Musashi binding elements in Zika and related Flavivirus 3′UTRs: A comparative study in silico

Abstract

Zika virus (ZIKV) belongs to a class of neurotropic viruses that have the ability to cause congenital infection, which can result in microcephaly or fetal demise. Recently, the RNA-binding protein Musashi-1 (Msi1), which mediates the maintenance and self-renewal of stem cells and acts as a translational regulator, has been associated with promoting ZIKV replication, neurotropism, and pathology. Msi1 predominantly binds to single-stranded motifs in the 3′ untranslated region (UTR) of RNA that contain a UAG trinucleotide in their core. We systematically analyzed the properties of Musashi binding elements (MBEs) in the 3′UTR of flaviviruses with a thermodynamic model for RNA folding. Our results indicate that MBEs in ZIKV 3′UtRs occur predominantly in unpaired, single-stranded structural context, thus corroborating experimental observations by a biophysical model of RNA structure formation. statistical analysis and comparison with related viruses show that ZIKV MBEs are maximally accessible among mosquito-borne flaviviruses. Our study addresses the broader question of whether other emerging arboviruses can cause similar neurotropic effects through the same mechanism in the developing fetus by establishing a link between the biophysical properties of viral RNA and teratogenicity. Moreover, our thermodynamic model can explain recent experimental findings and predict the Msi1-related neurotropic potential of other viruses.

Full text

1
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
Musashi binding elements in Zika
and related Flavivirus 3′UTRs: A
comparative study in silico
Adriano de Bernardi Schneider
1 & Michael T. Wolnger
2
Zika virus (ZIKV) belongs to a class of neurotropic viruses that have the ability to cause congenital
infection, which can result in microcephaly or fetal demise. Recently, the RNA-binding protein
Musashi-1 (Msi1), which mediates the maintenance and self-renewal of stem cells and acts as a
translational regulator, has been associated with promoting ZIKV replication, neurotropism, and
pathology. Msi1 predominantly binds to single-stranded motifs in the 3′ untranslated region (UTR)
of RNA that contain a UAG trinucleotide in their core. We systematically analyzed the properties
of Musashi binding elements (MBEs) in the 3′UTR of aviviruses with a thermodynamic model for
RNA folding. Our results indicate that MBEs in ZIKV 3′UTRs occur predominantly in unpaired, single-
stranded structural context, thus corroborating experimental observations by a biophysical model
of RNA structure formation. Statistical analysis and comparison with related viruses show that ZIKV
MBEs are maximally accessible among mosquito-borne aviviruses. Our study addresses the broader
question of whether other emerging arboviruses can cause similar neurotropic eects through the same
mechanism in the developing fetus by establishing a link between the biophysical properties of viral
RNA and teratogenicity. Moreover, our thermodynamic model can explain recent experimental ndings
and predict the Msi1-related neurotropic potential of other viruses.
Flaviviruses are an emerging group of arboviruses belonging to the Flaviviridae family. Researchers have been
describing recent outbreaks of these viruses that have not been previously detected for decades1–3.
e genus Flavivirus comprises more than 70 species that are mainly transmitted by mosquitoes and ticks,
typically classied into four groups: Mosquito-borne aviviruses (MBFVs), tick-borne aviviruses (TBFVs),
insect-specic aviviruses (ISFVs), that do not have vertebrate hosts, and no known arthropod vector aviviruses
(NKVs), which typically infect bats and rodents. Flaviviruses represent a global health threat, including emerging
and re-emerging human pathogens such as Dengue (DENV), Yellow fever (YFV), Japanese encephalitis (JEV),
West Nile (WNV), Tick-borne encephalitis (TBEV) and Zika (ZIKV) viruses4,5.
Initially isolated in 1947 from a sentinel rhesus macaque in the Ziika forest, Uganda, ZIKV has not been
associated with severe disease, apart from skin rashes, body pain, and fever. Likewise, ZIKV has been circulat-
ing across equatorial zones in Africa and Asia for 60 years, until the rst outbreak was reported in Yap Island,
Micronesia in 2007. Subsequently, the virus spread eastwards to French Polynesia and other Pacic islands in
2013 and reached the Americas in 20156,7. ere are two main ZIKV lineages, the original African (type strain
MR766) and an Asian (type strain FSS13025)8,9, the latter also comprising American strains such as PE243.
Background
e 2015–2017 outbreak in the Americas raised the possibility of a link between ZIKV infection and congenital
abnormalities, which included placental damage, intrauterine growth restrictions, eye diseases and microcephaly
in children as well as acute motor axonal neuropathy-type Guillain-Barré syndrome in adults10. While MBFVs
are typically transmitted by host-vector interaction, vertical transmission from mother to child during pregnancy
via transplacental infection has been reported11.
e neurotropic potential of ZIKV-related aviviruses has been known since the 1970s, when Saint Louis
encephalitis virus (SLEV) has been attributed to a severe neurological disorder in infected mice12,13. Vertical
1Department of Medicine, University of California San Diego, 220 Dickinson St, Suite A, San Diego, CA, 92103, United
States of America. 2Department of Theoretical Chemistry, University of Vienna, Währingerstraße 17, 1090, Vienna,
Austria. Correspondence and requests for materials should be addressed to M.T.W. (email: michael.wolnger@
univie.ac.at)
Received: 8 November 2018
Accepted: 23 April 2019
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
transmission has been observed with JEV in mice14 and human15 and a case of human fetal infection have been
reported aer YFV vaccination16. Other transmission pathways of ZIKV include blood transfusions and sexual
transmission17,18. Despite enormous eorts in studying ZIKV infections in the last years, the biological reasoning
and mechanisms behind arbovirus congenital neurotropism remain elusive.
Flavivirus genome organization. Flaviviruses have the structure of an enveloped sphere of approximately
50 nm diameter. ey are single-stranded positive-sense RNA viruses of 10–12 kb in size, and their genomic RNA
(gRNA) encodes a single open reading frame (ORF) anked by highly structured untranslated regions (UTRs).
Upon translation of the ORF, a polyprotein is produced which is processed by viral and cellular enzymes, yielding
structured (C, prM, E) and unstructured proteins (NS1, NS2A, NS2B, NS3, NS4A, 2K, NS4B, NS5). Both avivi-
rus UTRs are crucially related to regulation of the viral life cycle, mediating processes such as genome circulari-
zation, viral replication and packaging19–22.
Flaviviruses hijack the host mRNA degradation pathway. e central role of avivirus 3′UTR in
modulating cytopathicity and pathogenicity became apparent when an accumulation of both gRNA and viral
long non-coding RNA (lncRNA) has been observed upon infection. ese lncRNAs, also known as subgenomic
aviviral RNAs (sfRNAs)23, are stable decay intermediates derived from exploiting the host’s mRNA degradation
machinery24.
sfRNAs are produced by partial degradation of viral gRNA by Xrn1, a host 5′-3′ exoribonuclease that is asso-
ciated with the endogenous mRNA turnover machinery25,26. e enzyme stalls at highly conserved RNA struc-
tures in the viral 3′UTR, so-called Xrn1-resistant RNAs (xrRNAs), resulting in sfRNAs of variable lengths27,28.
Xrn1-resistant RNAs and sfRNAs appear to be ubiquitously present in many flaviviruses. They have been
described in MBFVs, including DENV29, YFV30, JEV31, and ZIKV32, TBFVs23,33, and recently in ISFVs and
NKVs34,35. ere is typically more than one xrRNA, given the diverse molecular architecture of dierent avivirus
3′UTRs. Pseudoknot interactions have been proposed in some, but not all avivirus xrRNAs32,36. While they may
form transiently under certain conditions28, conclusive validation of their ubiquitous presence is missing. Hence,
we will exclude them in this work. Earlier studies in our group have identied conserved RNA structural elements
in viral 3′UTRs37–41, some of which have later been attributed to xrRNA functionality23. Stem-loop (SL) as well
as dumbbell (DB) structures are found in 3′UTRs of aviviruses in single or double copies (Fig.1) and have been
associated with quantitative protection of downstream viral RNA42.
e inhibition of Xrn1 by viral RNA yields sfRNAs that aect many cellular processes, both in the vector
and the host43. In mosquitoes, sfRNA interacts directly with the predominant innate immune response pathway,
RNA interference (RNAi), by serving as a template for microRNA (miRNA) biogenesis44. Conversely, in host
cells sfRNA modulates the anti-viral interferon response45, e.g., by binding proteins to inhibit the translation
of interferon-stimulated genes46. Moreover, sfRNA has been shown to inhibit Xrn1 and Dicer activity, thereby
altering host mRNA levels47,48.
At the same time, a variety of host proteins bind the 3′UTR of aviviruses, thereby mediating viral replication,
polyprotein translation or the anti-viral immune response (see Table 1 in ref.43 for a comprehensive overview of
host proteins that bind avivirus 3′UTR/sfRNA). Although notoriously underrepresented in literature, one can
expect that many of these proteins also bind sfRNA due to sequence and structure conservation.
Subgenomic aviviral RNA interacts with Musashi. One of these groups of host factors is the Musashi
(Msi) protein family. Msi is a highly conserved family of proteins in vertebrates and invertebrates that act as a
translational regulator of target mRNAs and is involved in cell proliferation and dierentiation. While the two
Msi paralogs in mammals, Musashi-1 (Msi1) and Musashi-2 (Msi2), are expressed in stem cells49–51 and overex-
pressed in tumors and leukemias52, they are absent in dierentiated tissue. Moreover, Msi1 is involved in the regu-
lation of blood-testis barrier proteins and spermatogenesis in mice53. Musashi proteins have two RNA recognition
motif (RRM) domains, whose sequence specicity has been determined by an in vitro selection method and NMR
spectroscopy51,54,55. e trinucleotide sequence UAG, whose thermodynamic binding specicity was determined
Figure 1. Schematic representation of the ZIKV 3′UTR. Conserved RNA elements include two stem-loop
structures (SL1 and SL2), a
Ψ
DB and canonical DB element as well as the terminal 3′ stem-loop structure (3-
SL). Positions of Musashi-binding UAG motifs in the Asian/American ZIKV lineage are highlighted in orange.
Possible pseudoknot interaction sites (sketched in light blue) do not overlap with potential Musashi binding
sites.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
by uorescence polarization assays, has been identied as core Musashi binding element (MBE). Nucleotides
enclosing the main MBE recognition motif make minor contributions to binding anity56. While earlier SELEX
experiments identied the binding aptamer sequence (G/A)UnAGU (n = 1 − 3)51, iCLIP experiments with Msi1
in human glioblastoma cells conrmed the preferential binding of Msi1 to single-stranded (stem-loop) UAG
sequences in 3′UTRs, but not in coding regions57. Zearfoss et al.56 observed that both GUAGU and AUAGU are
recognized by mouse Msi1, whereas Drosophila Msi1 has a higher anity for GUAGU. NMR-derived structures
of the two Msi1 RNA recognition motifs in complex with RNA also show that both RNA-binding domains bind
GUAGU (PDB IDs 2RS2 and 5X3Z).
In summary, there is a strong consensus in the literature that UAG is central to all proposed Musashi binding
motifs. erefore, we focus our calculations around this trinucleotide, and provide evidence that the availability
of UAG in pentanucleotides expands to the accessibility of the entire motif.
Musashi is involved in avivirus neurotropism. An interesting, yet understudied hypothesis is the
possibility that the stem cell regulator protein Musashi could be related to ZIKV tropism. Based on the iden-
tication of a MBE in the 3′UTR of the ZIKV genome10, de Bernardi Schneider et al.7 reported the presence
of the same element with a higher binding anity for human Msi1 in all ZIKV sequences that belong to the
Asia-Pacic-Americas clade in an in silico screen and implied that there could be a change of tropism for the
viral lineage. Chavali et al.58 tested the possibility of Msi1 interaction with the ZIKV genome in vivo and found
that Msi1 not only interacts with ZIKV, but also enhances viral replication. ey noted that ZIKV RNA could
compete with endogenous targets for binding Msi1 in the brain of the developing fetus, thereby dysregulating
the expression of genes required for neural stem cell development. Based on their data the authors concluded
that Msi1 is involved in ZIKV neurotropism and pathology and raised the question whether MBEs present in
other avivirus genomes could exhibit similar functionality. In a recent study, Platt et al.59 investigated whether
ZIKV-related arboviruses can cause congenital infection and fetal pathology in utero in immunocompetent mice.
ey tested two emerging neurotropic aviviruses, WNV, and Powassan virus (POWV), as well as two alpha-
viruses, Chikungunya virus (CHIKV) and Mayaro virus (MAYV). All four viruses caused placental infection,
however, only WNV and POWV resulted in fetal demise, indicating that ZIKV is not unique among aviviruses
in its capacity to be transplacentally transmitted and cause fetal neuropathology.
In this contribution, we systematically analyze the Musashi-related neurotropic potential of well-curated avi-
virus genomes in silico. We investigate structural features of MBEs in viral 3′UTRs by a thermodynamic model of
RNA structure formation and work out the biophysical properties of conserved RNA structures harboring MBEs
in order to build a theoretical ground for future in vivo studies.
Materials and Methods
Dataset. Sequence data for the present study was acquired from the public National Center for Biotechnology
Information (NCBI) refseq database (https://www.ncbi.nlm.nih.gov/refseq/) on 15 December 2017. We ltered
for all complete viral genomes under taxonomy ID 11051 (genus Flavivirus), resulting in 72 genomes, 51 of which
had 3′UTR sequences and annotation available (Table1).
e core Musashi binding element is only three nucleotides long, hence one can expect to observe a certain
number of UAG trinucleotides by chance in any viral 3′UTR. Table1 shows the number of MBEs present in
3′UTR regions of viral genomes analyzed here as well as the ratio
R O
UAGUAG
=
/
EUAG
, i.e., observed versus
expected frequencies. Assuming that all four nucleotides (A, U, G, C) occur independently and with equal prob-
ability, the expected probability to observe a subsequence of length l is equal to (1/4)l. For
=l3
, this is equal to
1/64. More realistically, the frequency of each nucleotide
∈iAUGC{, ,,}
in an RNA sequence of length L is
=FN
ii
/
L
, where, Ni is the nucleotide count of i. For any trinucleotide XYZ, the expected trinucleotide frequency
EXYZ is then computed from mononucleotide frequencies as
EFFF
XYZXYZ
=∗∗
.
e refseq genome for Spondweni virus (SPONV, accession number NC_029055.1) does not include a 3′UTR
sequence. Since SPONV is phylogenetically closely related to ZIKV60, we were looking to include this sequence
into our analysis. Nikos Vasilakis (Univ. of Texas Medical Branch, Galveston, TX, USA) generously provided
SPONV sequence data. e 338 nt 3′UTR sequence of the SA-Ar strain (see Supplementary Material) has been
added to the set of avivirus sequences analyzed here.
Kama virus (KAMV) does not contain UAG trinucleotides in the 3′UTRs, consequently it has been discarded
from our dataset. e remaining virus species contain between 1 and 19 MBEs in their 3′UTRs.
Opening energy directly relates to single-strandedness. e biophysical model employed here is
based on a description of RNA at the level of secondary structures, building upon the thermodynamic nearest
neighbor energy model as implemented in the ViennaRNA Package61. is allows for computing equilibrium
properties of RNA such as the single most stable, minimum free energy (MFE) structure, as well as the partition
function

. e latter makes an evaluation of the thermodynamic ensemble of RNA structures available and is
dened as the sum over all Boltzmann factors of individual structures s
e
(1)
s
Es RT()/

∑
=
−
where E(s) is the free energy of the structure, R the universal gas constant and T the thermodynamic temperature
of the system. e equilibrium probability of a secondary structure s is then dened as

=.
−
ps e
()
(2)
Es RT()/
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
Group Accession
number Acronym Scientic name 3′UTR
length MBE
count RUAG
MBFV NC_009026.2 AR OAV Aroa virus 421 7 1.18
MBFV NC_012534.1 BAGV Bagaza virus 566 13 1.40
MBFV NC_017086.1 CHAOV Chaoyang virus 326 6 1.25
MBFV NC_001477.1 DENV1 Dengue virus 1 462 7 1.10
MBFV NC_001474.2 DENV2 Dengue virus 2 451 10 1.47
MBFV NC_001475.2 DENV3 Dengue virus 3 440 7 1.14
MBFV NC_002640.1 DENV4 Dengue virus 4 384 5 0.94
MBFV NC_016997.1 DONV Donggang virus 343 10 1.75
MBFV NC_009028.2 ILHV Ilheus virus 388 9 1.87
MBFV NC_001437.1 JEV Japanese encephalitis virus 582 15 1.64
MBFV NC_012533.1 KEDV Kedougou virus 390 2 0.38
MBFV NC_009029.2 KOKV Kokobera virus 558 8 0.98
MBFV NC_000943.1 MVEV Murray Valley encephalitis virus 614 12 1.25
MBFV NC_032088.1 NMV New Mapoon virus 546 14 1.71
MBFV NC_033715.1 NOUV Nounané virus 347 7 1.29
MBFV NC_018705.3 NTAV Ntaya virus 565 15 1.67
MBFV NC_008719.1 SEPV Sepik virus 459 6 0.91
MBFV NC_007580.2 SLEV Saint Louis encephalitis virus 549 7 0.88
MBFV NC_034151.1 THOV T’Ho virus 556 13 1.45
MBFV NC_015843.2 TMUV Tembusu virus 618 16 1.56
MBFV NC_006551.1 USUV Usutu virus 665 19 1.62
MBFV NC_012735.1 WESSV Wesselsbron virus 478 6 0.90
MBFV NC_009942.1 WNV1 West Nile virus lineage 1 631 17 1.70
MBFV NC_001563.2 WNV2 West Nile virus lineage 2 573 16 1.83
MBFV NC_002031.1 YFV17D Yellow fever virus 17D 508 4 0.61
MBFV NC_035889.1 ZIKV-BR Zika virus - Asian-American lineage 429 5 0.92
MBFV NC_012532.1 ZIKV-UG Zika virus - African lineage 428 6 1.09
MBFV NC_029055.1 SPONV Spondweni virus 338a5a0.99
MBFV NC_026623.1 APCV Cacipacore virus N/A N/A N/A
MBFV NC_034018.1 YAOV Yaounde virus N/A N/A N/A
MBFV NC_033693.1 BOUV Bouboui virus N/A N/A N/A
MBFV NC_030289.1 EHV Edge Hill virus N/A N/A N/A
MBFV NC_033699.1 JUGV Jugra virus N/A N/A N/A
MBFV NC_033697.1 SABV Saboya virus N/A N/A N/A
MBFV NC_033698.1 UGSV Uganda S virus N/A N/A N/A
TBFV NC_004355.1 ALKV Alkhumra hemorrhagic fever virus 320 2 0.46
TBFV NC_006947.1 KSIV Karshi virus 381 2 0.38
TBFV NC_003690.1 LGTV Langat virus 568 3 0.40
TBFV NC_001809.1 LIV L ouping ill virus 500 3 0.49
TBFV NC_005062.1 OHFV Omsk hemorrhagic fever virus 410 3 0.60
TBFV NC_003687.1 POWV Powassan virus 480 5 0.76
TBFV NC_027709.1 SGEV Spanish goat encephalitis virus 493 5 0.83
TBFV NC_001672.1 TBEV Tick-borne encephalitis virus 764 6 0.55
TBFV NC_023424.1 TYUV Tyuleniy virus 273 1 0.22
TBFV NC_023439.1 KAMV Kama virus 282 0 —
TBFV NC_033721.1 MEAV Meaban virus N/A N/A N/A
TBFV NC_033726.1 SREV Saumarez Reef virus N/A N/A N/A
TBFV NC_033724.1 KADV Kadam virus N/A N/A N/A
TBFV NC_033723.1 GGV Gadgets Gully virus N/A N/A N/A
ISFV NC_012932.1 AEFV Aedes avivirus 942 10 0.78
ISFV NC_001564.2 CFAG Cell fusing agent virus 553 9 1.16
ISFV NC_008604.2 CxFV Culex avivirus 674 10 1.17
ISFV NC_012671.1 QBV Quang Binh virus 673 7 0.79
ISFV NC_005064.1 KRV Kamiti River virus 1208 13 0.82
ISFV NC_027819.1 MECDV Mercadeo virus 638 11 1.07
Continued
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
e partition function

can be computed eciently via dynamic programming62 and allows calculation of
individual base pair probabilities, even for large sequences63. In this line, the accessibility (i.e., the probability that
a region
i j…
along the RNA is single-stranded) can be derived from the partition function (Eq.1)64. Likewise, the
opening energy (i.e., the free energy required to force the region to be single-stranded) can be computed as
∆=−.GRTPln (unpaired) (3)
open
e opening energy of a region within an RNA is directly related to local RNA secondary structure. In this
line, low opening energy is a reliable indicator for single-strandedness. We employ the sliding window approach
of RNAplfold63 to compute local pairing probabilities of UAG trinucleotide motifs to assess the likelihood of
single-strandedness of and around MBEs. RNAplfold is part of the ViennaRNA Package61 and can compute
the accessibilities or single-strandedness of all intervals of an RNA in cubic time65. We select 97 nt windows
upstream and downstream of MBEs in viral 3′UTRs and compute local pairing probabilities for base pairs within
100 nt windows. Opening energies for trinucleotides are then evaluated from averaged pairing probabilities with
RNAplfold.
e signicance of a calculated MBE opening energy is assessed by comparison with a large number of ran-
domized sequences of the same length and same base or dinucleotide composition. We compute the opening
energies of trinucleotides both in a genomic as well as a shued sequence context and apply a z score statistics.
e normalized z score is dened as
μ
σ
=
−
z
EXYZ()
(4)
open
where Eopen(XYZ) is the opening energy of trinucleotide XYZ in its genomic context, μ and σ are the mean and
standard deviations, respectively, of the opening energies of XYZ computed over a large sample of randomized
sequences. Randomization with regard to keeping sequence composition is achieved here by applying dinucle-
otide shuing to the 97 nt windows upstream and downstream of MBEs, while keeping XYZ in place. e same
idea applies to calculations of pentanucleotide motifs.
e approach outlined above is implemented in the Perl utility plfoldz.pl, which is available from https://
github.com/mtw/plfoldz. e script employs the ViennaRNA61 scripting language interface for thermodynam-
ics calculations, the ViennaNGS66 suite for extraction of genomic loci and the uShue Perl bindings67 for k-let
shuing. e tool reports for each requested trinucleotide the opening energy in a genomic context as well
as an opening energy z score obtained from n shuing events of upstream and downstream sequences. Here,
n = 10,000 dinucleotide shuing events were used.
Characteriztaion of MBEs within xrRNAs. To localize MBEs within homologous substructures in avi-
virus 3′UTRs we constructed infernal68 covariance models for conserved xrRNA elements. e structural RNA
Group Accession
number Acronym Scientic name 3′UTR
length MBE
count RUAG
ISFV NC_021069.1 MSFV Mosquito avivirus 674 9 1.01
ISFV NC_034242.1 OCFVPT Ochlerotatus caspius avivirus 148 2 0.81
ISFV NC_027817.1 PaRV Parramatta River virus 629 12 1.21
ISFV NC_030401.1 HANV Hanko virus N/A N/A N/A
ISFV NC_033694.1 PCV Palm Creek virus N/A N/A N/A
NKV NC_008718.1 ENTV Entebbe bat virus 155 2 1.06
NKV NC_027999.1 EPEV Paraiso Escondido virus 316 2 0.37
NKV NC_004119.1 MMLV Montana myotis leukoencephalitis
virus 460 9 1.18
NKV NC_003635.1 MODV Modoc virus 366 9 1.42
NKV NC_005039.1 YOKV Yokose virus 429 9 1.21
NKV NC_026624.1 SOKV Sokolu k virus N/A N/A N/A
NKV NC_003676.1 APOIV Apoi virus N/A N/A N/A
NKV NC_026620.1 JUTV Jutiapa virus N/A N/A N/A
NKV NC_029054.2 POTV Potiskum virus N/A N/A N/A
NKV NC_034007.1 PPBV Phnom Penh bat virus N/A N/A N/A
NKV NC_003675.1 RBV Rio Bravo virus N/A N/A N/A
NKV NC_003996.1 TABV Tamana bat virus N/A N/A N/A
Table 1. Viral genomes analyzed in this study. Flaviviruses are categorized into the groups mosquito-borne
aviviruses (MBFV), tick-borne aviviruses (TBFV), insect-specic aviviruses (ISFV) and no known vector
aviviruses (NKV). 3′UTR lengths, number of Musashi binding elements found in 3′UTRs and relative MBE
abundance RUAG are listed. RUAG values above 1 indicate relative enrichment of UAG trinucleotides, whereas
values below 1 indicate relative depletion. a3′UTR length and MBE count of SPONV SA-Ar strain. N/A3′UTR
partial or not available in the refseq data set.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
alignments underlying the infernal models were computed with locarna69 and further analyzed with RNAalifold61
and RNAaliSplit70.
Results
MBEs are highly accessible in ZIKV 3′UTRs. e Musashi family of proteins preferentially bind sin-
gle-stranded UAG motifs in 3′UTRs57. To evaluate the thermodynamics of Msi-UAG anity more broadly, we set
out to analyze the single-strandedness of all possible trinucleotides in ZIKV genomes. To this end, we computed
the opening energies of all trinucleotide motifs present in the coding sequence (CDS) and 3′UTR of the African
(ZIKV-UG) and Asian/American (ZIKV-BR) Zika strains. A z score was calculated for each occurrence of tri-
nucleotide XYZ according to Eq.4, thereby normalizing the opening energy of XYZ in its genomic context with
n = 10,000 dinucleotide-shued upstream and downstream regions of 97 nt, using 100 nt windows in RNAplfold.
Negative opening energy z scores indicate increased accessibility, i.e., UAG trinucleotides in viruses with over-
all low z scores are likely to occur in an unpaired structural context within the 3′UTR. rough the distribution
of z scores, sorted by median z score (Fig.2) we were able to see three aspects standing out. First, the distribution
of z scores is markedly divergent among CDS and 3′UTR. e interquartile ranges of opening energy z scores are
homogeneous within the CDS region, while dispersion is varied within the 3′UTR. We hypothesize that this is
caused by a dierent sequence composition that manifests in highly variable opening energies. It could, however,
also be an artifact of the dierent sample sizes based on the divergent trinucleotide count in CDS and 3′UTR,
respectively. Second, UAG is the most accessible trinucleotide in the 3′UTR of ZIKV-BR and among the highest
accessible trinucleotides in the 3′UTR of ZIKV-UG. is is striking as it corroborates previous experimental
evidence of Musashi anity to ZIKV58 by means of a thermodynamic model, thus underlining a possible role of
Msi1 in ZIKV neurotropism. Moreover, the UAG trinucleotide is neither enriched nor depleted in the 3′UTRs
of ZIKV-BR and ZIKV-UG (Table1). ird, the canonical start codon AUG appears to the far right end of the
scale in both ZIKV-BR and ZIKV-UG 3′UTRs, i.e., it is among the least accessible trinucleotides. is suggests
Figure 2. Distribution of z scores of opening energies for trinucleotides found in the coding region (CDS,
le) and 3′UTR (right) of ZIKV from Brazil (top) and Uganda (bottom), sorted by median opening energy z
score. e MBE motif is highlighted in blue and shows low overall z scores in the 3′UTR, indicating that this
trinucleotide is more likely to appear in a single-stranded structural context. Contrary, the canonical start codon
AUG (highlighted in orange) shows high opening energy z scores, indicating reduced accessibility within the
3′UTR region. Data for trinucleotides AUU, UAA, UCG and UUU are omitted because they only occur once
within the 3′UTR of ZIKV-BR. Likewise, trinucleotides CGU, GUA, UAA and UAU are omitted in the ZIKV-UG
3′UTR plot.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
evolutionary pressure on keeping the start codon in a paired structural context within the 3′UTR, thereby prohib-
iting accessibility to ribosomes and disabling undesirable leaky translation start from these AUG triplets.
We also tested the accessibility of larger Musashi recognition motifs. To this end, we employed the same
approach outlined above for all pentanucleotides found in the 3′UTRs of ZIKV-BR and ZIKV-UG, respectively.
e distributions of opening energy z scores (Supplementary Data, Fig.S1) are in good agreement with our
results derived for trinucleotides as well as previous experimental data, suggesting that core a UAG is among the
most accessible motifs. In particular, our data shows that NUAGN is the most accessible pentanucleotide in the
3′UTR of ZIKV-BR and among the highest accessible pentanucleotides in the 3′UTR of ZIKV-UG, similar to the
situation found for trinucleotides. UAG appears to be conserved in an unpaired structural context not only by
itself but also in a larger sequence context of enclosing nucleotides, which exhibit high accessibility upon the pres-
ence of a central UAG Musashi recognition element. is nding is in line with the reported Musashi recognition
pentamers GUAGU and AUAGU.
MBE accessibility in related viruses. To assess the Musashi-related neurotropic potential of other avi-
viruses, we evaluated the accessibilities of MBEs in related species. To this end, all 435 UAG trinucleotide motifs
within 3′UTRs in the refseq dataset were identied, grouped by vector specicity and subjected to the computa-
tional approach outlined above (97 nt upstream/downstream windows, n = 10,000 dinucleotide shuings).
Msi1 preferentially binds single stranded RNA57, consequently UAG motifs that contribute with low z scores
have a high anity for Msi1 binding. Within the MBFV group, the Asian/American lineage Zika virus (ZIKV-BR)
has the lowest median z score, followed by Saint Louis encephalitis virus (SLEV), Nounané virus (NOUV) and
the African lineage Zika virus (ZIKV-UG). Among others, two lineages of West Nile virus (WNV1, WNV2) and
Yellow fever virus (YFV) appear with a negative median z score. ZIKV-BR turns out to be the only isolate among
MBFVs that has just negative z score values in our simulations, i.e., all UAG motifs within the 3′UTR of the
Brazilian ZIKV isolate appear in an unpaired structural context. Likewise, Karshi virus (KSIV), Alkhumra hem-
orrhagic fever virus (ALKV) and Langat virus (LGTV) have a strictly negative z scores distribution among the
TBFV group. POWV, Omsk hemorrhagic fever (OHFV) and Louping ill virus (LIV) have negative mean open-
ing energy z scores. Interestingly, UAG trinucleotides are relatively depleted in all TBFV species analyzed here
(Table1). Among NKVs, Montana myotis leukoencephalitis virus (MMLV) and Entebbe bat virus (ENTV) show
negative mean opening energy z scores. Culex avivirus (CxFV), Cell fusing agent virus (CFAG), Parramatta
River virus (PaRV) and Ochlerotatus caspius avivirus (OCFVPT) tend to have singe-stranded MBEs among the
ISFVs. Here, OCFVPT is the only isolate with a strictly negative z score distribution (Fig.3).
e number of UAG trinucleotide motifs in 3′UTRs of the refseq dataset lies between 1 and 19 (Table1). e
overall range of opening energy z scores is not equal among dierent avivirus groups. While the lower bound
is between −1.65 and −1.93 among all groups, MBFVs and ISFVs show markedly higher upper bounds than
TBFVs and NKVs, respectively. Absolute values of computed MBE opening energy z scores are listed in Table2.
Conserved xrRNAs contain MBEs. Several species appear to the right of the plots in Fig.3 due to the
sorting by median z score. However, they comprise a non-negligible number of accessible MBEs, as indicated by
negative opening energy z scores. Examples are (re-)emerging species like JEV and Usutu virus (USUV), which
contain 15 and 19 MBEs, respectively.
To investigate this further, we assigned each MBE in our dataset to one of the conserved elements stem-loop
(SL), dumbbell (DB) and 3′ stem-loop (3SL) (Fig.1) by means of covariance models. Analysis of RNA sequence
and structure conservation revealed that the majority of virus isolates in our dataset contain only a single UAG
motif within their SL and 3SL elements. Conversely, DB elements, which are conserved in MBFVs and NKVs
(Fig.4), stand out among conserved RNA structures in avivirus 3′UTRs. ey contain a pair of MBEs, separated
by a 4 nt spacer, within a perfectly conserved sequence motif of approx. 20 nt length in their distal stem-loop
structure. We hypothesize that this pair of conserved UAG motifs interacts with the two RNA-binding domains in
Musashi proteins. Figure5 shows the consensus secondary structure of avivirus DB elements.
Discussion
Our ndings lead to the conclusion that the accessibility of UAG motifs calculated through opening energies in
Flavivirus 3′UTRs is indicative of the Musashi-related neurotropic potential of virus species. Our computational
analyses show that there is little dierence in the distribution of opening energies for all trinucleotides within the
polyprotein (CDS) region of ZIKV. When comparing CDS and 3′UTR regions, we see a dierence in behavior,
as dierent trinucleotides do possess dierent opening energies, UAG being highly accessible in ZIKV. Although
it is not possible to quantify the impact of the accessibility on the patient phenotype, it is interesting to see the
UAG motifs in the Brazilian ZIKV isolate more accessible than in the Ugandan ZIKV isolate. is result raises the
question once again if the increased pathogenicity seen in ZIKV today is due to changes in the sequence over time
or simply lack of better surveillance71.
Previous experiments lead toward the idea that ZIKV is unique among aviviruses regarding the clinical out-
comes resulting from congenital infection72,73. Although our results indicate that this may be true for well-studied
viruses such as DENV and WNV, other viruses which have not caused recent outbreaks may have been neglected.
Looking in depth at other viruses, Nounané virus (NOUV), a dual-host aliated insect-specic avivirus
is found among the viruses with high MBE accessibility. NOUV was isolated in Cote d’Ivoire in 2004 from
Uranotaenia mashonaensis, a Culicidae mosquito not known to harbor aviviruses before74. While replication has
been tested in human and non-human cell lines, vertebrate infection and pathogenesis could not be observed75.
Within the TBFV serocomplex, KSIV has the lowest overall MBE opening energies. Originally isolated
from Ornithodoros papillipes ticks in Uzbekistan in 197276 it currently does not present history of infection in
humans. Conversely, Powassan virus (POWV), another TBFV with negative MBE opening energy, rst isolated
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
in Powassan, Ontario, Canada in 1959 from a child who died of acute encephalitis77 can cause transplacental
infection59 and has been associated with severe neuropathology and death in mice and human78.
Given that UAG can be regarded as the primary Msi1 binding motif, we can argue that ZIKV has the highest
anity for binding Msi1 among all MBFVs. Platt et al.59 showed that besides ZIKV, the neurotropic aviviruses
WNV and POWV, as well as the alphaviruses Chikungunya (CHIKV) and Mayaro (MAYV) infect placenta and
fetus in immunocompetent, wild-type mice. However, only WNV was shown to infect the placenta and the fetal
central nervous system, causing injury to the developing brain.
Congenital infection in humans is documented for WNV, JEV, YFV, and ZIKV. CHIKV and MAYV did not
show this behavior. In this line, our results are in agreement with experimental studies that reported teratogenic-
ity for SLEV12,13, WNV59,79, YFV16,80 and POWV59.
Bizarre neurological manifestations were also observed in patients infected by Ntaya virus (NTAV)81, a neuro-
tropic virus from the Japanese encephalitis serocomplex, as well as WNV in humans82,83 and in mice84, USUV85,86
and DENV87,88. e fact that these viruses line up more on the positive side of the opening energy plots in Fig.3
does not mean that they should not be neurotropic. It merely highlights that there might be additional mecha-
nisms causing neuropathogenicity.
MBEs are conserved in avivirus 3′UTR elements. Flavivirus DB elements do not only show structural
conservation over the MBFV and NKV serocomplexes, but even maintain their primary sequence within a region
Figure 3. Distribution of MBE opening energy z scores in avivirus 3′UTRs, grouped by vector specicity
and sorted by median z scores. Top le: MBFVs, top right: ISFVs, bottom le: TBFVs and bottom right: NKVs.
e Asian/American lineage ZIKV-BR isolate has the lowest median z score among all MBFV. Alkhurma virus
(ALKV), Ochlerotatus caspius avivirus (OcFV), ENTV and EPEV contain only two MBEs on the 3′UTR.
Tyuleniy virus (TYUV) was excluded as it only contains a single MBE.
Tot a l SL DB
min max min max min max
MBFV −1.82 6.86 −1.35 2.10 −1.42 2.23
ISFV −1.93 5.14 — — — —
TBFV −1.65 1.97 — — — —
NKV −1.79 2.02 — — −1.79 2.02
Table 2. Distribution ranges of opening energy z scores for MBEs in the 3′UTR of aviviruses. e minimal
and maximal z score is listed for all UAG motifs (total) within a the 3′UTR, and only for those that overlap one
of the conserved xrRNA elements SL and DB. Dashes indicate that SL/DB elements are not conserved.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

9
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
of approx. 20 nt of the distal stem-loop (Figs4 and 5). e combination of covariation and primary sequence
conservation within a single RNA element underlines the importance of DB elements in avivirus pathogenicity.
It could also be indicative of a special role of DB element regions in the minus-strand synthesis during avivirus
replication.
UAG trinucleotides are the core nucleotides within MBE motifs to contribute the highest binding energy56.
Our current analysis underlines that there seems to be evolutionary pressure on keeping UAG motifs within the
DB elements unpaired. In ZIKV we see that not only the UAGs within DB elements but also those that overlap
with SL elements show negative opening energy z-score.
Msi1 presence in dierent cells. e presence of Msi1 proteins in both sperm and neural precursor cells
highlights the importance of studying the Msi1-MBE interaction in aviviruses. Given that Msi1 has been shown
to enhance ZIKV replication58, this interaction could be a critical reason why ZIKV persists in sperm for a long
time aer the individual has been infected89,90, allowing the virus to be transmitted sexually and also why the virus
would harbor itself in neuronal cells, allowing it to interfere with and dysregulate neurodevelopment.
Figure 4. Structural alignment of conserved dumbbell (DB) elements in the 3′UTR of MBFV and NKV
aviviruses. Several species have two copies of DB elements in their 3′UTR, indicated by DB.1 and DB.2
in the sequence identier. Coordinates are given relative to the 3′UTR start. A consensus structure in dot-
bracket notation is plotted on top of the alignment. Gray bars at the bottom indicate almost perfect sequence
conservation within the distal stem-loop sub-structure (positions 60–80). Two conserved MBE motifs in the
central multiloop and distal stem loop are highlighted in blue.
Figure 5. Consensus secondary structure of the avivirus DB element with MBEs highlighted in blue. Figure
generated from the MSA in Fig.4 with R2R94. Structure and color annotation inferred by R2R. Nucleotide
symbols represent conserved nucleotides. Circles represent columns in the MSA that are typically or always
present but do not conserve nucleotide identity. Red, black and gray colors indicate the level of nucleotide
conservation, in decreasing order.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

10
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
A possible role of Musashi in the flavivirus life cycle. Msi1, which binds to the 3′UTR of target
mRNAs, has been shown to repress translation initiation by competing with the translation initiation factor
eIF4G for binding to poly(A)-binding protein (PABP), thereby inhibiting the assembly of the 80S ribosomal
unit91. Ribosome proling experiments have corroborated this down-regulatory eect of Msi1, while keeping
mRNA levels92. is allows for a speculative explanation of the ndings by Chavali et al.58, i.e., that Msi1 enhances
ZIKV replication, and a possible role of Msi1 in the viral life cycle: Flaviviruses need to “donate” a few copies of
the quasispecies ensemble for Xrn1 degradation and subsequent sfRNA production. In this line, Msi1 could serve
as an agent that provides a reasonable amount of gRNAs that are not translated but subject to Xrn1 degradation.
e resulting sfRNAs can then down-regulate the host response46,93.
Conclusion
We studied a specic aspect of avivirus congenital pathogenicity, i.e., the neurotropic eect inferred by the
presence of MBEs in the 3′UTR of avivirus genomes. Employing an established biophysical model of RNA
structure formation, we analyzed the thermodynamic properties of MBEs in silico. Our results underline experi-
mental studies suggesting that ZIKV is not alone in its capacity to cause severe neuropathology to infants through
the MBE mechanism. While several tick-borne and mosquito-borne avivirus species like Karshi virus (KSIV),
Alkhumra hemorrhagic fever virus (ALKV) or Nounané virus (NOUV) line up with ZIKV in our theoretical
model, their tropism might have been overseen due to the lack of reported signicant outbreaks. However, some
of them appear to have similar neurotropic potential and thus might be potent emerging pathogens.
e approach presented here could in principle be used for developing a tool to predict the Musashi-related
neurotropic potential of novel viruses or (re-)emerging strains of known viruses. Combination of opening energy
z scores with large scale epidemiologic data could be employed in a machine learning framework that also con-
siders structural conservation and homology of avivirus 3′UTR elements. Such a tool could play a role in cate-
gorizing viruses.
Data Availability
e plfoldz.pl Perl Utility for computing RNA opening energy z scores is available from https://github.com/mtw/
plfoldz.
References
1. Tognarelli, J. et al. A report on the outbrea of Zia virus on Easter Island, South Pacic, 2014. Arch Virol 161, 665–668 (2016).
2. Malone, . W. et al. Zia virus: medical countermeasure development challenges. PLoS Neglect Trop D 10, e0004530 (2016).
3. Hotez, P. J. & Murray, . O. Dengue, West Nile virus, Chiungunya, Zia and now Mayaro? PLoS Neglect Trop D 11, e0005462
(2017).
4. Gubler, D., uno, G. & Maro, L. Flaviviruses. Fields Virology 1, 1153–1252 (2007).
5. Weaver, S. C. et al. Zia virus: History, emergence, biology, and prospects for control. Antivir e s 130, 69–80 (2016).
6. Song, B.-H., Yun, S.-I., Woolley, M. & Lee, Y.-M. Zia virus: history, epidemiology, transmission, and clinical presentation. J
Neuroimmunol 308, 50–64 (2017).
7. de Bernardi Schneider, A. et al. Molecular evolution of Zia virus as it crossed the Pacic to the Americas. Cladistics 33, 1–20 https://
doi.org/10.1111/cla.12178 (2017).
8. Gong, Z., Xu, X. & Han, G.-Z. e diversication of Zia virus: Are there two distinct lineages? Genome Biol Evol 9, 2940–2945
(2017).
9. Simonin, Y., van iel, D., Van de Perre, P., ocx, B. & Salinas, S. Dierential virulence between Asian and African lineages of Zia
virus. PLoS Negl Trop D 11, e0005821 (2017).
10. lase, Z. A. et al. Zia fetal neuropathogenesis: etiology of a viral syndrome. PLoS Neglect Trop D 10, e0004877 (2016).
11. Platt, D. J. & Miner, J. J. Consequences of congenital Zia virus infection. Curr Opin Virol 27, 1–7 (2017).
12. Andersen, A. & Hanson, . Experimental transplacental transmission of St. Louis encephalitis virus in mice. Infect Immun 2,
320–325 (1970).
13. Andersen, A. & Hanson, . Intrauterine infection of mice with St. Louis encephalitis virus: immunological, physiological,
neurological, and behavioral eects on progeny. Infect Immun 12, 1173–1183 (1975).
14. Fujisai, Y., Miura, Y., Sugimori, T., Muraami, Y. & Miura, . Experimental studies on vertical infection of mice with Japanese
encephalitis virus. IV. Eect of virus strain on placental and fetal infection. NatL I Anim Health Q 23, 21–26 (1983).
15. Chaturvedi, U. et al. Transplacental infection with Japanese encephalitis virus. J Infect Dis 141, 712–715 (1980).
16. Tsai, T., Paul, ., Lynberg, M. & Letson, G. Congenital yellow fever virus infection aer immunization in pregnancy. J Inf Dis 168,
1520–1523 (1993).
17. Foy, B. D. et al. Probable non–vector-borne transmission of Zia virus, Colorado, USA. Emerg Infect Dis 17, 880 (2011).
18. Musso, D. et al. Potential sexual transmission of Zia virus. Emerg Infect Dis 21, 359 (2015).
19. Hahn, C. S. et al. C onserved elements in the 3′ untranslated region of avivirus NAs and potential cyclization sequences. J Mol Biol
198, 33–41 (1987).
20. Muhopadhyay, S., uhn, . J. & ossmann, M. G. A structural perspective of the avivirus life cycle. Nat ev Microbiol 3, 13 (2005).
21. Villordo, S. M., Alvarez, D. E. & Gamarni, A. V. A balance between circular and linear forms of the dengue virus genome is crucial
for viral replication. NA 16, 2325–2335 (2010).
22. de Borba, L. et al. Overlapping local and long-range NA-NA interactions modulate dengue virus genome cyclization and
replication. J Virol 89, 3430–3437 (2015).
23. Pijlman, G. P. et al. A highly structured, nuclease-resistant, noncoding NA produced by aviviruses is required for pathogenicity.
Cell Host Microbe 4, 579–591 (2008).
24. Aiyama, B. M., Eiler, D. & ie, J. S. Structured NAs that evade or confound exonucleases: function follows form. Curr Opin
Struct Biol 36, 40–47 (2016).
25. Jones, C. I., Zabolotsaya, M. V. & Newbury, S. F. e 5′-3′ exoribonuclease Xrn1/Pacman and its functions in cellular processes and
development. Wires NA 3, 455–468 (2012).
26. Antic, S., Wolnger, M. T., Sucha, A., Hosiner, S. & Dorner, S. General and miNA-mediated mNA degradation occurs on
ribosome complexes in Drosophila cells. Mol Cell Biol MCB–01346, https://doi.org/10.1128/MCB.01346-14 (2015).
27. Fun, A. et al. NA structures required for production of subgenomic avivirus NA. J Viro l 84, 11407–11417 (2010).
28. Chapman, E. G., Moon, S. L., Wilusz, J. & ie, J. S. NA structures that resist degradation by Xrn1 produce a pathogenic Dengue
virus NA. elife 3, e01892 (2014).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

11
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
29. Liu,  . et al. Identication and characterization of small sub-genomic NAs in dengue 1–4 virus-infected cell cultures and tissues.
Biochem Bioph es Co 391, 1099–1103 (2010).
30. Silva, P. A., Pereira, C. F., Dalebout, T. J., Spaan, W. J. & Bredenbee, P. J. An NA pseudonot is required for production of yellow
fever virus subgenomic NA by the host nuclease XN1. J Vi rol 84, 11395–11406 (2010).
31. Fan, Y.-H. et al. Small noncoding NA modulates Japanese encephalitis virus replication and translation in trans. Virol J 8, 492
(2011).
32. Aiyama, B. M. et al. Zia virus produces noncoding NAs using a multi-pseudonot structure that confounds a cellular
exonuclease. Science aah3963 (2016).
33. Schnettler, E. et al. Induction and suppression of tic cell antiviral NAi responses by tic-borne aviviruses. Nucleic Acids es 42,
9436–9446 (2014).
34. MacFadden, A. et al. Mechanism and structural diversity of exoribonuclease-resistant NA structures in aviviral NAs. Nat
Commun 9, 119 (2018).
35. Ochsenreiter, ., Hofacer, I. L. & Wolnger, M. T. Functional NA Structures in the 3′UT of Tic-borne, Insect-specic and No
nown Vector Flaviviruses. Viruses 11, 298 (2019).
36. Sztuba-Solinsa, J. et al. Structural complexity of Dengue virus untranslated regions: cis-acting NA motifs and pseudonot
interactions modulating functionality of the viral genome. Nucleic Acids es 41, 5075–5089 (2013).
37. auscher, S., Flamm, C., Mandl, C. W., Heinz, F. X. & Stadler, P. F. Secondary structure of the 3′-noncoding region of avivirus
genomes: comparative analysis of base pairing probabilities. NA 3, 779–791 (1997).
38. Hofacer, I. L. et al. Automatic detection of conserved NA structure elements in complete NA virus genomes. Nucleic Acids es
26, 3825–3836 (1998).
39. Witwer, C., auscher, S., Hofacer, I. L. & Stadler, P. F. Conserved NA secondary structures in picornaviridae genomes. Nucleic
Acids es 29, 5079–5089 (2001).
40. Hofacer, I. L., Stadler, P. F. & Stocsits, . . Conserved NA secondary structures in viral genomes: a survey. Bioinformatics 20,
1495–1499 (2004).
41. urner, C., Witwer, C., Hofacer, I. L. & Stadler, P. F. Conserved NA secondary structures in aviviridae genomes. J Gen Virol 85,
1113–1124 (2004).
42. ie, J. S., abe, J. L. & Chapman, E. G. New hypotheses derived from the structure of a aviviral Xrn1-resistant NA: Conservation,
folding, and host adaptation. NA Biology 12, 1169–1177 (2015).
43. oby, J. A., Pijlman, G. P., Wilusz, J. & hromyh, A. A. Noncoding subgenomic avivirus NA: multiple functions in West Nile
virus pathogenesis and modulation of host responses. Viruses 6, 404–427 (2014).
44. Hussain, M. et al. West Nile virus encodes a microNA-lie small NA in the 3′ untranslated region which up-regulates GATA4
mNA and facilitates virus replication in mosquito cells. Nucleic Acids es 40, 2210–2223 (2011).
45. S chuessler, A. et al. West Nile virus noncoding subgenomic NA contributes to viral evasion of the type I interferon-mediated
antiviral response. J Viro l 86, 5708–5718 (2012).
46. Manoaran, G. et al. Dengue subgenomic NA binds TIM25 to inhibit interferon expression for epidemiological tness. Science
350, 217–221 (2015).
47. Moon, S. L. et al. A noncoding NA produced by arthropod-borne aviviruses inhibits the cellular exoribonuclease XN1 and alters
host mNA stability. NA (2012).
48. Schnettler, E. et al. Non-coding avivirus NA displays NAi suppressor activity in insect and mammalian cells. J Vi rol JVI–01104
(2012).
49. Saaibara, S.-I. et al. Mouse-Musashi-1, a neural NA-binding protein highly enriched in the mammalian CNS stem cell. Dev Biol
176, 230–242 (1996).
50. Saaibara, S.-I., Naamura, Y., Satoh, H. & Oano, H. NA-binding protein Musashi2: developmentally regulated expression in
neural precursor cells and subpopulations of neurons in mammalian CNS. J Neurosci 21, 8091–8107 (2001).
51. Imai, T. et al. e neural NA-binding protein Musashi1 translationally regulates mammalian numb gene expression by interacting
with its mNA. Mol Cell Biol 21, 3888–3900 (2001).
52. haras, M. G. et al. Musashi-2 regulates normal hematopoiesis and promotes aggressive myeloid leuemia. Nat Med 16, 903 (2010).
53. ErLin, S. et al. Musashi-1 maintains blood–testis barrier structure during spermatogenesis and regulates stress granule formation
upon heat stress. Mol Biol Cell 26, 1947–1956 (2015).
54. Ohyama, T. et al. Structure of Musashi1 in a complex with target NA: the role of aromatic stacing interactions. Nucleic Acids es
40, 3218–3231 (2012).
55. Iwaoa, . et al. Structural Insight into the ecognition of r(UAG) by Musashi-1 BD2, and Construction of a Model of Musashi-1
BD1-2 Bound to the Minimum Target NA. Molecules 22, 1207 (2017).
56. Zearfoss, N. . et al. A conserved three-nucleotide core motif defines Musashi NA binding specificity. J Biolo Chem 289,
35530–35541 (2014).
57. Uren, P. J. et al. NA-binding protein Musashi1 is a central regulator of adhesion pathways in glioblastoma. Mol Cell Biol 35,
2965–2978 (2015).
58. Chavali, P. L. et al. Neurodevelopmental protein Musashi 1 interacts with the Zia genome and promotes viral replication. Science
eaam9243 (2017).
59. Platt, D. J. et al. Zia virus-related neurotropic aviviruses infect human placental explants and cause fetal demise in mice. Sci Transl
Med 10, eaao7090 (2018).
60. Haddow, A. D. et al. Genetic Characterization of Spondweni and Zia Viruses and Susceptibility of Geographically Distinct Strains
of Aedes aegypti, Aedes albopictus and Culex quinquefasciatus (Diptera: Culicidae) to Spondweni Virus. PLoS Neglect Trop D 10,
e0005083 (2016).
61. Lorenz, . et al. ViennaNA Pacage 2.0. Algorithm Mol Biol 6, 26 (2011).
62. McCasill, J. S. e equilibrium partition function and base pair binding probabilities for NA secondary structure. Biopolymers 29,
1105–1119 (1990).
63. Bernhart, S. H., Hofacer, I. L. & Stadler, P. F. Local NA base pairing probabilities in large sequences. Bioinformatics 22, 614–615
(2005).
64. Lorenz, ., Wolnger, M. T., Tanzer, A. & Hofacer, I. L. Predicting NA secondary structures from sequence and probing data.
Methods 103, 86–98, https://doi.org/10.1016/j.ymeth.2016.04.004 (2016).
65. Bernhart, S. H., Mücstein, U. & Hofacer, I. L. NA Accessibility in cubic time. Algorithm Mol Biol 6, 3 (2011).
66. Wolnger, M. T., Fallmann, J., Eggenhofer, F. & Amman, F. ViennaNGS: A toolbox for building ecient next-generation sequencing
analysis pipelines. F1000esearch 4:50, https://doi.org/10.12688/f1000research.6157.2 (2015).
67. Jiang, M., Anderson, J., Gillespie, J. & Mayne, M. uShue: a useful tool for shuing biological sequences while preserving the -let
counts. BMC Bioinformatics 9, 192 (2008).
68. Nawroci, E. P. & Eddy, S. . Infernal 1.1: 100-fold faster NA homology searches. Bioinformatics 29, 2933–2935, https://doi.
org/10.1093/bioinformatics/btt509 (2013).
69. Will, S., eiche, ., Hofacer, I. L., Stadler, P. F. & Bacofen, . Inferring noncoding NA families and classes by means of genome-
scale structure-based clustering. PLoS Comp Biol 3, e65 (2007).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

12
Scientific REPORTS | (2019) 9:6911 | https://doi.org/10.1038/s41598-019-43390-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
70. Wolfinger, M. T. Bio::NA::NAaliSplit 0.09, https://doi.org/10.5281/zenodo.2532826, Https://github.com/mtw/Bio-NA-
NAaliSplit (2019).
71. de Bernardi Schneider, A. & Wolnger, M. T. Preventing disease outbreas with computational biology, how far can we go? NCT
CBNW Newsletter 58, https://doi.org/10.5281/zenodo.1463018 (2018).
72. ichard, A. S. et al. AXL-dependent infection of human fetal endothelial cells distinguishes Zia virus from other pathogenic
aviviruses. Proc Natl Acad Sci USA 201620558 (2017).
73. aooza-Mwesige, A., Mohammed, A. H., ristensson, ., Juliano, S. L. & Lutwama, J. J. Emerging viral infections in Sub-Saharan
Africa and the Developing nervous System: A Mini eview. Front Neurol 9, 82 (2018).
74. Junglen, S. et al. A new avivirus and a new vector: characterization of a novel avivirus isolated from uranotaenia mosquitoes from
a tropical rain forest. J Virol 83, 4462–4468 (2009).
75. Huhtamo, E. et al. Novel aviviruses from mosquitoes: Mosquito-specic evolutionary lineages within the phylogenetic group of
mosquito-borne aviviruses. Virology 464, 320–329 (2014).
76. Lvov, D. et al. ”arshi” virus, a new avivirus (Togaviridae) isolated from Ornithodoros papillipes (Birula, 1895) tics in Uzbe SS.
Arch Virol 50, 29–36 (1976).
77. McLean, D. & Donohue, W. Powassan virus: isolation of virus from a fatal case of encephalitis. Canad Med Assoc J 80, 708 (1959).
78. Piantadosi, A. et al. Emerging cases of Powassan virus encephalitis in New England: clinical presentation, imaging, and review of the
literature. Clin Infect Dis 62, 707–713 (2015).
79. O’Leary, D. . et al. Birth outcomes following West Nile Virus infection of pregnant women in the United States: 2003–2004.
Pediatrics 117, e537–e545 (2006).
80. Nishioa, D. A. et al. Yellow fever vaccination during pregnancy and spontaneous abortion: a case-control study. Trop Med Int
Health 3, 29–33 (1998).
81. Woodru, A., Bowen, E. & Platt, G. Viral infections in travellers from tropical Africa. Br Med J 1, 956–958 (1978).
82. for Disease Control, C. (CDC, P. et al. Intrauterine West Nile virus infection–New Yor, 2002. MMW. Morb Mortal W 51, 1135
(2002).
83. Alpert, S. G., Fergerson, J. & Noël, L.-P. Intrauterine West Nile virus: ocular and systemic ndings. Am J Ophthalmol 136, 733–735
(2003).
84. Julander, J. G. et al. Treatment of West Nile virus-infected mice with reactive immunoglobulin reduces fetal titers and increases dam
survival. Antivir es 65, 79–85 (2005).
85. Salinas, S. et al. Deleterious eect of Usutu virus on human neural cells. PLoS Neglect Trop Dis 11, e0005913 (2017).
86. Bassi, M. ., S empere, . N., Meyn, P., Polace, C. & Arias, A. Extinction of Zia virus and Usutu virus by lethal mutagenesis reveals
dierent patterns of sensitivity to three mutagenic drugs. Antimicrob Agents Ch AAC–00380 (2018).
87. Yin, X., Zhong, X. & Pan, S. Vertical transmission of dengue infection: the rst putative case reported in China. ev Inst Med Trop
SP 58 (2016).
88. anjan, ., umar, . & Nagar, N. Congenital dengue infection: Are we missing the diagnosis? Pediatr Infect Dis J 8, 120–123 (2016).
89. Saaibara, S.-I. et al. NA-binding protein Musashi family: roles for CNS stem cells and a subpopulation of ependymal cells
revealed by targeted disruption and antisense ablation. Proc Nat Acad Sci USA 99, 15194–15199 (2002).
90. D’Ortenzio, E. et al. Evidence of sexual transmission of Zia virus. New Engl J Med 374, 2195–2198 (2016).
91. awahara, H. et al. Neural NA-binding protein Musashi1 inhibits translation initiation by competing with eIF4G for PABP. J Cell
Biol 181, 639–653 (2008).
92. atz, Y. et al. Musashi proteins are post-transcriptional regulators of the epithelial-luminal cell state. Elife 3 (2014).
93. Clare, B., oby, J., Sloncha, A. & hromyh, A. Functional non-coding NAs derived from the avivirus 3′ untranslated region.
Virus  es 206, 53–61 (2015).
94. Weinberg, Z. & Breaer, . . 2-software to speed the depiction of aesthetic consensus NA secondary structures. BMC
Bioinformatics 12, 3 (2011).
Acknowledgements
We thank Nikos Vasilakis for providing Spondweni virus sequences. We further thank Ivo Hofacker for fruitful
discussions. is work was partly funded by the Austrian science fund FWF project F43 “RNA regulation of the
transcriptome”.
Author Contributions
A.B.S. and M.T.W. conceived the study, conducted the in silico experiments, analysed the results and wrote the
manuscript. Both authors reviewed the manuscript.
Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-43390-5.
Competing Interests: e authors declare no competing interests.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2019
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com