Natural infections are the only model available to study endogenous transplacental transmission of N. caninum. In naturally infected heifers, the detection of an increase in the levels of specific antibodies has been related to a reactivation on the chronic infection during gestation (61, 62). Following these findings, Rosbottom et al. (63) monitored monthly fluctuations of N. caninum-antibodies during gestation in naturally infected cows, which were euthanized when recrudescence was detected (mid-gestation) and used to study the immune response, pathological changes and parasite loads in the placenta (63). This study showed that all the fetuses were alive, but parasite had already invaded the placenta and reached the fetal tissues when recrudescence was detected. Necrosis in the fetal villi and CD4+ and CD8+ T cells infiltration was also observed, with an increase of expression of pro-inflammatory (IFN-γ, IL-12, and TNF-α) and regulatory (IL-4 and IL-10) cytokines. Authors stated that the response in the placenta was not polarized toward either a Th1 or a Th2 phenotype, which may have an important role in controlling parasite multiplication.

Unlike neosporosis, the horizontal transmission is the main route that T. gondii exploits to infect small ruminants in the environment. Previous works carried out with naturally infected animals focused on the study of the consequences of the infection, with little information about the interactions of the parasite with the placenta. All these studies described the presence of multifocal areas of necrosis, commonly associated with the infiltration of non-purulent lymphoid in placentomes when abortion occurred (34).

Primoinfection models are the only experimental approach available to study bovine neosporosis, and they only mimic the exogenous transplacental route of transmission of N. caninum. In these models the time of gestation at which challenge is done determines the occurrence of fetal death or parasite vertical transmission, as well as the severity of pathological and immunological changes in the placenta (46).

Intravenous inoculation of heifers with tachyzoites of virulent isolates at the first term of gestation (65–70 days) is generally associated with a rapid fetal death due to the relative immunological immaturity of the fetus (46), and thus this model can be used to study pathological and immunological changes occurring in placentomes when fetal death occurs (Figure 3B). The results derived from these studies are quite homogeneous, with placentomes from aborted animals widely colonized by tachyzoites and displaying severe necrotic and extensive inflammatory lesions. With regard to cytokines expression at the maternal-fetal interface, an exacerbated pro-inflammatory response (particularly IFN-γ) and high expression of anti-inflammatory/regulatory cytokines (such as IL-10 or IL-4) were observed (76, 77, 87, 88). Moreover, a positive association between high number of T lymphocytes in placenta (CD3+, CD4+, CD8+, and γδ) and occurrence of abortion has been demonstrated by immunohistochemistry (IHC) (89). More recently, gene expression of the endosomal toll-like receptors (TLRs)-3, TLR-7, and TLR-8 by quantitative reverse transcription PCR (RT-qPCR) was described in the caruncle from cattle experimentally infected with N. caninum at day 70 of gestation and euthanized 1 month later (90), suggesting that the initial recognition of the parasite would occur in the maternal part of the placenta. Sequential euthanasia and sampling of cattle (Figure 3B) after intravenous challenge at early gestation, showed the presence of moderate to severe pathological changes and tachyzoites within areas of villous necrosis at 14 days post-infection (dpi) (79, 91). At 28 dpi, fetal death has occurred, and cotyledons were detached from caruncles, detecting N. caninum antigen in the cotyledons. When the immune response was studied by IHC, infiltration of CD4+ cells and NKp46+ cells was detected in the caruncles at 14 dpi, whereas infiltration of γδ TCR+ cells were observed from 28 dpi onwards (91). In-situ hybridization experiments demonstrated a more severe infiltration of IFN-γ expressing cells at 42 dpi, whereas the infiltrate of IL-4 expressing cells was scarce at 28 and 42 dpi, mainly in the caruncle (92). All these studies support the hypothesis that infection in early in gestation triggers a Th-1 type response in the placenta, which together with parasite replication and tissue destruction, may contribute to fetal death.

Experimental infections at mid gestation (110–140 days of gestation) can induce either foetopathy or birth of congenitally infected calves. It has been suggested that this model can be useful to understand why some dams abort and some do not. Here, the euthanasia of dams at a specific time of the experiment (Figure 3B) has been the experimental design most exploited so far. When the immune response was determined in placentomes from infected dams carrying live fetuses, Th1, Th2, and Treg cytokines were up-regulated, while TGF-β was down-regulated. This cytokine expression pattern could be beneficial to fetal survival, but could also promote transplacental transmission (93). When the immune response was determined in placentomes from aborted dams or dams carrying non-viable fetuses at the euthanasia, Th2 and Treg cytokines were down-regulated (71). Recently, cytokines expression pattern in Nc-Spain7 infected animals at mid-gestation, carrying non-viable fetuses and euthanized at 20 dpi, showed an up-regulation of IL-8, iNOS, and TNF-α, previously associated with placental inflammation, luteolysis and abortion (16). Other studies have described a reduced expression of SERPINA14 (linked to maternal immunosuppression during pregnancy) in aborting dams, and a negative relationship with IFN-γ expression in cotyledon samples (94). A recent work estimated the relative cell densities at the fetal-maternal interface from dams infected at day 110 of gestation, demonstrating lower proportions of bi- and mono-nucleate trophoblast cells after N. caninum infection at 42 dpi, and suggesting that this could be mediated by the Th1-type response of the placenta, promoting the destruction of fetal-genotyped cells (72). When infection dynamics were studied at mid gestation by serial euthanasia, the parasite was detected by PCR and IHC in the placenta as early as 10 dpi (53). Fetal death was detected from 20 dpi onwards (53, 80, 91, 95). Focal lesions and immune cell infiltration were observed in some placentomes at 14 and 28 dpi, and such lesions were resolved at later time points (80, 96). In non-aborting animals, there was no evidence of parasite antigen or infiltration of immune cells at 28, 42, and 56 dpi, suggesting that the parasite had not reached the placenta (91). In situ hybridization showed a scarce to mild infiltrate of IFN-γ, IL-4 expressing cells in the maternal caruncle at 14 and 28 dpi (92). Altogether these results suggest that infection at mid-gestation can be partially controlled by some dams, although the factors governing this remain unknown.

Experimental infection of cattle in the last third of gestation leads to the birth of healthy but congenitally infected calves. This model is useful to understand the basis of transplacental transmission and fetal survival. In placentomes obtained from animals euthanized 21 dpi, N. caninum was sporadically detected by PCR, very mild lesions with focal necrosis were found, and a modest expression of Th1 and Th2 cytokines was detected by RT-qPCR (87, 88). The presence of CD4+ and CD8+ cells was relatively low, and they were sparsely disseminated (88). Examination of placental tissues at different times post-infection evidenced the presence of parasites and focal necrotic lesions at 28 dpi, the infiltration of inflammatory cells at 42 dpi, and fibrosis at 56 dpi, which denoted lesion resolution. Animals euthanized at 14, 28, 42, and 56 dpi contained a scarce infiltrate of IFN-γ and IL-4 expressing cells, mainly in the caruncle (92). These findings indicate that infection at late gestation is less pathogenic at the maternal-fetal interface, which could partially explain the reduction on the severity of the clinical outcome after infection.

Pathological changes of N. caninum-challenged pregnant cattle at 70 days of gestation and previously immunized with live or inactivated experimental vaccines were also studied at the maternal-fetal interface. Heifers were culled at day 104 of gestation and placentomes were examined to evaluate lesions and local cellular immune responses using histopathology, IHC and RT-qPCR. Vaccination with live tachyzoites induced protective immunity, and thus, subsequent challenge only caused minimal inflammation, milder immune cell infiltration and a lower expression of cytokines. In contrast, animals vaccinated with inactivated vaccines displayed an up-regulation of Th1 and Th2 cytokines and a strong cellular immune response after challenge (97).

The influence of the N. caninum isolate on the pathological and immunological changes occurring in placentomes has been also studied in bovine experimental models. Inoculation of Nc-Spain1H and Nc-1 tachyzoites at early of gestation, resulted in death in three out of five fetuses from Nc-1 infected dams whereas that Nc-Spain1H did not have the ability to induce foetopathy. At the maternal-fetal interface, Nc-1 induced the most severe histopathological lesions and parasite DNA was also more frequently detected (52). In other studies using the same model, the high virulent isolate Nc-Spain7 induced fetal death earlier, severe placental damage and higher parasite load in caruncles than the moderate virulence Nc-1 (76) or Nc-Spain8 (77) isolates. In addition, higher IFN-γ expression was induced by Nc-Spain7, which could be detrimental for gestation, whereas a higher IL-10 upregulation was observed in Nc-Spain8 infected, which would have anti-inflammatory effects (77). More recently, we used a model of serial euthanasia for studying initial events of infection (10 and 20 dpi) at mid-gestation (53) to study the mechanisms that enable some isolates to be more effectively transmitted or cause fetal death. Infection with the low virulence Nc-Spain1H isolate triggered an early (10 dpi) innate immune recognition in the bovine placenta, characterized by upregulation of genes involved in pathogen recognition (e.g., TLR-2, TLR-3, TLR-8, TLR-9, and NOD2), inducing a solid Th1 response which were counterbalanced by a higher expression of anti-inflammatory and regulatory cytokines, minimizing placenta pathology. Conversely, signaling through pattern recognition receptors (PRRs, i.e., TLR-9, TLR-8) or cytokines (i.e., IL6, TGF-β1, IL-17A) was impaired at the placenta by the high-virulence isolate (Nc-Spain7), which may indicate the existence of an evasion mechanism that may favor an early multiplication and dissemination to the fetus. At the later stage (20 dpi), PRRs activation and a dramatically enhanced expression of the pro-inflammatory cytokines were observed. Altogether, this study demonstrated that the early activation of the innate immune response is crucial for the adequate control of N. caninum at the placenta, whereas an exacerbated pro-inflammatory immune response can lead to severe tissue damage, resulting in abortion. We went deeper in the study of the pathogenesis of the infection with these isolates of differing virulence using this model, by means of studying some components of ECM involved in maintaining placental homeostasis by IHC and RT-qPCR. Here, the low-virulence isolate promoted an ECM remodeling and tissue repair processes. However, in the placentome from animals infected by the highly virulent isolate, a profound alteration in ECM organization was observed, characterized by loss of fibronectin, vimentin and collagen in necrotic foci detected by IHC and downregulation of metalloproteases MMP-2, MMP-13, MMP-14, and their inhibitors. Recently, the lectin-binding pattern in the placentas of cows infected experimentally with N. caninum was studied by IHC techniques (98), suggesting the importance of the changes occurred in the ECM during N. caninum infection. In the last years, high-throughput technology has been used to generate relevant information on the immunological and cellular hallmarks of the host-parasite interaction at the placenta, aiming to facilitate further investigation on the key pathways influencing protection or abortion. Proteomic profiles comparison of the cotyledon and caruncle samples from previous experimental infections (53) has highlighted complement and coagulation routes, oxidation-reduction processes and ECM reorganization, among other biological processes, as key points in the pathogenic mechanism of N. caninum abortion (99). These mechanisms could play an important role and should be investigated in depth.

Regardless the gestation term at which challenge was done (Figure 3), most of the infections documented so far have extended the length of the experiments until the occurrence of abortion, usually around 30 dpi. These studies described an extensive damage of the placenta caused by proliferating parasites, with lesions similar to those described in natural infections (multiple small foci of necrosis and non-purulent inflammation) (34, 44, 100). In these abortions, parasitic debris were observed in the maternal caruncle, whereas parasitic vacuoles were found in the trophoblasts of fetal cotyledon. This suggests that parasites are first recognized in the maternal part of the placenta where they stimulate the maternal immune response, and then spread to the fetus through infection of the fetal trophoblast villi, where they remain viable (69).

As opposed to the well-known “classical” abortion, early abortions occurring during the acute phase of the infection have been also described in some occasions, although they have not been taken into consideration until recent days (44, 64, 82, 101, 102). These works described early abortions in ewes infected at different stages of gestation, with different isolates (M4, ME49, and TgShSp1) and doses of inoculation. Although there was a correlation between the dose of infection and the rate of early abortions, a dose of just 50 sporulated oocysts was capable to cause a high rate of abortions in the acute phase of the disease and even 10 sporulated oocysts of the ME49 isolate caused early abortion in one animal (64, 82). In all cases, parasites were virtually not found in placental or fetal tissues, indicating that the pathogenic mechanism triggering early abortion is different from that observed in later abortions. Histopathological analyses revealed thrombosis and infarcts in the placentomes and ischemic lesions (periventricular leukomalacia) in the fetal brain, probably as a consequence of placental lesions (47, 82, 103). IHC analyses showed an increase of macrophages in caruncular septa, which could indicate that early abortions are immune-mediated. Nevertheless, more studies are necessary to corroborate this hypothesis and to find out the origin of the placental thrombosis observed. In this sense, experimental models of serial euthanasia at early stages post-infection along with high-throughput approaches to dissect host-parasite interactions could prove useful.

Recent attempts to reproduce the classical late abortions described in natural infections have failed, and early abortions have turned into the predominant clinical presentation of experimental toxoplasmosis despite they have been virtually neglected from the literature until now. Specifically, infection with 2,000 and 500 sporulated oocysts resulted in almost 100% of early abortions. This rate was significantly reduced with lower challenge doses but, surprisingly, this reduction was not counteracted by an increase on the rates of late abortions, and infections with 50 sporulated oocysts or less resulted in the birth of healthy lambs, stillbirths and very few late abortions (64, 82). Therefore, the development of a model able to mimic the classical late abortions described in natural infections is an urgent need. Factors such as breed, previous immunization or even individual susceptibility should be considered in new experiments.

A number of studies aimed to investigate T. gondii infection dynamics through models of serial euthanasia (44, 47). However, these works failed to describe the mechanisms underlying early abortions (infected ewes were euthanized from 10 dpi onwards once most of those abortions had occurred) and focused on the late ones. Although in very few animals, T. gondii was detected in placenta as early as 10–12 dpi; in addition, percentages of detection and parasite loads were higher as the infection progressed. Development of lesions was observed as soon as 10 dpi, but only in ewes subcutaneously inoculated with tissue cyst at 60 or 90 days of gestation (44), an uncommon way to acquire the infection in the nature. These lesions were characterized by several foci of necrosis that became larger as the infection progressed. In ewes orally infected with sporulated oocysts, typical lesions were found in placenta from 26 dpi, always associated to the presence of the parasite, and were characterized by necrotic foci in the caruncular part and infiltration of macrophages and T lymphocytes within the fetal chorionic villi (69, 82).

Similarly to bovine neosporosis, the time of gestation at which infection occurs also influences lesion severity, parasite distribution and expression of cytokines at the placenta, but very few studies have addressed these matters. As early as 1986, Buxton et al., analyzed the placental lesions developed in pregnant ewes after subcutaneously inoculation of 200 tissue cysts of the M1 isolate at 40, 60, and 90 days of gestation and at the infection outcome (44). This study described the occurrence of necrotic lesions in caruncular septa, which were more severe in ewes infected at mid-gestation and in those where fetuses were dead. More recently, similar studies were done using pregnant ewes orally inoculated at early- (40 days of gestation), mid- (90 days of gestation), and late-gestation (120 days of gestation) with sporulated oocysts of M4 isolate (17, 47, 69, 100). Infection at mid-gestation also resulted in the most severe lesions and highest parasite burdens, both in the maternal and fetal placenta. However, ewes infected at late-gestation showed the earliest development of lesions and colonization of the placenta by the parasite (47). Castaño et al. hypothesized that these variations could be due to differences in the immune modulation along the pregnancy. While pro-inflammatory cytokines (IFN-γ and TNF-α) were equally increased during the three terms of gestation, IL-4 was mainly increased at the first and second terms, and IL-10 at the last term of gestation. Besides, a regulation of IL-12 expression was not observed at any term (17). This imbalance on the Th1 and Th2 responses could play a key role on the pathogenesis of toxoplasmosis, but this issue remains controversial among the scientific community.

During the course of a natural infection, parasite survival and propagation heavily rely on successful invasion and proliferation within receptive host cells as part of the parasite's lytic cycle (130). Thus, the study of such processes in target cells appears as an ideal tool to study host-parasite interactions at the placental barrier. Our previous work demonstrated that N. caninum actively replicates in caruncular and trophoblast cells, confirming the high tropism of the parasite for the placenta observed in vivo (56). However, we found interesting differences in the behavior of N. caninum between both cell lines. Immunofluorescence assays showed that parasites displayed a lower adhesion capacity to caruncular cells, and consequently, the infection rates and percentages of cells with multi-infection were lower in BCEC-1 cells. In addition, parasite proliferation capacities, measured by RT-qPCR, were also lower in those cells. These results reflected that susceptibility to infection is higher in the fetal part of the placenta, which is widely supported by previous results obtained from experimental bovine infections (16, 76, 77). In view of these, we proposed that the maternal part of the placenta acts as a barrier for parasite infection, reducing its adhesion and subsequent intracellular multiplication.

On the other hand, pathogenesis of N. caninum infection is heavily influenced by the virulence of the isolate employed. We hypothesized that differences in lytic cycle progression among different isolates could explain the differences described in vivo using a bovine model (52, 76, 77). In an attempt to address this possibility, the lytic cycle of the high-virulence isolate Nc-Spain7 and the low-virulence isolate Nc-Spain1H was compared using the BCEC-1 and F3 cell lines (56). Differences between isolates were remarkable in trophoblast cells, where Nc-Spain7 displayed higher invasion and proliferation rates than Nc-Spain1H. By contrast, the behavior of both isolates was very similar in caruncular epithelial cells (Figure 4). This higher proliferation capacity in trophoblast cells could explain why in heifers experimentally infected with the Nc-Spain7 isolate show higher parasite burdens in fetal tissues, and consequently, more severe clinical signs and histopathological lesions, resulting in higher ratios of fetal death and abortion.

It has been widely demonstrated that parasites and many other pathogens can manipulate host-cell machinery for their own benefit in order to perpetrate the infection by scavenging nutrients from the host, or even by inhibiting apoptosis or host immune responses. In addition, it has been shown that the pathogenesis of bovine neosporosis is not only related to parasite multiplication itself, but also to the host immune response triggered following infection (131). Therefore, the study of the host-cell modulation upon N. caninum infection could contribute to deepen on the knowledge of the disease.

The application of classical techniques such as RT-qPCR and commercial bovine cytokines ELISA tests to this in vitro infection model have proven to be very useful to study the response triggered by N. caninum in bovine target cells. This was revealed by studying the expression levels of cytokines, chemokines and other immune-related elements in F3 and BCEC-1 cell cultures infected with high- (Nc-Spain7) and low-virulence (Nc-Spain1H) isolates at early (4 h) and later (24 h) stages of infection (57). Similarly to what has been proven using in vivo models, this work described the development of a typical pro-inflammatory Th1 response by the bovine host against the parasite. Surprisingly, and despite the different phenotypes observed between BCEC-1 and F3 cells in terms of parasite adhesion, infection rates, and proliferation capacities, few differences were found in their responses against the infection. The main difference was the up-regulation of TLR2 in infected BCEC-1 cells, not found in F3 cells. This phenomenon could be in part responsible of the resistant phenotype found in caruncular cells, as TLR activation is crucial for initiating the immune responses against intracellular parasites such as N. caninum. However, the different TLR-2 activation did not influence the expression of TNF-α, IL-8, and IL-6, usually triggered by transcription factors also activated by this receptor. Regardless of TLR-2 expression levels, the pro-inflammatory cytokines TNF-α and IL-8 were in general up-regulated upon infection in both cell lines. Surprisingly, IL-6 expression levels were diminished in infected cultures of BCEC-1 and F3 cells. These results deserve further investigation, since a marked IL-6 up-regulation has been described upon N. caninum infection in vivo (16, 93, 132). On the other hand, TGF-β1 was found down-regulated in BCEC-1 and F3 cells. It is postulated that this factor might contribute to fetal death during infections, as it neutralizes the pro-inflammatory responses induced by Th1-type cytokines whose activity is detrimental for the gestation. When immune modulation was individually compared after infection with high- or low-virulence isolates, only differences on TLR-2 and TNF-α expression levels were found, with the Nc-Spain1H isolate inducing higher expression of TLR-2 and TNF-α than the Nc-Spain7. This could be translated into a better control of parasite proliferation by the cells infected with the Nc-Spain1H isolate, or the opposite in those infected with the Nc-Spain7 one (57).

Despite this model has demonstrated to be useful to unravel some aspects of N. caninum pathogenesis, its use dismisses the close interaction of caruncular and trophoblast cells with other immune cells present in the placenta. This is clearly reflected by the lack of detection or down-regulation of cytokines such as IL-17, IFN-γ, IL-4, IL-10, and IL-12, whose relevance in the response against N. caninum in bovine models has been long-established (16, 93, 132). These limitations could potentially be overcome by exogenous stimulation of F3 and BCEC-1 cultures with cytokines such as IFN-γ present at the placenta, or even by co-culturing these cell lines with other relevant bovine immune cells such as macrophages. Additional alternatives such as the use of placental explants are discussed below.

Although the use of classical methods has provided relevant information on the expression and abundance of many genes of interest upon infection with N. caninum, they fail to provide a comprehensive view of the global changes that infection might trigger on host target cells. In this regard, recent advances in high-throughput technologies have provided an opportunity to perform a global analysis of potential changes related to infection using trophoblast cells (58, 86). Infections of F3 cells with high- (Nc-Spain7) and low-virulence (Nc-Spain1H) isolates demonstrated that the parasite modulates processes such as the ECM reorganization, cholesterol biosynthesis and the transcription factor AP-1 network after just 8 hpi. Specific isolate-modulated processes were not identified, although infection with the low-virulence isolate exerted a higher modulation of the host cell (207 differentially expressed genes) compared to the high-virulence one (126 differentially expressed genes) (58). Subsequently, and in order to depict the infection dynamics throughout the tachyzoite lytic cycle, the proteome of N. caninum-infected F3 cells was studied by LC/MS-MS during early infection (8 hpi), parasite multiplication (36 hpi), and egress (56 hpi). Proteome changes revealed protein synthesis, protein turnover, and metabolism as the main host pathways disturbed by infection. Similar findings have been reported for other parasites, as manipulation of host metabolic events determines infection establishment and the initiation of host immune responses (58, 133–135). As observed in the transcriptomic study, infection with the low-virulence isolate Nc-Spain1H exerted a higher perturbation on F3 cell proteome with the exception of those profiles enriched in mitochondrial proteins, which were equally altered by high- and low-virulence isolates. This reflects the importance to govern this organelle during infection, which has been broadly demonstrated not only for N. caninum, but also for T. gondii (136–139). Nevertheless, the levels of perturbation of the host cell depending on the isolate's virulence deserves further investigation, as this could be a strategy to achieve a more successful transmission (higher perturbation by low-virulence isolates), or to avoid detection by the host's immune response (lower perturbation by high-virulence isolates). Unfortunately, there is a lack of similar studies performed in bovine caruncular cells, where the parasite has shown an uncommon behavior not described in other cells lines (49, 140). This could provide valuable information about the interactions between N. caninum and the maternal side of the placenta.