCSG7
vector using Gibson assembly. Cloning was conrmed by
Sanger sequencing. The nal pS AR S-CoV-2-Nsp15_6 ×His
vector encoded the full-length Nsp15 protein fused to an N-
terminal hexahistidine tag via a TEV protease cleavage site.
E. coli NiCo21(DE3) cells (New England Biolabs) were trans-
formed with this plasmid. For large-scale expression and pu-
rication, a 3 l culture of LB medium was grown at 37
◦C in
a LEX bioreactor (Epiphyte3) in the presence of 100 μg / ml
ampicillin. Once the culture reached OD
600
∼1.0, asks were
moved to an 18
◦C bioreactor bath and supplemented with
0.1% glucose and 40 mM K
2
HPO
4
(nal concentration). Pro-
tein expression was induced by the addition of 0.2 mM IPTG
for 16 h at 18
◦C. Bacterial cells were harvested by centrifu-
gation at 7000g and cell pellets were resuspended in 40 ml
lysis buffer (50 mM HEPES, 500 mM NaCl, 5% [v / v] glyc-
erol, 20 mM imidazole, 10 mM β-mercaptoethanol, pH 8.0)
per liter of culture and lysed using a CF1 high-pressure ho-
mogenizer. Cellular debris was removed by centrifugation at
25 000g for 40 min at 4
◦C. The supernatant was ltered
through a 0.45 μm lter, mixed with 2 ml of Ni
2+ Sepharose
equilibrated with lysis buffer, and the suspension was added
to a gravity-ow column. Unbound proteins were removed
by washing with 40 ml of lysis buffer. Bound proteins were
eluted with 10 ml of lysis buffer supplemented with 500 mM
imidazole pH 8.0. A nal purication was performed using
a Superdex 200 column equilibrated in lysis buffer in which
10 mM β-mercaptoethanol was replaced by 1 mM TCEP.
Fractions containing Nsp15 were collected. Lysis buffer was
replaced with storage buffer (150 mM NaCl, 20 mM HEPES
pH 7.5, 1 mM TCEP) via repeated concentration and dilution
using a 30 kDa MWCO lter (Amicon-Millipore). The nal
protein sample was concentrated to 1 mg / ml, aliquoted, snap
frozen in liquid nitrogen and stored at –80
◦C until further
use.
High-throughput screen for Nsp15 inhibitors using
an Aurora Nsp15 sensor
Small molecules from a 1000-member fragment screen library
(Maybridge) were transferred to the wells of 384-well plates
using an Echo 550 liquid handler. Nsp15 protein in 1 ×Nsp15
buffer (50 mM KCl, 20 mM HEPES pH 7.4, 5 mM MnCl
2
,
0.003% (v / v) Tween20) was then added using a CERTUS
Flex liquid handler. After mixing, the Aurora Nsp15 sensor
(in 1 ×Nsp15 buffer) was added using a CERTUS Flex liq-
uid handler. Reactions were mixed again. Final concentrations
were 25 μM Aurora Nsp15 sensor, 400 nM Nsp15 protein,
1 ×Nsp15 buffer (50 mM KCl, 20 mM HEPES pH 7.4 and
5 mM MnCl
2
), 0.003% (v / v) Tween20, and 200 μM small
molecule from the fragment screen library in a volume of 20
μl. After incubating at room temperature for 1 h to allow
Nsp15 to cleave and activate the Aurora Nsp15 sensor, 80 μl
of 1 ×Aurora reaction mixture (50 mM KCl, 20 mM HEPES
pH 7.4, 1.25 mM ZnCl
2
, 6.25% (v / v) DMSO and 18.75 μM
4-MUP) was added using a CERTUS Flex liquid handler. The
ZnCl
2 in this buffer inhibited the Nsp15 protein while acti-
vating Aurora for catalysis. Final concentrations were 5 μM
Aurora Nsp15 sensor, 80 nM Nsp15 protein, 40 μM small
molecule from the fragment screen library, 1 ×Aurora / Nsp15
buffer (50 mM KCl, 20 mM HEPES pH 7.4, 1 mM ZnCl
2
1 mM MnCl
2
), 0.0006% Tween20, 5% (v / v) DMSO, and
15 μM 4-MUP in a volume of 100 μl. The reaction mixture
was incubated at room temperature for 4 h, and uorescence
was measured using a Tecan Spark plate reader. Fluorescence
was measured in a black at bottom 384-well plate (Corn-
ing). Analysis of uorescence production was performed as
described in the section ‘Calculation of signal to noise ratios’.
A counter screen was also performed to conrm that the in-
hibitors identied in the initial screen inhibit the Nsp15 pro-
tein rather than Aurora. The counter screen was performed as
described above, but Aurora 2 was used instead of the Aurora
Nsp15 sensor.
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 5
High-throughput screen for Nsp15 inhibitors using
a FRET assay
Small molecules from a 1000-member fragment screen li-
brary (Maybridge) were transferred to the wells of 384-well
plates using an Echo 550 liquid handler. Nsp15 protein in
1 ×Nsp15 buffer (50 mM KCl, 20 mM HEPES pH 7.4,
5 mM MnCl
2
, 0.003% (v / v) Tween20) was then added us-
ing a CERTUS Flex liquid handler. After mixing, the FRET
substrate (5
-FAM-AAArUAA-BHQ1-3
) in 1 ×Nsp15 buffer
was added using a CERTUS Flex liquid handler. Reactions
were mixed again. Final concentrations were 25 μM FRET
substrate, 400 nM Nsp15 protein, 1 ×Nsp15 buffer (50 mM
KCl, 20 mM HEPES pH 7.4 and 5 mM MnCl
2
), 0.003%
(v / v) Tween20, and 200 μM small molecule from the frag-
ment screen library in a volume of 20 μl. Fluorescence was
measured every 20 min for 1 h in a black at bottom 384-
well plate (Corning) using a Tecan Spark plate reader with
the following settings: excitation 485 ( ±5) nm, emission 527
( ±5) nm, gain 134, 30 ashes, Z position calculated from the
well.
Analysis of data from high-throughput screens
Wells containing 1 mM ZnCl
2 (which inhibited Nsp15 at
this concentration) served as negative controls, and were used
to determine background levels of uorescence. The average
value of this background was subtracted from the uorescence
values obtained from all other wells. Wells containing aliquots
of DMSO alone rather than DMSO plus small molecule were
used as positive controls. After subtraction of the background,
the average value of these positive controls was dened as
100% Nsp15 activity. Activity of Nsp15 in the presence of
small molecules from the fragment screen library was calcu-
lated relative to this positive control value. Z- factors ( 22 ) were
calculated for each 384-well plate to determine the quality of
the screen. Z -factors were calculated using the equation Z -
factor = 1 –[3( σp + σn
)] / ( μp - μn
) where μp is the mean
uorescence of the positive control, μn is the mean uores-
cence of the negative control, σp
is the standard deviation of
the mean uorescence of the positive control, and σn is the
standard deviation of the mean uorescence of the negative
control.
Calculation of IC50 values
IC50 values were measured for small molecules that strongly
inhibited Nsp15 in high-throughput screens. Solutions con-
taining different concentrations of these inhibitors were trans-
ferred to the wells of 384-well plates using an Echo 550 liquid
handler. To obtain the same volume in each well, the drops
containing small molecules were backlled with DMSO to
200 nl. For each inhibitor characterized, IC50 values were
measured using both the Aurora Nsp15 sensor and using the
FRET assay. After determining the relative activity of Nsp15
at each concentration of inhibitor, IC50 values were calculated
using Prism 9 software (GraphPad).
NMR experiments
HPLC-puried DNA was purchased from GENERI
BIOTECH s.r.o. DNA was re-suspended in Milli-Q wa-
ter, heated at 65
◦C for 2 min, cooled at room temperature for
5 min, and 5 ×Aurora buffer was then added. Concentrations
at this point were 15 μM DNA, 200 mM KCl, 50 mM HEPES,
pH 7.4 and 1 mM ZnCl
2
. Samples were concentrated using
Ultra-Amicon Centrifugal Filter Units (cutoff 3 kDa) to 500
μM DNA and a 1.5 molar excess of 4-MUP, D
2
O and DSS
were added. Final concentration were 500 μM DNA, 200 mM
KCl, 50 mM HEPES, pH 7.4, 1 mM ZnCl
2
, 750 μM 4-MUP,
10% (v / v) D
2
O and a trace amount of DSS. NMR experi-
ments were performed on a Bruker Avance III HD 850 MHz
system equipped with an inverse triple resonance cryo-probe.
Spectral analyses were performed using TOPSPIN (Bruker)
software ( 23 ).
Results and discussion
Discovery of the uorescent deoxyribozyme Aurora
Supernova is a deoxyribozyme recently discovered in our lab-
oratory (Figure 1 A) ( 20 ). It transfers the phosphate group
from the 1,2-dioxetane substrate CDP-Star (Figure 1 B) to
its own 5
hydroxyl group, which triggers a chemically ini-
tiated electron exchange luminescence reaction and a ash of
blue light ( 24–26 ). Deoxyribozymes that use substrates which
generate orthogonal signals when they are dephosphorylated
would bring new functionality to the toolkit of functional
DNA parts. An example of such a substrate is the coumarin
4-MUP (Figure 1 C) ( 18 ). Dephosphorylation of 4-MUP yields
the uorescent compound 4-MU (Table 1 ) ( 27 ,28 ), and a de-
oxyribozyme that promotes this reaction could in principle
be used to generate a uorescent signal. To search for such
a deoxyribozyme, a library was generated by randomly mu-
tagenizing the sequence of Supernova at a rate of 21% per
position (Figure 1 D). We used Supernova as the starting point
for our library because this deoxyribozyme catalyzes a phos-
phoryl transfer reaction using a substrate with some similar-
ities to 4-MUP (compare Figure 1 B and C). After incubat-
ing with 4-MUP, library members containing a 5
phosphate
group were tagged by ligation, puried by PAGE , and am-
plied by PCR (Figure 1 E). After four rounds of selection
activity was detected (Figure 1 F), and after one more round
the library was characterized by high-throughput sequencing.
Sequences from the evolved library with high read numbers
could phosphorylate themselves in the presence of 4-MUP
and also generate uorescence ( Supplementary Figure S1 and
Supplementary Tabl e S1 ). However, most appeared to be
structurally and functionally distinct from Supernova (see also
reference ( 29 )). We initially appreciated this point by compar-
ing the mutational distances of sequences from Supernova in
a library separately challenged with two different substrates
(Figure 1 D). When a selection was previously performed to
identify variants in this library that used CDP-Star (i.e. the
original substrate) with improved efciency ( 20 ), the average
mutational distance of sequences in the evolved pool from Su-
pernova was 18.42 (Figure 1 G, blue peak). In contrast, the av-
erage mutational distance of variants in this library that used
4-MUP was 30.68 (Figure 1 G, orange peak; compare also to
Figure 2 a of reference ( 29 )), suggesting that deoxyribozymes
that use CDP-Star and 4-MUP form different structures. Anal-
ysis of individual sequences from the evolved library pro-
vided additional support for this idea: most contained mu-
tations that were not consistent with the sequence require-
ments of Supernova ( Supplementary Figure S2 ). The substrate
specicities of these new deoxyribozymes also differed from
that of Supernova. For example, the deoxyribozyme Aurora
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
6 Nucleic Acids Research , 2024
A
G3’
CT
G
G
G
T
G
A
T
C
C
C
C
G
G
G
G
G
G
G
G
T
AG
A
AA
A
A
T
G
A
5’
A
A
A
A
A
A
A
T
G
G
A
AG
Incubate
with
4-MUP
Tag
by
ligation
PCR and
strand
separation
Purify
by
PAG E
P
PP
P
Mutational distance from Supernova
10 20 30 40 50 60 70
10
8
6
4
2
0
Percent of sequences
CDP-Star
4-MUP
SN
R
U
SNAur Aur
Sub
CDP-Star 4-MUP
-+-+-+-+
12345
0
10
20
30
40
Round
% reacted
Selection
4-MUP
-
-
-
-
OCH3
CDP-Star 4-MUP Supernova
library
Selection
CDP-Star Selection
4-MUP
BCD
E
FG
H
Luminescent
DNAzymes
(previous work)
Fluorescent
DNAzymes
(this work)
Figure 1. Identication of deoxyribozymes that generate uorescence. ( A ) Secondary str uct ure of Supernova, a chemiluminescent deoxyribozyme
previously isolated in our group. ( B ) Chemical str uct ure of CDP-Star, the substrate used by Supernova. ( C ) Chemical str uct ure of 4-MUP, the substrate
used in this study. ( D ) Wo rk ow of a previous selection (in which deoxyribozymes that react with the original CDP-Star substrate were isolated from a
library of variants of Supernova) and the selection performed here (in which deoxyribozymes that react with the substrate 4-MUP were isolated from the
same library). ( E ) Articial e v olution protocol to identify deoxyribozymes that phosphor ylate themselves in the presence of 4-MUP. ( F ) Progress of the
selection for deoxyribozymes that can react with 4-MUP. ( G ) Distribution of mutational distances of sequences in a library of Superno v a v ariants relativ e
to Superno v a itself after selection f or the abilit y to react with CDP-St ar (blue) or 4-MUP (orange). ( H ) T he substrate specicities of Superno v a and Aurora
are orthogonal. Superno v a (labeled ‘SN’) and Aurora 1 (labeled ‘A u r ’ ) were each incubated separately with CDP-Star (left) or 4-MUP (right). Time points
w ere analyz ed using the ligation assa y. See Supplementary Tab l e S1 f or the sequence of Aurora 1.
1 could use 4-MUP but not CDP-Star as a substrate, whereas
Supernova reacted efciently with CDP-Star but not 4-MUP
(Figure 1 H). Similar results were obtained in a complemen-
tary study in which we selected for library members that re-
act with the colorimetric substrate pNPP ( 19 ). These results
demonstrate that our method can be used to identify new u-
orescent deoxyribozymes. They also suggest that, despite be-
ing isolated from a library based on Supernova, most variants
in this library that use 4-MUP as a substrate form structures
that are distinct.
The catalytic core of Aurora is a 47-nucleotide
bulged hairpin
We chose the uorescent deoxyribozyme from the initial se-
lection with the highest activity for further characterization
(Figure 2 A). We named this sequence Aurora 1 full-length,
and the catalytic motif encoded by this sequence Aurora. One
Ta b l e 1. Fluorescent properties of 4-MU, the uorescent product gener-
ated by Aurora. Values we re measured at pH 10 ( 27 , 28 )
Property Val u e
Maximum excitation wavelength ( λmax
) 360 nm
Maximum emission wavelength ( λem
) 450 nm
Extinction coefcient ( ε ) 17 000 M
−1
cm
−1
Quantum yield ( ) 0.63
Brightness ( ε ×) 10 710 M
−1
cm
−1
goal was to characterize the secondary structure, sequence re-
quirements, and minimized catalytic core of Aurora. Another
was to identify variants with improved catalytic efciencies.
To address both of these goals, we synthesized a second li-
brary by randomly mutating the sequence of Aurora 1 full-
length at a rate of 21% per position ( 29–31 ). At this rate of
mutagenesis, all possible variants within about four mutations
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 7
A
New
fluorescent
DNAzymes
Pick single
DNAzyme
Library of
Aurora
variants
Selection
4-MUP
Active
variants
of Aurora
Analysis of
structure
Full-length
version of
Aurora 1
Random
mutagenesis
Secondary
structure Sequence
requirements
Aurora
1
Aurora
2
B
C
D
G
1H (ppm)
15 14 13 11 10 98
12
Library of
Supernova
variants
Selection
4-MUP
4-MUP concentration (μM)
100
10-1
10-2
10-3
10-4
10-5
10-6
10-7
0.1 1 10 100 1000
kobs (min-1)
Aurora 1
Aurora 2
02
Conservation (bits)
E
C
C
C
G
CG
G
G
G
T
G
T A
CG
CG
G
T
C
A
T
G
T
G
T
A
T
G
T
3’ G
G
T
AG
A
G
5’
A
G
6
4
12
45
17
22
27 29
34
40
A
A
G
G
A
AG
G
Aurora
1
F
C
C
T
G
CG
G
G
G
T
G
T A
CG
CG
G
T
C
A
T
G
T
G
T
A
T
G
T
3’ G
G
T
AG
A
G
5’
A
G
6
4
12
45
17
22
27 29
34
40
G
T
A
T
T
TA
A
Aurora
2
471011114293223 2617 20 41 44 47
Del Del
35 38
0.0
0.5
1.0
1.5
Info (bits)
0
30
60
90
Relative % activity
120
TA CA TG C G
17 and 40
Figure 2. Sequence requirements and secondary str uct ure of the uorescent deoxyribozyme Aurora. ( A ) Evolutionary lineage of Aurora 1 (the minimized
catalytic core of the initial isolate of Aurora) and Aurora 2 (an optimized variant isolated from a randomly mutagenized library based on Aurora 1
full-length). ( B ) Sequence logo generated from analysis of variants of Aurora using high-throughput sequencing. ( C ) Proton NMR spectrum of the 17C
40G varia nt of Aurora ( Supplementar y Ta b le S1 ) showing chemical shifts consistent with base pairs. ( D ) Double-mutant cycle showing that positions 17
and 40 interact in a w a y that is consistent with base pairing. ( E ) Secondary str uct ure model of Aurora 1. Base pairs are shown using solid black lines,
interactions supported by mutual information analysis are shown in maroon, and the degree of conser vation at each position is indicated by blue
shading. ( F ) Secondary str uct ure model of A urora 2. Positions that differ from A urora 1 are shown in orange. ( G ) Cat alytic activit y of Aurora 1 and 2 o v er a
range of 4-MUP concentrations as measured using a ligation assay.
of the starting sequence were expected to be present in the li-
brary ( Supplementary Table S2 ) ( 32 ). Catalytically active vari-
ants were then identied by articial evolution and character-
ized by high-throughput sequencing ( Supplementary Figure S3
and Supplementary Tab l e s S3 –S5 ). Initial analysis of these se-
quences revealed two highly conserved regions (corresponding
to nucleotides 1–10 and 43–79) separated by 32 less conserved
positions (Figure 2 B). The catalytic activity of a 47 nucleotide
deoxyribozyme in which the nucleotides at positions 11–42,
80–85, and in the 3
primer binding site were deleted (called
Aurora 1) was similar to that of the full-length sequence
( Supplementary Figure S4 ). The proton NMR spectrum of the
17C 40G mutant of Aurora 2 suggests that Aurora forms a
structure containing multiple Watson–Crick base pairs (Figure
2 C). Consistent with this observation, comparative sequence
analysis ( 31 , 33 , 34 ) revealed four pairs of covarying positions
in the deoxyribozyme (positions 11 and 47, 12 and 46, 17
and 40, and 26 and 30) with mutational patterns consistent
with those expected of base pairs. Mutagenesis experiments in
which these putative base pairs were disrupted by point muta-
tions and restored by compensatory mutations generally pro-
vided strong additional support for the proposed interactions
(Figure 2 D and Supplementary Figure S5 ). When interpretated
in the context of the conserved nucleotides that ank these
base pairs, the 11–47, 12–46, 17–40 and 26–30 constraints
suggest that Aurora forms an 11-base pair hairpin interrupted
by an asymmetric bulge (Figure 2 E). This hairpin contains
two irregular features: a TT mismatch (which could reect
a noncanonical interaction) and a highly conserved bulged
guanine. The hairpin is capped by a three-nucleotide loop
formed by positions 27, 28 and 29. One of the most highly
enriched mutations identied in the selection (29 G to A) oc-
curred in this loop ( Supplementary Figure S3 ). In addition,
several correlations identied by mutual information analysis
(including 22–29, 22–27, 23–26, 26–29 and 22–30; Figure 2 E-
F and Supplementary Figure S6 ) suggest that this loop inter-
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
8 Nucleic Acids Research , 2024
acts with the conserved asymmetric bulge formed by positions
20–23 and 33–37 rather than extending into solution. Covari-
ation analysis suggests that nucleotides at the 5
end of Aurora
(which include the phosphorylation site) do not form canoni-
cal base pairs with one another or with the rest of the deoxyri-
bozyme. However, a network of correlations among positions
4, 6 and 45 (including an AT to GA covariation between po-
sitions 4 and 45 which is one of the strongest in the dataset) is
consistent with a tertiary interaction that anchors the 5
end
of Aurora to the rest of the catalytic core. It is possible that
this interaction helps to position the nucleophilic 5
hydroxyl
group in the vicinity of the phosphate group of 4-MUP. Several
additional correlations (including the 16–34, 6–10, 6–46, 6–
12, 6–9, 16–37 and 6–42 pairs; Supplementary Figure S6 ) are
consistent with the idea that the 5
end of Aurora is also facing
the asymmetric bulge. If this is the case, the overall architec-
ture of Aurora is likely a bent hairpin in which nucleotides
distant in both the primary sequence and secondary structure
converge on this bulge.
Aurora generates a robust uorescent signal
To evaluate the extent to which these articial evolution ex-
periments yielded improved variants of Aurora, we compared
the catalytic activity of the initial isolate (Aurora 1; Fig-
ure 2 E) with that of the variant with the highest read num-
ber from the randomly mutagenized library (Aurora 2; Fig-
ure 2 F) (see also Figure 2 A for more information about the
evolutionary lineage of these deoxyribozymes). Each vari-
ant was characterized in the context of the 47-nucleotide
minimized catalytic core, and measurements were performed
over a range of 4-MUP concentrations using a ligation assay
(which measures the extent of self-phosphorylation). These
experiments revealed that the catalytic activity of Aurora 2
was more than 100-fold higher than that of Aurora 1 at
some substrate concentrations (Figure 2 G). Most mutations
in Aurora 2 occurred in either the asymmetric bulge (posi-
tions 20–23 and 33–37) or the loop (positions 27–29; Fig-
ure 2 F), highlighting the importance of these parts of the de-
oxyribozyme. Surprisingly, 4-MUP concentration affected ac-
tivity in a slightly cooperative way, and evidence for coop-
erativity was also observed in proton NMR experiments in
which Aurora folding was characterized as a function of 4-
MUP concentration ( Supplementary Figure S7 ). This could
indicate that Aurora contains multiple binding sites for 4-
MUP, or a single site that binds multiple 4-MUP molecules.
At saturating substrate concentrations, the rate of Aurora 2
of 0.18 min
−1 (Figure 2 G) was similar to the k
cat of Super-
nova ( 20 ) of 0.15 min
−1
. The concentration of substrate at
which activity was half maximal (30 μM for Aurora and
150 μM for Supernova) was also comparable for these two
deoxyribozymes.
In a complementary series of experiments, we investigated
the extent to which Aurora enhances uorescence. When us-
ing an experimental setup in which 4-MUP and buffer was
mixed with Aurora 2 and uorescence was continuously mon-
itored using a plate reader (Figure 3 A), signal to noise ratios
of 10-fold were obtained in minutes and 100-fold in hours
(Figure 3 B). A stable signal was observed over the course of
this experiment, indicating that the product is relatively sta-
ble under these conditions. When using a discontinuous setup
in which reactions were quenched with base before measure-
ment (which can enhance uorescence and also increase the
stability of the uorescent product ( 35 )), signal to noise ra-
tios were about 6-fold higher, and values exceeding 700-fold
could be achieved (Figure 3 C). In both assays, the uorescent
signal generated by Aurora 2 was at least 10-fold higher than
that of Aurora 1 ( Supplementary Figure S8 ). Maximum signal
to noise ratios in both the absence of base (continuous assay)
and the presence of base (discontinuous assay) were similar
to those obtained from samples containing synthetic 4-MU at
the same concentration as that of 4-MUP used in our assays
( Supplementary Figure S9 ). This indicates that Aurora gener-
ates the maximum possible signal to noise ratio for 4-MUP
in solution, although it is possible that higher signal to noise
ratios could be achieved by deoxyribozymes that enhance the
uorescence of 4-MU when it is bound to the deoxyribozyme
( 3 , 4 , 12–14 ).
Aurora requires multiple zinc ions for structure and
function
Although our selection experiments provided extensive infor-
mation about the sequence requirements of Aurora, they re-
vealed little about how external factors (such as components
of the buffer) inuence the reaction. Such factors can signi-
cantly affect signal to noise ratios, and can also provide clues
about catalytic mechanisms. We were especially interested in
the effects of metal ions on the reaction because they can play
both structural and catalytic roles in ribozymes and deoxyri-
bozymes ( 36 ,37 ). Our survey of reaction conditions revealed
both differences and similarities between Aurora and Super-
nova ( Supplementary Figures S10 - S20 ) ( 20 ,38 ) as well as be-
tween Aurora and the colorimetric deoxyribozyme Apollon
( 19 ). An important difference was that Aurora appears to re-
quire monovalent ions for activity ( Supplementary Figures
S10 - S12 ) while Supernova ( 20 ,38 ) and Apollon ( 19 ) do not.
On the other hand, Aurora ( Supplementary Figure S13 - S14 ),
Supernova ( 20 ,38 ) and Apollon ( 19 ) each require zinc. The de-
pendence of catalytic rate on zinc concentration is also highly
cooperative for these three deoxyribozymes ( Supplementary
Figure S14 , ( 38 ,19 ), suggesting that multiple zinc ions are
needed for function. This is intriguing because zinc ions play
catalytic roles in some protein enzymes (such as alkaline phos-
phatase ( 39 ,40 )) that catalyze reactions similar to that pro-
moted by Aurora. To determine whether these metal ion re-
quirements in part reect structural roles, proton NMR was
used to directly monitor the effects of different metal ions
on deoxyribozyme folding. The results of these experiments
were similar to those that used catalytic activity as a readout.
For example, chemical shifts consistent with canonical base
pairs were observed in a buffer that contained zinc and potas-
sium, but not in buffers that lacked either zinc or potassium
( Supplementary Figure S10 ). NMR experiments also provided
additional evidence that zinc affects Aurora folding in a highly
cooperative way ( Supplementary Figure S20 ). Zinc also plays
an important role in deoxyribozymes that cleave DNA ( 41–
43 ), suggesting a more general role for this ion in the con-
text of nucleic acid enzymes that promote phosphoryl trans-
fer reactions ( 44 ). Tak e n together, these experiments indicate
that both zinc and a monovalent metal ion (but not neces-
sarily potassium) are needed for Aurora folding and function.
They also highlight possible mechanistic similarities among
the chemiluminescent, uorescent, and colorimetric deoxyri-
bozymes recently identied in our group.
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 9
Continuous
assay
0 50 100 150 200 250
Time (minutes)
0
50
100
150
Signal to noise
0
200
400
600
0 50 100 150 200 250
Time (minutes)
Signal to noise
800 Discontinuous
assay
RNase
Inhibitor
Aurora sensor Aurora
0
50
100
150
200
250
Signal to noise
rC
5’
Inactive
form of
Aurora
Active
form of
Aurora
5’
OH
Aurora
catalysis
rC
5’
P P
5’
P
Cleavage
of Aurora
Turnover
of RNase
RNase
RNase RNase
5’
OH
A
B
C
D
E
F
G
Continuous
assay Discontinuous
assay
No
quenching Quench with
base
4-MUP
P
5’
P
5’ 3’
0 0.1 1 10 100 1000 10000
10
100
1000
RNase A concentration (nM)
Signal to noise
Background
Figure 3.
Aurora generates a robust uorescent signal. ( A ) Wor kow of continuous and discontinuous assays using Aurora. ( B ) Example of a continuous
assay in which the reaction is continually monitored in a plate reader. ( C ) Example of a discontinuous assay in which time points are quenched with base
before measuring uorescence. ( D ) Design of a ribonuclease sensor based on Aurora that is activated by RNA cleavage. ( E ) Amplication of the
single-turno v er signal generated by Aurora in the presence of a ribonuclease that promotes a multiple turno v er reaction. ( F ) An Aurora sensor with the
architecture shown in panel D is activated by RNase A, but not when a ribonuclease inhibitor is present. In contrast, the catalytic activity of Aurora itself
is not affected by either RNase A or this ribonuclease inhibitor. ( G ) The Aurora sensor detects ribonuclease A with a limit of detection of 100 pM.
R eactions w ere incubated f or 4 h in the presence of the indicated concentration of RNase A, and af ter quenc hing with base, uorescence was
measured using a plate reader. The green box indicates the average plus or minus three standard deviations of the background signal to noise ratio
measured in the absence of RNase A. See Supplementary Figure S23 for more information about the detection limit of the sensor. Experiments shown
in panels B and C w ere perf ormed using Aurora 2, which those in panels F and G w ere perf ormed using the sensor sho wn in Supplementary Figure S22 .
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
10 Nucleic Acids Research , 2024
Engineered forms of Aurora can detect ligands and
enzymes in solution
Vari a n t s of Aurora that only generate uorescence in the
presence of an input of interest could be useful for applica-
tions such as high-throughput screening and diagnostics. This
is especially true for variants that can be activated in solu-
tion without the need for wash steps or biochemical puri-
cations. To determine whether the catalytic activity of Au-
rora can be modulated by ligands, we used rational design
to construct a programmable sensor that only produces uo-
rescence in the presence of specic oligonucleotide sequences
( Supplementary Figure S21 ). This sensor produced signi-
cantly more uorescence in the presence of the target than
in its absence, could be programmed to detect a range of tar-
gets, and was only activated by oligonucleotides that it was
designed to detect ( Supplementary Figure S21 ). However, its
sensitivity was low, with a limit of detection of approximately
1 μM of target ( Supplementary Figure S21 ). This is likely re-
lated to catalytic turnover because, unlike classical enzymes, a
single molecule of Aurora can only generate one molecule of
uorescent product.
To improve sensitivity, we investigated whether it was pos-
sible to link the single-turnover signal generated by Aurora to
the catalytic activity of an enzyme that itself catalyzes a mul-
tiple turnover reaction. Because Aurora is made of DNA, we
expected that this type of coupling would be easiest to achieve
using enzymes that modify nucleic acids, and set out to de-
velop a variant of Aurora that is activated by enzymes that
cleave RNA. Our sensor was constructed by fusing a short
DNA oligonucleotide containing a ribonucleotide at its 3
end
to the 5
end of Aurora (Figure 3 d and Supplementary Figure
S22 ). Because Aurora uses its 5
hydroxyl group as the nucle-
ophile in the reaction, this modication was expected to abol-
ish catalytic activity and eliminate the production of uores-
cence. In the presence of a ribonuclease that cleaves RNA at in-
ternal sites to generate 3
phosphate (or 2
-3
cyclic phosphate)
and 5
hydroxyl termini, however, the RNA linkage should be
cleaved, which will regenerate the 5
end of Aurora and restore
catalytic activity (Figure 3 D and Supplementary Figure S22 ).
Because protein ribonucleases are generally capable of multi-
ple turnover catalysis, this architecture was also expected to
amplify the single-turnover signal generated by Aurora (Fig-
ure 3 E). We tested our sensor using ribonuclease A. This ac-
tivated the sensor and enhanced uorescence more than 10-
fold (Figure 3 F). Furthermore, the detection limit of the sensor
under these conditions ( ∼100 pM; dened here as the mini-
mum concentration of RNase A that gives a signal ≥3 stan-
dard deviations higher than that of the average background
value measured in the absence of RNase A) was approxi-
mately 10 000-fold lower than our oligonucleotide sensor
(compare Figures 3 G and Supplementary Figure S21 , and see
also Supplementary Figure S23 ). This dramatic increase in sen-
sitivity is likely due to the high turnover number of RNase A.
To further probe the mechanism of this sensor, we investigated
whether activation was affected by RiboLock, a commercially
available inhibitor of RNase A. RiboLock had no effect on Au-
rora itself (Figure 3 F, right), but prevented the Aurora sensor
from being activated by RNase A (Figure 3 F, left). This pro-
vided additional evidence that the sensor is activated by RNA
cleavage. Ta k e n together, these experiments show that assays
which use a covalently blocked form of Aurora to detect a
multiple-turnover enzyme can be orders of magnitude more
sensitive than those that use unmodied Aurora. They also
indicate that such a sensor can be used to detect the presence
of ribonuclease inhibitors in a sample.
Using Aurora to identify Nsp15 inhibitors in a
high-throughput screen
Because assays using Aurora sensors can be performed rapidly
and inexpensively, they appear to be well-suited for applica-
tions such as high-throughput screens. To further investigate
this idea, we investigated whether our Aurora sensor could be
used to identify inhibitors of the S AR S-CoV-2 endoribonucle-
ase Nsp15. This ribonuclease cleaves 3
of pyrimidines (pref-
erentially uridines) to generate 2
-3
cyclic phosphate and 5
hydroxyl termini ( 45 ). It helps to prevent host recognition by
degrading double-stranded viral intermediates ( 45 ), and in-
hibitors could potentially be useful as antiviral agents ( 46 ).
Pilot experiments showed that, as was the case for RNase A,
it was possible to construct a version of Aurora that was ac-
tivated by Nsp15 ( Supplementary Figures S22 and S24 ). To
perform a screen using this sensor, a master mix containing
Nsp15 and buffer was aliquoted into the wells of 384 well
plates, each of which contained a different compound from
a 1000-member fragment-based small molecule library (Fig-
ure 4 A). A second master mix containing the Aurora sensor
was then added to each well. After a short incubation to al-
low Nsp15 to cleave the RNA linkage and activate the sensor,
zinc and 4-MUP were added to initiate deoxyribozyme catal-
ysis. After another incubation, uorescence was measured in
a plate reader (Figure 4 B). In wells containing compounds
that do not inhibit Nsp15, cleavage of the RNA linkage in
the Aurora sensor by Nsp15 was expected to activate the
sensor and lead to production of a uorescent signal (Fig-
ure 4 B, black points). In contrast, RNA cleavage should not
occur and uorescence should not be produced in wells con-
taining compounds that inhibit either Nsp15 or Aurora itself
(Figure 4 B, orange points). To distinguish compounds that in-
hibit Nsp15 from those that inhibit Aurora, a counterscreen
was performed using Aurora rather than the Aurora sensor.
A graph comparing these two screens revealed that none of
the hits identied in the initial screen inhibited Aurora in the
counterscreen (Figure 4 C). This indicates that these hits are
Nsp15 rather than deoxyribozyme inhibitors, and also that
they do not quench uorescence of 4-MU itself. To compare
these results to those obtained using standard methods, the
screen was repeated using a FRET assay in which Nsp15 was
incubated with library members and a DNA substrate con-
taining a uorophore at one end and a quencher at the other
(Figure 4 D). Cleavage by Nsp15 was expected to result in
an increase in uorescence, while the uorescence in wells
containing Nsp15 inhibitors was expected to remain at back-
ground levels. The results of this FRET screen were virtually
identical to those obtained using the Aurora sensor (Figure
4 E). Several hits were further characterized as a function of
concentration using both the Aurora sensor and the FRET
assay. The most potent of these compounds inhibited Nsp15
with an IC50 of 12 μM in an assay that used the Aurora sensor
and 11 μM in an assay that used a FRET assay (Figure 4 F).
Other hits inhibited Nsp15 with IC50 values ranging from
7.9 to 121 μM ( Supplementary Figure S25 ). These experi-
ments indicate that Aurora sensors can be used in combination
with small-molecule libraries to rapidly identify inhibitors in
high-throughput screens.
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 11
Measure
fluorescence
Inhibitor
Compound
inhibits
Nsp15
Compound
does not inhibit
Nsp15
rU
5’
rU
5’
Aliquot into wells
containing Nsp15
and library member
5’
P
P
A
B
C
D
F
P
Nsp15 Nsp15
Aurora
sensor
E
Nsp15
FRET
assay
Diffusion
F Q
rU
F Q F Q
% activity (FRET)
0.01 0.1 1 10 100 1000
0.01
0.1
1
10
100
1000
% activity (Aurora sensor)
0 20406080100120140160
% activity (screen)
Z factor = 0.91
0
20
40
60
80
100
120
140
160
% activity (counterscreen)
0 200 400 600 800 1000
0
20
40
60
80
100
120
Compound number
% activity (screen)
Possible
inhibitors Aurora sensor
FRET assay
NH2
N
% activity
0.1 1 10 100 1000
0
20
40
60
80
100
120
Inhibitor concentration (μM)
Inhibitor structure
Figure 4. Identication of small-molecule inhibitors of the SARS-CoV-2 ribonuclease Nsp15 using a uorescent Aurora sensor. ( A ) Workow of
high-throughput screen to identify Nsp15 inhibitors. ( B ) Effect of each compound in the 10 0 0-member library on the uorescence of the Aurora sensor.
Potential inhibitors are shown in orange. ( C ) Identication of inhibitors and false positives. The x -axis of the graph shows the uorescent signal
generated by the Aurora sensor in the presence of Nsp15 and different compounds in the librar y, while the y -axis shows the uorescent signal
generated by Aurora itself in the presence of Nsp15 and the same compounds. Po in ts with high uorescence values on both the x -axis and the y -axis
(shown in black) correspond to wells containing compounds that inhibit neither Nsp15 nor Aurora. Point s with low uorescence values on the x -axis and
a high uorescence value on the y -axis (shown in orange) correspond to wells containing compounds that inhibit Nsp15 but not Aurora. ( D ) Workow of
a FRET assay for ribonuclease activity. ( E ) Comparison of the results of a high-throughput screen for Nsp15 inhibitors using the Aurora sensor ( x -axis)
with a screen of the same library using a FRET assay ( y -axis). ( F ) Example of an Nsp15 inhibitor identied in the screen. This compound inhibits Nsp15
with an IC50 value of 12 μM when measured using the Aurora sensor and 11 μM when measured using the FRET assay.
Conclusions
In this study, we developed a new way to generate uorescence
using a deoxyribozyme called Aurora and a coumarin sub-
strate called 4-MUP. Our approach offers a number of advan-
tages when compared to other methods of generating uores-
cent signals. Both Aurora and 4-MUP are stable, inexpensive
and widely available. The workow is simple, and formation
of the uorescent product can be monitored in solution and in
real time without the need for wash steps or biochemical pu-
rications. The signal to noise ratio of the uorescent signal is
also higher than that produced using widely used methods like
molecular probes. A second goal of this study was to establish
that these deoxyribozymes can be used for real-world applica-
tions. As an initial proof of concept for this idea, we showed
that Aurora can be readily converted by rational design into
a sensor that only generates uorescence in the presence of an
input. Our most sensitive sensor could detect ribonucleases
with a limit of detection of approximately 100 pM, which
compares favorably with the detection limits of many ho-
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
12 Nucleic Acids Research , 2024
mogenous assays that use aptamers in combination with more
expensive signaling elements such as uorophores, dyes, quan-
tum dots, or gold nanoparticles ( 47–50 ). Although selection
was not used to optimize this sensor, it could in principle be
utilized to improve its performance or to develop sensors that
detect other target molecules ( 9 ,51–53 ). After verifying that
this sensor worked, it was used to identify inhibitors of the
Nsp15 ribonuclease from S AR S-CoV-2 in a high-throughput
screen. Our assay could readily distinguish between reactions
that contained active ribonuclease and those that did not ( Z -
factor = 0.91). It did not produce false positives from com-
pounds that inhibit Aurora rather than Nsp15, although we
note that the frequency of such false positives will depend on
the properties of the library. It also yielded results that were
similar to those obtained when the library was screened in
parallel using a more standard FRET assay ( 54 ,55 ). While
our assay is comparable to those which use FRET in terms of
both simplicity and workow, reagent costs are several fold
lower and signal to noise ratios are considerably higher. We
anticipate that further optimization of Aurora using methods
such as recombination ( 56 ) and secondary structure libraries
( 57 ) in combination with single-step ( 58 ) and conventional
selections will continue to decrease costs and increase signal
to noise ratios, which will in turn increase the utility of Au-
rora for applications such as high-throughput screening and
diagnostics. In a more general sense, our work highlights the
potential of functional DNA molecules as widely applicable
uorescent tools.
Data availability
The data underlying this article will be shared on reasonable
request to the corresponding author.
Supplementary data
Supplementary Data are available at NAR Online.
A c kno wledg ements
We thank colleagues in our group and at the IOCB for useful
discussions.
Funding
GA
ˇ
CR [24-11210S to E.A.C.]; OP JAK project RNA for ther-
apy [CZ.02.01.01 / 00 / 22_008 / 0004575] co-nanced by the
European Union; GAUK [337022 to M.V.]. Funding for open
access charge: IOCB (my research institute).
Conict of interest statement
We have led a patent application (EP 23196423.0) covering
the work described in this manuscript.
References
1. Tsi en ,R .Y. (1998) The green uorescent protein. Annu. Rev.
Biochem., 67 , 509–544.
2. Shimomura,O. (2005) The discovery of aequorin and green
uorescent protein. J. Microsc., 217 , 1–15.
3. Jaffrey,S.R. (2018) RNA-based uorescent biosensors for
detecting metabolites in vitro and in living cells. Adv. Pharmacol.,
82 , 187–203.
4. Lu, X. , Kong, K.Y.S. and Unrau, P. J . (2023) Harmonizing the
growing uorogenic RNA aptamer toolbox for RNA detection
and imaging. Chem. Soc. Rev., 52 , 4071–4098.
5. Zhou, J. and Rossi, J. (2017) Aptamers as targeted therapeutics:
current potential and challenges. Nat. Rev. Drug Discov., 16 ,
181–202.
6. Silverman,S.K. (2016) Catalytic DNA: scope, applications, and
biochemistry of deoxyribozymes. Trends Biochem. Sci , 41 ,
595–609.
7. Micura, R. and Höbartner, C. (2020) Fundamental studies of
functional nucleic acids: aptamers, riboswitches, ribozymes and
DNAzymes. Chem. Soc. Rev., 49 , 7331–7353.
8. Curtis,E.A. (2022) Pushing the limits of nucleic acid function.
Chemistry , 28 , e202201737.
9. Levy, M. and Ellington, A.D. (2002) ATP-dependent allosteric
DNA enzymes. Chem. Biol., 9 , 417–426.
10. Cheglakov, Z. , Weizmann, Y. , Beissenhirtz, M.K. and Willner, I.
(2006) Ultrasensitive detection of DNA by the PCR-induced
generation of DNAzymes: the DNAzyme primer approach. Chem.
Commun., 30 , 3205–3207.
11. Cho, E.J. , Ya ng, L. , Levy, M. and Ellington, A.D. (2005) Using a
deoxyribozyme ligase and rolling circle amplication to detect a
non-nucleic acid analyte, AT P. J. Am. Chem. Soc., 127 ,
2022–2023.
12. Sando, S. , Narita, A. and Aoyama, Y. (2007) Light-up
Hoechst–DNA aptamer pair: generation of an aptamer-selective
uorophore from a conventional DNA-staining dye.
ChemBioChem , 8 , 1795–1803.
13. Kato, T. , Shimada, I. , Kimura, R. and Hyuga, M. (2016) Light-up
uorophore–DNA aptamer pair for label-free turn-on aptamer
sensors. Chem. Commun., 52 , 4041–4044.
14. Kolpashchikov, D.M. and Spelkov, A.A. (2021) Binary (Split)
light-up aptameric sensors. Angew. Chem. Int. Ed., 60 ,
4988–4999.
15. Travascio, P. , Li, Y. and Sen, D. (1998) DNA-enhanced peroxidase
activity of a DNA aptamer-hemin complex. Chem. Biol., 5 ,
505–517.
16. Kosman, J. and Juskowiak, B. (2020) Bioanalytical application of
peroxidase-mimicking DNAzymes: status and challenges. Adv.
Biochem. Eng. Biotechnol., 170 , 59–84.
17. Nakayama, S. and Sintim, H.O. (2009) Biomolecule detection with
peroxidase-mimicking DNAzymes; expanding detection modality
with uorogenic compounds. Mol. BioSyst., 6 , 95–97.
18. Robinson, D. and Willcox, P. (1969) 4-Methylumbelliferyl
phosphate as a substrate for lysosomal acid phosphatase.
Biochim. Biophys. Acta (BBA) - Enzymol., 191 , 183–186.
19. Vol e k, M. , Kurfürst, J. , Kožíšek, M. , Srb, P. , Veverka, V. and
Curtis,E.A. (2024) Apollon: a deoxyribozyme that generates a
yellow product. Nucleic Acids Res.,
https:// doi.org/ 10.1093/ nar/ gkae490 .
20. Svehlova, K. , Lukšan, O. , Jakubec, M. and Curtis, E.A. (2022)
Supernova: a deoxyribozyme that catalyzes a chemiluminescent
reaction. Angew. Chem. Int. Ed., 61 , e202109347.
21. Kim, Y. , Jedrzejczak, R. , Maltseva, N.I. , Wilamowski, M. ,
Endres, M. , Godzik, A. , Michalska, K. and Joachimiak, A. (2020)
Crystal structure of Nsp15 endoribonuclease NendoU from
S AR S-CoV-2. Protein Sci., 29 , 1596–1605.
22. Zhang, J.-H. , Chung, T. D. Y. and Oldenburg, K.R. (1999) A simple
statistical parameter for use in evaluation and validation of high
throughput screening assays. SLAS Discov. , 4 , 67–73.
23. Lee, W. , Tonelli, M. and Markley, J.L. (2015) NMRF AM-SP ARKY:
enhanced software for biomolecular NMR spectroscopy.
Bioinformatics , 31 , 1325–1327.
24. Bronstein, I. , Edwards, B. and Voyta, J.C. (1989) 1,2-dioxetanes:
novel chemiluminescent enzyme substrates. Applications to
immunoassays. J. Biolumin. Chemilumin., 4 , 99–111.
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
Nucleic Acids Research , 2024 13
25. Trayhurn, P. , Thomas, M.E.A. , Duncan, J.S. , Black, D. , Beattie, J.H.
and Rayner,D.V. (1995) Ultra-rapid detection of mRNAs on
Northern blots with digoxigenin-labelled oligonucleotides and
‘CDP-Star’, a new chemiluminescent reagent. Biochem. Soc.
Trans., 23 , 494S.
26. Hananya, N. and Shabat, D. (2017) A glowing trajectory between
bio- and chemiluminescence: from luciferin-based probes to
triggerable dioxetanes. Angew. Chem. Int. Ed Engl., 56 ,
16454–16463.
27. Sun, W. C . , Gee, K.R. and Haugland, R.P. (1998) Synthesis of novel
uorinated coumarins: excellent UV-light excitable uorescent
dyes. Bioorg. Med. Chem. Lett., 8 , 3107–3110.
28. Lavis, L.D. and Raines, R.T. (2008) Bright ideas for chemical
biology. ACS Chem. Biol., 3 , 142–155.
29. Curtis, E.A. and Bartel, D.P. (2005) New catalytic structures from
an existing ribozyme. Nat. Struct. Mol. Biol., 12 , 994–1000.
30. Ellington, A.D. and Szostak, J.W. (1990) In vitro selection of RNA
molecules that bind specic ligands. Nature , 346 , 818–822.
31. Ekland, E.H. and Bartel, D.P. (1995) The secondary structure and
sequence optimization of an RNA ligase ribozyme. Nucleic Acids
Res., 23 , 3231–3238.
32. Knight, R. and Yarus, M. (2003) Analyzing partially randomized
nucleic acid pools: straight dope on doping. Nucleic Acids Res.,
31 , e30.
33. Gutell, R.R. , Power, A. , Hertz, G.Z. , Putz, E.J. and Stormo, G.D.
(1992) Identifying constraints on the higher-order structure of
RNA: continued development and application of comparative
sequence analysis methods. Nucleic Acids Res. , 20 , 5785–5795.
34. Gutell,R.R. (2014) Ten lessons with Carl Woes e about RNA and
comparative analysis. RNA Biol , 11 , 254–272.
35. Mead, J.A. , Smith, J.N. and Williams, R.T. (1955) Studies in
detoxication. 67. The biosynthesis of the glucuronides of
umbelliferone and 4-methylumbelliferone and their use in
uorimetric determination of beta-glucuronidase. Biochem. J., 61 ,
569–574.
36. Sigel, R.K.O. and Pyle, A.M. (2007) Alternative roles for metal ions
in enzyme catalysis and the implications for ribozyme chemistry.
Chem. Rev., 107 , 97–113.
37. Zhou, W. , Saran, R. and Liu, J. (2017) Metal sensing by DNA.
Chem. Rev., 117 , 8272–8325.
38. Jakubec, M. , Pšenáková, K. , Svehlova, K. and Curtis, E.A. (2022)
Optimizing the chemiluminescence of a light-producing
deoxyribozyme. ChemBioChem , 23 , e202200026.
39. Holtz, K.M. and Kantrowitz, E.R. (1999) The mechanism of the
alkaline phosphatase reaction: insights from NMR,
crystallography and site-specic mutagenesis. FEBS Lett., 462 ,
7–11.
40. McCall, K.A. , Huang, C. and Fierke, C.A. (2000) Function and
mechanism of Zinc Metalloenzymes. J. Nutr., 130 , 1437S–1446S.
41. Chandra, M. , Sachdeva, A. and Silverman, S.K. (2009)
DNA-catalyzed sequence-specic hydrolysis of DNA. Nat. Chem.
Biol., 5 , 718–720.
42. Gu, H. , Furukawa, K. , Weinberg, Z. , Berenson, D.F. and
Breaker,R.R. (2013) Small, highly active DNAs that hydrolyze
DNA. J. Am. Chem. Soc., 135 , 9121–9129.
43. Zhang, C. , Li, Q. , Xu, T. , Li, W. , He, Y. and Gu, H. (2021) New
DNA-hydrolyzing DNAs isolated from an ssDNA library carrying
a terminal hybridization stem. Nucleic Acids Res. , 49 , 6364–6374.
44. Yarus,M. (1993) How many catalytic RNAs? Ions and the
Cheshire cat conjecture. FASEB J. , 7 , 31–39.
45. Deng, X. and Baker, S.C. (2018) An ‘old’ protein with a new story:
coronavirus endoribonuclease is important for evading host
antiviral defenses. Virology , 517 , 157–163.
46. Faheem, n. , Kumar, B.K. , Sekhar, K.V.G.C. , Kunjiappan, S. ,
Jamalis, J. , Balaña-Fouce, R. , Te kwa n i, B.L. and
Sankaranarayanan,M. (2020) Druggable targets of S AR S-CoV-2
and treatment opportunities for COVID-19. Bioorg. Chem., 104 ,
104269.
47. Cho, E.J. , Lee, J.-W. and Ellington, A.D. (2009) Applications of
aptamers as sensors. Annu. Rev. Anal. Chem. (Palo Alto Calif) , 2 ,
241–264.
48. Sassolas, A. , Blum, L.J. and Leca-Bouvier, B.D. (2010)
Homogeneous assays using aptamers. Analyst , 136 , 257–274.
49. Elowe, N.H. , Nutiu, R. , Allali-Hassani, A. , Cechetto, J.D. ,
Hughes, D.W. , Li, Y. and Brown, E.D. (2006) Small-molecule
screening made simple for a difcult target with a signaling
nucleic acid aptamer that reports on deaminase activity. Angew.
Chem. Int. Ed., 45 , 5648–5652.
50. Hafner, M. , V ianini, E. , Albertoni, B. , Marchetti, L. , Grüne, I. ,
Gloeckner, C. and Famulok, M. (2008) Displacement of
protein-bound aptamers with small molecules screened by
uorescence polarization. Nat. Protoc., 3 , 579–587.
51. Koizumi, M. , Soukup, G.A. , Kerr, J.N.Q. and Breaker, R.R. (1999)
Allosteric selection of ribozymes that respond to the second
messengers cGMP and cAMP. Nat. Struct. Mol. Biol., 6 ,
1062–1071.
52. Robertson, M.P. and Ellington, A.D. (1999) In vitro selection of an
allosteric ribozyme that transduces analytes to amplicons. Nat.
Biotechnol., 17 , 62–66.
53. Piganeau, N. , Jenne, A. , Thuillier, V. and Famulok, M. (2000) An
allosteric ribozyme regulated by doxycyline. Angew. Chem. Int.
Ed Engl., 39 , 4369–4373.
54. Canal, B. , Fujisawa, R. , McClure, A.W. , Deegan, T.D. , Wu, M. ,
Ulferts, R. , Weissmann, F. , Drury, L.S. , Bertolin, A.P. , Zeng, J. , et al.
(2021) Identifying S AR S-CoV-2 antiviral compounds by screening
for small molecule inhibitors of nsp15 endoribonuclease.
Biochem. J., 478 , 2465–2479.
55. Choi, R. , Zhou, M. , Shek, R. , Wilson, J.W. , T illery, L. , Craig, J.K. ,
Salukhe, I.A. , Hickson, S.E. , Kumar, N. , James, R.M. , et al. (2021)
High-throughput screening of the ReFRAME, Pandemic Box, and
COVID Box drug repurposing libraries against S AR S-CoV-2
nsp15 endoribonuclease to identify small-molecule inhibitors of
viral activity. PLoS One , 16 , e0250019.
56. Curtis, E.A. and Bartel, D.P. (2013) Synthetic shufing and in vitro
selection reveal the rugged adaptive tness landscape of a kinase
ribozyme. RNA , 19 , 1116–1128.
57. Sgallová, R. and Curtis, E.A. (2021) Secondary structure libraries
for articial evolution experiments. Molecules , 26 , 1671.
58. Streckerová, T. , Kurfürst, J. and Curtis, E.A. (2021) Single-round
deoxyribozyme discovery. Nucleic Acids Res. , 49 , 6971–6981.
Received: February 12, 2024. Revised: May 7, 2024. Editorial Decision: May 15, 2024. Accepted: May 23, 2024
©The Author(s) 2024. 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-NonCommercial License
(https: // creativecommons.org / licenses / by-nc / 4.0 / ), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For
commercial re-use, please contact journals.permissions@oup.com
Downloaded from https://academic.oup.com/nar/advance-article/doi/10.1093/nar/gkae467/7690888 by guest on 14 June 2024
