Mouse Models of Zika Virus Transplacental Transmission

Abstract

Seven years after the onset of the Zika virus (ZIKV) epidemic in the Americas, longitudinal studies are beginning to demonstrate that children infected in utero and born without severe birth defects exhibit motor skill deficits at up to 3 years of age. Long term health and socioeconomic impacts of fetal ZIKV infection appear imminent. ZIKV continues to circulate in low levels much as the virus did for decades prior to the 2015 epidemic, and the timing of the ZIKV outbreak is unknown. Thus, in the continued absence of ZIKV vaccines or antivirals, small animal models of ZIKV transplacental transmission have never been more necessary to test antiviral strategies for both mother and fetuses, and to elucidate mechanisms of immunity at the maternal-fetal interface. Here we review the state of ZIKV transplacental transmission models, highlight key unanswered questions, and set goals for the next generation of mouse models.

1. Introduction

Zika virus (ZIKV) is a member of the Flaviviridae family and was first isolated from a sentinel rhesus macaque in Uganda in 1947 (Dick, 1952). ZIKV has a single-stranded positive-sense RNA genome encoding 3 structural proteins (envelope, membrane, capsid) and 7 non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, NS5) (Shankar et al., 2017). Other arboviral flaviviruses of high epidemiological relevance include dengue virus (DENV), West Nile virus (WNV), yellow fever virus, and Japanese encephalitis virus. The first known route of ZIKV transmission to humans was via Aedes aegypti mosquitoes; however, increasing evidence suggests that sexual and vertical transmission are important contributors to the spread of ZIKV (Cauchemez et al., 2016; de Oliveira et al., 2016; D’Ortenzio et al., 2016a; Medina et al., 2019; Moreira et al., 2017; Russell et al., 2017; Vermillion et al., 2017; Wu et al., 2016). Human cases of ZIKV infection were sporadic before the first reported outbreak in the Yap Islands of Micronesia in 2007. Since then, multiple outbreaks have occurred, including one in French Polynesia in 2013 and the first emergence in the Americas in 2015, which resulted in the Brazilian epidemic (Musso and Gubler, 2016). ZIKV infection is typically asymptomatic or causes mild illness that can include fever, muscle aches, headache, conjunctivitis, and rash. However, the outbreaks in French Polynesia and Brazil revealed the ability of ZIKV to cause severe complications, including Guillain–Barré syndrome in adults (Cao-Lormeau et al., 2016) and, in the case of women infected during pregnancy, congenital Zika syndrome manifesting as microcephaly and neurological defects in neonates (Bhatnagar et al., 2017; de Oliveira et al., 2016; Johansson and Hills, 2016; Satterfield-Nash et al., 2017). In a mixed cohort of case-patients from the United States, Brazil, and Colombia, 7 out of 8 microcephalic infants harbored ZIKV RNA in the cerebral cortex, and 9 of 12 women who miscarried, all of whom had been infected during the first trimester, had detectable ZIKV RNA in Hofbauer cells of the fetal chorionic villi (Bhatnagar et al., 2017). A Brazilian cohort study found that 19 children born with microcephaly had evidence of congenital ZIKV infection, as defined by the presence of serum ZIKV-specific IgM in sera and/or anti-ZIKV neutralizing antibodies (Abs) at 1 to 7 months of age (Satterfield-Nash et al., 2017). These observations suggest that infection of not only the mother but also the fetus is a key component of ZIKV-induced fetal pathology. Recent studies have also shown that motor deficits, cognitive defects, and learning retardation may result from in utero ZIKV infection, even if the newborns were asymptomatic (Lopes Moreira et al., 2018; Nielsen-Saines et al., 2019; Satterfield-Nash et al., 2017; Wheeler, 2018). In one case report, a congenitally infected male with microcephaly did not show evidence of neurological abnormalities until 67 days after birth (Oliveira et al., 2016).

Vertical transmission of ZIKV can occur through multiple routes; in addition to intrapartum and postpartum transmission via contact with blood or the mucous membranes during parturition and breastfeeding (Dupont-Rouzeyrol et al., 2016), transplacental infection in utero is also possible (Langerak et al., 2022; Mysorekar, n.d.; Vermillion et al., 2017). Although ZIKV-induced fetal damage can occur throughout the first, second, and third trimesters in human pregnancies (referred to as early, mid-, and late-gestation in mice), mathematical modelling and both mouse and human studies support an association between more severe fetal outcomes (i.e., microcephaly and intrauterine growth restriction [IUGR]) and ZIKV infection during the first trimester (before week 14) in humans and during early/mid-gestation in mice (Cauchemez et al., 2016; de Oliveira et al., 2016; Jagger et al., 2017a; Vermillion et al., 2017; Yockey et al., 2016).

The risk of another ZIKV epidemic increases with the rise in global temperatures and the gradual migration of mosquitoes into temperate regions(Whitehorn and Yacoub, 2019). There is currently no cure or vaccine for ZIKV infection. A safe and effective vaccine that prevents infection of all age groups, including fetuses, is urgently needed to avoid the potentially devastating consequences of infection. Understanding the mechanisms of ZIKV vertical transmission and how fetal damage can be averted are critical to the development of a ZIKV vaccine, and the immense contribution of animal models to this effort cannot be overstated. Although several species, including non-human primates (NHPs) and chicken embryos, have been used for this purpose, mice remain the animal model of choice due to their cost-effective husbandry, ease of handling, large litter sizes, short gestation period, and ease of genetic manipulation. The extensive body of accrued knowledge on mouse physiology and the availability of experimental reagents for mice also facilitate the development of models to study transplacental transmission of ZIKV, as outlined in this review (Figure 1).

Figure 1.

Strategies for modelling of ZIKV transplacental transmission in mice. These include immunocompetent hSTAT2 KI mice; mice deficient in IFN-I signaling such as Ifnar1^−/−, Ifnar1^−/−Ifngr1^−/− (AG129), and Irf3^−/−Irf7^−/− mice; mice treated with an anti-IFNAR1 monoclonal Ab (MAR1–5A3) for transient blocking of IFN-I signaling; direct intrauterine inoculation of ZIKV; infection at a pre-placentation timepoint (before E10.5); and use of a high viral titer or a highly virulent ZIKV strain.

Figure creation: Figure 1 was created with Biorender.com.

2. Mouse models for studying transplacental transmission of ZIKV

2.1. Wild-type mouse models

The past several years has seen intense effort to develop mouse models of ZIKV infection and pathogenesis. The general consensus among researchers is that immunocompetent wild-type (WT) mice are relatively resistant to ZIKV infection and only poorly recapitulate the patterns of ZIKV pathogenesis commonly observed in humans. One reason for this is that interspecies differences in the major antiviral type I interferon (IFN-I) pathway enable ZIKV to evade clearance in humans but not in mice. For instance, ZIKV NS5 binds to and targets human, but not mouse, STAT2 for proteasomal degradation (Grant et al., 2016). Similarly, ZIKV NS2B3 protease mediates the cleavage of human STING but not the mouse counterpart, thereby evading cGAS/STING signaling-induced upregulation of IFN-Is, IFN-IIIs, and IFN-stimulated genes (ISGs)(Ding et al., 2018; “Species-specific disruption of STING-dependent antiviral cellular defenses by the Zika virus NS2B3 protease,” n.d.). However, the relative inability of ZIKV to replicate in mice may be mouse strain-dependent. For example, infection of non-pregnant C57BL/6J mice with the Brazilian strain ZIKV^BR resulted in poor viral replication(Larocca et al., 2016), consistent with other reports (Cugola et al., 2016; Miner et al., 2016); in contrast, ZIKV^BR replicated efficiently in non-pregnant BALB/c and SJL mice, albeit without inducing severe disease (Larocca et al., 2016). In another study, infection of pregnant SJL dams with a high titer of ZIKV^BR resulted in transplacental transmission, as confirmed by detection of ZIKV genomic RNA in newborn pups, including the brains, and resulted in IUGR (Cugola et al., 2016). In the same study, however, neither vertical transmission nor major growth defects of pups were observed when C57BL/6J dams were infected with ZIKV^BR (Cugola et al., 2016). Additional work will be needed to clarify whether the inability of ZIKV to infect WT mice is reproducible across multiple mouse strains and, if not, to uncover the source of the inconsistency. Interestingly, Szaba et al. proposed that placental pathology, not ZIKV infection per se, was the key determinant of adverse fetal outcomes based on their studies using a WT pregnancy model (Szaba et al., 2018). Inoculation of pregnant WT C57BL/6J dams with ZIKV strain PRVABC59 at embryonic development day 9.5 (E9.5) resulted in fetal demise, IUGR, and resorption at E17.5. However, ZIKV genomic RNA was detected only in maternal spleens, not in the maternal sera nor other tissues, and only sporadically in the placentas and fetuses (Szaba et al., 2018). It is unclear how ZIKV-induced placental damage could have occurred if there was low ZIKV replication, and this aspect of the hypothesis warrants further investigation.

In general, most WT mouse models, including those for pregnancy studies, have been unable to provide robust and consistent phenotypes of ZIKV replication and induced pathologies. This continues to be a key challenge for advancement of the field. Various methods are currently in use to either increase mouse susceptibility to infection or to increase ZIKV virulence/titer in order to recapitulate ZIKV infection and pathogenesis in both non-pregnant and pregnant mice. One promising immunocompetent mouse model of ZIKV infection developed by Gorman et al. is C57BL/6J mice with human STAT2 (hSTAT2) knockin (KI) at the mouse Stat2 locus (Gorman et al., 2018). Infection of non-pregnant hSTAT2 KI mice with a mouse-adapted ZIKV strain (Dakar-MA) resulted in 30% fatality, and higher levels of ZIKV RNA were detected in the spleens and brains of these mice than in WT C57BL/6J mice infected with non-mouse-adapted ZIKV Dakar. Moreover, infection of pregnant hSTAT2 mice (x hSTAT2 sires) with ZIKV Dakar-MA at E6.5 resulted in higher levels of infection in maternal spleens, sera, placentas, and fetal heads compared with a ZIKV Dakar-infected WT × WT cross, highlighting the promise of hSTAT2 mice for use in studies of ZIKV vertical transmission (Gorman et al., 2018).

2.2. Mice with genetic deficiencies in IFN signaling

Mice deficient in the type I IFN receptor (Ifnar1^−/−) or both type I and type II IFN receptors (Ifnar1^−/−Ifngr1^−/− [AG129]) are more susceptible to ZIKV infection and the development of severe neurological phenotypes than other mouse models (Rossi et al., 2016). Duggal et al. infected pregnant dams of a AG129 × AG129 cross with a low dose (10³ plaque forming units [PFU]; low and high ZIKV challenge doses are generally considered to be in the range of 10³ PFU and 10⁶ PFU, respectively) of ZIKV PRVABC59 at E3.5 and compared various routes of inoculation (s.c., intravaginal [ivag.], and sexually) (Duggal et al., 2018). They detected high levels of ZIKV RNA in the fetuses regardless of the infection route, suggesting that mice deficient in both IFN-I and IFN-II signaling are highly permissive to ZIKV vertical transmission (Duggal et al., 2018). Yockey et al. examined dams of an Irf3^−/−Irf7^−/− × Irf3^−/−Irf7^−/− cross inoculated ivag. with 2.5 × 10⁴ PFU of ZIKV FSS13025 and also detected high ZIKV RNA levels in the fetuses (Yockey et al., 2016), consistent with the findings with AG129 mice. IRF3 and IRF7 are master regulators of type I IFN production, and mice deficient in these genes lack the ability to turn on many ISGs. These studies provide further evidence that mice with attenuated IFN-I signalling can support ZIKV replication and that such models can be used to study ZIKV vertical transmission. Nevertheless, the use of immunocompromised mouse models of ZIKV infection during pregnancy has received some criticism due to the loss of normal antiviral immunity and the inability to mount a normal immune response to vaccination in a physiologically relevant manner. IFN-I signaling plays crucial roles in antiviral responses of both mice and humans; for example, IFN-I pathway signalling leads to upregulation of ISGs that mediate cell apoptosis, cross-presentation of viral peptides, inhibition of viral replication, and overall control of the viral infection (Lukhele et al., 2019). This innate IFN-mediated response is also necessary for robust activation of cellular and humoral immune responses, which are critical immune components for successful vaccination.

IFN signalling also seems to be important for a healthy pregnancy. Mice deficient in Ifnar1 lack spiral artery remodeling; yet interestingly, these pups are otherwise healthy and experience only mild growth restriction (Murphy et al., 2009). IFN-I signaling can also play a pathogenic role in ZIKV infection during pregnancy and was shown to mediate the severe fetal pathology observed in Ifnar1^+/− but not Ifnar1^−/− fetuses in an Ifnar1^−/− dam × Ifnar1^+/− sire cross, including high rates of fetal resorption and IUGR that was associated primarily with retardation of placental development (Yockey et al., 2018). Notably, fetuses of women with disorders associated with IFN-1 overexpression have abnormal fetal brain development (Crow, 2015), and elevated IFN-1 levels in pregnant systemic lupus erythematosus patients is associated with an increased risk of preeclampsia (Andrade et al., 2015).

These observations suggest that mating of dams and sires with defective IFN-I signaling would provide a powerful method to study fetal/placental immunity in ZIKV infection. For example, crosses of Ifnar^−/− dams × WT sires gives rise to fetuses with one functioning copy of Ifnar1 and permits ZIKV replication in maternal tissues and some degree of vertical transmission. Yockey et al. found higher levels of ZIKV RNA in fetal tissues resulting from an Ifnar1^−/− × WT C57BL/6NCrl cross as compared to a WT C57BL/6NCrl × C57BL/6NCrl cross, but those levels were still lower than those detected in fetuses from an Irf3^−/−Irf7^−/− × Irf3^−/−Irf7^−/− cross (Yockey et al., 2016). Miner et al. infected dams of an Ifnar1^−/− × WT C57BL/6J cross with 10³ focus-forming units (FFU) of ZIKV strain H/PF/2013 and observed IUGR and fetal demise accompanied by a high ZIKV RNA load in fetal tissues (Miner et al., 2016). This hemizygous mating strategy better allows the study of fetal immunity than a fully Ifnar1^−/− cross. Recent evidence has shown that patients with severe COVID-19 were more likely to have autoantibodies that impair type I IFN immunity, mimicking the outcomes of inborn errors of type I IFN signaling (Bastard et al., 2020). Although a similar association has not yet been reported for ZIKV-infected individuals, the observation does support the human relevance of studies with Ifnar1^+/− mice harboring partial impairment of the type I IFN response. One important factor to consider with the use of mice with compromised IFN-I signaling, however, is that maternal immunity plays a critical role in physiological changes during pregnancy, and other confounding factors may contribute to fetal outcome. Moreover, such models will limit vaccine testing studies that require the dam to be immunized before pregnancy.

2.3. Models with transient blocking of IFN signaling using anti-IFNAR1 monoclonal antibody

To overcome the limitations of mice genetically deficient in IFN signaling, some studies have employed transient IFNAR1 blockade of immunocompetent mice by injection of the blocking monoclonal Ab (mAb) MAR1–5A3. This treatment allows flaviviruses to replicate in immunocompetent mice without having a major impact on CD8 T cell priming, although WNV-specific CD8 T cells were found to have depressed cytokine function 4 days after infection (Pinto et al., 2011). The half-life of MAR1–5A3 in WT mice administered at 2 mg/mouse intraperitoneally (i.p.) is reported to be about 5.2 days (Sheehan et al., 2006), whereas the time to peak ZIKV viremia is approximately 3 days post-infection (Liu et al., 2021), supporting the utility of MAR1–5A3 in flavivirus replication models. With the use of a low dose of MAR1–5A3 (0.2 mg/mouse), Sheehan et al. reported a shorter half-life for MAR1–5A3 in WT mice (1.8 days) compared with Ifnar1^−/− mice (7.5 days), most likely due to absorption of the mAb by IFNAR1-positive cells(Sheehan et al., 2006). Lazear et al. tested the effects of 750 μg, 1 mg, and 2 mg of MAR1–5A3 administered i.p. to WT C57BL/6J mice 1 day before infection with 10³ FFU of ZIKV H/PF/2013 (Lazear et al., 2016). They found that mice treated with 1 mg or 2 mg of MAR1–5A3 had detectable ZIKV at 3 days post-infection and recapitulated the phenotype of Ifnar1^−/− mice (Lazear et al., 2016). A number of studies have subsequently employed MAR1–5A3 to permit ZIKV infection of immunocompetent pregnant and non-pregnant mice (Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016).

Miner et al. treated syngeneically pregnant WT C57BL/6J dams with 1 mg MAR1–5A3 i.p. on E5.5 followed by subcutaneous (s.c.) infection with 10³ FFU of ZIKV H/PF/2013 on E6.5 or E7.5 (Miner et al., 2016). At E13.5 and E15.5, mild IUGR was observed and ZIKV RNA was detectable in fetal and maternal tissues. However, the ZIKV infection level was lower and fetal pathology was milder in the MAR1–5A3-treated WT mouse model compared with the Ifnar1^−/− × WT C57BL/6J model. In the same study, it was found that injection of MAR1–5A3 at doses greater than 0.5 mg/mouse had a dose-dependent effect on ZIKV viral load in fetal heads but not in the placentas (Miner et al., 2016). Sapparapu et al. treated syngeneically pregnant WT C57BL/6J dams with 1 mg MAR1–5A3 i.p. 1 day before and 1 day after infection with 10³ FFU of ZIKV Dakar-MA s.c. on E6.5. In this case, high viral load was detected in both the fetal brains and placentas at E13.5 (Sapparapu et al., 2016). Richner et al. employed MAR1–5A3 treatment to study the effects of ZIKV prME RNA vaccination of female mice before ZIKV challenge⁴⁴. WT C57BL/6J female mice were immunized with a prime-boost vaccine regimen that induced high levels of serum neutralizing Abs. The immunized females were then mated to WT males, treated with 2 mg MAR1–5A3 on E5, and infected with 10⁵ FFU ZIKV Dakar-MA on E6. At E13, the unvaccinated mice exhibited high levels of ZIKV RNA in fetal heads and placentas, placental insufficiency, and high rates of fetal resorption, whereas none of these phenotypes was observed in the vaccinated mice (Richner et al., 2017). Importantly, these studies demonstrate the advantage of the MAR1–5A3/WT mouse strategy in allowing a robust maternal immune response to vaccination while concomitantly permitting ZIKV replication that can recapitulate fetal pathology.

2.4. Influence of route of ZIKV inoculation on mouse models of transplacental transmission

2.4.1. Intravenous and subcutaneous inoculation

The route of virus delivery has a critical influence on the spatial and temporal effects of the virus on the host animal. Intravenous (i.v.) inoculation of ZIKV is commonly used to mimic the blood meal by which the mosquito vector transmits ZIKV to humans (Cugola et al., 2016; Regla-Nava et al., 2018; Szaba et al., 2018); moreover, i.v. administration allows faster systemic dissemination of the virus and is likely to permit a more widespread and severe disease phenotype in the mouse, which is especially important if mice that are highly susceptible to ZIKV infection such as Ifnar1^−/− are not used. The i.v. route also permits increased antigen presentation at relevant lymph nodes, resulting in faster immune responses and, possibly, more rapid virus clearance(Coffey et al., 2017). Cugola et al. showed that inoculation of WT × WT SJL mice with an extremely high dose (2 × 10¹¹ PFU) of ZIKV^BR (Paraiba 2015) by the i.v. route recapitulated the IUGR and microcephaly phenotype as seen in ZIKV-infected human newborns (Cugola et al., 2016). In another study, Szaba et al. observed that i.v. inoculation of pregnant dams of a WT × WT C57BL/6J cross with a relatively high dose (3.4 × 10⁵ PFU) of ZIKV PRVABC59 resulted in a low ZIKV burden in the dams, yet fetal demise and IUGR were still observed, suggesting that a high maternal viral load is not necessary for fetal pathology (Szaba et al., 2018). Further studies are needed to determine the influence of ZIKV strain, dose and mouse strain on the maternal and fetal phenotypes.

In addition to the i.v. route, s.c. inoculation may also mimic the natural route of infection, as a large proportion of viral inoculum deposited by mosquito bite are detected extravascularly (Dudley et al., 2017). Virus spread throughout the host is likely to be slower following s.c. inoculation compared with the i.v. route, largely because the virus must be disseminated through the draining lymph nodes. Nevertheless, s.c. inoculation has been used successfully in various studies to observe fetal pathology upon ZIKV infection, with methods spanning the use of different mouse models (Ifnar1^−/− dams (Miner et al., 2016; Sapparapu et al., 2016), WT C57BL/6J dams (Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016), AG129 dams (Duggal et al., 2018), hSTAT2 KI dams (Gorman et al., 2018)), viral titers (10³ to 10⁶ FFU/PFU) (Duggal et al., 2018; Gorman et al., 2018; Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016), ZIKV strains (H/PF/2013 (Miner et al., 2016), ZIKV^BR [Paraiba 2015] (Sapparapu et al., 2016), Dakar-MA (Gorman et al., 2018; Richner et al., 2017; Sapparapu et al., 2016), PRVABC59 (Duggal et al., 2018)), and the use of MAR1–5A3 IFNAR1 blocking mAb (Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016).

2.4.2. Intravaginal inoculation and male-to-female sexual transmission

Sexual transmission has been estimated to account for about 2% of ZIKV infections in the United States(D’Ortenzio et al., 2016a), and multiple case reports confirm this route of transmission(D’Ortenzio et al., 2016b; Foy et al., n.d.; Russell et al., 2017; Venturi et al., 2016). ZIKV has been shown to persist in the testes of rhesus macaques for up to 28 days post-infection and in the semen of men for up to 62 days after onset of symptoms of ZIKV infection (Atkinson et al., 2016; Osuna et al., 2016). Sexual transmission of ZIKV is an important route because it encourages the geographical spread of ZIKV even to places at low risk of mosquito transmission. Intravaginal (ivag.) inoculation of mice is used to simulate the sexual transmission of ZIKV, and inoculation after mating of uninfected mice also allows ZIKV infection at different gestation timepoints to be studied. Yockey et al. showed that ivag. infection of pregnant dams of a WT C57BL/6NCrl × C57BL/6NCrl cross at E4.5 or E8.5 resulted in mild IUGR or no fetal pathology, respectively²³. Comparing WT and IFN signaling-deficient mice, these authors showed that ZIKV replication in the vagina and fetal tissues was highest in an Irf3^−/−Irf7^−/− × Irf3^−/−Irf7^−/− cross, followed by an Ifnar1^−/− × WT cross, and then a WT × WT cross (Yockey et al., 2016). The observation that fetal pathology was present despite little to no maternal viremia in the WT × WT cross prompted the authors to suggest that a direct transmission route from the vagina to the intrauterine space could exist. In a non-pregnant LysMCre⁺ Ifnar1^fl/fl C57BL/6J mouse model, which lacks IFNAR expression on myeloid cells (macrophages, neutrophils, some dendritic cells), Tang et al. showed that ivag. inoculation of ZIKV FSS13025 during the diestrus phase resulted in self-limiting infection, with ZIKV RNA detectable in the serum at 3, 5, and 7 days post-infection (Tang et al., 2016). Unlike Ifnar1^−/− mice, the LysMCre⁺ Ifnar1^fl/fl mice have IFN-I-competent T cell and B cells, which makes them a more relevant model for the study of immune responses to ZIKV infection. Thus, it may be preferable to use ivag. inoculation and the LysMCre⁺ Ifnar1^fl/fl mouse model for studies evaluating vaccine-elicited immunity to ZIKV.

To simulate sexual transmission of ZIKV in parallel with conception, Uraki et al. mated ZIKV^MEX-infected Ifnar1^−/− male mice with naïve Ifnar1^−/− females and observed that ZIKV particles were detectable in the fetal brains and that IUGR and ocular malformations were present (Uraki et al., 2017). Duggal et al. also showed that infection through sexual transmission resulted in more severe maternal and fetal pathology than when infection was through the ivag. or s.c. routes, both of which resulted in pathology of similar severity (Duggal et al., 2018).

2.4.3. Intrauterine inoculation

Vermilion et al. showed that direct intrauterine inoculation of WT pregnant dams (CD1 × CD1) at E10 with ZIKV IBH 30656, FSS13025, ZIKV^BR, or PRVABC59 resulted in detectable ZIKV RNA levels in fetal tissues and also led to fetal demise (Vermillion et al., 2017). This method of inoculation would be useful for studying fetal ZIKV infection in WT mice, but may be less representative of natural routes of infection because it circumvents maternal infection.

Collectively, these studies demonstrate the impact of the route of ZIKV inoculation on fetal outcomes in mouse models and highlight the importance of experimental context when selecting the most appropriate route. However, because most cases of ZIKV vertical transmission require the mother to be viremic, i.v. inoculation is the fastest way to ensure systemic dissemination of ZIKV in the dam. Introducing ZIKV into the maternal bloodstream also exploits the fact that the developing placenta is in contact with maternal blood, thereby providing a direct route for placental and, possibly, fetal infection.

2.5. Other factors influencing ZIKV models of transplacental transmission

2.5.1. High titer/highly virulent ZIKV strains

Not surprisingly, a high inoculating titer of ZIKV is likely to result in higher maternal viremia than lower titers. However, given that mosquito-transmitted ZIKV titers are estimated to range from 10^2.5 to 10^6.2 PFU (Dudley et al., 2017), it is also important to strike a balance between achieving viral replication in the model and ensuring clinical translatability. The number of ZIKV particles administered to mice is also restricted by the viral titer and the practical volume for a particular route of inoculation. Multiple ZIKV strains of African and Asian lineages are available, but they differ in virulence. Aubry et al. compared seven diverse ZIKV strains and found that the African lineage caused more severe fetal pathology than the Asian lineage in a SWISS × SWISS cross (Aubry et al., 2021). As mentioned above, a high titer of ZIKV^BR (2 × 10¹¹ PFU) can result in fetal infection, IUGR, and microcephaly in a SJL × SJL mating cross even when infection occurs post-placentation (E10 to E13) (Cugola et al., 2016). Most other studies of ZIKV utilize titers ranging from 10³ to 10⁵ PFU or FFU and require other strategies to permit fetal infection and pathology (Duggal et al., 2018; Lopez et al., 2022; Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016; Szaba et al., 2018; Vermillion et al., 2017; Yockey et al., 2016).

2.5.2. Pre-placentation infection

ZIKV infection during the first trimester of human pregnancy is associated with the highest rates of fetal damage and microcephaly (Cauchemez et al., 2016; de Oliveira et al., 2016). In mice, the highest rates of ZIKV replication in the placenta occur in early gestation and decline thereafter(Jagger et al., 2017b). Inoculation of MAR1–5A3-treated pregnant dams of a WT C57BL/6J × C57BL/6J cross with ZIKV^BR (Paraiba 2015) s.c. at E6 caused placental insufficiency and fetal demise, whereas infection at E9 caused fetal growth abnormalities, and infection at E12 resulted in generally normal-sized, viable fetuses even though ZIKV RNA was detected in the fetal heads²⁴. However, ZIKV burden in maternal tissues was similar for all three infection protocols, suggesting that placental maturity could explain the differences in fetal outcomes(Jagger et al., 2017a). Another study found that ivag. inoculation of dams of an Ifnar1^−/− × WT cross at E4.5 resulted in 100% fetal demise, but this was not seen when inoculation occurred at E8.5 (Yockey et al., 2016). Mouse placentation is typically considered to be complete at E10.5, when the labyrinth zone, spongiotrophoblasts, and trophoblast giant cells are established and make up the definitive placenta (Rossant and Cross, 2001; Szaba et al., 2018; Woods et al., 2018). To recapitulate the real-world infection situation and to use a model with high permissivity to ZIKV replication, most studies infect the pregnant dams at a pre-placentation timepoint (Duggal et al., 2018; Lopez et al., 2022; Miner et al., 2016; Richner et al., 2017; Sapparapu et al., 2016; Szaba et al., 2018; Yockey et al., 2016). I.v. inoculation of pregnant CD-1 mice with ZIKV IBH 30656 at E10 (complete placentation) vs E14 (end of placenta growth in mice) found that even in a post-placentation context, inoculation at an earlier rather than later timepoint resulted in higher levels of ZIKV replication in the placenta and more severe fetal damage (Vermillion et al., 2017).

It is important to note that mouse and human pregnancies have temporal and physiological differences that may influence study outcomes and conclusions. Mouse gestation is an average of 21 days in length compared with 9 months in humans; moreover, the placenta is formed much later in mouse gestation (E10.5–12.5) than in human gestation (day 21) (Ander et al., 2019; Rossant and Cross, 2001; Szaba et al., 2018; Woods et al., 2018). Although both mice and humans have hemochorial placentas with chorionic villi separated from maternal blood via the trophoblast cells, mice have three layers of trophoblasts (hemotrichorial) while humans have only one layer (hemomonochorial) (Ander et al., 2019). It is not well understood how these differences affect ZIKV propensity for fetal infection in mice compared with humans. In addition, because ZIKV affects fetal brain development, an area of intense interest in the field, the temporal differences between mouse and human neurogenesis should be noted. Mouse fetal brain development starts at E10 and continues after parturition. At birth, the developmental stage of the pup brain is analogous to that of the human fetal brain in mid-gestation(Semple et al., 2013). Hence, it is possible that ZIKV infection of pregnant dams before E10 might affect progenitor cells and disproportionately impair neurogenesis in mice compared with humans.

2.5.3. Sequential DENV→ZIKV infection models

The four serotypes of DENV (DENV 1–4) and ZIKV are closely related flaviviruses, co-circulate in many regions, and are transmitted predominantly by Aedes aegypti mosquitoes. DENV outbreaks have occurred regularly since the 19^th century and, notably, they have steadily increased in severity with time (Brathwaite Dick et al., 2012; Gubler, 2006). The first large ZIKV outbreak was reported only in 2007 (Musso and Gubler, 2016), and consequently, many ZIKV-infected patients have pre-existing immunity to one or more DENV serotypes (Gordon et al., 2019; Reynolds et al., 2020; Rodriguez-Barraquer et al., 2019). This is a significant observation because pre-existing poorly neutralizing cross-reactive Abs to DENV or ZIKV can exacerbate the severity of a subsequent DENV infection, possibly leading to severe dengue and even death through a phenomenon known as Ab-dependent enhancement of infection (ADE) (Elong Ngono and Shresta, 2019; Katzelnick et al., 2020). Indeed, Brown et al. showed that passive transfer of sera from DENV-immune individuals to pregnant dams of a Stat2^−/− × Stat2^−/− cross followed by intradermal inoculation of ZIKV PRVABC59 on E6.5 resulted in IUGR and fetal demise at E13.5, suggesting a pathogenic effect of DENV-elicited Abs upon ZIKV infection during pregnancy (Brown et al., 2019). The same group also demonstrated enhancement of ZIKV pathogenesis in non-pregnant mice passively transferred with DENV-immune sera (Bardina et al., 2017). However to date, evidence supporting ADE of ZIKV infection and pathogenesis in DENV-exposed humans is lacking (Halai et al., 2017; Michlmayr et al., 2020; Santiago et al., 2019; Terzian et al., 2017) albeit one study has suggested an association between maternal Ab features and congenital Zika syndrome (Robbiani et al., 2019).

On the contrary, passive transfer studies using strongly neutralizing DENV-ZIKV cross-reactive mAbs have shown protective effects of Abs against ZIKV infection in pregnant mice. For instance, Kam et al. administered human DENV E-protein-specific mAbs with high ZIKV neutralizing capacities at 0, 1, and 3 days after infection of Ifnar1^−/− dams with 10⁷ PFU of ZIKV H/PF/2013 at E10.5 and found that ZIKV loads in fetal organs and IUGR were reduced in animals receiving the cross-reactive mAbs as compared to control Abs (Kam et al., 2017). Fernandez et al. also reported that human DENV E-dimer-specific mAbs protected Ifnar1^−/− pregnant dams (x WT sires) from infection with ZIKV^BR (Paraiba 2015) and dramatically improved fetal survival (Fernandez et al., 2017). In addition to Abs, DENV-ZIKV cross-reactive CD8⁺ T cells have been shown to protect against ZIKV in both non-pregnant (Wen et al., 2017a, 2017b) and pregnant mice (Regla-Nava et al., 2018). Consistent with these mouse studies demonstrating a protective role for DENV-ZIKV cross-reactive Abs and CD8⁺ T cells, multiple human cohort studies have suggested that prior DENV exposure confers cross-protection against subsequent ZIKV infection (Gordon et al., 2019; Pedroso et al., 2019; Rodriguez-Barraquer et al., 2019). Based on these findings, an argument can be made that mouse models of primary ZIKV infection (Table 1), although undoubtedly crucial to elucidating ZIKV-specific transmission and fetal pathology, are less epidemiologically relevant than the DENV→ZIKV sequential infection model.

Table 1.

Mouse models of transplacental transmission during primary ZIKV infection

Mating pairs (dam × sire)	Notable interventions	Route of infection	Day of infection	ZIKV titer and strain	Key findings		Ref.
Ifnar1−/− C57BL/6J × WT C57BL/6J		s.c.	E6.5 or E7.5	103 FFU H/PF/2013	Both models: IUGR Ifnar1−/− dam: Fetal demise ZIKV RNA detected in maternal spleens, brains, sera, fetal heads and placentas greater for Ifnar1−/− dams than for WT dams		(Miner et al., 2016)
WT C57BL/6J × WT C57BL/6J	IFNAR1 blocking mAb (E5.5)	s.c.	E6.5 or E7.5	103 FFU H/PF/2013
Ifnar1−/− C57BL/6J × WT C57BL/6J		s.c.	E6.5	103 FFU ZIKVBR (Paraiba 2015)	Fetal demise		(Sappar apu et al., 2016)
WT C57BL/6J × WT C57BL/6J	IFNAR1 blocking mAb (E5.5 and E7.5)	s.c.	E6.5	103 FFU Dakar-MA	ZIKV RNA detected in placentas and fetal brains
Irf3−/− Irf7−/−C57BL/6J × Irf3−/− Irf7−/− C57BL/6J		ivag.	E4.5 or E8.5	2.5 × 104 PFU FSS13025	Infection at E4.5: IUGR E8.5: Normal-sized fetuses	ZIKV RNA detected in vaginas, fetuses, and placentas greater for Irf3−/− Irf7−/− dams followed by > Ifnar1−/− dams followed by WT dams	(Yockey et al., 2016)
Ifnar1−/− C57BL/6J × WT C57BL/6J		ivag.	E4.5 or E8.5	2.5 × 104 PFU FSS13025	Infection at E4.5: Fetal demise E8.5: IUGR
WT C57BL/6NCrl × WT C57BL/6NCrl		ivag.	E4.5 or E8.5	2.5 × 104 PFU FSS13025	Infection at E4.5: IUGR E8.5: Normal-sized fetuses
AG129 X AG129		s.c., ivag.	E3.5	103 PFU PRVABC59	Similar ZIKV titers in maternal brains, ovaries and uteri across routes of inoculation. Infection rate and ZIKV RNA levels in fetuses greater by sexual transmission than by ivag. or s.c.		(Duggal et al., 2018)
Naïve AG129 × ZIKV-infected AG129	Sexual transmission	M (ZIKV-infected) →F (naïve) sexual transmission	7 days before mating
hSTAT2 KI C57BL/6J × hSTAT2 KI C57BL/6J		s.c.	E6.5	106 FFU Dakar-MA	ZIKV RNA detected in maternal serum and spleens, placentas, fetal heads		(Gorma n et al., 2018)
Naïve Ifnar1−/− C57BL/6J × ZIKV- infected Ifnar1−/− C57BL/6J	Sexual transmission	M (ZIKV-infected) →F (naïve) sexual transmission	3 days before mating	105 PFU ZIKVMEX	IUGR and ocular malformations; ZIKV virions detected in fetal brains		(Uraki et al., 2017)
WT C57BL/6J × WT C57BL/6J	IFNAR1 blocking mAb (E5)	s.c.	E6	105 FFU Dakar-MA	Placenta insufficiency and fetal demise; ZIKV RNA detected in maternal spleens, brains, placentas, and fetuses		(Richner et al., 2017)
WT SJL × WT SJL	High viral dose	i.v.	E10–13	2 × 1011 ZIKVBR (Paraiba 2015)	IUGR, cortical and ocular malformations; ZIKV RNA detected in fetuses		(Cugola et al., 2016)
WT C57BL/6J × WT C57BL/6J
CD1 × CD1	Mini-laparotomy; intrauterine inoculation	Intrauterine	E10 or E14	106 TCID50 units IBH 30656, FSS13025, ZIKVBR (Paraiba 2015), PRVABC59	E10: fetal demise, placenta defect, cortical malformations, ZIKV RNA and infectious virus particles detected in placentas and fetal heads, E14: fetal demise, ZIKV RNA detected in placentas		(Vermill ion et al., 2017)
WT C57BL/6J × WT C57BL/6J		i.v.	E9.5	3.4 × 105 PFU PRVABC59	IUGR and fetal demise; low ZIKV RNA levels detected in maternal and fetal tissues.		(Szaba et al., 2018)
WT C57BL/6J × WT C57BL/6J		ivag.	E7	103 FFU H/PF/2013	Healthy pregnancies; ZIKV RNA detected in vaginal washes but not in maternal sera		(Lopez et al., 2022)

Abbreviations: E, embryonic day; F, female; FFU, focus-forming unit; hSTAT2, human STAT2; Ifnar1, IFN α/β receptor subunit 1; i.p., intraperitoneal; i.v., intravenous; ivag, intravaginal; IUGR, intrauterine growth restriction; KI, knockin; M, male; mAb; monoclonal antibody; PFU, plaque-forming unit; Ref., reference; s.c., subcutaneous; WT, wild-type.

At the time of writing, only two studies have examined ZIKV infection during pregnancy using a DENV→ZIKV sequential infection model (Table 2). Both studies used similar protocols in which primary infection with a DENV2 strain occurred at least 3 weeks before mating and was followed by infection with an Asian lineage ZIKV strain pre-placentation at E7/7.5. Nevertheless, a direct comparison between the two studies is difficult because one (Rathore et al., 2019) used high titers (10⁶ PFU) of both DENV and ZIKV and WT C57BL/6J × C57BL/6J mice whereas the second (Regla-Nava et al., 2018) used 10³/10⁴ FFU of DENV/ZIKV and either Ifnar1^−/− × WT C57BL/6J mice or MAR1–5A3-treated C57BL/6J mice. Interestingly, Rathore et al. found that prior DENV2 immunity exacerbated ZIKV-induced fetal pathology, whereas Regla-Nava et al. demonstrated a protective effect of prior DENV2-immunity, suggesting a potentially important role for IFN-I or levels of DENV and ZIKV replication in the protective vs pathogenic involvement of pre-existing DENV immunity on fetal outcomes.

Table 2.

Mouse models of transplacental transmission during sequential DENV → ZIKV infection

Mating pairs (dam × sire)	Notable interventions	Route of infection	Day of infection	Order of infection, virus strain, and titer	Key findings	Ref.
C57BL/6J × C57BL/6J		i.p.	3 weeks before mating	1°: DENV2 Eden2 1 × 106 PFU	ZIKV RNA load detected in fetal tissues and maternal spleens greater for DENV-immune dams than for DENV-naive dams. Higher incidence of ZIKV-associated microcephaly among fetuses of DENV-immune dams than of DENV-naïve dams.	(Ratho re et al., 2019)
		i.p.	E7	2°: ZIKV H/PF/2013 1 × 106 PFU
Ifnar1−/− × WT C57BL/6J		i.p.	3 to 5 weeks before mating	1°: DENV2 S221 1 × 103 FFU	Same observations for both protocols: ZIKV RNA levels were reduced in fetal tissues and maternal spleens of DENV-immune dams compared with DENV-naïve dams. Fetuses of DENV-immune dams were protected from IUGR and resorption in a CD8 T cell-dependent manner.	(Regla-Nava et al., 2018)
		i.v.	E7.5	2°: ZIKV FSS13025 1 × 104 FFU
C57BL/6J × C57BL/6J	IFNAR1 blocking mAb (1 day before infection)	i.v.	3 to 5 weeks before mating	1°: DENV2 S221 1 × 104 FFU
	IFNAR1 blocking mAb (E6.5)	i.v.	E7.5	2°: ZIKV FSS13025 1 × 104 FFU

Abbreviations: 1°, primary; 2°, secondary; E, embryonic day; FFU, focus-forming unit; Ifnar1, IFN α/β receptor subunit 1; i.p., intraperitoneal; i.v., intravenous; IUGR, intrauterine growth restriction; mAb, monoclonal antibody; PFU, plaque-forming unit; Ref., reference; WT, wild-type.

3. Conclusion

There was a mere 8 year interval between the first large ZIKV outbreak on the Yap Islands and the ZIKV epidemic in South America; since then, ZIKV has spread silently and the timing of the next ZIKV epidemic is unknown. There is evidence that rising global temperatures have contributed to encroachment of mosquito vectors into more temperate zones, putting billions more people at risk for infection. This has created an imperative and urgent need to develop a ZIKV vaccine that can protect people of all ages, including unborn fetuses, from infection. Current models of ZIKV transplacental transmission are still greatly dependent on mice with genetic or acquired Ifnar1 deficiency, and given that the IFN-I response plays a crucial role in ensuring a healthy pregnancy, the use of such models may be confounding the fetal outcome. Other modeling strategies described in this review may better recapitulate the human infection and reduce the dependence on immunocompromised mice; for example, the use of different routes of infection, ZIKV strains of differing virulence, variations in viral titers, and optimization of the infection timing could all be explored in immunocompetent mice. The hSTAT2 model of Gorman et al. is also a promising advance in efforts to generate immunocompetent mouse models of ZIKV infection. Because ZIKV mediates STING cleavage in humans, a hSTING/hSTAT2 double-KI mouse model might allow for a more robust ZIKV-induced disease phenotype. The real-world relevance of DENV/ZIKV co-circulation must be addressed in the development of mouse models, particularly because the DENV→ZIKV sequential infection pregnancy models tested to date typically build on knowledge gleaned from primary ZIKV infection models and have conflicting conclusions, as noted. These findings highlight the complexity underlying the cross-reactive immune responses between DENV and ZIKV that hinder ZIKV vaccine development. There is a pressing need to better understand how sequential infection affects fetal outcomes during ZIKV infection. Several key questions remain to be addressed. For example, when during gestation does the placenta provide peak protection against ZIKV infection, given that placental development continues throughout pregnancy? To what extent does fetal/placental immunity contribute to fetal protection against ZIKV? How do Asian and African ZIKV lineages differ in pregnancy outcomes, given their differences in virulence? Do allogeneic mouse pregnancies (mimicking the human situation) alter ZIKV transplacental transmission? How does the route of maternal infection (sexually vs mosquito-transmitted) affect vertical transmission and fetal outcome? Overall, it seems clear that mouse models of transplacental transmission will allow us to elucidate the mechanisms of in utero ZIKV transmission and to advance our understanding of maternal and fetal immunity, with the ultimate goal of designing ZIKV vaccines and therapeutics that can avert the potentially devasting consequences of maternal ZIKV infection.