Novel isoguanine derivative of unlocked nucleic acid—Investigations of thermodynamics and biological potential of modified thrombin binding aptamer

Article abstract of DOI:10.1371/journal.pone.0197835

Thrombin binding aptamer (TBA), is a short DNA 15-mer that forms G-quadruplex structure and possesses anticoagulant properties. Some chemical modifications, including unlocked nucleic acids (UNA), 2⁰-deoxy-isoguanosine and 2⁰-deoxy-4-thiouridine were previously found to enhance the biological activity of TBA. In this paper, we present thermodynamic and biological characteristics of TBA variants that have been modified with novel isoguanine derivative of UNA as well as isoguanosine. Additionally, UNA-4-thiouracil and 4-thiouridine were also introduced simultaneously with isoguanine derivatives. Thermodynamic analysis indicates that the presence of isoguanosine in UNA or RNA series significantly decreases the stability of G-quadruplex structure. The highest destabilization is observed for substitution at one of the G-tetrad position. Addition of 4-thiouridine in UNA or RNA series usually decreases the unfavorable energetic cost of the presence of UNA or RNA isoguanine. Circular dichroism and thermal denaturation spectra in connection with thrombin time assay indicate that the introduction of UNA-isoguanine or isoguanosine into TBA negatively affects G-quadruplex folding and TBA anticoagulant properties. These findings demonstrate that the highly-ordered structure of TBA is essential for inhibition of thrombin activity.

Full text of DOI:10.1371/journal.pone.0197835

RESEARCH ARTICLE
Novel isoguanine derivative of unlocked
nucleic acid—Investigations of
thermodynamics and biological potential of
modified thrombin binding aptamer
Weronika Kotkowiak1, Tomasz Czapik2, Anna Pasternak1*
1 Department of Nucleic Acids Bioengineering, Institute of Bioorganic Chemistry, Polish Academy of
Sciences, Noskowskiego, Poznan, Poland, 2 Department of Structural Chemistry and Biology of Nucleic
Acids, Institute of Bioorganic Chemistry, Polish Academy of Sciences, Noskowskiego, Poznan, Poland
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
* apa@ibch.poznan.pl
Abstract
Thrombin binding aptamer (TBA), is a short DNA 15-mer that forms G-quadruplex structure
and possesses anticoagulant properties. Some chemical modifications, including unlocked
nucleic acids (UNA), 20-deoxy-isoguanosine and 20-deoxy-4-thiouridine were previously
found to enhance the biological activity of TBA. In this paper, we present thermodynamic
and biological characteristics of TBA variants that have been modified with novel isoguanine
derivative of UNA as well as isoguanosine. Additionally, UNA-4-thiouracil and 4-thiouridine
were also introduced simultaneously with isoguanine derivatives. Thermodynamic analysis
indicates that the presence of isoguanosine in UNA or RNA series significantly decreases
the stability of G-quadruplex structure. The highest destabilization is observed for substitu-
tion at one of the G-tetrad position. Addition of 4-thiouridine in UNA or RNA series usually
decreases the unfavorable energetic cost of the presence of UNA or RNA isoguanine. Circu-
lar dichroism and thermal denaturation spectra in connection with thrombin time assay indi-
cate that the introduction of UNA-isoguanine or isoguanosine into TBA negatively affects G-
quadruplex folding and TBA anticoagulant properties. These findings demonstrate that the
highly-ordered structure of TBA is essential for inhibition of thrombin activity.
OPEN ACCESS
Citation: Kotkowiak W, Czapik T, Pasternak A
(2018) Novel isoguanine derivative of unlocked
nucleic acid—Investigations of thermodynamics
and biological potential of modified thrombin
binding aptamer. PLoS ONE 13(5): e0197835.
https://doi.org/10.1371/journal.pone.0197835
Editor: Vittorio de Franciscis, Consiglio Nazionale
delle Ricerche, ITALY
Received: March 9, 2018
Accepted: May 9, 2018
Published: May 24, 2018
Copyright: © 2018 Kotkowiak et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Introduction
Anticoagulants belong to a class of compounds that influence blood coagulation by prolonging
clotting time. From the medical point of view, anticoagulants prevent the formation of blood
clots or inhibit the growth of formed clots, thus decreasing the risk of emboli formation in ves-
sels of organs [1,2]. Anticoagulants are commonly used as drugs for the prevention and/or
treatment of heart attack, stroke, deep vein or stent thrombosis, pulmonary embolism, cancer
(as supportive therapy) and other pathologies [3,4]. Unfortunately, currently used anticoagu-
lants show many side effects which include prolonged bleeding or even fatal bleeding.
Funding: This publication was supported by the
Polish Ministry of Science and Higher Education,
under the KNOW program and by Polish National
Science Center grant UMO-2013/10/E/NZ1/00741
and UMO-2017/25/B/NZ7/00127 to AP and UMO-
2015/19/N/NZ5/02246 to WK. The funders had no
role in study design, data collection and analysis,
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
1 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
decision to publish, or preparation of the
manuscript.
Anticoagulants may reduce the activity of clotting factors directly or indirectly. One of the
direct thrombin inhibitors is thrombin binding aptamer (TBA), a DNA 15-mer [5,6]. This oli-
gonucleotide forms an intramolecular, antiparallel G-quadruplex structure with a chair-like
conformation containing two G-tetrads connected by three edge-wise loops (Fig 1A). Accord-
ing to NMR and X-ray studies, TBA interacts with the positively charged exosite I of thrombin
via two shorter TT loops, thus inhibiting enzyme activity [6,7]. Previous reports indicate that
there are a few chemical modifications, among those unlocked nucleic acids (UNA), 4-thio-20-
deoxyuridine and 20-deoxy-isoguanosine, that are able to increase thermodynamic and/or
anticoagulant properties of the aptamer [8–10].
Competing interests: The authors have declared
that no competing interests exist.
Isoguanosine (iG) is an unnatural nucleoside that pairs with isocytidine (iC) via three
hydrogen bonds with the strength comparable to regular G-C interactions [11]. Interestingly,
hybridization of isoguanosine with cytidine (iG-C) was found to be isoenergetically similar to
A-U base pair. The strong interactions between iG and iC can be potentially applied in the
quantitative hybridization assay to prevent non-specific signal amplification, in PCR tech-
niques of DNA containing unnatural base pairs, in nanotechnology to construct more intricate
structures or in antisense therapy where iG-iC base pair might increase the specificity of apta-
mers and decoys [12,13].
4-Thiouridine (s4U) is a photoreactive uridine analogue that was found to naturally occur
in bacterial and archaeal tRNAs and play a role in protecting cells from near-UV light [14].
Moreover, in eukaryotes this thiolated nucleoside is successfully applied as metabolic label,
upon its random incorporation during transcription [15]. Previously reported thermodynamic
studies indicate that s4U slightly increases thermodynamic stability of A-U base pair [16].
Notably, the presence of s4U significantly stabilizes G-U wobble base pair, thus decreasing
base pairing specificity of A-U vs. G-U. Due to natural occurrence of 4-thiouridine and its
crosslinking properties, s4U was used as molecular tool in many analytical strategies [15].
It was found that certain UNA residues at position T7 of TBA increase thermodynamic sta-
bility of G-quadruplex and improve anticoagulant potential of the aptamer [8]. Similarly, sub-
stitution of 4-thio-20-deoxyuridine at positions T3, T7, T9 and T13 in TBA was found to inhibit
thrombin-catalyzed fibrin clotting 2-fold compared to unmodified TBA [9]. Moreover, based
on colorimetric assay, Nallagatla et al. performed screening of the oligonucleotide library con-
taining all possible substitutions of 20-deoxy-isoguanosine and indicated that its presence at
positions G1, G8 or G10 in TBA improves the aptamer-thrombin interaction [10].
Stimulated by the previously published results, we decided to identify potential of novel
chemical modifications that alter both sugar and heterocyclic base moieties known to be
Fig 1. Structure of thrombin binding aptamer (A) and novel isoguanine derivative of UNA (B).
https://doi.org/10.1371/journal.pone.0197835.g001
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
2 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
favorable for TBA (50GGTTGGTGTGGTTGG30) biological activity. Recently, we reported anti-
coagulant and antiproliferative properties of modified TBA variants containing a 4-thiouracil
derivative of UNA (UNA-s4U) [17]. Herein, we present chemical synthesis of UNA-isogua-
nine (iGU) a novel UNA derivative (Fig 1B), its incorporation into TBA and a comprehensive
thermodynamic and biological characteristics of novel TBA variants. Moreover, we also ana-
lyzed TBA variants with isoguanosine as well as with mixed 4-thiouridine and isoguanosine
substitutions in RNA (s4UR or iGR) or UNA (s4UU or iGU) series.
Materials and methods
Chemical synthesis of UNA-isoguanine phosphoramidite
The synthesis of protected UNA-isoguanine phosphoramidite (Fig 2) was performed accord-
ing to general procedure of the synthesis of UNA phosphoramidites with some modifications
[18]. 2,6-Diaminopurine riboside (1) was treated with lithium nitrite and acetic acid to yield
isoguanosine (2). Treatment of compound (2) with N,N-dimethylformamide dimethyl acetal
resulted in the N6 protected derivative (3) which was converted with dimethoxytrityl chloride
into 50-O-4,40-dimethoxytrityl derivative (4). Compound (4) was treated with sodium period-
ate followed by sodium borohydride to produce UNA derivative (5). Selective benzoylation of
(5) at O20 was achieved by addition of benzoyl chloride at decreased temperature accordingly
to the previously published protocol [18]. Treatment of the resulting compound (6) with
2-cyanoethyl-N,N,N0,N0-tetraisopropylphosphordiamidite gave the final UNA-isoguanine in
overall yield ca. 17%. Protocol for the synthesis of UNA-isoguanine phosphoramidite is
described in S1 File. The 1H and 13C NMR spectra were recorded on a 500 MHz (11.74 T)
AVANCE III Bruker spectrometer. Thin layer chromatography was conducted on Merck silica
gel 60 F254 glass plates. Silica gel column chromatography was performed using Merck TLC
gel H 60.
Fig 2. Chemical synthesis of UNA-isoguanine phosphoramidite. Reagents: (i) LiNO2, AcOH, H2O; (ii) N,N-dimethylformamide dimethyl acetal, DMF; (iii) DMTCl,
Py; (iv) NaIO4, 1,4-dioxane/water; NaBH4, 1,4-dioxane/water; (v) BzCl, DCM, -70˚C; (vi) 2-cyanoethyl-N,N,N0,N0-tetraisopropylphosphordiamidite, tetrazole, MeCN.
https://doi.org/10.1371/journal.pone.0197835.g002
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
3 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
Chemical synthesis of oligonucleotides
All oligonucleotides were synthesized on an automated RNA/DNA synthesizer using standard
phosphoramidite chemistry [19]. The deprotection and purification of unmodified oligoribo-
nucleotides or TBA variants containing UNA-iG and RNA-iG were performed according to
general procedures. Deprotection of TBA variants modified simultaneously with RNA-s4U
and RNA-iG or UNA-s4U and UNA-iG was performed by the treatment with methanol/
aqueous ammonia solution (1:1 v/v) at room temperature for 48 h. The purification was
accomplished according to previously published procedures [20]. The composition of all oligo-
nucleotides was confirmed by MALDI-TOF mass spectrometry.
UV melting analysis
Oligonucleotides were dissolved in a buffer containing 100 mM potassium chloride, 20 mM
sodium cacodylate and 0.5 mM Na2EDTA, pH 7.0. Single-stranded oligonucleotide concentra-
tions were calculated based on the absorbance measured above 80˚C [21] and the extinction
coefficients were calculated, using Oligo Calculator (ribotask.com). The samples were dena-
tured at 95˚C for 5 min and then cooled to room temperature overnight. The measurements
were performed for nine different concentrations of oligonucleotide in the concentration
range of 10−4–10-6 M. Absorbance versus temperature curves were obtained using the UV
melting method at 295 nm in the temperature range of 4–90˚C with a heating rate of 0.2˚C/
min on a JASCO V-650 spectrophotometer equipped with a thermoprogrammer. Melting
curves were analyzed using the MeltWin 3.5 software by van’t Hoff analysis [22]. The thermo-
dynamic parameters were determined by nonlinear curve fitting, averaging ΔH˚ and ΔS˚ from
fits of individual melting curves at different concentrations. For TBA variants presented in S1
Table the alternative method was additionally applied i.e. plots of 1/Tm vs lnCT were fitted to
the following equation:
TÀm1 ¼ ðR=DHꢀÞlnCT þ DSꢀ=DHꢀ
where CT is the total oligonucleotide strand concentration and R is the gas constant. Both
methods assume a two-state model. Agreement within 15% of ΔH˚ obtained from both meth-
ods was indicative of the two-state approximation.
Melting temperatures calculated for a 10−4 M concentration of oligonucleotide are denoted
by TM, and melting points for any other concentration of oligonucleotide are denoted by Tm.
Circular dichroism spectra
CD spectra were recorded on a JASCO J-815 spectropolarimeter. The oligonucleotides were
dissolved in a buffer containing 100 mM potassium chloride, 20 mM sodium cacodylate and
0.5 mM Na2EDTA, pH 7.0, to achieve a sample concentration of 3.0 μM. All samples were
denatured at 95˚C for 5 min and then slowly cooled to room temperature overnight prior to
data collection. The spectra were recorded in triplicate at 37˚C in the 205–320 nm wavelength
range. The buffer spectrum was subtracted from the sample spectra. The data analysis was per-
formed using the Origin 8.0 software.
Thermal difference spectra
The TDS measurements were performed on a JASCO V-650 spectrophotometer equipped
with a thermoprogrammer. The oligonucleotides were dissolved in a buffer containing 100
mM potassium chloride, 20 mM sodium cacodylate and 0.5 mM Na2EDTA, pH 7.0, to achieve
a sample concentration of 0.1 μM. Absorbance spectra were recorded in triplicate at 4˚C and
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
4 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
90˚C in the 220–335 nm wavelength range. The scan speed was 1000 nm/min with a 1 nm data
interval. Thermal difference spectra were obtained by subtraction of the low temperature absor-
bance spectra from the high temperature absorbance spectra using the Origin 8.0 software. The
differential spectra were normalized by dividing the data by its maximum value [23].
Thrombin time assay
The thrombin time assay was performed using commercially available Dia-TT kit
(DIAGON1) and coagulometer K-3002 Optic. Each reaction mixture consisted of TBA vari-
ant (0.33 μM) dissolved in 100 μl Dia-TT reagent. The reaction mixture was incubated at 37˚C
for 5 min and put into sample well of coagulometer K-3002 Optic. Next, the 100 μl of citrate
plasma was added. Each TBA variant was analyzed using plasma samples derived from five
healthy volunteers. The anticoagulant effect was obtained by subtraction of time needed for
plasma clotting in the presence of the aptamers from time needed for clotting of poor plasma.’
Results and discussion
Thermodynamic analysis of TBA variants containing UNA-iG or RNA-iG
TBA variants containing single and multiple substitution of modified residues were synthe-
sized and analyzed to determine the influence of RNA-iG or UNA-iG on the G-quadruplex
thermodynamic stability, folding topology and biological properties. The positions of modifi-
cation G1, G8 and G10 were selected according to a previous report indicating these positions
as the most favorable for anticoagulant properties of TBA modified with 20-deoxy-isoguano-
sine [10].
Thermodynamic analysis showed that the presence of UNA-iG or RNA-iG at positions G1
and G10, involved in G-quartet formation, significantly destabilize G-quadruplex structure
(Table 1). Modification of G-tetrad at terminal position (G1) by UNA-iG or RNA-iG decreases
G-quadruplex thermodynamic stability in a comparable manner (ΔΔG˚37 = 3.2–3.3 kcal/mol).
In contrast, the presence of UNA-iG within internal position G10 of the G-quartet destabilizes
G-quadruplex more than RNA-iG (ΔΔG˚37 = 3.39 kcal/mol for UNA-iG and 2.13 kcal/mol for
RNA-iG). Despite previous reports suggesting that mixed G-quartets containing both isogua-
nosine and guanosine can be formed [24] in this case, large destabilization of the G-quadru-
plex structure suggests that the unfavorable effect might be a consequence of a change in the
spatial arrangement of the amino and carbonyl functional groups within RNA-iG and UNA-
iG as well as the anti-conformation of N-glyosidic bond relative to syn-oriented 20-deoxygua-
nosines at G1 and G10. Nevertheless, in the future further structural studies are needed to con-
firm the above theory.
The substitution of TBA loop at position G8 by UNA-iG or RNA-iG also destabilizes
G-quadruplex structure, but the effect is reduced in comparison to positions G1 and G10 that
are involved in G-quartet formation. The presence of UNA-iG at G8 decreases G-quadruplex
thermodynamic stability by 1.95 kcal/mol, whereas RNA-iG causes destabilization mounting
to 1.23 kcal/mol. According to some structural analysis of unmodified TBA, the carbonyl
group of 20-deoxyguanosine at G8 may form a hydrogen bond with the 20-deoxyguanosine at
G-quartet [25–28]. Therefore, the unfavorable effect induced by the presence of UNA-iG or
RNA-iG at G8 might result from H-bond disruption between the nucleobase at G8 and the G-
quadruplex core. This is most likely due to opposite carbonyl and amino group arrangement
in isoguanosine relative to guanosine. Notably, UNA-iG at position G8 decreases TBA thermo-
dynamic stability more than RNA-iG at the same position. The increased flexibility of UNA-
iG at G8 compared to RNA-iG might additionally hinder a more favorable nucleobase orienta-
tion contributing to a higher destabilization for the UNA than RNA modification. A similar
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
5 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
Table 1. Thermodynamic stability of TBA variants modified with RNA-iG (iGR) or UNA-iG (iGU)a.
Position of
modification
Sequence
(50-30)
Average of curve fits
ΔG˚37 (kcal/mol)
TM
(˚C)
ΔΔG˚37 (kcal/mol)
ΔTM (˚C)
ΔΔG˚37 (kcal/mol)
ΔTM (˚C)
GGTTGGTGTGGTTGG
iGRGTTGGTGTGGTTGG
iGUGTTGGTGTGGTTGG
GGTTGGTiGRTGGTTGG
GGTTGGTiGUTGGTTGG
GGTTGGTGTiGRGTTGG
GGTTGGTGTiGUGTTGG
iGRGTTGGTiGRTGGTTGG
iGUGTTGGTiGUTGGTTGG
GGTTGGTiGRTiGRGTTGG
GGTTGGTiGUTiGUGTTGG
iGRGTTGGTGTiGRGTTGG
iGUGTTGGTGTiGUGTTGG
iGRGTTGGTiGRTiGRGTTGG
iGUGTTGGTiGUTiGUGTTGG
-1.74 0.02
1.49 0.48
1.55 0.07
-0.51 0.02
0.21 0.01
0.39 0.05
1.65 0.33
1.36 0.14
n.d.
50.7
19.9
0
3.23
3.29
1.23
1.95
2.13
3.39
3.10
-
0
-30.8
-30.0
-9.1
-15.6
-18.8
-26.9
-28.4
-
G1
G1
G8
G8
G10
0
0.06
0
0
0.8
0
20.7
41.6
35.1
0.72
0
-6.5
0
31.9
G10
23.8
1.26
-8.1
G1, G8
G1, G8
G8, G10
G8, G10
G1, G10
G1, G10
G1, G8, G10
G1, G8, G10
22.3
<15.0
34.1
-
-
0.31 0.01
n.d.
2.05
-
-16.6
-
<15.0
<15.0
<15.0
<15.0
<15.0
-
-
-
-
-
-
-
-
-
-
n.d.
-
-
n.d.
-
-
n.d.
-
-
n.d.
-
-
a buffer: 100 mM KCl, 20 mM sodium cacodylate, 0.5 mM EDTA(Na)2, pH 7.0
n.d.–not determined, detailed thermodynamic analysis is presented in S3 Table
https://doi.org/10.1371/journal.pone.0197835.t001
trend was observed for G10 but not for the G1 position. This is probably due to the internal (G8
and G10) vs. terminal (G1) location, which is less and more tolerant towards structural changes,
respectively [29,30].
As expected, the simultaneous substitution of UNA-iG or RNA-iG at more than one of the
TBA G1, G8 and G10 positions caused significant destabilization that prevented thermody-
namic analysis of most multi-modified G-quadruplex variants (Table 1).
Notably, the Gibbs free energy of almost all analyzed TBA variants possesses positive values,
with the exception of the oligonucleotide substituted by RNA-iG at G8. This indicates that the
unfolded G-quadruplex form is the predominant species at physiological temperature for the
majority of modified oligonucleotides.
Thermodynamic analysis of TBA variants containing UNA-iG or RNA-iG
and UNA-s4U or RNA-s4U
In the light of reports that indicate 20-deoxy-isoguanosine, 20-deoxy-4-thiouridine and UNA
as the residues having potential to improve the anticoagulant properties of TBA [8–10], we
decided to analyze TBA variants that contain simultaneously UNA-iG and UNA-s4U as well
as RNA-iG and RNA-s4U. UNA-s4U or RNA-s4U were substituted at positions T3, T7, T9, T12
or T13 and UNA-iG or RNA-iG at G1, G8 and G10 in TBA. These positions were selected based
on earlier publications, which indicated the positions as most biologically favorable [9,10].
The presence of RNA-s4U at positions T9 or T12 with RNA-iG at G1 mitigated an unfavor-
able thermodynamic effect triggered by isoguanosine having a ΔΔG˚37 = 1.83 kcal/mol for G1,
T9 and 1.85 kcal/mol for G1, T12 (Table 2). On the contrary, the presence of UNA-s4U at T3 or
T7 and UNA-iG at G1 as well as RNA-s4U at T7 and RNA-iG at G1 does not significantly reduce
destabilization caused by isoguanine derivatives. Moreover, for TBA variants containing UNA-
iG or RNA-iG at G1 together with UNA-s4U at T9, T12 or RNA-s4U at T7 the destabilization of
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
6 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
Table 2. Thermodynamic stability of TBA variants modified with RNA-iG and RNA-s4U (iGR,s4UR) or UNA-iG and UNA-s4U (iGU,s4UU)a.
Position of
modification
Sequence
(50-30)
Average of curve fits
ΔG˚37 (kcal/mol)
TM
(˚C)
ΔΔG˚37 (kcal/mol) ΔTM ΔΔG˚37 (kcal/mol) ΔTM
(˚C)
(˚C)
GGTTGGTGTGGTTGG
iGRGs4URTGGTGTGGTTGG
iGUGs4UUTGGTGTGGTTGG
iGRGTTGGs4URGTGGTTGG
iGUGTTGGs4UUGTGGTTGG
iGRGTTGGTGs4URGGTTGG
iGUGTTGGTGs4UUGGTTGG
iGRGTTGGTGTGGs4URTGG
iGUGTTGGTGTGGs4UUTGG
iGRGTTGGTGTGGTs4URGG
iGUGTTGGTGTGGTs4UUGG
GGs4URTGGTiGRTGGTTGG
GGs4UUTGGTiGUTGGTTGG
GGTTGGs4URiGRTGGTTGG
GGTTGGs4UUiGUTGGTTGG
GGTTGGTiGRs4URGGTTGG
GGTTGGTiGUs4UUGGTTGG
GGTTGGTiGRTGGs4URTGG
GGTTGGTiGUTGGs4UUTGG
GGTTGGTiGRTGGTs4URGG
GGTTGGTiGUTGGTs4UUGG
GGs4URTGGTGTiGRGTTGG
GGs4UUTGGTGTiGUGTTGG
GGTTGGs4URGTiGRGTTGG
GGTTGGTGs4URiGRGTTGG
GGTTGGTGTiGRGs4URTGG
GGTTGGTGTiGUGs4UUTGG
GGTTGGTGTiGRGTs4URGG
iGRGs4URTGGs4URGs4URGGTs4URGG
GGs4URTGGs4URiGRs4URGGTs4URGG
GGs4URTGGs4URGs4URiGRGTs4URGG
-1.74 0.02
1.39 0.06
1.48 0.07
2.18 0.22
1.33 0.06
0.09 0.01
2.21 0.19
0.11 0.02
2.04 0.11
n.d.
50.7
29.0
20.6
17.1
22.7
36.0
20.0
35.9
19.4
<15.0
<15.0
44.1
35.2
44.4
36.1
44.5
30.4
44.1
35.4
33.6
28.0
38.5
22.3
38.8
42.6
38.7
21.3
30.6
24.7
42.0
38.7
<15
<15
0
0
G1, T3
G1, T3
G1, T7
G1, T7
G1, T9
G1, T9
G1, T12
G1, T12
G1, T13
G1, T13
T3, G8
T3, G8
T7, G8
T7, G8
G8, T9
G8, T9
G8, T12
G8, T12
G8, T13
G8, T13
T3, G10
T3, G10
T7, G10
T9, G10
3.13
3.22
3.92
3.07
1.83
3.95
1.85
3.78
-
-21.7
-30.1
-33.6
-28
0
0.09
0
0
-8.4
0
-0.85
0
5.6
0
-14.7
-30.7
-14.8
-31.3
-
2.12
0
-16.0
0
1.93
-
-16.5
-
n.d.
-
-
-
-
-0.81 0.03
0.19 0.02
-0.88 0.04
0.10 0.07
-0.81 0.03
0.71 0.03
-0.85 0.04
0.17 0.02
0.36 0.10
0.95 0.01
-0.19 0.04
1.87 0.17
-0.21 0.02
-0.54 0.02
-0.20 0.02
1.61 0.09
0.60 0.02
0.99 0.08
-0.50 0.09
-0.17 0.04
n.d.
0.93
1.93
0.86
1.84
0.93
2.45
0.89
1.91
2.10
2.69
1.55
3.61
1.53
1.2
-6.6
-15.5
-6.3
-14,6
-6.2
-20.3
-6.6
-15.3
-17.1
-22.7
-12.2
-28.4
-11.9
-8.1
-12.0
-29.4
-20.1
-26.0
-8.7
-12.0
-
0
0
1
-8.9
0
0
0.98
0
-8.3
0
1.52
0
-14.1
0
1.02
0
-8.7
0
0.59
0
-5.6
0
2.06
-16.2
G10, T12
G10, T12
G10, T13
G1, T3, T7, T9, T13
T3, T7, G8, T9, T13
T3, T7, T9, G10, T13
1.54
3.35
2.34
2.73
1.24
1.57
-
0
0
1.81
-17.4
G1, T3, T7, G8, T9, G10, T13 iGRGs4URTGGs4URiGRs4URiGRGTs4URGG
G1, T3, T7, G8, T9, G10, T13 iGUGs4UUTGGs4UUiGUs4UUiGUGTs4UUGG
-
-
-
-
n.d.
-
-
a buffer: 100mM KCl, 20mM sodium cacodylate, 0.5 mM EDTA(Na)2, pH 7.0
n.d.–not determined, detailed thermodynamic analysis is presented in S4 Table
https://doi.org/10.1371/journal.pone.0197835.t002
G-quadruplex structure was even higher relative to the TBA modified with a single isoguanosine
in UNA or RNA series at G1 (ΔΔG˚37 = 3.8–4.0 kcal/mol). Unfortunately, we were unable to
analyze the thermodynamics of TBA variants with substitutions at positions T13 and G1 by
UNA-s4U and UNA-iG or RNA-s4U and RNA-iG due to extremely large destabilization.
Simultaneous substitutions of TBA at positions G8 by UNA-iG or RNA-iG and T3, T7, T9
or T12 by UNA-s4U or RNA-s4U usually caused an increase in G-quadruplex thermodynamic
stability compared to TBA modified with single UNA-iG or RNA-iG at G8 (Table 2). TBA
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
7 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
variants with RNA-iG at G8 and RNA-s4U at T3, T7, T9 or T12 were found to be the most sta-
ble. In contrast, the most unfavorable effects were observed for TBA variants containing
UNA-iG at G8 and UNA-s4U at T9 or T13 as well as RNA-iG at G8 and RNA-s4U at T13.
Similarly, substitution of positions T3, T7, T9 or T12 by UNA-s4U or RNA-s4U reduced
some unfavorable thermodynamic effects caused by UNA-iG or RNA-iG at position G10.
However, the presence of UNA-s4U at T3 and T12 together with UNA-iG at G10 causes signifi-
cant destabilization (ΔΔG˚37 = 3.61 kcal/mol for T3 and G10 and 3.35 kcal/mol for T12 and
G10). Interestingly, introduction of UNA-s4U at position T7, T9 or T13 and UNA-iG at G10
changed intramolecular to intermolecular folding of TBA (S1 Table). Thus, the direct compar-
ison of Gibbs free energy of those variants was not possible. Nevertheless, the analysis of TM
values indicates these variants do not form G-quadruplex structure at physiological tempera-
ture. A similar trend of molecularity change was observed with UNA-iG residue at G1, G8 or
G10 and four UNA-s4U residues at T3, T7, T9 and T13.
In general, the thermodynamic effects observed for multiple modifications of TBA are the
sum of single modification effects of UNA/RNA-iG and UNA/RNA-s4U. These findings stay
in great accordance with the influence of UNA-s4U and RNA-s4U on TBA thermodynamic
stability recently reported by our group [17].
CD spectroscopy
Circular dichroism is a commonly used technique to investigate structural geometry of various
nucleic acid structures. Particularly, when studying G-quadruplexes, it allows differentiation
between antiparallel and parallel folding topology, thus one can identify any alteration in this
molecular process. The specific CD pattern might be a result of the RNA or DNA strands ori-
entation, but also the stacking arrangement of guanines within G-tetrads [31]. Usually, parallel
G-quadruplexes are characterized by CD spectra containing positive band near 265 nm and a
negative peak around 240 nm, whereas antiparallel G-quadruplexes exhibit positive signal
around 295 nm and a negative band near 265 nm [32].
CD spectrum of unmodified TBA exhibits two positive bands i.e. a lower intensity peak
near 245 nm and higher intensity band around 295 nm as well as negative signal near 265 nm
(Fig 3). The CD pattern of TBA is consistent with previous reports indicating antiparallel G-
quadruplex folding topology. On the contrary, the TBA variants substituted by UNA-iG or
RNA-iG alone or in the presence of UNA-s4U or RNA-s4U exhibit a significant decrease in all
CD band intensities or even loss of CD signals characteristic for antiparallel G-quadruplex.
This shows that UNA-iG as well as RNA-iG has a negative influence on G-quadruplex forma-
tion, as indicated by thermodynamic analysis.
Thermal difference spectra
Thermal difference spectra (TDS) are often used to supplement circular dichroism studies.
This is a fast and cost-effective method to analyze differences in UV spectra. Absorbance val-
ues of nucleic acid structures below and above the melting point correspond to folded and
unfolded nucleic acid state, respectively [23]. The subtraction of low temperature spectra from
the high temperature spectra results in the pattern which is specific for a given nucleic acid
structure. TDS of unmodified TBA exhibits a pattern characteristic for antiparallel G-quadru-
plex with two positive bands near 240 nm and 270 nm and two negative signals around 260
nm and 295 nm (Fig 4). Analogously to CD spectra, the TDS bands were significantly flattened
for TBA variants containing only UNA-iG or RNA-iG substitution as well as for multiple sub-
stitutions of oligonucleotides with UNA-iG or RNA-iG with the addition of UNA-s4U or
RNA-s4U, thereby confirming CD analysis and thermodynamic results.
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
8 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
Fig 3. Representative circular dichroism spectra of TBA (black, solid line), and TBA variants substituted by RNA-
iG (red, solid line), UNA-iG (blue, solid line), RNA-iG and RNA-s4U (red, dashed line), UNA-iG and UNA-s4U
(blue, dashed line).
https://doi.org/10.1371/journal.pone.0197835.g003
Thrombin time assay
Thrombin time assay is a commonly known test to diagnose blood coagulation disorders but it
is also routinely used for analysis of anticoagulant potential of different compounds. Thrombin
time (TT) is the time needed to form a clot within plasma samples after adding an excess of
thrombin. The differences in thrombin time of plasma with and without anticoagulant can be
Fig 4. Representative thermal difference spectra of TBA (black, solid line), and TBA variants substituted by RNA-
iG (red, solid line), UNA-iG (blue, solid line), RNA-iG and RNA-s4U (red, dashed line), UNA-iG and UNA-s4U
(blue, dashed line).
https://doi.org/10.1371/journal.pone.0197835.g004
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
9 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
10 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
Fig 5. Anticoagulant activity of TBA variants modified with RNA-iG or UNA-iG (A), RNA-iG and RNA-s4U (B),
UNA-iG and UNA-s4U (C).
https://doi.org/10.1371/journal.pone.0197835.g005
named anticoagulation effect (AE). Despite the fact that previous report indicated that the
presence of 20-deoxy-isoguanosine improves TBA-thrombin interactions [10], our studies
show that isoguanosine analogues in UNA and RNA series do not have such potential. All ana-
lyzed TBA variants containing UNA-iG or RNA-iG appeared to be poor anticoagulants exhib-
iting AE values that were at least 2.4 times lower compared to unmodified TBA (Fig 5 and S2
Table). The further addition of UNA-s4U or RNA-s4U did not result in the improvement of
TBA anticoagulant properties. The results clearly indicate that the ability to form G-quartet is
pivotal for the anticoagulant properties of the aptamer. Some published works also designate
G-tetrad formation as the crucial factor for TBA activity [33,34]. It was previously reported
that the analogues of guanine, such as 7-deazaguanine, evidently decrease thrombin binding as
the residues are unable to form H-bonds within G-tetrad and consequently disrupt TBA struc-
ture [33,35]. On the other hand, there are examples of TBA variants, modified within loop
region, that show considerable inhibitory activity despite the low thermodynamic stability
[36]. In this case, however, it seems the thrombin acts as a molecular chaperone by stabilizing
G-quadruplex structure and consequently an anticoagulant activity of TBA is retained [36,37].
Importantly, the formation of thermodynamically stable G-quadruplex structure of TBA with
disrupted loop geometry could be detrimental for TBA anticoagulant properties [38]. Taken
together, the literature data indicate that thermodynamic stability is not the only factor that
determines inhibitory activity of TBA. Nevertheless, the results presented herein as well as
above mentioned literature suggests that G-quartet formation is one of the major factors
required for retaining TBA anticoagulant properties.
The studies described herein, clearly demonstrate that the presence of isoguanosine in both
UNA and RNA series is detrimental for G-quadruplex formation and biological function of
TBA. Substitution at positions G1, G8 or G10 significantly decreases thermodynamic stability
of G-quadruplex structure. CD and TDS analyses indicate that modified monomer substitu-
tion is unfavorable for G-quadruplex folding. These substitutions promoted a significant
decrease in the anticoagulant properties of TBA variants compared to unmodified TBA. This
study enhances our understanding of how UNA-iG and RNA-iG affect the thermodynamics
and biological function of TBA. This study also shows influence of novel isoguanine derivative
of UNA on the thermodynamic stability and folding topology of G-quadruplexes in general.
Supporting information
S1 File. Chemical synthesis of UNA-isoguanine phosphoramidite.
(DOCX)
S1 Table. Thermodynamic parameters of intermolecular G-quadruplex formation of TBA
variants modified with UNA-iG (iGU) and UNA-s4U (s4UU).
(DOCX)
S2 Table. The anticoagulant properties of TBA and modified TBA variants.
(DOCX)
S3 Table. Thermodynamic parameters of G-quadruplex formation of TBA variants modi-
fied with RNA-iG (iGR) or UNA-iG (iGU).
(DOCX)
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
11 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
S4 Table. Thermodynamic parameters of G-quadruplex formation of TBA variants modi-
fied with RNA-iG and RNA-s4U(iGR,s4UR) or UNA-iG and UNA-s4U (iGU,s4UU).
(DOCX)
Author Contributions
Conceptualization: Weronika Kotkowiak, Anna Pasternak.
Formal analysis: Weronika Kotkowiak.
Funding acquisition: Weronika Kotkowiak, Anna Pasternak.
Investigation: Weronika Kotkowiak.
Methodology: Weronika Kotkowiak, Anna Pasternak.
Supervision: Anna Pasternak.
Validation: Tomasz Czapik.
Visualization: Weronika Kotkowiak.
Writing – original draft: Weronika Kotkowiak, Anna Pasternak.
Writing – review & editing: Weronika Kotkowiak, Anna Pasternak.
References
1. Herrera S, Comerota AJ. Embolization during treatment of deep venous thrombosis: incidence, impor-
tance, and prevention. Techniques in vascular and interventional radiology. 2011; 14(2):58–64. https://
doi.org/10.1053/j.tvir.2011.01.002 PMID: 21550507
2. Amin A. Oral anticoagulation to reduce risk of stroke in patients with atrial fibrillation: current and future
therapies. Clinical Interventions in Aging. 2013; 8:75–84. https://doi.org/10.2147/CIA.S37818 PMID:
23378750
3. Fareed J, Hoppensteadt DA, Fareed D, Demir M, Wahi R, Clarke M, et al. Survival of heparins, oral anti-
coagulants, and aspirin after the year 2010. Seminars in thrombosis and hemostasis. 2008; 34(1):58–
73. https://doi.org/10.1055/s-2008-1066025 PMID: 18393143
4. Ahrens I, Peter K, Lip GY, Bode C. Development and clinical applications of novel oral anticoagulants.
Part II. Drugs under clinical investigation. Discovery medicine. 2012; 13(73):445–50. PMID: 22742650
5. Bock LC, Griffin LC, Latham JA, Vermaas EH, Toole JJ. Selection of single-stranded DNA molecules
that bind and inhibit human thrombin. Nature. 1992; 355:564–6. https://doi.org/10.1038/355564a0
PMID: 1741036
6. Russo Krauss I, Merlino A, Randazzo A, Novellino E, Mazzarella L, Sica F. High-resolution structures of
two complexes between thrombin and thrombin-binding aptamer shed light on the role of cations in the
aptamer inhibitory activity. Nucleic Acids Res. 2012; 40(16):8119–28. https://doi.org/10.1093/nar/
gks512 PMID: 22669903
7. Macaya RF, Schultze P, Smith FW, Roe JA, Feigon J. Thrombin-binding DNA aptamer forms a unimo-
lecular quadruplex structure in solution. Proc Natl Acad Sci U S A. 1993; 90:3745–9. PMID: 8475124
8. Pasternak A, Hernandez FJ, Rasmussen LM, Vester B, Wengel J. Improved thrombin binding aptamer
by incorporation of a single unlocked nucleic acid monomer. Nucleic Acids Res. 2011; 39(3):1155–64.
https://doi.org/10.1093/nar/gkq823 PMID: 20870750
9. Raviv SM, Horvath A, Aradi J, Bagoly Z, Fazakas F, Batta Z, et al. 4-Thio-deoxyuridylate-modified
thrombin aptamer and its inhibitory effect on fibrin clot formation, platelet aggregation and thrombus
growth on subendothelial matrix. Journal of Thrombosis and Homeostasis. 2008; 6(10):1764–71.
10. Nallagatla SR, Heuberger B, Haque A, Switzer C. Combinatorial synthesis of thrombin-binding apta-
mers containing iso-guanine. J Comb Chem. 2009; 11(3):364–9. https://doi.org/10.1021/cc800178m
PMID: 19243167
11. Roberts C, Bandaru R, Switzer C. Theoretical and Experimental Study of Isoguanine and Isocytosine:
Base Pairing in an Expanded Genetic System. Journal of the American Chemical Society. 1997; 119
(20):4640–9.
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
12 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
12. Hirao I. Unnatural base pair systems for DNA/RNA-based biotechnology. Curr Opin Chem Biol. 2006;
10(6):622–7. https://doi.org/10.1016/j.cbpa.2006.09.021 PMID: 17035074
13. Collins ML, Irvine B, Tyner D, Fine E, Zayati C, Chang C, et al. A branched DNA signal amplification
assay for quantification of nucleic acid targets below 100 molecules/ml. Nucleic Acids Res. 1997; 25
(15):2979–84. PMID: 9224596
14. Liu Y, Zhu X, Nakamura A, Orlando R, Soll D, Whitman WB. Biosynthesis of 4-thiouridine in tRNA in the
methanogenic archaeon Methanococcus maripaludis. The Journal of biological chemistry. 2012; 287
(44):36683–92. https://doi.org/10.1074/jbc.M112.405688 PMID: 22904325
15. Schmid K, Adobes-Vidal M, Helm M. Alkyne-Functionalized Coumarin Compound for Analytic and Pre-
parative 4-Thiouridine Labeling. Bioconjugate Chem. 2017; 28(4):1123–34.
16. Testa SM, Disney MD, Turner DH, Kierzek R. Thermodynamics of RNA-RNA duplexes with 2- or 4-
thiouridines: Implications for antisense design and targeting a group I intron. Biochemistry. 1999; 38
(50):16655–62. PMID: 10600128
17. Kotkowiak W, Lisowiec-Wachnicka J, Grynda J, Kierzek R, Wengel J, Pasternak A. Thermodynamic,
anticoagulant and antiproliferative properties of thrombin binding aptamer analogues containing novel
UNA derivative. Molecular Therapy—Nucleic Acids. 2018; 10, 304–16. https://doi.org/10.1016/j.omtn.
2017.12.013 PMID: 29499943
18. Langkjær N, Pasternak A, Wengel J. UNA (unlocked nucleic acid): A flexible RNA mimic that allows
engineering of nucleic acid duplex stability. Bioorg Med Chem. 2009; 17(15):5420–5. https://doi.org/10.
1016/j.bmc.2009.06.045 PMID: 19604699
19. McBride LJ, Caruthers MH. An investigation of several deoxyribonucleoside phosphoramidites useful
for synthesizing deoxyoligonucleotides. Tetrahedron Lett. 1983; 24:245–8.
20. Lopez-Gomollon S, Nicolas FE. Purification of DNA Oligos by denaturing polyacrylamide gel electro-
phoresis (PAGE). Methods in enzymology. 2013; 529:65–83. https://doi.org/10.1016/B978-0-12-
418687-3.00006-9 PMID: 24011037
21. Borer PN. Optical properties of nucleic acids, absorption and circular dichroism spectra. In: Fasman
GD, editor. CRC Handbook of Biochemistry and Molecular Biology: Nucleic Acids, 3rd edn. 1. Cleve-
land, OH: CRC Press; 1975. p. 589–95.
22. SantaLucia JJ. The use of spectroscopic techniques in the study of DNA stability. Michael GG, editor:
Oxford University Press; 2000. 363 p.
23. Mergny J-L, Li J, Lacroix L, Amrane S, Chaires JB. Thermal difference spectra: a specific signature for
nucleic acid structures. Nucl Acids Res. 2005; 33(16):e138-. https://doi.org/10.1093/nar/gni134 PMID:
16157860
24. Roberts C, Chaput JC, Switzer C. Beyond guanine quartets: cation-induced formation of homogenous
and chimeric DNA tetraplexes incorporating iso-guanine and guanine. Chem Biol. 1997; 4(12):899–
908. PMID: 9427655
25. Padmanabhan K, Padmanabhan KP, Ferrara JD, Sadler JE, Tulinsky A. The structure of alpha-throm-
bin inhibited by a 15-mer single-stranded DNA aptamer. Journal of Biological Chemistry. 1993; 268
(24):17651–4. PMID: 8102368
´
26. Aviño A, Portella G, Ferreira R, Gargallo R, Mazzini S, Gabelica V, et al. Specific loop modifications of
the thrombin-binding aptamer trigger the formation of parallel structures. The FEBS Journal. 2014; 281
(4):1085–99. https://doi.org/10.1111/febs.12670 PMID: 24304855
27. Zeng X, Zhang L, Xiao X, Jiang Y, Guo Y, Yu X, et al. Unfolding mechanism of thrombin-binding apta-
mer revealed by molecular dynamics simulation and Markov State Model. Scientific Reports. 2016;
6:24065. https://doi.org/10.1038/srep24065 PMID: 27045335
28. Virgilio A, Petraccone L, Vellecco V, Bucci M, Varra M, Irace C, et al. Site-specific replacement of the
thymine methyl group by fluorine in thrombin binding aptamer significantly improves structural stability
and anticoagulant activity. Nucleic Acids Res. 2015; 43(22):10602–11. https://doi.org/10.1093/nar/
gkv1224 PMID: 26582916
29. Pasternak A, Wengel J. Thermodynamics of RNA duplexes modified with UNA (unlocked nucleic acid)
monomers. Nucleic Acids Res. 2010; 38(19):6697–706. https://doi.org/10.1093/nar/gkq561 PMID:
20562222
30. Kotkowiak W, Kotkowiak M, Kierzek R, Pasternak A. Unlocked nucleic acids: implications of increased
conformational flexibility for RNA/DNA triplex formation. The Biochemical Journal. 2014; 464(2):203–
11. https://doi.org/10.1042/BJ20141023 PMID: 25226286
31. Masiero S, Trotta R, Pieraccini S, De Tito S, Perone R, Randazzo A, et al. A non-empirical chro-
mophoric interpretation of CD spectra of DNA G-quadruplex structures. Org Biomol Chem. 2010; 8
(12):2683–92. https://doi.org/10.1039/c003428b PMID: 20440429
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
13 / 14

Thrombin binding aptamer modified with novel isoguanine derivative of UNA
´
´
´
´
´
´
32. Tothova P, Krafčıkova P, Vıglasky V. Formation of Highly Ordered Multimers in G-Quadruplexes. Bio-
chemistry. 2014; 53(45):7013–27. https://doi.org/10.1021/bi500773c PMID: 25347520
33. Avino A, Fabrega C, Tintore M, Eritja R. Thrombin binding aptamer, more than a simple aptamer: chem-
ically modified derivatives and biomedical applications. Current pharmaceutical design. 2012; 18
(14):2036–47. PMID: 22376107
34. Tatarinova O, Tsvetkov V, Basmanov D, Barinov N, Smirnov I, Timofeev E, et al. Comparison of the
‘Chemical’ and ‘Structural’ Approaches to the Optimization of the Thrombin-Binding Aptamer. PloS one.
2014; 9(2):e89383. https://doi.org/10.1371/journal.pone.0089383 PMID: 24586736
35. Krawczyk SH, Bischofberger N, Griffin LC, Law VS, Shea RG, Swaminathan S. Structure-Activity Study
of Oligodeoxynucleotides Which Inhibit Thrombin. Nucleosides and Nucleotides. 1995; 14(3–5):1109–
16.
36. Tsvetkov V, Varizhuk A, Pozmogova G, Smirnov I, Kolganova N, Timofeev E. A universal base in a spe-
cific role: tuning up a thrombin aptamer with 5-nitroindole. Scientific Reports. 2015; 5:1–11.
37. Nagatoishi S, Isono N, Tsumoto K, Sugimoto N. Loop residues of thrombin-binding DNA aptamer
impact G-quadruplex stability and thrombin binding. Biochimie. 2011; 93(8):1231–8. https://doi.org/10.
1016/j.biochi.2011.03.013 PMID: 21511000
38. Zavyalova E, Taqiltsev T, Reshetnikov R, Arutyunyan A, Kopylov A. Cation Coordination Alters the Con-
formation of a Thrombin-Binding G-Quadruplex DNA Aptamer That Affects Inhibition of Thrombin.
Nucleic acid therapeutics. 2016; 26(5):299–308. https://doi.org/10.1089/nat.2016.0606 PMID:
27159247
PLOS ONE | https://doi.org/10.1371/journal.pone.0197835 May 24, 2018
14 / 14