Content uploaded by Ethan Will Taylor
Author content
All content in this area was uploaded by Ethan Will Taylor on Oct 10, 2017
Content may be subject to copyright.
D
I
S
C
L
A
I
M
E
R
This paper was submitted to the Bulletin of the World Health Organization and was posted to
the Zika open site, according to the protocol for public health emergencies for international
concern as described in Christopher Dye et al. (http://dx.doi.org/10.2471/BLT.16.170860).
The information herein is available for unrestricted use, distribution and reproduction in any
medium, provided that the original work is properly cited as indicated by the Creative Commons
Attribution 3.0 Intergovernmental Organizations licence (CC BY IGO 3.0).
RECOMMENDED
C
I
T
A
T
I
O
N
Taylor EW, Ruzicka JA. Antisense inhibition of selenoprotein synthesis by ZIKV may
contribute to neurological disorders and microcephaly by mimicking SePP1 knockout and the
g
enetic
d
ise
a
se PCCA.
[
Submitte
d
]
. Bull Wo
r
l
d
He
a
lth Or
g
a
n. E-
p
ub: 13 Jul
y
2016. doi:
htt
p
://
d
x.doi.o
r
g
/10.2471/BLT.16.182071
Antisense inhibition of selenoprotein synthesis by Zika virus may contribute
to neurological disorders and microcephaly by mimicking selenoprotein P
knockout and the genetic disease progressive cerebello-cerebral atrophy
Ethan Will Taylora and Jan A. Ruzickab
aDept. of Chemistry and Biochemistry, University of North Carolina at Greensboro, 435 Patricia A.
Sullivan Science Building, PO Box 26170, Greensboro, NC 27402-6170, United States of America.
bDept. of Basic Pharmaceutical Sciences, Fred Wilson School of Pharmacy.
High Point University, One University Parkway, High Point, NC 27268, United States of America.
Correspondence to Ethan Taylor (email: ewtaylor@uncg.edu) or Jan Ruzicka (jruzicka@highpoint.edu)
(Submitted: 7 July 2016 – Published online: 13 July 2016)
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
2
Abstract
Objective: Selenium status plays a major role in health impacts of various RNA viruses. We recently
reported potential antisense interactions between viral mRNAs and host mRNAs of the antioxidant
selenoprotein thioredoxin reductase (TR). Here, we examine possible targeting of selenoprotein mRNAs
by Zika virus (ZIKV) as a pathogenic mechanism, because microcephaly is a key manifestation of
Progressive Cerebello-Cerebral Atrophy (PCCA), a genetic disease of impaired selenoprotein synthesis.
Methods: Potential antisense matches between ZIKV and human selenoprotein mRNAs were initially
identified via nucleotide BLAST searches, using ZIKV genomic RNA as a probe. The strongest antisense
matches of ZIKV regions, against human TR1 and selenoprotein P (SePP1), were validated by algorithms
for prediction of RNA hybridization and microRNA/target duplexes. The ZIKV-SePP1 interaction was
further assessed by gel shift assay.
Findings: Computationally, ZIKV has regions of extensive (~30bp) and stable (ΔE < −50kcal/mol)
antisense interactions with mRNAs of both TR1 and SePP1, a selenium carrier protein essential for
delivery of selenium to the brain. The ZIKV/SePP1 hybridization was experimentally confirmed at the
DNA level using synthetic oligonucleotides.
Conclusion: Antisense inhibition of TR may be a general RNA virus strategy to favor RNA synthesis over
DNA. ZIKV-mediated antisense inhibition of SePP1 and TR1 in fetal brain could mimic SePP1 knockout in
mice, contribute to neuronal cell death, and mimic the genetic disease PCCA, characterized by brain
atrophy and microcephaly. When given to nursing mothers, sodium selenite can counteract neurological
deficits in SePP1 knockout mice, suggesting that a similar approach might help reduce ZIKV-induced
human fetal abnormalities.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
3
Introduction
Among the many possible causes of microcephaly, one of the rarest is the genetic disease known as
Progressive Cerebello-Cerebral Atrophy (PCCA)a, which was first reported in 2003 by Ben-Zeev et al. as a
previously undescribed autosomal recessive syndrome with an unknown genetic basis [1]. The syndrome
was identified in children of Sephardic Jewish descent in Iraq and Morocco, and is characterized by
progressive microcephaly, profound mental retardation, and severe quadriplegic spasticity. The
underlying genetic basis was later identified by Agamy et al. [2] as a defect in selenium metabolism,
disrupting the sole route to the biosynthesis of the 21st amino acid, selenocysteine (Sec), and thus
making it impossible for affected individuals to synthesize selenoproteins. Specifically, the mutations
responsible for this disease were found to be in the SepSecS gene, which encodes O-phosphoseryl-
tRNA:selenocysteinyl-tRNA synthase, a selenium transferase that catalyzes the final step in the
biosynthesis of Sec. Unique among the amino acids, Sec is synthesized while bound to its cognate tRNA,
by conversion of O-phosphoseryl-tRNASec to selenocysteinyl-tRNASec. Homozygous individuals, lacking a
functional copy of SepSecS, have therefore lost the unique catalytic benefits of Sec in selenoenzymes, at
best having a serine hydroxyl group in place of the selenol of an active site Sec, leading to a drastic or
total loss of catalytic activity in thioredoxin reductases, Se-dependent glutathione peroxidases,
iodothyronine deiodinases, and other human selenoproteins.
The essential role of selenium and selenoproteins in mammalian brain function and development is well
established (as briefly reviewed by Agamy et al. [2] and others [3-5]), and is critically dependent upon
selenoprotein P (SePP1), a selenium carrier protein that is secreted by the liver and serves to transport
selenium to other tissues [6]. Comparison of whole body knockout versus selective liver knockout of the
SePP1 gene shows that the brain is able to (and indeed must) synthesize its own version of the protein,
in order to redistribute selenium within the brain [3, 7, 8]. After SePP1 uptake into cells of the blood
brain barrier via its receptor ApoER2 [9, 10], selenium is cleaved from endocytosed SePP1 molecules by
Sec lyase, so that selenium can then be dispersed within the brain, at least in part by incorporation into
newly synthesized SePP1 that is in turn taken up by neurons [3, 5, 8].
This mechanism for SePP1-dependent transport of selenium into the brain is consistent with results
showing that total knockout of SePP1 leads to a substantial reduction of selenoprotein synthesis within
the brain [11], explaining the severe neurological and developmental deficits seen in SePP1–/– mice.
a PCCA is also known as pontocerebellar hypoplasia type 2D (PCH2D), Online Mendelian Inheritance in Man
(OMIM) Phenotype MIM # 613811 Gene locus MIM # 613009.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
4
Because biosynthesis of the entire selenoproteome within the brain is critically dependent on SePP1, it
is not surprising that knockout of SePP1 in the brain produces a neurodegenerative phenotype under
low selenium conditions that is similar to that observed by neuron-specific knockout of tRNASec (a
condition analogous to the genetic defect of PCCA). Either SePP1 knockout under low selenium
conditions [10], or tRNASec knockout alone [12], will result in total or near-abrogation of brain
selenoprotein synthesis.
This raises the possibility that knockdown of SePP1 during fetal development by other mechanisms, such
as antisense inhibition, might also be able to mimic the effects of SePP1 or tRNASec knockout, and
produce neurodegenerative and neurodevelopmental symptoms similar to PCCA, including
microcephaly, particularly if accompanied by selenium deficiency.
Significantly, we recently demonstrated potential virus-host RNA antisense interactions between Ebola
and HIV-1 viral mRNAs and the mRNAs of several isoforms of human thioredoxin reductase (TR), which
raises the possibility that this novel mechanism might contribute to the well-established role for
selenium in the pathogenesis of a number of RNA virus infections [13]. Hence, we investigate here the
possibility that ZIKV mRNA might target one or more human selenoprotein mRNAs by antisense, which
could lead to knockdown of the target selenoprotein levels by either mRNA degradation, or, more likely,
by inhibition of protein synthesis at the ribosomal level.
Although RNA viruses lacking a DNA stage generally replicate their RNA exclusively in the cytoplasm, it is
well established that Flaviviruses in the same genus as ZIKV form replication complexes in the nucleus as
well as in the cytosol of infected cells, with about 20% of viral RNA polymerase activity in the nucleus
[14]. Thus, the possibility that ZIKV mRNA might be able to interact with unspliced cellular pre-mRNAs in
the nucleus must be seriously considered. Formation of a virus-host antisense complex involving an
intron of a cellular RNA transcript could lead to the inhibition of mRNA splicing and maturation, and thus
serve as a mechanism for gene silencing.
In addition, Flaviviruses are known to produce a non-coding "subgenomic Flavivirus RNA" (sfRNA),
corresponding more or less to the entire 3' untranslated region (3’UTR) of the mRNA [15], and which
contains a number of highly conserved RNA secondary structures [16], some of which are quite dynamic
and alter their conformation and interactions with different regions of the viral RNA during replication of
the viral genome [17]. There is also some evidence that sfRNA may play various roles in the regulation
and inhibition of cellular RNA processing [17-19]. Here we show that a region near the 5' end of the
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
5
predicted ZIKV sfRNA may target an intron in the TR1 primary transcript, potentially leading to
knockdown of TR1 gene expression in infected cells. This is of particular interest in the case of ZIKV,
because it has been shown that TR1 inhibition in neuronal cells leads to apoptotic cell death [20], which
has also been demonstrated to be a consequence of ZIKV infection of cultured neurons [21].
Thus, virus-mediated antisense knockdown of host SePP1 and TR1 could together contribute to the most
important ZIKV-associated pathologies.
Methods
Computational methods for identification of virus-host RNA antisense interactions: Potential
antisense matches between regions of ZIKV and host selenoprotein mRNAs were initially identified via
nucleotide BLAST searches (http://blast.ncbi.nlm.nih.gov/Blast.cgi). A complete 2015 Brazilian ZIKV
genomic mRNA sequence (see following section) was used as a probe for Nucleotide BLAST (blastn
option) with default search parameters, initially against the Reference RNA sequence database
(refseq.rna), restricted to Homo sapiens (taxid:9606) with selenoprotein as a search term. Because ZIKV
is known to replicate in the nucleus, the search was later extended to include genomic sequences
(corresponding to unspliced RNA; see below). The initial mRNA search lead to the identification of the
core region of the antisense match to a coding region of SePP1 mRNA shown in Fig. 1. The RNAHybrid
program (http://bibiserv2.cebitec.uni-bielefeld.de/rnahybrid) [22] was then used to assess the match
using an algorithm and parameters designed for actual RNA:RNA interactions, and to investigate the
possibility that, with some mismatches and bulges characteristic of dsRNA, the match at the RNA level
might be extended in the 5’ and 3′ directions beyond the core region identified by BLAST. A more
stringent method for accurate prediction of RNA-RNA interactions was then applied: the IntaRNA
program (http://rna.informatik.uni-freiburg.de/IntaRNA/Input.jsp) [22-24]. This algorithm factors in not
only the hybridization energy of the interacting pair of RNAs, but also takes into consideration the
“unfolding energy” required to overcome and outweigh the stability of any internal RNA secondary
structures within the two individual RNA strands. The program will only identify a match if the net inter-
strand binding energy is substantially lower (i.e., more stable) than the sum of the internal folding
energies of the individual RNA strands. Using input fragments of up to 2000 bases in length (the
maximum supported by the program) IntaRNA was in each case able to identify as the single most
significant match an antisense interaction that was essentially identical to that found by the combined
BLAST-RNAHybrid approach.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
6
Reference sequences used in antisense sequence analysis and oligonucleotide design: The ZIKV
Brazilian strain SPH2015, Genbank accession number KU321639, was used as a reference sequence for
ZIKV, as it was the only complete Brazilian genomic sequence available at the time the study was
initiated. Using methods described below, the most significant antisense match found to a region of a
human selenoprotein was to selenoprotein P (SePP1). The initial SePP1 hit was to a region of mRNA
common to all of its transcript variants; the sequence numbering used in the figures and text is from
transcript variant 6 (the longest variant), Genbank # NM_005410.2. Because the antisense match to the
SePP1 mRNA was at the very beginning of an exon, to explore the possibility that this match might
extend into the upstream intron in unspliced RNA, that search was repeated using genomic DNA instead
of the mRNA database, leading to the identification of an extended match spanning the intron-exon
boundary in the SePP1 genomic sequence, Genbank DQ022288.1. An antisense match of a 3’ noncoding
region of ZIKV to an intron of thioredoxin reductase 1 (TR1) was also identified in the search vs. human
genomic DNA; the TR1 sequence numbering used in the figures and text is from the complete TR1 gene
sequence, Genbank # DQ157758.1.
Experimental confirmation of the predicted ZIKV-SePP1 antisense interaction at the DNA level: An
electrophoretic mobility shift assay was used to confirm the predicted interaction shown in Fig. 1 via
demonstration of in vitro DNA hybridization of the cognate ZIKV:SePP1 pair of sequences. The procedure
was essentially as described previously [13]. Briefly, synthetic single stranded ssDNA oligomers
(Integrated DNA Technologies, Inc., Coralville, IA) were obtained, corresponding precisely to the ZIKV
and SePP1 sequence fragments shown in Fig. 1A. An additional oligo with a random sequence of
identical base composition to the SePP1 fragment was used as a control. Prior to gel electrophoresis,
oligos (~1 μg each in 10 μl PBS), either singly or in all three possible pairs, were incubated at 37°C for 15
hours, followed by cooling to room temperature over 1 hour. See legend to Fig. 1 for details about the
arrangement of lanes on the gel. Bands were separated on a 4% agarose gel and visualized using
ethidium bromide.
Results
The most significant BLAST-generated antisense match to a human selenoprotein identified using the
entire ZIKV mRNA as a probe is to a coding region of SePP1 mRNA. The core match identified using
BLAST involves ZIKV bases 9719-9740 (ATATGGGAAAAGTTAGGAAGGA) with 21/22 bases exactly
matching bases in the minus sense of SePP1, with a single A base forming an unpaired bulge near the
center of an extended double helix, observable in the computed hybridized RNA as the lower portion of
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
7
the more extended structure shown in Fig. 1A. By an iterative process with several slightly extended
versions of this region of ZIKV as queries against the entire SePP1 mRNA, the RNAHybid program was
able to identify the structure shown in Fig. 1A as a possible more extended antisense interaction
between these two RNAs, as the minimum free energy result obtained when using the ZIKV 54mer
shown as a query vs. the entire SePP1 mRNA.
Using fragments up to 2000 bases long from both ZIKV and SePP1 spanning the regions of interest as
input, the IntaRNA program produced an essentially identical (but slightly truncated) interaction as the
single most significant predicted RNA:RNA interaction (included as Fig. S1 in online Supplemental
Materials). This result is important, as it indicates that even in the context of much larger lengths of
RNA, these specific regions are potentially able to interact and form a hybrid virus-host dsRNA structure
that is considerably more thermodynamically stable than the pair of individual mRNAs even when they
are folded into their respective minimum energy states.
The potential of these two regions of the ZIKV and SePP1 mRNAs to interact was assessed at the DNA
level via a gel shift assay, using synthetic oligos corresponding exactly in size and sequence to the
regions shown in Fig. 1A, with a random oligo used as a control. Each of the three oligos (ZIKV, SePP1
and random) was run in an individual lane, and the three possible pairwise combinations were also run
in three additional lanes (Fig 1B). Single-stranded DNA migrates to the bottom of the gel; only if the
oligos interact to form dsDNA will a higher mass band be observed. Only the combination of ZIKV plus
the native unscrambled SePP1 oligo form a band running at the expected mass for ~50bp dsDNA. The
other pairs, ZIKV plus random control and SePP1 plus random control, still run as ssDNA, indicating a
failure to hybridize, and demonstrating the specificity of the predicted ZIKV-SePP1 antisense interaction
under these experimental conditions.
After these refinements of the extent of the region of strongest antisense interaction between the ZIKV
and SePP1 mRNAs, yielding the interaction shown in Fig. 1A, a 52 base region of ZIKV was used as a
BLAST query against the entire set of human mRNA (this region was identical to the 54 base ZIKV
segment shown in fig 1A, minus the unpaired single bases at the 5’ and 3’ ends). The results of this
BLAST search (included as Table S1 in online Supplemental Materials) show that the antisense match
between this region of ZIKV and SePP1 is the very top hit in the entire database of human mRNA.
Furthermore, most of the other highly ranked hits are not +/- strand antisense matches, but +/+
similarities, indicating possible sequence homology at the nucleic acid or protein level. This result
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
8
strongly suggests that the antisense pairing of these two mRNA regions of ZIKV vs. SePP1 could take
place in vivo, as it would be stronger than any competing RNA:RNA interactions.
As mentioned in the introduction, the fact that Flaviviruses like ZIKV can form active replication
complexes in the nucleus [14] raises the possibility that ZIKV mRNA might be able to interact with
unspliced cellular pre-mRNAs. However, our initial search conducted vs. the human cellular mRNA
database would fail to reveal potential virus-host antisense interactions targeting an intron of a cellular
RNA transcript.
Consideration of this possibility led to the realization that the ZIKV/SePP1 interaction shown in Fig. 1A,
which lies at the very beginning of an exon in SePP1, can be extended (after a short unpaired region) at
the 3’ end of the ZIKV sequence in question, into an upstream intron of SePP1. This expanded antisense
interaction is shown as the RNAHybrid-generated structure in Fig. 2A, where an arrow indicates the
intron/exon boundary in SePP1. This possibility was further validated using the IntaRNA program, which
output an essentially identical RNA-RNA interaction to that shown in Fig. 2A when up to 2000 base long
regions of ZIKV and genomic SePP1 were used as input. This result is shown as Fig. 3A; the full details
and raw IntaRNA output are included as Fig. S2 in the online Supplemental Materials.
When the sequence of the entire ZIKV 3’UTR (corresponding more or less to the expected ZIKV
subgenomic sfRNA sequence) was used as a BLAST query against the set of complete human
selenoprotein genes (i.e., including introns), an antisense match vs. a TR1 intron forming the core of Fig
2B was identified. Following the same procedures detailed above for the ZIKV-SePP1 mRNA match
(RNAHybrid followed by IntaRNA analysis), the result shown in Figs. 2B and 3B were obtained (i.e., the
initial BLAST hit was slightly extended into a larger antisense interaction; the full details and IntaRNA
output are included as Fig. S3 in the online Supplemental Materials). This finding suggests that the ZIKV
3’ subgenomic sfRNA might target unspliced TR1 pre-mRNA in the nucleus, potentially resulting in
knockdown of TR1 mRNA and protein synthesis.
Discussion
Selenium is an essential dietary trace element, having the lowest concentration in the earth’s crust of
any nutrient element, and whose bioavailability in the food chain varies widely by geographical location
and agricultural practices [25]. Perhaps the most dramatic example of its critical importance for living
organisms is evidence that periods of severe selenium depletion in Earth’s oceans correlate closely with
three major mass extinction events between 500 and 200 million years ago [26]. In regard to our results
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
9
involving ZIKV, it is significant that dietary selenium status has been associated with the pathogenesis
and risk of disease progression for a number of viruses, almost all of which are RNA viruses (a point
which we will revisit below). The most extensive evidence is related to HIV-1, for which a negative
correlation between selenium status and disease progression or mortality has been firmly established
(e.g., [27, 28]). Significant clinical benefits of selenium supplementation in HIV/AIDS cohorts have been
demonstrated [29-32]. Other established cases of chemoprotective antiviral effects of dietary selenium
include reduction in incidence of mammary tumors caused by MMTV, a retrovirus [33], the incidence of
liver cancer and hepatitis in China linked to hepatitis B virus [34], and Keshan disease myocarditis linked
to coxsackievirus, which becomes more virulent when combined with selenium deficiency [35]; similar
effects have been reported for influenza virus [36, 37]. An underappreciated example of the
pharmacological use of selenium vs. a viral infection is a reported case in which an Asian viral
hemorrhagic fever was treated with oral sodium selenite (2 mg per day for 9 days), giving an overall 80%
reduction in mortality [38].
The mechanisms underlying all the reported clinical impacts of selenium status on viral disease are
probably complex and multifactorial, but selenium is certainly not an antiviral in the conventional sense
of a potent inhibitor of viral replication. Typically, selenium seems to work by decreasing the harmful
effects of viruses on the host, rather than blocking viral replication, because there are only a few reports
of some direct inhibition of viral replication by selenium and other antioxidants (e.g., [39]), and some of
that effect may be due to indirect mechanisms, such as NF-κB inhibition in the case of HIV-1.
The results reported here suggesting that ZIKV virus (like several other pathogenic RNA viruses [13]),
may engage in antisense interactions with the mRNA or pre-mRNA of isoforms of TR, a ubiquitous
selenoprotein oxidoreductase, may provide a new explanation for the observed interactions between
dietary selenium and the pathogenesis of viruses with RNA genomes. Antisense targeting of host TR
isoforms leading to decreased TR protein levels could be an effective strategy for many RNA viruses.
DNA precursors can only be made from RNA precursors, and the key enzyme involved in DNA synthesis,
ribonucleotide reductase (RR), uses thioredoxin (Trx) as the primary hydrogen donor for the reduction in
mammalian cells [40]. Because they are required to regenerate reduced Trx, TR enzymes are important
for sustaining the conversion of RNA precursors into DNA precursors. Note that, even with near-
complete knockdown of TR1, an adequate basal level of RR activity (e.g. for DNA repair) could still be
maintained in the cell by the glutaredoxin (Grx) system, as Grx can substitute for Trx; however, Trx is the
favored S phase electron donor for RR, and the Grx system only produces about 10% as much
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
10
deoxyribonucleotide product as the Trx system (see Fig. 4 of Avval and Holmgren, 2009 [40]). Thus, viral
antisense inhibition of TR protein synthesis is an ideal strategy for mammalian RNA viruses to slow down
RR and to decrease the rate at which ribonucleotides are diverted for deoxyribonucleotide synthesis,
thereby facilitating viral RNA synthesis.
In addition to our published results showing antisense targeting of TR isoforms by HIV-1 and Ebola [13],
and the current results for ZIKV, we have found that a number of other RNA viruses show potential
antisense anti-TR interactions, including avian influenza and mumps virus; some of these are shown in
Fig. S4 in the online Supplemental Materials.
The potential virus-host “natural antisense” interactions that we have identified for ZIKV and other RNA
viruses could have several possible biochemical outcomes, the most likely being inhibition of protein
synthesis from the target mRNA (either via mRNA degradation, or via inhibition at the ribosomal level
without mRNA degradation [41]). In the case of HIV-1 and Ebola, we have also proposed that, via host
mRNA “antisense tethering interactions”, viruses may capture functional RNA sequence motifs, e.g., to
enable the expression of hidden viral selenoprotein genes, by recoding a conserved 3’-terminal UGA
stop codon to be translated as Sec [13]. Thus, viral RNA antisense interactions with selenoprotein
mRNAs could directly perturb host selenium biochemistry by various mechanisms.
For the current study, the most significant finding for ZIKV is its potential antisense targeting of SePP1,
given the critical role of that selenoprotein in particular in brain function and development (as reviewed
in the Introduction). Our results beg the question, could ZIKV-mediated perturbation of selenium
biochemistry via downregulation of SePP1 and other selenoproteins like TR in the brain contribute to
neuronal loss, cerebral atrophy and microcephaly?
We propose that ZIKV-mediated antisense inhibition of SePP1 and TR in fetal brain could mimic effects
of SePP1 knockout in mice. Combined with low Se status, this could also mimic the genetic disease
PCCA, via decreased brain selenoprotein synthesis. Selenoprotein downregulation could also contribute
to other ZIKV-induced neurological symptoms; in addition to the evidence from SePP1 knockout studies
reviewed in the Introduction, selenoprotein expression has been shown to be required for proper
neuronal development and prevention of seizures and neurodegeneration [42]. Furthermore, SePP1 in
particular has been shown to be a survival factor for embryonic neuronal cells [43].
In regard to the ZIKV-specific mechanism of neuronal pathogenesis, it has been reported that ZIKV
infects neural progenitor cells and induces apoptosis and cell-cycle dysregulation [21]. This could be a
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
11
direct consequence of viral antisense-mediated TR downregulation, because inhibition of TR has been
shown to induce apoptosis in neuronal cell lines [20].
If our proposed selenium-based mechanism for ZIKV pathogenesis proves to be valid, one reason for
optimism is that Schweizer and coworkers found that supplementation of maternal drinking water with
sodium selenite was able to prevent or reduce the neurological and growth deficits in newborn SePP1
knockout mice, presumably by supplying non-protein bound selenium directly to the brain via mother’s
milk, as a small molecule, either as selenite itself, or a metabolite [7]. This is consistent with evidence
that both the absence of selenoprotein P and low dietary selenium have to be present for neuronal
dysfunction to occur [10, 44]. It is important to note that selenite was found to be more effective than
organic forms like selenomethionine in countering neurological defects and weight loss induced by
SePP1 knockout in mice [44].
In the light of this mechanism, it is possible that selenium deficiency in pregnant mothers might be a risk
factor for fetal microcephaly and related neurological birth defects. Although selenium deficiency per se
may not be widespread in Brazil, it has been documented, e.g., in children [45]. However, there is
another related environmental risk factor that has been associated with microcephaly in the past, and
might be exacerbated by ZIKV-induced SePP1 knockdown: mercury toxicity. There is a well-documented
antagonism between mercury and selenium [46, 47], with the latter able to mitigate mercury toxicity via
an SePP1-related mechanism. SePP1 has been shown to decrease mercury toxicity and bioavailability
via chelation of multiple mercury-selenium complexes by a single molecule of SePP1 [48]. ZIKV-induced
SePP1 knockdown could impair this defense, and increase susceptibility to mercury-associated neonatal
neuronal damage and microcephaly. High mercury levels exist in some Amazonian soils, rivers and fish,
and ingestion of high levels of mercury may lead to a functional selenium deficiency, e.g., decreased
selenoprotein expression [47]. This might help to explain why relatively higher rates of ZIKV-associated
microcephaly have been observed in Brazil than in other some countries in the region.
Summary and Conclusions
We have presented evidence that several RNA viruses, including ZIKV (this work), HIV-1 and Ebola [13],
may engage in antisense interactions with mRNAs of TR, an essential antioxidant selenoprotein. We
propose here for the first time that TR inhibition may be a general RNA virus strategy to favor RNA
synthesis over DNA, potentially shedding new light on the role of dietary selenium status in viral
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
12
pathogenesis that has been documented for many human and animal RNA viruses, including HIV-1, HCV
and HBV, coxsackievirus, influenza, and viral hemorrhagic fever.
Most significantly, our results suggest that ZIKV may also target SePP1, a selenium carrier protein, which
could disrupt selenoprotein synthesis in the brain. ZIKV-mediated antisense inhibition of SePP1 and TR in
fetal brain could mimic the effects of SePP1 knockout in mice, contribute to neuronal cell death, and
mimic the genetic disease PCCA, which has symptoms of brain atrophy and microcephaly. SePP1
knockdown could also increase susceptibility to mercury toxicity via a well-established mechanism.
Lastly, when given to nursing mothers, the ability of sodium selenite to counter neurological deficits in
SePP1 knockout mice suggests a similar approach with ZIKV-infected pregnant human mothers,
particularly in the first trimester, might help to reduce the risk of ZIKV-induced fetal abnormalities.
Conflict of Interest
The authors have no conflicts of interest to declare.
Supplemental Material: The figures S1-S4 and Table S1 cited in the text can be obtained by email from
the authors, or downloaded as a merged PDF file from:
http://www.researchgate.net/profile/Ethan_Taylor2/publications
Acknowledgments
This research was supported by a gift from the Dr. Arthur and Bonnie Ennis Foundation, Decatur, IL, to
E.W.T., and by High Point University research startup funds for J.A.R.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
13
References
1. Ben-Zeev B, Hoffman C, Lev D, Watemberg N, Malinger G, Brand N, et al. Progressive
cerebellocerebral atrophy: a new syndrome with microcephaly, mental retardation, and spastic
quadriplegia. J Med Genet. 2003 Aug;40(8):e96.
2. Agamy O, Ben Zeev B, Lev D, Marcus B, Fine D, Su D, et al. Mutations disrupting selenocysteine
formation cause progressive cerebello-cerebral atrophy. Am J Hum Genet. 2010 Oct 8;87(4):538-44.
3. Richardson DR. More roles for selenoprotein P: local selenium storage and recycling protein in the
brain. Biochem J. 2005 Mar 1;386(Pt 2):e5-7.
4. Schweizer U, Brauer AU, Kohrle J, Nitsch R, Savaskan NE. Selenium and brain function: a poorly
recognized liaison. Brain Res Brain Res Rev. 2004 Jul;45(3):164-78.
5. Cardoso BR, Roberts BR, Bush AI, Hare DJ. Selenium, selenoproteins and neurodegenerative
diseases. Metallomics. 2015 Aug;7(8):1213-28.
6. Hill KE, Wu S, Motley AK, Stevenson TD, Winfrey VP, Capecchi MR, et al. Production of selenoprotein
P (Sepp1) by hepatocytes is central to selenium homeostasis. J Biol Chem. 2012 Nov
23;287(48):40414-24.
7. Schweizer U, Michaelis M, Kohrle J, Schomburg L. Efficient selenium transfer from mother to
offspring in selenoprotein-P-deficient mice enables dose-dependent rescue of phenotypes
associated with selenium deficiency. Biochem J. 2004 Feb 15;378(Pt 1):21-6.
8. Schweizer U, Streckfuss F, Pelt P, Carlson BA, Hatfield DL, Kohrle J, et al. Hepatically derived
selenoprotein P is a key factor for kidney but not for brain selenium supply. Biochem J. 2005 Mar
1;386(Pt 2):221-6.
9. Burk RF, Hill KE, Olson GE, Weeber EJ, Motley AK, Winfrey VP, et al. Deletion of apolipoprotein E
receptor-2 in mice lowers brain selenium and causes severe neurological dysfunction and death
when a low-selenium diet is fed. J Neurosci. 2007 Jun 6;27(23):6207-11.
10. Burk RF, Hill KE, Motley AK, Winfrey VP, Kurokawa S, Mitchell SL, et al. Selenoprotein P and
apolipoprotein E receptor-2 interact at the blood-brain barrier and also within the brain to maintain
an essential selenium pool that protects against neurodegeneration. FASEB J. 2014 Aug;28(8):3579-
88.
11. Hoffmann PR, Hoge SC, Li PA, Hoffmann FW, Hashimoto AC, Berry MJ. The selenoproteome exhibits
widely varying, tissue-specific dependence on selenoprotein P for selenium supply. Nucleic Acids
Res. 2007;35(12):3963-73.
12. Wirth EK, Bharathi BS, Hatfield D, Conrad M, Brielmeier M, Schweizer U. Cerebellar hypoplasia in
mice lacking selenoprotein biosynthesis in neurons. Biol Trace Elem Res. 2014 May;158(2):203-10.
13. Taylor EW, Ruzicka JA, Premadasa L, Zhao L. Cellular Selenoprotein mRNA Tethering via Antisense
Interactions with Ebola and HIV-1 mRNAs May Impact Host Selenium Biochemistry. Curr Top Med
Chem. 2016;16(13):1530-5.
14. Uchil PD, Kumar AV, Satchidanandam V. Nuclear localization of flavivirus RNA synthesis in infected
cells. J Virol. 2006 Jun;80(11):5451-64.
15. Pijlman GP, Funk A, Kondratieva N, Leung J, Torres S, van der Aa L, et al. A highly structured,
nuclease-resistant, noncoding RNA produced by flaviviruses is required for pathogenicity. Cell Host
Microbe. 2008 Dec 11;4(6):579-91.
16. Funk A, Truong K, Nagasaki T, Torres S, Floden N, Balmori Melian E, et al. RNA structures required
for production of subgenomic flavivirus RNA. J Virol. 2010 Nov;84(21):11407-17.
17. Bidet K, Garcia-Blanco MA. Flaviviral RNAs: weapons and targets in the war between virus and host.
Biochem J. 2014 Sep 1;462(2):215-30.
18. Clarke BD, Roby JA, Slonchak A, Khromykh AA. Functional non-coding RNAs derived from the
flavivirus 3' untranslated region. Virus Res. 2015 Aug 3;206:53-61.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
14
19. Schnettler E, Sterken MG, Leung JY, Metz SW, Geertsema C, Goldbach RW, et al. Noncoding
flavivirus RNA displays RNA interference suppressor activity in insect and Mammalian cells. J Virol.
2012 Dec;86(24):13486-500.
20. Seyfried J, Wullner U. Inhibition of thioredoxin reductase induces apoptosis in neuronal cell lines:
role of glutathione and the MKK4/JNK pathway. Biochem Biophys Res Commun. 2007 Aug
3;359(3):759-64.
21. Tang H, Hammack C, Ogden SC, Wen Z, Qian X, Li Y, et al. Zika Virus Infects Human Cortical Neural
Progenitors and Attenuates Their Growth. Cell Stem Cell. 2016 May 5;18(5):587-90.
22. Rehmsmeier M, Steffen P, Hochsmann M, Giegerich R. Fast and effective prediction of
microRNA/target duplexes. RNA. 2004 Oct;10(10):1507-17.
23. Busch A, Richter AS, Backofen R. IntaRNA: efficient prediction of bacterial sRNA targets
incorporating target site accessibility and seed regions. Bioinformatics. 2008 Dec 15;24(24):2849-56.
24. Wright PR, Georg J, Mann M, Sorescu DA, Richter AS, Lott S, et al. CopraRNA and IntaRNA: predicting
small RNA targets, networks and interaction domains. Nucleic Acids Res. 2014 Jul;42(Web Server
issue):W119-23.
25. Haug A, Graham RD, Christophersen OA, Lyons GH. How to use the world's scarce selenium
resources efficiently to increase the selenium concentration in food. Microb Ecol Health Dis. 2007
Dec;19(4):209-28.
26. Long JA, Large RR, Lee MSY, Benton MJ, Danyushevsky LV, Chiappe LM, et al. Severe selenium
depletion in the Phanerozoic oceans as factor in three global mass extinction events. Gondwana Res.
2015;In Press.
27. Baum MK, Shor-Posner G, Lai S, Zhang G, Lai H, Fletcher MA, et al. High risk of HIV-related mortality
is associated with selenium deficiency. J Acquir Immune Defic Syndr Hum Retrovirol. 1997 Aug
15;15(5):370-4.
28. Constans J, Pellegrin JL, Sergeant C, Simonoff M, Pellegrin I, Fleury H, et al. Serum selenium predicts
outcome in HIV infection. J Acquir Immune Defic Syndr Hum Retrovirol. 1995 Nov 1;10(3):392.
29. Baum MK, Campa A, Lai S, Sales Martinez S, Tsalaile L, Burns P, et al. Effect of micronutrient
supplementation on disease progression in asymptomatic, antiretroviral-naive, HIV-infected adults
in Botswana: a randomized clinical trial. JAMA. 2013 Nov 27;310(20):2154-63.
30. Hurwitz BE, Klaus JR, Llabre MM, Gonzalez A, Lawrence PJ, Maher KJ, et al. Suppression of human
immunodeficiency virus type 1 viral load with selenium supplementation: a randomized controlled
trial. Arch Intern Med. 2007 Jan 22;167(2):148-54.
31. Jiamton S, Pepin J, Suttent R, Filteau S, Mahakkanukrauh B, Hanshaoworakul W, et al. A randomized
trial of the impact of multiple micronutrient supplementation on mortality among HIV-infected
individuals living in Bangkok. AIDS. 2003 Nov 21;17(17):2461-9.
32. Kamwesiga J, Mutabazi V, Kayumba J, Tayari JC, Uwimbabazi JC, Batanage G, et al. Effect of selenium
supplementation on CD4+ T-cell recovery, viral suppression and morbidity of HIV-infected patients
in Rwanda: a randomized controlled trial. AIDS. 2015 Jun 1;29(9):1045-52.
33. Schrauzer GN, Molenaar T, Kuehn K, Waller D. Effect of simulated American, Bulgarian, and
Japanese human diets and of selenium supplementation on the incidence of virally induced
mammary tumors in female mice. Biol Trace Elem Res. 1989 Apr-May;20(1-2):169-78.
34. Yu SY, Zhu YJ, Li WG. Protective role of selenium against hepatitis B virus and primary liver cancer in
Qidong. Biol Trace Elem Res. 1997 Jan;56(1):117-24.
35. Beck MA, Kolbeck PC, Shi Q, Rohr LH, Morris VC, Levander OA. Increased virulence of a human
enterovirus (coxsackievirus B3) in selenium-deficient mice. J Infect Dis. 1994 Aug;170(2):351-7.
36. Beck MA, Nelson HK, Shi Q, Van Dael P, Schiffrin EJ, Blum S, et al. Selenium deficiency increases the
pathology of an influenza virus infection. FASEB J. 2001 Jun;15(8):1481-3.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
15
37. Yu L, Sun L, Nan Y, Zhu LY. Protection from H1N1 influenza virus infections in mice by
supplementation with selenium: a comparison with selenium-deficient mice. Biol Trace Elem Res.
2011 Jun;141(1-3):254-61.
38. Hou JC. Inhibitory effect of selenite and other antioxidants on complement-mediated tissue injury in
patients with epidemic hemorrhagic fever. Biol Trace Elem Res. 1997 Jan;56(1):125-30.
39. Hori K, Hatfield D, Maldarelli F, Lee BJ, Clouse KA. Selenium supplementation suppresses tumor
necrosis factor alpha-induced human immunodeficiency virus type 1 replication in vitro. AIDS Res
Hum Retroviruses. 1997 Oct 10;13(15):1325-32.
40. Zahedi Avval F, Holmgren A. Molecular mechanisms of thioredoxin and glutaredoxin as hydrogen
donors for Mammalian s phase ribonucleotide reductase. J Biol Chem. 2009 Mar 27;284(13):8233-
40.
41. Zeng Y, Yi R, Cullen BR. MicroRNAs and small interfering RNAs can inhibit mRNA expression by
similar mechanisms. Proc Natl Acad Sci U S A. 2003 Aug 19;100(17):9779-84.
42. Wirth EK, Conrad M, Winterer J, Wozny C, Carlson BA, Roth S, et al. Neuronal selenoprotein
expression is required for interneuron development and prevents seizures and neurodegeneration.
FASEB J. 2010 Mar;24(3):844-52.
43. Yan J, Barrett JN. Purification from bovine serum of a survival-promoting factor for cultured central
neurons and its identification as selenoprotein-P. J Neurosci. 1998 Nov 1;18(21):8682-91.
44. Hill KE, Zhou J, McMahan WJ, Motley AK, Burk RF. Neurological dysfunction occurs in mice with
targeted deletion of the selenoprotein P gene. J Nutr. 2004 Jan;134(1):157-61.
45. Vieira Rocha A, Cardoso BR, Cominetti C, Bueno RB, de Bortoli MC, Farias LA, et al. Selenium status
and hair mercury levels in riverine children from Rondonia, Amazonia. Nutrition. 2014 Nov-
Dec;30(11-12):1318-23.
46. Chen C, Yu H, Zhao J, Li B, Qu L, Liu S, et al. The roles of serum selenium and selenoproteins on
mercury toxicity in environmental and occupational exposure. Environ Health Perspect. 2006
Feb;114(2):297-301.
47. Falnoga I, Tusek-Znidaric M. Selenium-mercury interactions in man and animals. Biol Trace Elem Res.
2007 Dec;119(3):212-20.
48. Suzuki KT, Sasakura C, Yoneda S. Binding sites for the (Hg-Se) complex on selenoprotein P. Biochim
Biophys Acta. 1998 Dec 8;1429(1):102-12.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
16
Figure 1. Predicted antisense interaction between complementary
regions of human selenoprotein P (SePP1) mRNA and Zika virus mRNA.
A: The hybrid dsRNA secondary structure shown and computed interaction energy
(as well as those in Fig. 2) were generated using the RNAHybrid 2.2 program.
When used as a query vs. the entire SePP1 mRNA, the illustrated (green) 54 base
region of Brazilian ZIKV strain SPH2015 (bases 9706-9759) yields the match shown
as the minimum free energy (MFE) antisense interaction, to a (red) 47-base region
of SePP1 (bases 541-587; these sequence ranges include the unpaired single base
overhangs shown at each end).
B: Target-specific in vitro DNA hybridization of the predicted ZIKV-SePP1 antisense
pairing was confirmed at the DNA level by gel shift assay, using DNA
oligonucleotides corresponding exactly to the sequences shown in panel A. The
left three lanes contain only a single (unpaired) ssDNA oligo, as follows: Zika = the
54 base fragment from panel A, SePP1 = the 47 base fragment from panel A,
Random = a 47 base randomly shuffled version of the SePP1 fragment. The right
3 lanes are from incubations of pairwise combinations of those 3 oligos: Zika +
SePP1 (Z+S), Zika + Random (Z+R) and SePP1 + Random (S+R). Of these, only the
Zika + unshuffled SePP1 hybridize to form a slower moving dsDNA band that
migrates as expected for ~50bp dsDNA (size markers not shown; for dsDNA
ladder, see original uncropped gel photo, Fig. S5 in online Supplemental
Materials). The Z+R and S+R combinations still migrate as single stranded DNA,
demonstrating the specificity of the Z+S antisense interaction.
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
17
Figure 2. Predicted antisense interactions between Zika virus mRNA and regions either
in introns or spanning RNA slice sites of human selenoprotein mRNAs.
A: Spanning an intron/exon junction of SePP1: Because Flaviviruses like ZIKV replicate in
the nucleus as well as in the cytosol, ZIKV mRNA could also have the opportunity to
interact with unspliced cellular pre-mRNAs. The ZIKV/SePP1 interaction shown in Fig. 1
can be extended by another ~30 bases at the 3’ end of the ZIKV sequence, into an intron
of SePP1, shown as the RNAHybrid-generated structure A above. The black arrow
indicates the intron/exon boundary in SePP1. The right side of the structure pictured here
is identical to that shown in Fig. 1A, but rotated counterclockwise. Formation of such a
structure might inhibit the splicing and maturation of SePP1 mRNA. An essentially
identical antisense complex is predicted using the IntaRNA progam, differing only in that
the terminal 7 base pairs at the bottom of the structure shown above are not included in
the IntaRNA prediction (see Fig. 3A).
B: Targeting an intron of TR1: Both the RNAHybrid and IntaRNA programs predict the
antisense interaction shown here between a 3’ noncoding region of ZIKV mRNA (near the
beginning of the well documented non-coding “subgenomic Flavivirus RNA” region) and
an intron of TR1. An identical antisense complex is predicted using the IntaRNA progam
(see Fig. 3B).
Zika-mediated antisense inhibition of selenoprotein synthesis and microcephaly, E.W. Taylor and J.A. Ruzicka, 2016
18
9006 9076
| INTRON >< EXON |
SePP1 5'-C A GG U AUAUCAUCUU G G-3'
CUUUUUCUUAG UGU CCGUCU GU GGUUU C CUUUU UCCUUCCUAACUUU CCCAUAU
A |||:|||::|| ||| ||:||: || ||||| | ||:|| |||||||||||||| |||||||
GAAGAAGGGUC ACA GGUAGG CA CCAAA G GAGAA AGGAAGGAUUGAAA GGGUAUA
ZIKV 3'-U A G U ACUC G U CACAC A G-5'
| |
9788 9718
56096 56133
| |
TR1 intron 5'-U A UUU U A-3'
CUG UAC GUUUUCCCCAAGCUGU GGCUGAUUA
B ||| :|| |:|||||||||||||| ||||||:||
GAC GUG CGAAAGGGGUUCGACA CCGACUGAU
ZIKV sfRNA 3'-C U C-5'
| |
10438 10405
Figure 3. IntaRNA-predicted RNA-RNA interactions between Zika virus mRNA and regions of unspliced
selenoprotein pre-mRNAs, confirming matches shown in Figure 2.
A. Predicted interaction between a 3’ region of the ZIKV mRNA (polymerase coding region) and a region
of SePP1 pre-mRNA, spanning an intron/exon boundary in SePP1. This is essentially the same structure
predicted by RNAHybrid shown in Fig. 2A, but slightly truncated at the SePP1 3’ end. The SePP1
intron/exon boundary is indicted by the >< symbol above the sequence. For this analysis, a 500 base
fragment of ZIKV (9,501-10,000) was scanned vs. a 2,000 base fragment (8,001-10,000) of the SePP1
genomic sequence. The computed hybridization energy for the structure shown is -59.6 kcal/mole, with
a net energy of -25.7 kcal/mole, after subtraction of unfolding energies of 10.0 and 23.9 kcal/mole for
the respective SePP1 and ZIKV RNA regions. The underlined region indicates the core of the antisense
pairing of these two RNAs that was first identified by BLAST search as a 21/22 identity plus/minus
sequence match, which is substantially extended by IntaRNA into the intron at left.
B. Predicted interaction between a region of the ZIKV non-coding “subgenomic flavivirus RNA” (sfRNA)
and an intronic region of the human thioredoxin reductase 1 (TR1) pre-mRNA. For this analysis, the ZIKV
~300 base 3’UTR region was scanned vs. a 2000 base fragment (55001-57000) of the TXNRD1 genomic
sequence. The computed hybridization energy for the structure shown is -48.0 kcal/mole, with a net
energy of -24.8 kcal/mole, after subtraction of unfolding energies of 7.6 and 15.6 kcal/mole for the
respective TR1 and ZIKV RNA regions. The underlined region indicates the core of the antisense pairing
of these two RNAs that was first identified by BLAST search as a 22/24 identity plus/minus sequence
match, which is only slightly extended by IntaRNA.
