Until now, mortality in mice has been considered the main parameter for the virulence evaluation of T. gondii strains, and it was established as the ratio between casualties and the number of infected animals challenged with the strain under study. Different subspecies of mouse have been used for T. gondii infection in vivo modelling, such as Swiss Webster, CD-1, C57BL/6, BALB/c or Kunming strains, among many others (Wang et al., 2013a; Taniguchi et al., 2018; Fukumoto et al., 2020; Uzelac et al., 2020). Toxoplasma gondii infection has been widely studied using cell-culture derived tachyzoites (or zoites grown in the mouse peritoneal cavity) that are intraperitoneally (IP) or subcutaneously (SC) inoculated into naïve laboratory mice (Howe et al., 1996). Despite not constituting a natural infection route, this model has the advantages of reproducibility, ease of inoculation, and accurate administration of challenge dose (Sibley et al., 1999). However, variants of this procedure using other parasite stages have been also implemented, including IP injection of bradyzoites contained in tissue cysts (Taniguchi et al., 2018; Gatkowska et al., 2019), or per os (PO) inoculation of tissue cysts (McLeod et al., 1984; Khan et al., 2007; Taniguchi et al., 2018; Arcon et al., 2021) and oocysts (Yang et al., 2017; Chiebao et al., 2021). These variants are inherently less reproducible due to the variable bradyzoite content of a given tissue cyst, or the difficulties in guaranteeing the oral dosage, as well as less feasible due to the complexity of oocysts production; but on the other hand, oocyst- or bradyzoite-induced infections are much more representative of what occurs in nature (Sibley et al., 1999). Consistent with the T. gondii life cycle, oocyst-mediated infections are known to be more pathogenic than bradyzoite- and finally, tachyzoite-induced infections (Dubey et al., 1977; Dubey et al., 1981; Dubey, 2006; Saraf et al., 2017). Logically, the doses tested (number of parasites inoculated) vary greatly between experimental designs depending on the parasite life cycle stage used, but they also do so between experiments that use the same parasite stage. All these mentioned factors (the route of inoculation, the parasite stage, the dose, and the host species or even subspecies) along with the number of passages in mice or cell culture, have been demonstrated to drastically affect the degree of parasite virulence (Saraf et al., 2017). Currently, there is a greater awareness that it is important to keep passages of evaluated isolates low before they become lab adapted, but the use of strains maintained for a long time under laboratory conditions (“laboratory strains”) or for which the passage number is not known still remains widespread in literature (Khan et al., 2014). Therefore, despite the numerous studies published on virulence evaluation ( Supplementary Table S1 ), the large diversity of conditions and methodologies implemented, with heterogeneous interpretations and fragmented data, makes it difficult to draw conclusions about the real T. gondii population structure in terms of virulence.

After initial attempts for normalization (Su et al., 2002; Taylor et al., 2006), the only available standard operating procedure for the evaluation of cumulative mortality rates was recently published (Saraf et al., 2017). According to the authors, cumulative mortality rate calculation implies the use of outbred mice (e.g., Swiss Webster [SW] or CD-1 mouse strains), at least three consecutive doses of IP inoculated tachyzoites, and the recording of casualties among those successfully infected animals by day 28 post-inoculation (dpi). It should be pointed out that in some contexts, animal welfare regulations prevent getting ethical approval when mouse assays include several inoculation groups. In most publications, mortality rate in mice is the only parameter evaluated, which implies an overly simplistic and narrow view of virulence. Until Saraf’s publication, there was no consensus in the literature about how the mortality rate in mice should be calculated, so it is frequent to find lethality estimations/assumptions based on animal casualties during isolation procedures (Pena et al., 2006; Clementino Andrade et al., 2013; Shwab et al., 2016; Vilares et al., 2017), and even valuable attempts have been made to quantify the parasites in the inoculum (Mercier et al., 2010). Despite being a substantial attempt to standardize the procedure, some limitations should be pointed out. A parameter originally valued in the strain’s virulence assessment was the median lethal dose (LD₅₀; Probit tests: https://probitanalysis.wordpress.com/2016/07/07/first-blog-post/; Finney, 1971; Shittu et al., 2020), especially since the discovery of absolute lethal doses (LD₁₀₀) of a single parasite in the case of the strains related to haplogroup 1 (Sibley and Boothroyd, 1992; Su et al., 2002; Salman et al., 2021). However, this calculation has fallen out of favour, similarly to duration of survival post infection analyses (Wang et al., 2013a; Costa Viegas de Lima et al., 2019). Accurately determined, these parameters could offer new insights into dose-dependency and infection dynamics.

The assessment of virulence in mice for a large number of T. gondii strains worldwide is contained in literature. The present section aims to critically examine the available phenotypic data, defined as mortality in mice, from T. gondii isolates worldwide, trying to provide an overview of the virulence profiles of T. gondii populations in the different continents. The PubMed database (https://pubmed.ncbi.nlm.nih.gov/) was searched combining the terms “Toxoplasma gondii”, “virulence characterization” and “pathogenicity”; 2644 investigations published until December 2021 were found. Only T. gondii isolate virulence studies involving assays measuring tachyzoite stage induced mortality in mice were considered, while studies focusing on in vitro assays or other parasite stages (oocysts or bradyzoites) were excluded. Infections based on these other stages account for only a minority of investigations, which is further limited by methodological variations (see previous section). Other rodent in vivo models (e.g., rats) were not included. Data from laboratory strains (e.g., RH, GT1, CTG, ME49, PRU, VEG) were not covered (except when obtained shortly after strain isolation), in order to better describe the real phenotypic diversity of the T. gondii populations globally. In total 62 studies were selected, involving 311 isolates (see Supplementary Table S1 ). Isolate IDs, host of origin, available genetic (e.g., ToxoDB#, ROP18/ROP5 alleles) and geographical (e.g., country, continent) features, experimental conditions for mouse mortality assessment, and other evaluated parameters, were extracted for each reference. Due to the heterogeneity of data presentation, mortality rate was re-calculated in each reference (when possible) based on Saraf et al. (2017) criteria (analysis of three sequential inoculation dosages, with the lowest dose resulting in only partial infection of the animals), and isolates were classified into “Highly virulent” (100% mortality rate), “Intermediate virulent” (99–30%) and “Non-virulent” (<30%) categories according to Su et al. (2002). Calculations considering 4 or more doses when available were also included. In the Figure 1 , only studies in which mortality in outbred mice was assessed implementing at least 3 doses of IP or SC inoculated tachyzoites (serial 10-fold dilutions from 1 to 10⁶ parasites/mouse), and 28-dpi animal monitoring were filtered (Saraf et al., 2017) (33 studies; 204 isolates), in order to map the T. gondii virulence diversity more accurately by increasing methodological homogeneity. Data is geographically biased; it is worth noting the sizable proportion of South American (including Caribbean) isolates (n=130) compared to the scarcity of those from Europe (n=17), North America (n=22) or Asia (n=34) and the extremely poor representation of the African continent (n=1). Generally, the T. gondii population in South America is mostly considered highly virulent, in association with a notably diverse, endemic genetic structure (Shwab et al., 2014; Shwad et al., 2018). However, present data show a proportion of South American highly virulent isolates that are not more common than those found in North America (37% vs. 45%) ( Figure 1A ). In both regions, almost one in four strains proved to be non-virulent (22-23%). In contrast, European isolates are broadly considered non-virulent, in relation with the prevalent genetic clonality found in the continent (involving mainly genotypes II and III) (Shwab et al., 2014; Shwab et al., 2018). Nonetheless, present data showed a not insignificant percentage of highly virulent strains in Europe (18%) ( Figure 1A ). Interestingly, the global/worldwide T. gondii virulence in mice distribution displays a certain balance between highly-, intermediate- and non-virulent strains (33%, 35% and 32%, respectively) ( Figure 1A ). Collected virulence data distribution within the different hosts of origin is also strongly biased; the vast majority of isolates assessed were obtained from domestic animals (n=129/204), while only 18 were isolated from wild animals ( Figure 1B and Figure S1 ). Ignoring this fact, phenotypic data extracted showed an almost equal proportion of the different virulence degrees in strains infecting animals. In detail, isolates from domestic animals seem to be slightly less virulent than isolates from wildlife, which could correspond with the apparent more virulent character of “wild” strains vs. “domestic” strains described in North America (Jiang et al., 2018). Data fromhuman isolates (n=57/204) is partially biased as most of them came from clinical cases, which could explain that only 5% of the strains evaluated showed a non-virulent profile, while more than a half presented a highly virulent character ( Figure 1B ). Nevertheless, although a certain phenotype is probably being selected in isolates of human origin, it should be pointed out that there is no clear link between virulence in mice and in humans. On the other hand, researchers deal with an extra issue when it comes to human isolates due to the uncertainty of the geographical origin of some infections, with importation or migration related cases, only solvable by comprehensive epidemiological surveys.

Present data collection probably provides a view of T. gondii worldwide population phenotypic diversity more realistic than previously described (Shwab et al., 2016; Shwab et al., 2018) based mainly on mortality rates calculated during the isolation process. The bioassay of digested tissue samples in mice cannot be considered as a method for virulence evaluation due to the lack of parasite quantification or control of inoculum composition, the assessment of a single “dose” and the widespread use of a very low numbers of mice (typically 3). Nonetheless, it should be noted that since standardized models are relatively expensive assays as they involve the use of a higher number of experimental animals (quite restrictive and regularized), it is common for the strains subjected to these experiments to have been pre-selected due to their more disparate behavior, which may result in an extra bias specially in underrepresented continents such as Europe. Data extracted from standardized mouse models ( Supplementary Table S1 ) identified 40.1% of highly virulent strains in the southern hemisphere, while this accounted for 25.3% in the northern hemisphere. This strongly contrasts with figures described in the Americas by Shwab et al. (2018), where 61% of 427 isolates from South/Central America were categorized as highly virulent in mice, while only 7% of 193 isolates from North America were found to be virulent.

Little is known about how parasite virulence in mice extrapolates to other relevant hosts. Since the first characterization studies of T. gondii strains, there has always been an interest in associating virulence in mice with an outcome in human clinical infections (Sibley and Boothroyd, 1992). However, there are only a few comparative studies available, which indicate that drastically different infection outcomes occur in different hosts challenged with the same isolate (Taniguchi et al., 2018; Hassan et al., 2019; Sánchez-Sánchez et al., 2019; Xia et al., 2020).

Current animal welfare policies aim to minimize the use of laboratory animals, replace them when possible, and to refine the experiments to reduce animal suffering. In this regard, important improvements can be introduced for the evaluation of virulence using the mouse model by assessing additional non-lethal parameters involved in virulence ( Table 1 ). A much more comprehensive view of the virulence of T. gondii strains should combine the study of infection effects on the host (e.g., lethality and tissue lesions) with other aspects inherent to the parasite’s own fitness. The oocyst production rate is a direct indicator of the parasite’s transmission success, and it has been studied in several publications in the past (Dubey et al., 2002; Dubey et al., 2003). The demonstrated loss of the capacity to produce oocysts, as a consequence of successive passages in cell culture or mice, is considered as a loss of virulence (Frenkel et al., 1976; Lindsay et al., 1991; Dubey et al., 1999). In particular, a study of oocyst shedding in domestic cats of several genetically diverse T. gondii strains from French Guiana showed difference of fecundity among them, which may potentially affect their transmission (Khan et al., 2014). However, the use of cat models has long been controversial and currently is not accepted by the wide scientific community, in ethical terms. Another useful way to measure parasite transmissibility is to evaluate parasite cystogenesis, the capacity and capability of T. gondii to form tissue cysts. The central nervous system (CNS) is the tissue par excellence to study the development of cysts during the chronic phase of the infection but the T. gondii tropism towards immune privileged organs also involves ocular or muscle tissues (Jiang et al., 2020; Yang et al., 2020; Dubey, 2021). Cysts can be quantified and even measured (diameter) by microscopic observation of fresh/unstained (Dubey et al., 2012) and immunostained fixed brain sections or brain homogenates (Masatani et al., 2020; Wang and Sibley, 2020). An indirect way to quantify the presence of cysts in the mouse brain is to measure the parasite burden in the CNS from 3 weeks pi by quantitative PCR (Fernández-Escobar et al., 2020; Fernández-Escobar et al., 2021; Salman et al., 2021). A literature review (Watson and Davis, 2019) compiled and described T. gondii experimental latent infections in murine models for the quantification of brain cysts, in order to find key factors on data variance and to propose optimized protocols; however, the conclusions were not as informative as expected because of the fragmentation of data gathered. Some non-archetypical strains showed decreased potential to develop into bradyzoites in vitro, bradyzoites with decreased resistance to pepsin treatment, and formation of a lower number and smaller tissue cysts, associated with a limited oral transmission (Fux et al., 2007).

As a reflection of virulence in vivo, parasite burdens in different organs have been determined after short times post-inoculation in many studies. Apart from CNS, parasite burdens in lung, spleen, kidney, liver, ocular tissues, mesenteric lymph nodes, diaphragm or even blood has been also studied in literature (Zenner et al., 1998; Djurković-Djaković et al., 2012; Hill and Su, 2012; Hill et al., 2012; Wang et al., 2013b; Hamilton et al., 2019; Fernández-Escobar et al., 2020; Fernández-Escobar et al., 2021). Real-time PCR was used to monitor the distribution of T. gondii in different murine tissues during the infection with 10² or 10⁶ tachyzoites (IP) of the RH strain, or 10 cysts (PO) of the ME49 strain (Djurković-Djaković et al., 2012); this study concluded that the level of parasite burden in the lungs seemed to be critical for mice survival/parasite virulence. This result agrees perfectly with what was observed previously (Loeuillet et al., 2019), where pneumomegaly, and the greater parasitic load and tissue destruction in the lungs were strongly associated with the hypervirulence of the non-canonical GUY008-ABE isolate (haplogroup 5, ToxoDB # unknown). On the other hand, some studies pointed out ocular tropism as a virulence related trait, with a more frequent parasite DNA detection in ocular tissues from mice infected with more virulent strains (Hamilton et al., 2019; Chiebao et al., 2021; Fernández-Escobar et al., 2021). These findings are in agreement with data from human ocular toxoplasmosis in Brazil, where lesions in the retina are the most common clinical manifestation (Dubey et al., 2012). From a practical point, all these parasite distribution assays could be carried out in parallel to mortality rate evaluation, by selecting a dose of interest for tropism evaluation (Fernández-Escobar et al., 2021). Histological and immunohistochemical examinations for describing pathological lesions in tissues after T. gondii infection have been carried out in several studies, and even though they are considered essential to demonstrate the definitive cause-effect in clinical diagnostic reports, for virulence assessment this technique has less discriminatory power (Yang et al., 2017; Hamilton et al., 2019; Fernández-Escobar et al., 2020; Chiebao et al., 2021).

Another interesting approach to assess differences in virulence of T. gondii strains is to study differential expression of cytokines and other markers of the immune response during infection. The levels of mRNA expression in spleen and mesenteric lymph nodes of IFNγ, IL-12, T-cells surface markers CD8, CD4 and CD25, as well as the receptor adapter MyD88 and the chemokine receptor CXCR3, were evaluated and compared during mice infection with the archetypal non-virulent M4 strain (type II variant, ToxoDB #3), a non-archetypal virulent strain (genotype BrI, #6) and a non-archetypal intermediate virulent isolate (genotype BrIII, #8) (Chiebao et al., 2021). In accordance with previous literature, the authors associate a strong and acute Th1 immune response (IFNγ, IL-12 higher levels) with highly lethal strains, whereas a longer and modulatory Th2 immune response was triggered by moderately virulent isolates (TLR-MyD88, CXCR3 expression). Similarly, serum levels of IFNγ, TNF-α, IL-12 p40, IL-10, IL-6, IL-4, and IL-2 were evaluated as virulence markers for the severity of toxoplasmic encephalitis (Araujo and Slifer, 2003).

The cumulative morbidity rate or the severity of clinical signs are additional non-lethal virulence parameters considered in literature. It requires trained personnel, cohesion in the evaluation criteria and, at best, a normalized clinical scoring criterion (Sánchez-Sánchez et al., 2019; Fernández-Escobar et al., 2020; Chiebao et al., 2021; Fernández-Escobar et al., 2021; Pena et al., 2021; Salman et al., 2021). The main clinical signs associated with T. gondii infections in mice are (in increasing order of severity) piloerection (ruffled coat), ascites, loss of body weight/condition, prostration, dyspnoea, and neurological signs such as motor incoordination, head tilting, or circling motion. The cumulative morbidity could be calculated based on the number of mice that present any clinical sign after inoculation divided by the number of infected mice (Fernández-Escobar et al., 2020). Another strategy to quantify such a subjective aspect is the representation of the daily scoring mean value variation and standard deviation according to the clinical findings; both aspects were successfully evaluated for mice experimentally infected with type II (#3), BrI (#9) or BrIII (#8) genotypes elsewhere (Chiebao et al., 2021).

In some Danish studies, levels of the acute-phase reactant haptoglobin in serum were considered for evaluation as an additional virulence parameter along mice infection with T. gondii (Jensen et al., 1998; Jungersen et al., 1999; Jungersen et al., 2002). Apparently, strains that caused more severe body weight loss also induced the highest serum haptoglobin and specific anti-T. gondii antibodies concentrations during the acute phase of the infection. Weight loss and anti-T. gondii IgG response have been used in other studies although, similarly to haptoglobin levels, not extensively, probably due to the excessive handling of animals required and a weak correlation with mortality rates (Kannan et al., 2010; Bezerra et al., 2019; Masatani et al., 2020; Salman et al., 2021; Yang et al., 2021). Interestingly, some studies also approached the analysis of behavioral changes in mice infected by different T. gondii reference strains, through the evaluation of learning and memory, locomotor activity, spatial working and aversion to feline odour, among other parameters (Kannan et al., 2010; Bezerra et al., 2019); however, such complex assessments require special devices to monitor mice activity. The study of behavioral changes as phenotypic markers associated with infection could be of special interest since the relationship between toxoplasmosis and human neurological disorders and psychiatric illnesses such as schizophrenia and bipolar disorder has been repeatedly demonstrated (Xiao et al., 2013; Chaudhury and Ramana, 2019; Wang et al., 2019).

Despite attempts to describe and incorporate new non-lethal virulence parameters, the main problem we face is the lack of homogeneity in the selection and in the experimental conditions between studies. Time points of infection, procedures and analytical methods vary completely between different investigations ( Supplementary Table S1 ). However, despite this lack of consensus protocols, it is undeniable that there is a growing interest, as shown within the literature, in going beyond the mere mortality rate calculation in the assessment of virulence for different T. gondii strains using the mouse model.

An additional approach to virulence degree prediction can be achieved based on molecular analyses. The CS3 locus has been described previously as a highly predictive marker of mortality in mice challenged with T. gondii isolates (Pena et al., 2008); high mortality rates associated with the type I or II alleles of the CS3 region, and low or null rates associated with the type III alleles have been reported (Pena et al., 2008; Wang et al., 2013b; Rocha et al., 2018). However, several subsequent findings contradicted such observations (Langoni et al., 2012; Rêgo et al., 2017; Fernández-Escobar et al., 2020). Currently, ROP18 and ROP5 are well-known virulence factors in Toxoplasma virulence in mice (Behnke et al., 2015; Xia et al., 2021), and their allelic combination was proposed as highly predictive of virulence in mice across globally distributed T. gondii isolates (Dubey et al., 2014; Shwab et al., 2016). As detailed in above studies, some allelic combination such as 2/2 or 4/4 has been found firmly associated with certain mortality degrees (0% or 100% lethality, respectively); however, other genetic combinations such as 3/1 or 4/3 are much less predictive, and ultimately, the combination 3/3 is the most unspecific profile due to its association with levels of mortality strongly varying from 100 to 0% (Shwab et al., 2016; Hamilton et al., 2019; Uzelac et al., 2020; Bernstein et al., 2021; Fernández-Escobar et al., 2021). In summary, although molecular tools seem useful to predict the virulence degree to some extent, additional genetic factors might be also involved. It should be taken into account that despite the large database used for these correlations at a global level (Shwab et al., 2016), with up to 240 records, it included mouse mortality data calculated at the time of strain isolation by bioassay in mice. In the light of gathered data ( Supplementary Table S1 ), a re-assessment of genotype-phenotype correlation, considering only mice mortality data obtained from standardised mouse infection models, could yield more reliable results. A continuous collection of ROP18 and ROP5 allelic types and mouse virulence data will facilitate future studies to identify other virulence genes, and ultimately improve the prediction power of genotyping for virulence.

The current animal welfare policies not only highlight the necessity to refine the in vivo procedures but also to minimize and replace the use of laboratory animals. In this context, the use of in vitro models represents an excellent alternative for the study of intracellular organisms such as T. gondii. The in vitro models allow the study of the host cell infection process by the tachyzoite stage, namely the lytic cycle, which mimics the dissemination of the parasite during the acute phase of the infection (Black and Boothroyd, 2000). The lytic cycle of T. gondii is a tightly regulated process, which includes adhesion to the host cell, invasion, PV formation, multiplication, and egress steps (Sibley, 2010). It is important to note that in vitro experiments reflect the behavior of the parasite in the absence of selective pressures during an infection of a host, which explains that the results may easily not correspond to what occurs in an in vivo assay. In the case of T. gondii, given its enormous plasticity represented in a vast host range, in vitro experiments can provide a less biased, host-centered view, although it will always be necessary to contextualize the results (Poulin and Combes, 1999).

Proliferative stages of the parasite have been cultured in vitro employing a variety of cell culture lines (e.g., HeLa, Vero, HFF, BeWo) and primary cell cultures (Scheidegger et al., 2005; Müller and Hemphill, 2013), and among them, target cells or tissues (e.g., trophoblast and nervous cells, dendritic cells [DCs] or macrophages) should be highlighted (Guimarães et al., 2008; Dellacasa-Lindberg et al., 2011; Mammari et al., 2014; Witola et al., 2014; Barbosa et al., 2015; da Silva et al., 2017; Pacheco et al., 2020). Most of publications on T. gondii implementing in vitro assays are focused on safety and efficacy assessment of potential antiparasitic drugs (Basto et al., 2017; Murata et al., 2017; Radke et al., 2018) or on demonstrating the role of different host and parasite effectors in the T. gondii lytic cycle (Camejo et al., 2014; Bai et al., 2018; Guo et al., 2019; Wang et al., 2020). In vitro models are considered also suitable first approaches to phenotypically characterize apicomplexan parasite strains (Regidor-Cerrillo et al., 2011; Dellarupe et al., 2014; Frey et al., 2016; Jiménez-Pelayo et al., 2017; García-Sánchez et al., 2019). However, only a small proportion of the publications addresses the virulence characterization of non-laboratory T. gondii isolates in vitro (Loeuillet et al., 2019; Sánchez-Sánchez et al., 2019; Bernstein et al., 2020; Uzelac et al., 2020; Fernández-Escobar et al., 2021; Salman et al., 2021).

A systematic review of literature (Contreras-Ochoa et al., 2012) compared studies that had used mouse and human glial cell cultures to determine T. gondii invasion and replication rates in these cells. The wide experimental heterogeneity found hampers drawing definitive conclusions but type II strains (ME49 [#1] and PRU [#3]) seem to be less invasive of nervous system-derived cells than type I (RH [#10] and BK [genotype # unknown]) strains. Several publications have characterized some in vitro virulence parameters of the par excellence T. gondii strain RH in different cell types, and it has been extensively used as an experimental control. However, the RH strain has been maintained and passed through lab mice or cell culture for several decades and its biological behaviour has drastically changed (Khan et al., 2009). Comparative studies of other different laboratory strains have been also published (Appleford and Smith, 1997; Diana et al., 2004; Fux et al., 2007; Lambert et al., 2009; Cañedo-Solares et al., 2013; Mammari et al., 2014). Overall, it is generally claimed that type I strains present enhanced proliferation capacity and lower host immune system stimulation than type II isolates (Mammari et al., 2014). Regarding the in vitro virulence assessment of non-laboratory (field) isolates, the number of studies is relatively low; however, its use as a complement to the evaluation of virulence in mice has increased in recent years (Loeuillet et al., 2019; Sánchez-Sánchez et al., 2019; Bernstein et al., 2020; Fukumoto et al., 2020; Uzelac et al., 2020; Fernández-Escobar et al., 2021; Salman et al., 2021).

Reviewing the T. gondii literature revealed that in vitro phenotypic evaluation is mostly based on parameters such as parasite invasion rate, proliferation kinetics, tachyzoite yield (TY), or on the assessment of plaque formation, tachyzoite-bradyzoite conversion and spontaneous cyst-formation (Regidor-Cerrillo et al., 2011; Li et al., 2014; Loeuillet et al., 2019; Sánchez-Sánchez et al., 2019; Uzelac et al., 2020; Fernández-Escobar et al., 2021; Salman et al., 2021). Data on the use of the main virulence parameters in recent remarkable investigations are summarized in Table 2 . Other interesting but less extended in vitro parameters can be found in the literature. For example, in a Chinese study (Zhang et al., 2013) activation and polarization of macrophages after infection with a highly virulent or a mildly virulent ToxoDB genotype #9 strains were studied in primary BMMφs and peritoneal Mφs from mice and Raw 264.7 cells (mouse macrophages), showing completely opposite phenotypes. In a complex study (Barragan and Sibley, 2002), in vitro migration of 20 T. gondii strains was measured using a transwell system based on polarized Madin-Darby canine kidney (MDCK) and human foreskin fibroblast (HFF) cell monolayers co-cultures, as well as in HFF cell monolayers covered with agarose. The transwell system consisted of the upper chamber with polarized MDCK and HFF cell monolayer on a filter and the lower chamber with a HFF cell monolayer. Freshly released parasites were added to the upper chamber of the transwell system and migration of parasites to the lower chamber HFF monolayer was quantified. Both approaches revealed a superior migratory capacity of Type I overtype II and type III strains. These in vitro assays need to be tested with more T. gondii strains to confirm their significance in predicting parasite virulence in the future.

One of the key issues is that there is no consensus in experimental conditions (e.g., multiplicity of infection [MOI], number of passages, time points for infection, cell culture lines, methods of analysis, among others), yielding non-comparable results ( Table 2 ) (Contreras-Ochoa et al., 2012).

In vitro assays offer only a partial view of the processes triggered during infection, since the interaction between the different cell types that shape an organ or the immune system response, for example, are not reflected. Meanwhile, ex vivo models that preserve the cellular architecture and function of certain organs (e.g., placenta) during the in vitro study period might be an interesting alternative (Fry et al., 2019). Noteworthy studies on the proliferation of T. gondii in explants and organoid-derived monolayers have been carried out (Scheidegger et al., 2005; Robbins et al., 2012; Ander et al., 2018; Holthaus et al., 2021). Practical applications of placental explants in the study of host (human)-parasite (T. gondii) interactions ex vivo were reviewed recently (Pastor-Fernández et al., 2021), but until now no study has been focused on virulence evaluation of T. gondii strains using ex vivo approaches. An investigation that deserves attention was carried out by Robbins et al. (2012), in which the preferential infection of different structural components of the placenta and the proliferation capacity in a human placental explant model were evaluated for three reference strains of T. gondii of different genetic types (I, II, and III); nevertheless, no statistical significance was observed.