Zika Virus–Infected Decidual Cells Elicit a Gestational Age–Dependent Innate Immune Response and Exaggerate Trophoblast Zika Permissiveness: Implication for Vertical Transmission

Flaviviruses are enveloped, positive‐stranded RNA viruses responsible for a variety of infections, including Zika virus (ZIKV), dengue fever, yellow fever, Japanese encephalitis, and West Nile fever (1). Studies reported a tendency of ZIKV to follow chikungunya epidemics (2, 3). Accordingly, a west-to-east chikungunya pandemic in 2013 was closely followed by a ZIKV outbreak. The Centers for Disease Control and Prevention recently reported 42,750 symptomatic ZIKV cases in the U.S. and its territories, including 7407 ZIKV-infected pregnant women delivering 283 live infants with ZIKV-associated birth defects and 17 pregnancy losses (4).

In addition to dissemination by Aedes mosquito bite (5, 6), the virus is acquired by sexual or vertical (mother-to-fetus) transmission (7–9). Although most ZIKV-infected adults are asymptomatic, the virus causes minor symptoms such as fever, joint pain, conjunctivitis and rash (10). Although ZIKV is associated with Guillain-Barré syndrome, its most severe CNS sequelae occur in fetuses and infants [e.g., microcephaly, cortical developmental malformations, retinopathy, and limb contractures (11, 12)]. The leading hypothesis for these findings is that ZIKV induces apoptosis in human fetal neuroprogenitor cells, resulting in developmental arrest of the CNS (13–15). Approximately 20% of newborns of affected mothers display such findings, with infections in the first and second trimester more strongly associated with fetal abnormalities (16–18). In humans, ZIKV infections are also linked to miscarriage, stillbirth, and fetal growth restriction (FGR) and are indicated in preterm birth (PTB), resulting from adverse pregnancy outcomes, such as chronic placentitis, abruption, and preeclampsia (11, 19), suggesting that ZIKV likely impairs placental functions. However, the biological mechanism(s) used by ZIKV to gain access to the placenta and the cause of the inverse relationship between the gestational age of infection and its severity remain unclear.

Fetal dissemination of any infectious agent requires transmission through the placenta, which is attached to the immunologically active uterine decidua. The decidua is comprised of decidualized endometrial stromal cells, glandular epithelium, placental-derived extravillous trophoblasts, blood vessels, and maternal immune cells dominated by decidual NK (dNK) cells and macrophages, with smaller percentages of dendritic cells and T lymphocytes (20, 21). Dynamic changes of immune cells occur during pregnancy at the maternal–fetal interface. The numbers of dNK cells reach a maximum during the first trimester but decline near term. Conversely, lower numbers of T cells are present in the first trimester than at term (22). As the dominant lymphocyte population in the decidua, dNK cells play a critical role in antiviral innate immunity as well as placental development by producing several soluble factors (23). In addition to dNK cells, decidual CD8⁺ T cells at the maternal–fetal interface are responsible for balancing the paradoxical requirements for induction of maternal–fetal tolerance and antiviral immunity (24).

Syncytiotrophoblasts (STBs), the outer cell layer of placental floating villi, are in direct contact with maternal blood and thus are assumed to be site of the ZIKV entry. However, STBs are ZIKV nonpermissive (resistant to ZIKV attachment and replication) (25), suggesting that maternal–fetal ZIKV transmission occurs at other placental site(s), including the anchoring villi, the sites of placental attachment to the uterine decidua. The absence of STBs and the presence of modestly more ZIKV-permissive cytotrophoblasts (CTBs) in the anchoring villi suggest that either decidual cells or CTBs or both are the most likely site(s) of ZIKV transmission to the fetus. Prolonged ZIKV survival is reported in human eye, testes, brain and placenta (7, 8, 26–28), all of which have a similar blood–tissue barrier and immune-privileged site, likely enabling them to serve as both sanctuary and reservoir to promote ZIKV persistence. We hypothesized that immunologically active decidual cells act as both a reservoir and source of ZIKV transmission to adjacent anchoring villi at the maternal–fetal interface. Therefore, the current study initially compared expression levels of potential mediators of ZIKV infection in human decidual stromal cell versus trophoblast cultures and then explored the ZIKV infection index in primary cultures of human endometrial stromal cells (HESCs), first trimester decidual cells (FTDCs), and term decidual cells (TDCs) as well as term CTBs.

Nitazoxanide displays potent actions against anaerobic bacteria, protozoa, and viruses (29). The U.S. Food and Drug Administration approved nitazoxanide for treatment of diarrhea and enteritis in adults and in children ≥12 mo (30, 31). When ingested with food, nitazoxanide is absorbed from the gastrointestinal tract and hydrolyzed in the plasma to form its active metabolite, tizoxanide, with serum levels attaining up to 10 μg/ml (27 μM) (32). Recent studies demonstrated that tizoxanide inhibits H1N1 virus replication and blocks influenza A virus by selectively targeting posttranslational viral hemagglutinin maturation (33), and nitazoxanide effectively inhibits the NS2B-NS3 protease activity of ZIKV between 1.1 and 15.9 μM (34). Thus, we further tested the inhibitory effect of tizoxanide against ZIKV infection in primary cultures of HESCs, FTDCs, TDCs, and term CTBs.

The study was approved by the University of South Florida Institutional Review Board and the Institutional Biosafety Committee (Pro00031487 and 1360, respectively) and the Yale School of Medicine Human Investigation Committee (22334). Placental specimens were collected after receiving written informed consent.

Previously isolated and cultured frozen leukocyte-free TDCs (35), FTDCs (21), and HESCs (36) at passage ≤4 were grown to confluence. TDCs were isolated from uncomplicated term pregnancies undergoing cesarean section. Briefly, after removal of the amniotic membrane, the decidua was scraped from the maternal surface of the chorionic membrane and enzymatically digested with a mixture of collagenase type IV and DNase 1 (Worthington Biochemical, Lakewood, NJ). The isolated cells were filtered through 100-, 70-, and 40-μm sieves and purified on a discontinuous Percoll gradient, as previously described (35). Similarly, FTDCs were isolated from placental samples obtained from voluntary termination of uncomplicated first trimester pregnancies by performing enzymatic digestion, consecutive filtration, and Percoll purification (20, 21). After two to three passages, cultured FTDCs and TDCs were analyzed by immunocytochemistry to check purity, found to be leukocyte free (<1%), vimentin positive (>98%), and cytokeratin negative. Trypan blue exclusion identified >90% of the isolated TDCs or FTDCs as viable. HESCs were obtained from premenopausal healthy women undergoing hysterectomy for benign conditions. Endometrial tissues were minced and digested by collagenase B and DNase 1 (Worthington Biochemical) and filtered through 70- and 40-μm sieves to exclude epithelium, as previously described (37). After the first passage, cell purity was determined by immunocytochemistry and found to contain vimentin-positive fibroblast-like cells (>98%) with <2% epithelial cells and <0.2% leukocytes. HESCs, FTDCs, and TDCs, induced or maintained under estradiol plus medroxyprogesterone acetate, displayed characteristic decidualization-related morphological and biochemical changes such as elevated prolactin levels. CTBs were isolated and cultured as previously described (38) from term uncomplicated placental specimens obtained after cesarean deliveries. Briefly, after removal of chorion and decidua, minced villous tissues were subjected to sequential trypsin/DNase 1 (Worthington Biochemical) enzymatic digestion followed by a discontinuous Percoll gradient. The cell suspension was then immunopurified by negative selection using anti-human CD9 and CD45 Abs to eliminate extravillous trophoblasts and immune cells, respectively. The viability of isolated CTBs was >90%. The purity of cultured CTB was assessed by flow cytometry, and a cytokeratin 7–positive/vimentin-negative/CD163-negative phenotype was confirmed in 90% of the cell population (38); or CTBs were assessed by quantitative real-time PCR (qPCR) for cytokeratin-7, a trophoblast cell marker, vimentin, a mesenchymal marker, and CD45, a pan-leukocyte marker. Purified CTBs were plated and maintained in DMEM/F12 supplemented with 10% FCS. (35). Under these conditions, CTBs spontaneously fuse to form multinucleated STBs within 72 h, based on morphological assessment of cell fusion as well as by measurement of human chorionic gonadotropin and syncytin 1 expression (39). HTR8/SV^neo cells (first trimester trophoblastic cell line; CRL3271) were purchased from American Type Culture Collection (Manassas, VA).

The Vero (African green monkey kidney; CCL-81) cell line and ZIKV (African MR766 strain; VR-84) were purchased from American Type Culture Collection. A total of 5 × 10⁵ Vero cells were seeded in T75 flasks (Thermo Fisher Scientific, Waltham, MA) and cultured in phenol-free basal media DMEM/F12 with 10% FBS (Thermo Fisher Scientific) and 1% penicillin/streptomycin (Life Technologies, Carlsbad, CA). Cells were infected with 3 μl of ZIKV MR766 strain diluted in 1 ml of DMEM/F12 media without any supplements for 1 h, shaking every 15 min at 37°C in 5% CO₂. After 1 h, 5 ml of DMEM/F12 with 10% FBS was added, and the cultures were incubated for 5 d for viral propagation, followed by passing the collected conditioned medium through a 40-μm filter, centrifuging the filtrate at 1500 rpm for 5 min, then passing the collected supernatant through a 0.22-μm filter. The collected supernatants were stored at −80°C as viral stocks for subsequent plaque-forming assays (40). Virus stocks underwent no more than two passages.

Vero cells plated in 12-well culture plates were grown to 90% confluence. On the next day, the cells were exposed to ZIKV infection using 10-fold dilution (1 × 10⁻¹–1 × 10⁻⁵) of viral stocks or mock infection with control media for 1 h at 37°C in 5% CO₂. After removal of the viral inoculum, cells were washed with 1× PBS and incubated in 1 ml of medium containing 10% FBS for 4 d to visualize viral plaque formation at 37°C. Thereafter, cells were fixed with 4% paraformaldehyde (PFA; Thermo Fisher Scientific) overnight and stained with 0.1% crystal violet (Sigma-Aldrich, St. Louis, MO). Crystal violet was removed by washing with tap water, and viral titers were calculated by counting PFU per milliliter, with multiplicity of infection (MOI) calculated as previously described (40).

Total RNAs from term primary CTBs or STBs cultures or from primary TDC (n = 3 per each cell type) cultures were isolated by miRNeasy Mini Kit (QIAGEN, Germantown, MD) and processed for array-based global transcriptome analysis using Illumina Human HT-12 v4.0 Gene Expression BeadChip (Illumina, San Diego, CA) at the Keck Biotechnology Resource Laboratory at Yale University (New Haven, CT). Specimens with an RNA integrity number value >8 were used for microarray analysis. Normalization for gene readouts and for each gene across chips was performed as described previously (36). Transcription level changes with ≥1.5-fold with a p value < 0.05 were considered differentially expressed genes (DEG) across cell groups. Further evaluation of microarray results by Ingenuity pathway analysis (IPA) software (QIAGEN) was performed using DEGs for comparison of molecular, functional, and biological networks among the cell groups. The microarray data can be found in National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE146980) under accession number GSE146980.

Aliquots of frozen primary HESCs, FTDCs, and TDCs at passages ≤4 were cultured in phenol-free basal medium (DMEM/F12; Thermo Fisher Scientific) with 10% charcoal-stripped calf serum (Thermo Fisher Scientific) and 1% penicillin/streptomycin complex (Life Technologies). Confluent cells were then cultured with 1 × 10⁻⁸ M estradiol (Sigma-Aldrich) plus 1 × 10⁻⁷ M medroxyprogesterone acetate (Sigma-Aldrich) to induce and/or maintain decidualization. After 7 d, cultures were washed with PBS, then trypsinized and seeded in eight-well chamber slides (3 × 10⁴ cells per well) or six-well culture plates (1 × 10⁵ cells per well). Chamber slides or culture plates of primary HESCs, FTDCs, and TDCs were each infected with either mock or ZIKV in parallel at an MOI of 0.1 or 1 and then shaken every 15 min to enhance viral adsorption at 37°C and 5% CO₂ for 1 h. Thereafter, initial viral inoculum was removed from the plates, cells were incubated with fresh DMEM/F12 media with or without 5 or 20 μM tizoxanide (kindly provided by Romark Laboratories, Tampa, FL) as a powder and dissolved in dimethyl sulfoxide (Sigma-Aldrich) for 96 h. Two hundred microliters of cell supernatant aliquots were removed daily from each well and stored at −80°C. At the end of day 4, eight-well chamber slides were fixed by 4% PFA for 20 min at 4°C and stored at 4°C for immunostaining. Six-well culture plates were washed with ice-cold PBS after collecting conditioned media supernatant (CMS). Both CMS and six-well plates were stored at −80°C for further RNA analysis.

HESC, FTDC, and TDC cultures grown on eight-well chamber slides infected with ZIKV ± tizoxanide, then fixed with 4% PFA for 20 min at 4°C were immunostained as previously described (36). Briefly, after permeabilization of fixed cells with 0.1% Triton X-100 (Thermo Fisher Scientific) in PBS for 10 min; blocking was performed using 1% BSA with 0.2% Tween 20 (Thermo Fisher Scientific) for 1 h. The cells were probed with a mouse monoclonal anti-Flavivirus viral envelope (E) protein Ab (Millipore, Burlington, MA) at 1:1000 dilution for 1 h, then rinsed 3× in PBS with 0.1% Tween 20 for 5 min, followed by incubation with Alexa Fluor–488 conjugated anti-mouse IgG Ab (Life Technologies) and nuclear staining with DAPI (Vector Laboratories, Burlingame, CA). Images were captured using an Axio Observer Z1 Fluorescence Microscope (ZEISS, Oberkochen, Germany) at original magnification ×20. The infection rate was detected by counting E protein immune–positive cells.

Total RNA isolation using a miRNeasy Mini Kit (QIAGEN) was followed by DNase I treatment (QIAGEN) of isolated total RNA to eliminate genomic DNA contamination. Reverse transcription was performed using the RETROscript Kit (Life Technologies) as described previously (20). qPCR was performed using TaqMan Gene Expression Assay Kits to detect expression levels of the viral attachment/entry genes AXL, GAS6, PROS1; transmembrane Ig and mucin (TIM) family members TIM1, TIM3, and TIM4; the antiviral genes IFN-stimulated gene (ISG) 15; IFN-induced with helicase C domain 1 (IFIH1); and retinoic acid–inducible gene I (RIG1); as well as proinflammatory cytokines and chemokines IL-1β, IL-6, IL-8, TNF-α, CCL-2, CCL-5, IFN-γ–induced protein 10 (IP-10, also known as CXCL10), and IFN-inducible T cell α chemoattractant (I-TAC, also known as CXCL11), (Applied Biosystems, Grand Island, NY). All reactions were performed in triplicate. Expression of the target mRNAs was normalized to β-actin levels, and the 2^−ΔΔCT method was used to calculate relative expression levels (37). Viral RNA was extracted from 200 μl of culture supernatant using Invitrogen PureLink Viral RNA/DNA Kit (Invitrogen, Carlsbad, CA). The cDNA was generated by random hexamers-directed reverse transcription following Promega ImProm-II Reverse Transcription System protocol (Promega, Madison, WI). ZIKV RNA copy numbers were quantified by qPCR targeting a conserved region of the E gene. To standardize the assay, serial dilutions of transcribed ZIKV MR766 genomic RNA (NR-50085; BEI Resources, Manassas, VA) were used to plot each standard curve (41). The reaction mixture included 1 μl of cDNA, 10 μl of Ssodvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA), and 0.5 μl of 10 μM forward/reverse primers as follows: forward (5′-TGCCCAACACAAGGTGAAGC-3′; nt 1196–1215) and reverse (5′-CTCTGTCCACTAAYGTTCTTTTGC-3′; nt 1251–1274) in a final volume of 20 μl. Reactions were run using the 7500 Real-Time PCR System (Applied Biosystems) with an initial denaturation step at 95°C for 30 s followed by 40 cycles each that included 15 s at 95°C and 30 s at 60°C. Data acquisition and analysis were carried out using the 7500 Software version 2.0.5.

CTBs isolated from term placenta (n = 5) were cultured on six-well plates for 2 d and were then infected with mock control or with ZIKV at MOI = 1 directly or indirectly with the equivalent MOI = 1 in CMS obtained from ZIKV-infected HESCs, FTDCs, or TDC cultures ±20 μM tizoxanide. In addition, CMS from mock-infected cultures was also used as control. At the end of 48 h, collected CMS and culture plates were stored at −80°C for RNA exaction and subsequent analysis. HTR8/SV^neo cells were also infected with mock control or with ZIKV at MOI = 1 directly or indirectly with the equivalent MOI = 1 in CMS obtained from ZIKV-infected HESCs, FTDCs, or TDC cultures.

Data were analyzed by SigmaStat version 3.0 software (Systat Software, San Jose, CA) using pairwise multiple comparisons by one-way ANOVA followed by the post hoc Student–Newman–Keuls if nonparametrically distributed. Comparison of two groups used a t test or a Mann–Whitney U test for parametric or nonparametric distribution, respectively.

Global microarray evaluation of primary cell monolayers (n = 3 per each) revealed that TDCs retain a pool of 6493 DEGs (4970 upregulated and 1523 downregulated) versus STBs. TDCs also display a pool of 6686 DEGs (3675 upregulated and 3011 downregulated) versus CTBs. Moreover, STBs sustain a pool of 3368 (104 upregulated and 3264 downregulated) DEGs versus CTBs. Among these DEGs, several transcripts are involved in viral attachment/entry, including AXL, a TAM (tyrosine kinases receptor family for TYRO3, AXL, and MERK) receptor that mediates viral attachment and the receptor ligands GAS6 and PROS1, which act as virus–cell bridging factors. Expression of these factors is significantly higher in TDCs versus STBs or CTBs (Table I). Bioinformatics analysis using IPA revealed that virus entry via endocytic pathways is among the top canonical pathways in TDCs versus STBs or CTBs (Supplemental Fig. 1). Specifically, compared with STBs or CTBs, 65 and 61 DEGs, respectively, in TDCs overlap those involved in virus entry via endocytic pathways (Supplemental Table I). Moreover, IPA “Disease and Biologic Functions” evaluation of DEGs in infectious diseases category also identified significantly higher viral infection and replication susceptibility for several different viruses, including RNA viruses in TDCs versus CTBs or STBs (Tables II, III). A similar IPA comparison revealed significantly lower viral infection vulnerability in STBs versus CTBs (Table IV). Taken together, these results indicate that at term, decidual cells exhibit the highest, whereas trophoblasts exhibit the lowest, viral infection susceptibility based on our microarray analysis. This prediction is confirmed by our observation that infected TDCs display higher ZIKV RNA copy numbers than CTBs.

Although ZIKV binds to several surface proteins that likely facilitate entry into cells (42), the precise ZIKV cell receptors and mechanisms that mediate entry remain unknown. Among TAM receptors, AXL mediates ZIKV entry into keratinocytes (43, 44). Thus, we sought to confirm the microarray results by qPCR analysis of AXL, as well as the other TAM receptor viral attachment factors, TYRO3 and MERTK. We also analyzed expression of the viral bridging molecules GAS6 and PROS1 in HESC, FTDC, TDC, CTB, and STB cultures. The qPCR results revealed that, compared with cultured TDCs, HESCs, and FTDCs exhibit significantly higher levels of AXL, GAS6, and PROS1, with negligible expression detected in CTB and STB cultures (Fig. 1). In contrast, TYRO3 and MERTK expression was very weak, with no significant difference evident among these cell types (Fig. 1A). In addition to the TAM family, we also analyzed T cell TIM family members, TIM1, 3, and 4 in these cells. TIM1 and TIM4 expression were weak in all cell types, whereas TIM3 expression was higher in CTBs versus other cell types (Supplemental Fig. 2).

To determine whether higher expression of ZIKV entry molecules in HESCs and FTDCs versus TDCs correlates with their ZIKV infection rates, decidualized primary cultured HESCs, FTDCs, and TDCs were inoculated in parallel with either media without virus (mock infection) or ZIKV at MOI = 1 for 96 h. The ZIKV infection rate was detected by counting cells immunopositive for E protein. Ab specificity was confirmed by positive E immunostaining of ZIKV-permissive Vero cells with no staining evident in mock-infected cells. As determined by E (+) cell numbers (Fig. 2A), the ZIKV infection rate was highest in FTDCs (73.98%), then in HESCs (70.8%), then in TDCs (65.5%, Fig. 2B), indicating that each cell type is highly susceptible to ZIKV infection. However, E protein immunofluorescence intensity was stronger in HESCs and FTDCs versus TDCs, likely reflecting higher ZIKV entry and/or replication in both HESCs and FTDCs (Fig. 2A).

To investigate whether higher E protein immunostaining intensity correlates with ZIKV replication in HESCs or FTDCs versus TDCs (Fig. 2A), qPCR was performed using a ZIKV-specific RNA primer in parallel experiments to quantify viral RNA levels. The presence of intracellular ZIKV RNA expression in HESCs, FTDCs, and TDCs confirms ZIKV replication in each cell type. However, higher intracellular ZIKV RNA copies were observed in ZIKV-infected FTDCs (∼7 × 10⁷ RNA copies) or in HESCs (∼5 × 10⁷) compared with TDCs (∼1.8 × 10⁷ ZIKV copies), corresponding to 3.9-fold or 2.7-fold higher intracellular ZIKV RNA copies in FTDCs and HESCs versus TDCs, respectively (Fig. 2C). Next, the quantity of viral progeny released into the media by each of these ZIKV-infected cell cultures was determined by assessing viral RNA copy numbers in supernatants collected at 96 h postinfection. As determined in the cell lysates, significantly higher ZIKV RNA copy numbers were detected in supernatants derived from ZIKV-infected FTDCs and HESCs versus TDCs (Fig. 2C). Moreover, comparison of ZIKV growth curves in supernatants collected between 24 and 96 h postinfection revealed enhanced viral production in HESCs and FTDCs versus TDCs, which attained statistical significance at 72 and 96 h postinfection (Fig. 2D). These results indicate that HESCs, FTDCs, and TDCs are highly permissive to ZIKV infection, that ZIKV replication and release correlates with the expression of viral attachment/entry molecules in these cells, and that viral replication efficiency is greater in first trimester versus TDCs.

To evaluate whether ZIKV induces a differential innate immune response among HESCs, FTDCs, and TDCs, expression levels of 1) IFN-induced antiviral response genes, ISG15, IFIH1, and RIG1 (Fig. 3A); 2) proinflammatory cytokine genes IL-1β, IL-6, IL-8, and TNF-α (Fig. 3B); and 3) chemokines CXCL10, CXCL11, CCL2, and CCL5 (Fig. 3C) were measured in primary cultures at 96 h post-ZIKV infection by qPCR. At an MOI of 0.1 or 1, ZIKV induced a significant increase in mRNA expression of these genes in all three cell types (Fig. 3). However, HESCs did not exhibit a concentration-dependent increase for any of these genes, whereas FTDCs responded with a concentration-dependent increase in IL-1β, TNF-α, and CXCL11 levels. Strikingly, TDCs displayed a concentration-dependent increase for each of these genes that was much more robust compared with either HESCs or FTDCs (Fig. 3).

We then tested the efficacy of tizoxanide, the active serum metabolite of nitazoxanide, in ZIKV-infected HESC, FTDC, and TDC cultures grown on chamber slides or six-well plates. Following ZIKV infection at MOI = 1 for 1 h, the cultures were treated with vehicle (control) or 5 or 20 μM tizoxanide for 96 h. Thereafter, collected culture supernatants and cells grown in six-well plates were processed for qPCR analysis, whereas cells grown on chamber slides were fixed in 4% PFA and immunostained with anti-ZIKV E protein. The qPCR analyses revealed that 5 μM tizoxanide lowers ZIKV RNA levels by 2.4-, 5.6-, and 2.8-fold in HESCs, FTDCs, and TDCs, respectively, compared with untreated ZIKV-infected cells, with 20 μM tizoxanide further reducing replication by 23-, 50-, and 15-fold in HESCs, FTDCs, and TDCs, respectively (Fig. 4A). Similarly, 5 and 20 μM tizoxanide also significantly lowered ZIKV RNA copy numbers in the culture supernatants (Fig. 4B). Analysis of immunofluorescence-labeled slides for ZIKV E protein further confirmed that 5 and 20 μM tizoxanide reduces both the incidence and intensity of ZIKV cell infections in a concentration-dependent manner (Fig. 4C, 4D).

To investigate whether decidual cells act as a source and/or promotor of ZIKV transmission to adjacent trophoblasts, cultures of primary human CTBs or HTR8/SV^neo, first extravillous trophoblastic cell line, were exposed to mock (control) media or directly infected with ZIKV (MOI = 1) or HESC- or FTDC- or TDC-derived CMS containing ZIKV at MOI = 1. Analysis by qPCR of cell lysates and supernatants obtained from these trophoblast cultures revealed that direct ZIKV infection of trophoblasts at MOI = 1 produced the minimal numbers of significant ZIKV (∼1 × 10² RNA copies) at 48 h (Fig. 5A). Conversely, infection via HESC-, FTDC-, or TDC-derived CMS resulted in significantly higher ZIKV copy numbers in both trophoblast lysates and supernatants at 48 h compared with ZIKV RNA copies in the initial inoculation (∼1.5–2 × 10³ viral RNA copies; Fig. 5B). Moreover, CTBs infected with FTDC-derived CMS exhibited the highest rate of ZIKV replication as well as viral release, whereas CTBs infected with TDC-derived CMS exhibited the lowest ZIKV replication and viral release (Fig. 5B). Tizoxanide significantly reduced ZIKV RNA copy numbers (∼2.5–3-fold) in both cell lysates and supernatants of all ZIKV-infected cultures (Fig. 5A, 5B), with the greatest relative suppression found in CTBs treated with FTDC CMS and the greatest absolute suppression seen in CTBs treated with TDC CMS (Fig. 5B). We also performed similar experiments using the extravillous trophoblast cell line HTR8/SV^Neo, which were infected with ZIKV (MOI = 1) or CMS obtained from TDCs, FTDCs, or HESCs for 48 h. These experiments found that HTR8/SV^neo cells are more susceptible to ZIKV infection (∼6.3 × 10⁴ RNA copies, Fig. 5C) compared with CTBs (∼1 × 10² RNA copies, Fig. 5A). These results confirmed previous findings (45, 46) indicating that term cytotrophoblasts are more resistant to ZIKV infection than first trimester extravillous trophoblasts (Fig. 6). However, infection via HESC-, FTDC-, or TDC-derived CMS resulted in significantly higher ZIKV copy numbers in HTR8/SV^neo cell lysates compared with direct ZIKV infection of HTR8/SV^neo cells at 48 h. Moreover, HTR8/SV^neo cells infected with CMS obtained from ZIKV-infected FTDCs or HESCs exhibited higher ZIKV replication compared with HTR8/SV^neo infected with TDC CMS (Fig. 5C). To explore whether increased ZIKV infection in CTBs treated with HESC, FTDC, or TDC CMS is related to changes in expression levels of TAM and TIM family members, qPCR was performed. The results indicate that CMS obtained from TDCs or FTDCs or HESCs did not significantly induce AXL, PROS1, or GAS6, as well as TIM1, 3, or 4 expression in CMS-treated CTBs versus control CTBs (Supplemental Fig. 3A, 3B).

In the human placenta, proliferating and invasive CTBs and Hofbauer cells in chorionic villi, as well as endothelial and amniotic epithelial cells, are permissive to ZIKV infection (42, 47, 48), whereas STBs lining the intervillous space are nonpermissive to ZIKV (25, 49). These findings raise the question of the source of ZIKV entry into the placenta. Given that CTBs exist in close proximity to both decidual cells and villous STBs in the anchoring villi that attach to the uterine decidua and that the anchoring villi are the only site of direct cell–cell interactions between maternal (decidual) and fetal (placental) cells (46, 49), we postulated that decidual cells may act as sites of initial ZIKV infection and subsequent transmission to the placenta.

The initial phase of ZIKV infection requires viral attachment and entry into the host cells mediated through binding to specific receptors, which triggers endocytosis via clathrin-coated pits (43, 44). Recent studies identified AXL and the bridging factors GAS6 and PROS1 as potential ZIKV entry mediators. AXL-expressing astrocytes, endothelial cells, skin cells, including immune cells, fibroblasts epidermal keratinocytes, glial cells, and sertoli cells (43, 50–53) are highly susceptible to ZIKV infection. Our microarray analysis revealed an extensive set of genes differentially expressed in TDCs compared with term trophoblasts that are involved in viral entry, replication and/or infection (Supplemental Fig. 1, Tables I, II), providing a potential explanation for the higher susceptibility of decidual cells compared with trophoblasts to ZIKV infection. Moreover, to our knowledge, this is first study demonstrating differential expression levels of AXL, GAS6, and PROS1 in HESCs and FTDCs versus TDCs, which is consistent with the clinical observation that first trimester ZIKV infection of the mother is associated with greater rates of fetal infection compared with infections at term. Furthermore, very low TYRO3 and MERTK expression in maternal cells suggests that their contribution to ZIKV infection at the maternal–fetal interface is negligible. Therefore, only AXL among the TAM members is likely the primary ZIKV entry molecule at the maternal–fetal interface. Weak expression of TIM1 and TIM4 in these cells eliminates the contribution of TIM family members in gestational age changes in ZIKV infection in decidual cells, whereas the highest TIM3 expression in CTBs suggests TIM3 as a potential ZIKV entry molecule for CTBs. Several previous studies reported DEGs between villus trophoblasts and extravillous trophoblasts obtained from first trimester placenta (54–56). Among these studies, only Telugu et al. (55) found that PROS1 and MERTK are 0.3- and 0.1-fold lower, respectively, whereas TYRO3 expression is 2.3-fold higher in the HLA-G⁺ (extravillous type) trophoblast versus TACSTD2⁺ trophoblasts. In comparison, our microarray and qPCR analyses did not reveal a significant alteration for the expression levels of these molecules between CTBs versus STBs, which are obtained from term placentas.

Indeed, we found that all decidual cell types are far more susceptible to ZIKV infection and replication than trophoblasts (Figs. 2, 5, 6). Although infection rates were found to be similar among decidualized endometrial stromal cells and both first trimester and TDCs, stronger immunofluorescence intensity for ZIKV E protein levels and higher RNA copy numbers in both intracellular lysates and supernatants were observed in FTDCs and HESCs versus TDCs (Fig. 2). Moreover, the growth kinetics of ZIKV were greater in FTDC and HESC cultures at 72–96 h postinfection compared with TDCs. Taken together, these results are consistent with the inverse relationship between gestational age and both the susceptibility to and the severity of ZIKV infections. Supporting our results, El Costa et al. (57) using ex vivo organ cultures obtained from first trimester placenta confirmed that mainly decidual fibroblasts and Hofbauer cells were infected. They also found that ZIKV replication in decidual tissue explants results in infectious virions able to infect an extravillous trophoblast cell line and concluded that the maternal decidua in early pregnancy serves as a replication platform for ZIKV before virus spread to the placenta through the invasive CTBs (57). Similarly, Tabata et al. (58) showed that ZIKV infection in early pregnancy targets proliferating cytotrophoblasts and Hofbauer cells, amplifying infection in basal decidua and chorionic villi and enabling transplacental transmission.

Both innate and adaptive immune responses are required for the efficient prevention or resolution of viral infections (59). Indeed, in mice, the maternal immune response is sufficient to prevent perinatal infections because only mice lacking a component of type 1 IFN-γ and -β signaling pathway(s) display increased ZIKV infection susceptibility and ZIKV-related complications (60). Several in vitro studies using either cell lines or primary human cells reported that ZIKV infection induces production of IFN type I (α, β), type II (γ), and type III (λ) as well as activation of several IFN-stimulated genes (43, 45, 61). Previous studies have also shown that ZIKV induces cytokines TNF-α, IL-1β, and IL-6 as well as chemokines IL-8, CXCL10, and CCL5 in human astrocytes (62) and in serum of patients infected with ZIKV (63). Previously, we reported expression of several of these cytokines and chemokines in human first trimester dNK cells and both FTDC and TDC cultures (21, 64, 65). The current study also determined that ZIKV infection triggers a gestational age–dependent upregulation of antiviral response genes (ISG15, RIG1, and IFIH1), proinflammatory cytokines (IL-1β, IL-6, IL-8, and TNF-α), and chemokines (CXCL10, CXCL11, CCL2, and CCL5). This augmented antiviral response to ZIKV infection correlated with lower ZIKV RNA copy numbers in TDC lysates and supernatants compared with HESCs or FTDCs. These findings provide yet another mechanism to account for the inverse relationship between gestational age and ZIKV susceptibility. Our findings are also confirmed by recent epidemiological studies showing that ZIKV infection of women during the first trimester resulted in higher severe congenital abnormalities and pregnancy complications than women infected during the second and third trimesters (66, 67). Although the magnitude of cytokine and chemokine response to ZIKV infection is lower in decidual cells from early versus late pregnancy, it is nevertheless substantial and likely has clinical importance. For example, ZIKV-induced CXCL10 and 11 expression in HESC and FTDC cultures may account for the association between ZIKV infections and the later development of both FGR and preeclampsia. Previously, we demonstrated (21) that cotreatment of FTDCs with IFN-γ and TNF-α or IL-1β synergistically upregulates expression of CXCL10 and 11, which normally recruit NK cells and activated T cells by binding to CXCR3. However, excessive increases in both chemokines downregulate CXCR3, which impedes NK cell migration. Preeclampsia, which is associated with chronic inflammation, is accompanied by decreased dNK cell numbers. Because dNK cells are important mediators of spiral artery remodeling, decreased dNK availability may contribute to the reduced uteroplacental blood flow seen in preeclampsia and FGR. Consistent with our hypothesis, CXCL10 levels are significantly elevated in first trimester plasma of patients destined to develop preeclampsia (21). Conversely, the robust proinflammatory response observed in ZIKV-infected TDCs, as manifested by higher expression of IL-1β, IL-6, IL-8, and TNF-α mRNAs, may help account for the association between ZIKV infection and inflammation-associated PTB (11, 19). We previously reported (65, 68) that IL-1β and IL-6 contribute to chorioamnionitis-associated preterm premature rupture of the membranes and PTB by suppressing decidual cell–expressed progesterone receptors (20, 65, 68) and by promoting the intense generation of extracellular matrix–degrading metalloproteases, causing fetal membrane weakening (20).

Nitazoxanide displays potent inhibitory effects against the replication of a broad range of RNA and DNA viruses, including respiratory syncytial virus, coronavirus, parainfluenza, norovirus, rotavirus, Japanese encephalitis virus, dengue, yellow fever, and HIV (29). Nitazoxanide is hydrolyzed in plasma to form its active metabolite, tizoxanide. Given this broad range of antiviral activity, we tested the anti-ZIKV activity of tizoxanide. The current results indicate that tizoxanide has the following effects: 1) reduces ZIKV replication in HESCs, FTDCs and TDCs, 2) reduces both the occurrence and intensity of ZIKV cell infections in a concentration-dependent manner, and 3) also inhibits ZIKV infection in cytotrophoblasts. Taken together, these results provide solid evidence for the potential use of tizoxanide to prevent ZIKV-induced fetopathy. Moreover, Cao et al. (69) demonstrated that pretreatment with either nitazoxanide or tizoxanide failed to inhibit ZIKV replication in the Vero cell line, whereas adding either drug postinfection exerts an anti-ZIKV effect, suggesting that these drugs inhibit infection after viral attachment.

By comparison with endometrial stromal or decidual cells, direct infection of CTBs with ZIKV yields low ZIKV replication as detected in our initial experiments using primary cultures of CTBs. However, ZIKV infection using CMS obtained from endometrial stromal or decidual cell cultures triggers up to 20-fold higher ZIKV replication in CTBs. Moreover, the ZIKV infection rate was higher in HTR8/SV^neo cells than CTBs, as also previously shown (45), and that CMS from ZIKV-infected endometrial stromal or decidual cell cultures further increased ZIKV replication in HTR8/SV^neo, indicating that decidual cells serve as a ZIKV reservoir and facilitate ZIKV infection of extravillous trophoblast and CTBs. However, decidual cell–induced ZIKV infection is unlikely to be mediated by induction of ZIKV-related viral attachment molecules in CTBs because we did not observe alteration of AXL, PROS1, and GAS6 as well as TIM family members expression in CTBs treated with CMS from ZIKV-infected endometrial stromal or decidual cell cultures.

In conclusion, our results revealed that stromal cells isolated from both nonpregnant and pregnant endometrium are highly permissive to ZIKV infection and likely serve as reservoirs for placental/fetal infection. Moreover, trimester-dependent responses of decidual cells to ZIKV help to explain why pregnant women are susceptible to ZIKV infection and why the subsequent effects are more detrimental in the first trimester than in late pregnancy. Last, the efficacy of tizoxanide in inhibiting ZIKV replication and viral release in decidual and cytotrophoblast cultures supports its use in attempting to block perinatal transmission of this virus, thereby averting its harmful effects on the fetus.

We thank Dr. S. Joseph Huang for microarray data normalization and preparation of relevant fold changes between groups.

The authors have no financial conflicts of interest.