Preventive effects of folic acid on Zika virus-associated poor pregnancy outcomes in immunocompromised mice.

Zika virus (ZIKV) infection may lead to congenital microcephaly and pregnancy loss in pregnant women. In the context of pregnancy, folic acid (FA) supplementation may reduce the risk of abnormal pregnancy outcomes. Intriguingly, FA may have a beneficial effect on the adverse pregnancy outcomes associated with ZIKV infection. Here, we show that FA inhibits ZIKV replication in human umbilical vein endothelial cells (HUVECs) and a cell culture model of blood-placental barrier (BPB). The inhibitory effect of FA against ZIKV infection is associated with FRα-AMPK signaling. Furthermore, treatment with FA reduces pathological features in the placenta, number of fetal resorptions, and stillbirths in two mouse models of in utero ZIKV transmission. Mice with FA treatment showed lower viral burden and better prognostic profiles in the placenta including reduced inflammatory response, and enhanced integrity of BPB. Overall, our findings suggest the preventive role of FA supplementation in ZIKV-associated abnormal pregnancy and warrant nutritional surveillance to evaluate maternal FA status in areas with active ZIKV transmission.


Author summary
Zika virus (ZIKV) infection in pregnant women may cause a broad spectrum of abnormal pregnancy outcomes. Folic acid (FA) is an essential water-soluble vitamin B required for placental and fetal development to avoid placental abruption, neural tube defects, intrauterine growth restriction, and pregnancy loss. However, the beneficial role of FA in improving clinical presentations with in utero transmission of ZIKV is unknown. We observe that FA treatment increases the level of maternal serum FA and improves fetoplacental outcomes in two mouse models of in utero transmission of ZIKV. FA blocks ZIKV transmission in a cell culture model of blood-placental barrier by inhibiting viral replication in endothelial cells. This study may serve as a scientific reference for public health practitioners to evaluate the FA status of pregnant women with ZIKV-infection and the correlation with their pregnancy outcomes.

Introduction
Zika virus (ZIKV), a mosquito-borne flavivirus, is a clinically important pathogen. The outbreak of ZIKV in Brazil has created public health of international concern in 2015-2016 [1]. Despite mild clinical symptoms, ZIKV infection in pregnant women may cause a broad spectrum of abnormal pregnancy outcomes including intrauterine growth restriction (IUGR), microcephaly, miscarriage, and stillbirth [1][2][3]. The presence of ZIKV in placenta indicates its potential transmission from mother to embryo [1]. The pathogenesis of ZIKV in the placenta has been suggested to mediate the adverse pregnancy outcomes [4][5][6]. Experimental studies reveal that ZIKV causes placental damage and dysfunction [7,8]. Notably, placental insufficiency or dysfunction, a clinical presentation of poor blood flow in the placenta, may lead to IUGR and pregnancy loss [9]. The placenta is a transient organ that facilitates the maternal-fetal exchange of gases, nutrients, hormones, and waste products through the blood-placental barrier (BPB). In addition, it provides a physical and immunological barrier against pathogen transmission from the mother to embryo [10]. The presence of pathogens in the intervillous maternal blood space may permit in utero viral transmission. However, to establish infection in the placenta and/or fetal organs, pathogens including viruses need to breach the BPB [11]. Thus, maintaining the placental function is critical for normal fetal development and preventing disease transmission.
In the context of pregnancy, folic acid (FA) supplementation is recommended for preventing neural tube defects in infants [12]. Notably, the coverage of FA supplementation among pregnant women in southern Brazil in 2013 was only 54.2% [13]. FA, a water-soluble B-vitamin, plays an important role in amino acid metabolism, cellular homeostasis, and DNA synthesis [14]. FA also plays an important role in maintaining vascular endothelial function by repressing nitric oxide production and homocysteine levels [15]. Moreover, its anti-inflammatory effect prevents bacteria-associated fetal growth restriction and abnormal pregnancy in mice [16]. Although the detail mechanisms remain unclear, FA signal transduction has various implications for feto-placental development and functions [17]. Mice deficient in folate receptor-α (FRα) are defective in early embryonic development [18]. Maternal exposure to FA antagonists increases the risk of placenta-mediated adverse pregnancy outcomes, including placental abruption, fetal growth restriction, and fetal death [19]. Furthermore, a high intake of supplemental FA greatly reduces the risk of spontaneous abortion [20]. Taking into account the role of FA in preventing poor pregnancy outcomes, we tested the potential effect of FA to alleviate the feto-placental pathology associated with ZIKV infection.

Folic Acid (FA) displays antiviral effects against ZIKV infection in human umbilical vein endothelial cells (HUVECs)
The largest blood-placental barrier is in the labyrinth that consists of endothelial and trophoblast placental cells [11]. Therefore, we first evaluated antiviral activities of FA against ZIKV infection in HUVECs and human trophoblast placental JEG-3 cells. Longer duration of pretreatment of FA greatly repressed the expression of ZIKV-NS3 and virus progeny production in HUVECs, while the shorter pretreatment time did not affect ZIKV replication (16 vs  Immunofluorescence microscopy was performed on cells immunostained for ZIKV-NS3 (green) and Hoechst for nuclei (blue) (A). Western blot analysis of the protein level of ZIKV-NS3 (B). Density ratios of respective protein and β-actin are shown in Western blot. Plaque-forming assay (PFA) of viral progeny production in culture supernatants (C). (D-F) Human trophoblast JEG-3 cells were pretreated with FA for 16 hr. Cells were infected with ZIKV in the presence of FA for 24 hr. Immunofluorescence microscopy was performed on cells immunostained for ZIKV-NS3 (green) and Hoechst for nuclei (blue) (D). Western blot analysis of the protein level of ZIKV-NS3 (E). Density ratios of ZIKV-NS3 and β-actin are shown in Western blot. Plaque-forming assay (PFA) of viral progeny production in culture supernatants (F). Data are mean (black bar) and individual values (n = 3 independent experiments). � P<0.05 and �� P<0.01 by Kruskal-Wallis, Bonferroni post-hoc test. https://doi.org/10.1371/journal.ppat.1008521.g001

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy

Antiviral effect of FA against ZIKV infection is associated with folate receptor-α (FRα)-AMPK signal transduction
Folate receptor (FR) and transporter (FOLT) facilitate cellular uptake of FA. Notably, folate receptor-α (FRα) displays a high affinity for binding and transporting physiologic levels of FA into cells than any other FR isoforms and FOLT [21,22]. FRα and FOLT were both expressed in HUVECs, while JEG-3 cells predominantly showed FOLT expression (Fig 2A). Low expression of FRα in JEG-3 cells also has been reported by a previous study [23]. Signal transduction of FA may implicate in adenosine-monophosphate activated protein kinase (AMPK) activation partly via an AMP-LKB1-dependent mechanism [24,25]. We observed that FA treatments increased the level of phosphorylated-AMPK-α (p-AMPKα) in HUVECs, but not in JEG-3 cells (Fig 2B). Notably, ZIKV infection downregulated the expression of p-AMPKα in both HUVECs and JEG-3 cells, suggesting that ZIKV may inhibit AMPK activity (Fig 2C and  2D). FA treatments could greatly rescue the expression of p-AMPKα in ZIKV-infected HUVECS but displayed a minor effect in ZIKV-infected JEG-3 cells (Fig 2C and 2D). Interferon-α (IFN-α) has been shown to selectively sensitize HUVECs to double-stranded RNAinduced apoptosis that may implicate in restricting viral infections [26]. FA treatments in ZIKV-infected HUVECs increased the levels of IFN-α as compared to solvent-treated cells To further demonstrate the role of AMPK and FRα in mediating the anti-ZIKV property of FA in HUVECs, we transiently depleted the protein expressions by transfecting small hairpin RNA-targeting AMPKα (shAMPKα) or FRα (shFRα). Transfection with shRNAs decreased the protein expression of AMPKα or FRα in HUVECs (S2C and S2D Fig). Treatment with 6.25 μM of FA greatly reduced the level of ZIKV-NS3 and viral progeny production in control shLacZ-HUVECs (Fig 2E-2G). Notably, FA displayed a minor antiviral effect against ZIKV infection in shAMPKα-and shFRα-HUVECs (Fig 2E-2G). Thus, the inhibitory effect of FA against ZIKV infection was associated with FRα-AMPKα signaling.

FA inhibits ZIKV-induced endothelial damage and reduces viral transmission in a cell culture model of blood-placental barrier (BPB)
To establish infection in the placenta and/or fetal organs, pathogens including viruses need to impair the BPB [11]. Vascular endothelial (VE)-cadherin, an adhesion molecule, plays a critical role in the formation of BPB [27]. Notably, reactive oxygen species (ROS) induces loss of surface expression of VE-cadherin. This, in turn, may weaken cell-cell junctions and damage endothelial permeability [28]. ZIKV infection significantly increased the level of intracellular ROS in HUVECs that could be reduced by the treatments of FA ( Fig 3A). Moreover, FA rescued the cell surface expression of VE-cadherin in ZIKV-infected HUVECs (Fig 3B). FA also may inhibit ZIKV-associated endothelial hyperpermeability because treatments with FA greatly reduced the passage of macromolecule in solute flux assay ( Fig 3C). In addition, treatment with antioxidant ebselen (EBS) that has been shown to reduce the levels of nitric oxide in sperms of ZIKV-infected mice [29] greatly repressed the level of ROS, recovered the expression of VE-cadherin, and improved cellular permeability in ZIKV-infected HUVECs (Fig 3A-3C). The disruption of cell-cell junction and endothelial permeability in HUVECs may not be limited to ZIKV. In agreement with the previous study [30]

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy Next, we evaluated whether FA may prevent viral transmission in a cell culture model of BPB (Fig 3D). HUVECs were pre-treated with 3.125 μM of FA for 16 hr in the presence of anti-folate receptor-α and transporter (anti-FRα + anti-FOLT) antibodies or control IgG.

FA alleviates feto-placental outcomes of ZIKV-infected pregnant AGB6 mice
We further tested whether FA could alleviate ZIKV-associated abnormal pregnancy outcomes in vivo. Immunocompromised pregnant mice have been used to model in utero transmission of ZIKV that show several pregnancy pathologies including placental damage, fetal resorption, and early postnatal death [8,31]. To evaluate the beneficial effect of FA on ZIKV-associated IUGR and fetal demise, we used pregnant interferon-α/β and -γ receptor-deficient AGB6 mice. Pregnant mice were treated with standard-dose FA (FA-S, 0.164 mg FA/kg body weight (bw)/mouse/oral/day), high-dose FA (FA-H, 0.328 mg FA/kg bw/mouse/oral/day) or phosphate-buffered saline (PBS) as solvent control at embryonic day 6.5-14.5 (E6.5-14.5) as outlined in Fig 4A. To get a sufficient number of developing embryos and fetal demise for further analysis, pregnant mice were subcutaneously infected in the footpad with ZIKV at E10.5 [32]. Despite ZIKV-infected pregnant mice did not show any observable symptoms with undetectable ZIKV-NS3 expression in the brain (S4A Fig), the levels of serum FA were significantly lower than mock-infected pregnant mice ( Fig 4B). Treatments with FA greatly increased the serum FA levels of ZIKV-infected pregnant mice ( Fig 4B).
ZIKV infection in AGB6 mouse model caused abnormal pregnancies including uterine deformity and fetal demise ( Fig 4C and S4B Fig). Fetal resorption was observed in more than 60% of the embryos of ZIKV-infected pregnant mice ( Fig 4D). Remarkably, FA treatments greatly alleviated uterine abnormality and reduced the number of resorbed embryos in ZIKVinfected pregnant mice (Fig 4C and 4D). In addition, ZIKV-infected pregnant mice showed abnormal morphology of developing embryos and placentae at E15.5, including fetal growth restriction, abnormal placenta position and shape, edema, and hemorrhage ( Fig 4E). Developing embryos of ZIKV-infected pregnant mice were significantly smaller than mock-infected embryos ( Fig 4F). FA treatments in ZIKV-infected pregnant mice greatly improved feto-placental morphological features including normal placental position, a lower degree of placental hemorrhage, and larger size of developing embryos (Fig 4E and 4F). ZIKV infection affected overall fetal brain development, including small forebrain, absence of neopallial cortex, and abnormal fourth ventricle ( Fig 4G). The levels of pro-inflammatory cytokines including interleukin 1β (IL-1β) and monocyte chemoattractant protein 1 (MCP-1) were upregulated in the ZIKV in the presence of FA for 24 hr. Immunofluorescence microscopy was performed on cells immunostained for ZIKV-NS3 (green) and Hoechst for nuclei (blue) (E). Western blot analysis of the protein level of ZIKV-NS3 (F). Triangle indicates time of exposure. Density ratios of ZIKV-NS3 and β-actin are shown in Western blot. PFA of viral progeny production in culture supernatants (G). Data are mean (black bar) and individual values (n = 3 independent experiments). �� P<0.01 by Mann-Whitney test. ns: not significant. https://doi.org/10.1371/journal.ppat.1008521.g002

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy ZIKV-infected fetal brains (S4C and S4D Fig). The embryos of ZIKV-infected pregnant mice receiving FA treatments showed improved fetal brain histology with a lower degree of inflammation ( Fig 4G, S4C and S4D Fig). Moreover, ZIKV infection caused necrotic lesions in the maternal placental decidua and junctional zone, with a massive hemorrhage in the labyrinth ( Fig 4H). ZIKV-infected pregnant mice receiving FA showed minor placental lesions in the junctional zone ( Fig 4H).

FA reduces the number of ZIKV-associated neonatal loss and improves the nutritional status of surviving pups
To study the effect of FA on the number of neonatal death and postnatal nutritional status, we performed in utero transmission of ZIKV by the use of anti-interferon-α/β receptor 1 (IFNAR1) antibody-treated wild-type (WT) C57BL/6 pregnant mice ( Fig 5A). Pregnant mice were treated with FA-S, FA-H or PBS at E2.5-18.5. IFNAR1 antibody was delivered to pregnant mice at E5.5 and E6.5, one day before and right after ZIKV infection at early placentation [33]. In this mouse model, ZIKV infection at E6 yields a higher viral load in the placenta than viral inoculation at E9 or E12 [34]. In this C57BL/6 mouse model, ZIKV infection caused a number of stillbirths and the low body weight of surviving pups [31]. In addition to postnatal observation, placentae were isolated at E13.5 for further analysis. First, we also evaluated the levels of maternal FA in ZIKV-infected pregnant mice. Consistent with AGB6 mouse model, ZIKV-infected C57BL/6 pregnant mice showed lower serum FA levels than mock pregnant mice at E18.5; FA treatments significantly increased the serum levels of FA in ZIKV-infected pregnant mice (Fig 5B). ZIKV infection has been shown to reduce the area of maternal-fetal exchange, the placental labyrinth [8]. FA treatments significantly alleviated the histopathology of ZIKV in terms of the size of the placental labyrinth ( Fig 5C and 5D, S4E Fig). We observed 48% death of pups within 12 hr after term delivery in ZIKV-infected mice, whereas FA-treated mice had less than 10% stillbirths (Fig 5E, S4F  and S4G Fig). Notably, ZIKV protein was detected in the brain of stillbirth pups (S4H Fig). To evaluate the nutritional status at a later age, we prospectively monitored the body weight of surviving pups in the weaning age up to 7-week-old. Although the surviving pups of ZIKV-infected dams gained weight throughout the monitoring period, their body weight was significantly lower than that of mock-infected mice (Fig 5F). Importantly, the body weight was higher for surviving pups of ZIKV-infected dams receiving FA than solvent treatment (Fig 5F). Taken together, these data suggest that FA alleviates pregnancy abnormality of ZIKV-infected pregnant mice.

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy

FA limits ZIKV replication in the placenta and improves the prognostic profile of placental dysfunction
Since FA alleviated ZIKV-induced pathological features in the placenta, we expected a lower viral burden in the placenta of FA-treated pregnant mice. Indeed, FA treatments greatly reduced the expression of ZIKV-NS3 protein in the placenta of infected pregnant mice (Fig 6A  and 6B), with no significant effect on maternal ZIKV viremia level (S5A and S5B Fig). Notably, FA treatments had no effect on the median survival time of ZIKV-infected non-pregnant mice (S5C and S5D Fig). These data suggest that FA may display tissue-specific effects. FRα abundantly expressed in the placenta and developing embryo; however, the expression is minimal in other tissues or organs [22,35]. It is well-known that viral infection triggers inflammatory responses. The elevated proinflammatory response is attributed to poor placental function and pregnancy outcomes [36]. Leukocyte infiltration and elevated levels of pro-inflammatory cytokines impair vascular endothelial growth in a model of placental insufficiency [36]. Notably, FA reduces the levels of proinflammatory cytokines that are associated with BPB dysfunction, including IL-1β, MCP-1, and TNF-α [37][38][39]. We observed that ZIKV-infected pregnant mice showed a systemic inflammation, as indicated by a significant increase of serum C-reactive protein (CRP) levels (S7A and S7B Fig). FA treatments could significantly reduce the serum levels of CRP in ZIKVinfected pregnant mice (S7A and S7B Fig). ZIKV infection also caused infiltration of CD45 and IL-1β-positive cells in the placental labyrinth, which could be inhibited by FA treatments (Fig 6C and 6D). Moreover, FA repressed ZIKV-induced inflammatory response in the placenta, including MCP-1, TNF-α, IL-6, and IL-10 levels (Fig 6E and 6F, S7C and S7D Fig).
ZIKV infection downregulated the placental expression of VE-cadherin (Fig 7A-7D, S7C and S7D Fig). Notably, FA treatments rescued the placental expression of VE-cadherin ( Fig  7A-7D, S7C and S7D Fig). Because increased levels of inflammatory cytokines and poor blood flow are associated with utero-placental hypoxia [40], we further evaluated the expression of hypoxia-inducible factor-1α (HIF-1α) in the placenta. ZIKV infection upregulated HIF-1α expression, an adverse response that could be alleviated by FA treatments (Fig 7C and 7D, S7C  and S7D Fig). HIF-1 has a role in hypoxia inducing apoptosis [41]. Consistently, the active form of caspase-3 in ZIKV-infected placentae was greatly increased and could be suppressed by FA treatments (Fig 7C and 7D, S7C and S7D Fig). Overall, our data suggest that FA treatment in ZIKV-infected pregnant mice could reduce viral burden in the placenta and improve the prognostic profile of placental insufficiency.

Discussion
The role of FA in viral infection is still limited. Our study reveals the beneficial effect of FA in preventing in utero ZIKV transmission and improving feto-placental disease prognosis. Direct

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy infection of BPB and/or pathogen-induced pro-inflammatory response may weaken the integrity of the maternal-fetal interface to permit in utero viral transmission [42]. A recent study shows that activation of AMPK inhibits ZIKV replication in HUVECs by inducing antiviral response, inhibiting the inflammatory response and/or viral-induced glycolysis that could be cell-or tissue-specific [43,44]. In this study, we report that FA recovered AMPKα activity, increased the level of interferon-α, and inhibited viral replication in ZIKV-infected HUVECs. Interestingly, FA remained potent in reducing viral load and alleviating pathological features in the placenta of interferon-α/β and -γ receptor-knockout AGB6 mice, suggesting interferonindependent modes of therapeutic action of FA in animals. Studies have documented the role of FA in suppressing endothelial oxidative stress and placental inflammatory response, which in turn alleviates the disease pathology [15,16]. Moreover, AMPK activation is not only necessary for the placental differentiation and growth but also regulating the inflammatory response in the placenta [45]. Intriguingly, both antiviral and anti-inflammatory effects of FA may play a role in limiting ZIKV transmission through maternal-fetal interface. Further studies are required to elucidate the role of FA in viral-induced glycolysis.
FRα is expressed abundantly and consistently in the syncytiotrophoblast of BPB during pregnancy [22]. Despite structural differences between human and mouse BPB, they share similar functions, proteomic profile, and physiological conditions [33]. The largest BPB is in the labyrinth that consists of syncytiotrophoblast-cytotrophoblast in humans and two syncytiotrophoblast layers-cytotrophoblast in rodents, respectively. The second maternal-fetal interface is extravillous trophoblast in humans or trophoblast giant cells in rodents [11,33]. The syncytiotrophoblast is a unique trophoblast that has more than one nucleus (multinucleate) and forms a continuous syncytial layer as a result of cell-cell fusion. In humans, the syncytiotrophoblast forms the outer surface of the villi and makes direct contact with maternal blood. The syncytiotrophoblast also plays a role in waste product removal and synthesis of pregnancy hormones [33].
The underlying mechanism of ZIKV-induced low level of maternal FA in our study is unclear. Intriguingly, ZIKV infection may reduce the dietary intake of FA due to loss of appetite and/or adversely affect FA transport and pool in the liver. Recent observations in humans indicate that ZIKV infection causes abnormal liver function and severe liver injury [46,47]. A case-control study in Brazil suggested that maternal use of FA might not be associated with ZIKV-related microcephaly [48]. However, this study did not evaluate the serum level of maternal FA that could be confounded by ZIKV infection and/or smoking during pregnancy. Pregnant women who smoked had significantly lower serum maternal FA levels than nonsmokers, although their FA intake was relatively the same [49]. Notably, the risk of ZIKV-associated microcephaly is 3-fold-increased in smoking pregnant women [48]. These studies may argue the use of a higher dose of FA supplement for smoking pregnant women with ZIKV infection. Moreover, a cross-sectional study in Southern Brazil in 2013 reported the use of FA during pregnancy was considerably poor (54.2%) [13], suggesting that the FA supplementation program needs to be revitalized.
As calculated by a published method [50], the equivalent dose of 0.164 mg FA/kg/day (FA-S) for mice is 0.0133 mg/kg/day for humans; thus, for a 60-kg person, the dose would be E5.5. Pregnant mice, at E6.5, were infected with ZIKV and subsequently treated with purified anti-mouse IFNAR1 antibody (0.5 mg/kg bw/mouse/ intraperitoneal). Placentae were isolated at E13.5. Daily monitoring was performed to observe the term delivery and the death of the newborn within 12 hours (stillbirth).

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy about 800 μg/day. The US Preventive Services Task Force recommends that women of childbearing age take 400-800 μg FA/day to prevent neural tube defects in infants [12]. A high dose of FA up to 5 mg/day has been proven safe in healthy women [51]; however, the beneficial effect of a high FA dose remains controversial. In preeclamptic pregnant women, daily supplementation with 1 and 5 mg FA does not significantly differ in affecting disease outcomes [52]. Importantly, in the presence of vitamin B12 deficiency, � 5 mg of FA intake may lead to adverse clinical presentations including anemia and cognitive impairment [53].
Overall, our findings reveal the beneficial effect of FA supplementation to alleviate the adverse pregnancy outcomes associated with ZIKV infection in immunocompromised AGB6 and WT C57BL/6 mouse models. Nutritional surveillance to evaluate maternal FA status in areas with active ZIKV transmission is required to revitalize FA supplementation program among women at childbearing for preventing ZIKV-associated abnormal pregnancy.

Ethics statement
The mouse experiments were conducted according to the guideline outlined by Council of Agriculture Executive Yuan, Republic of China. This animal protocol was approved by the Academia Sinica Institutional Animal Care and Use Committee (Protocol no. 16-06-966) and were performed in accordance with the guidelines. Infection was performed in mice under isoflurane anesthesia and all efforts were made to minimize animal suffering.

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy (Millipore). Cells were treated with solvent or FA at the indicated doses for 16 hr. Cells were infected with ZIKV (MOI = 1) for 24 hr in the absence or presence of FA. The apical chamber was replenished with culture medium containing FITC-CM-Dextran (70-kDA, Sigma, 53471) and the lower chamber was loaded HBSS with phenol red. Cells were incubated for 40 minutes. Fluorescence intensity of medium in the lower chamber was measured with a fluorescence microplate reader at 480/530 nm (excitation/emission). Quantitative assay for interferon-α in the cell culture media was performed according to the manufacturer's protocol (VeriKine Human IFN alpha ELISA kit, PBL Assay Science, 41100). Briefly, 100 μl of cell culture medium was used for a sandwich immunoassay. The absorbance of samples was determined by use of an ELISA reader (Molecular Devices) at 450 nm. The concentration of the sample was extrapolated from the standard curve.
For the cell culture model of the blood-placental barrier, HUVECS were grown in 1% gelatin-coated 24-well hanging inserts (Millipore). Cells were treated with 3.125 μm of FA for 16 hr in the presence or absence of 2 μg of antibodies against folate receptor-α (FRα) and folate transporter (FOLT) or control IgG. Cells were adsorbed with ZIKV (MOI = 1) for 2 hr and cocultured with JEG-3 cells for 24 hr. JEG-3 cells and culture supernatants were used for immunofluorescence, western blot, and plaque-forming assays.
The loss-of-function study was performed in HUVECs by transfecting pLKO.1 vector plasmid which carries a short hairpin RNA (shRNA) targeting the human FRα (TRCN0000060347, TRCN0000372330), human AMPKα (TRCN0000000859, TRCN0000000861), or LacZ (TRCN0000072223). pLKO.1-shRNAs were obtained from the Taiwan National RNAi Core Facility. Transfection was performed as described [56] with modification. Briefly, 0.8 μg of plasmid diluted in Opti-MEM (Gibco) was transfected into HUVECs by use of Lipofectamine 2000 (Invitrogen). Cells were incubated for 4 hr. Subsequently, the complete culture medium was added and the cells were incubated for 48 hr. Knockdown efficiency was checked at 48 hr after transfection. Transfected cells were used to evaluate the antiviral effect of FA.

Mouse study
The minimal sample size of each group was calculated by the use of a published method [58]. Mice under isoflurane anesthesia were infected and all efforts were made to minimize animal suffering. Seven-to 8-week-old male and female AGB6 mice with interferon-α/β and -γ receptor-knockout (National Health Research Institutes, Taiwan) or wild-type C57BL/6JNarl (WT-C57BL/6, National Laboratory Animal Center, Taiwan) mice were mated overnight. The day 0.5 of embryonic age (E0.5) was defined as the first observation of a vaginal plug. To study the effect of FA (Sigma, F7876) on feto-placental outcomes, each pregnant mouse was orally given 0.164 mg FA/kg body weight/day (standard dose of FA [FA-S]), 0.328 mg FA/kg body weight/day (high dose of FA [FA-H]), or solvent control (phosphate-buffered saline [PBS]) by the use of feeding needle (stainless steel, olive tip). FA treatments were given to pregnant AGB6 and WT-C57BL/6 mice at E6.5-14.5 and E2.5-18.5, respectively. Pregnant AGB6 mice were subcutaneously infected in the footpad with 1×10 2 plaque-forming units (PFU) of ZIKV per mouse. The culture supernatant of C6/36 mosquito cells was used for mock infection. Maternal blood sera and uteri of pregnant AGB6 mice were collected at E15.5. Pregnant WT mice were treated with purified anti-mouse IFNAR1 antibody (0.5 mg/kg body weight/mouse/ intraperitoneal) at E5.5. Pregnant mice were intravenously injected in the tail vein with 1×10 6

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy PFU of ZIKV per mouse, then treated with anti-mouse IFNAR1 antibody (0.5 mg/kg body weight/mouse/intraperitoneal) at E6.5. Placentae and maternal sera were collected at E13.5 and E18.5, respectively. Mice were monitored daily to observe the term delivery and death of newborns within 12 hr (stillbirth). To evaluate the antiviral effect of FA against ZIKV infection in vivo, groups of 7-to 8-week-old non-pregnant AGB6 mice were subcutaneously infected in the footpad with 1×10 2 PFU of ZIKV per mouse. Immediately after infection, mice were treated with FA-S, FA-H, or PBS orally. Thereafter, mice received the same treatments up to day 10 after infection. The mice were checked daily for mortality. Serum was collected on day 3 after infection by phlebotomy from a facial vein, then the viral load was determined by plaque-forming assay.

Statistical analysis
Data were compared by Kruskal-Wallis Bonferroni post-hoc or Mann-Whitney test. The statistical tests were two-tailed and significance was set at P < 0.05. For immunoblotting, the band density was quantified by using ImageJ (US National Institutes of Health) Supporting information S1 Fig. Antiviral effect of FA in vitro. (A-C) HUVECs were pretreated with FA for 2 hr. Cells were infected with ZIKV in the presence or absence of FA for 24 hr. Immunofluorescence microscopy was performed on cells immunostained for ZIKV-NS3 (green) and Hoechst for nuclei (blue) (A). Western blot analysis of the protein level of ZIKV-NS3 (B). Plaque-forming assay (PFA) of viral progeny production in culture supernatants (C). (D-F) JEG-3 cells were pretreated with FA for 2 hr. Cells were infected with ZIKV in the presence or absence of FA for 24 hr. Immunofluorescence microscopy was performed on cells immunostained for ZIKV-NS3 (green) and Hoechst for nuclei (blue) (D). Western blot analysis of the protein level of ZIKV-NS3 (

PLOS PATHOGENS
Folic acid prevents ZIKV-associated abnormal pregnancy (H) Western blot analysis of ZIKV-E in pooled brain lysates of stillbirth pups of C57BL/6 mouse, β-actin for loading control.