The cyst-forming apicomplexan parasite Toxoplasma gondii is the etiologic agent of toxoplasmosis, a worldwide zoonotic disease that can affect virtually all vertebrates, including approximately one third of humans. Toxoplasmosis is considered one of the most important foodborne and waterborne parasitic diseases in the world (Almeria and Dubey, 2021), causing important economic losses in the livestock sector derived from the parasite induced abortions, especially in sheep and goats, and posing also a risk to public health when infected animals are destined for human consumption (Stelzer et al., 2019). Thus, it represents a perfect example of the One Health concept (Aguirre et al., 2019). Among livestock, T. gondii infections are usually more prevalent in sheep and pigs than in cattle and poultry (Stelzer et al., 2019), being both animal species extremely relevant for toxoplasmosis epidemiology. Indeed, lamb and pork are considered to be the most relevant meat source of T. gondii infection for humans (Belluco et al., 2016, Belluco et al., 2018; Almeria and Dubey, 2021). A range of variables have been proposed as key factors in the development of the different clinical forms of toxoplasmosis, including the strain, life-cycle stage and infective dose of the parasite, the individual host species susceptibility, and the immune status of the host (Mukhopadhyay et al., 2020). Among them, the genetic and phenotypic diversity of T. gondii strains constitutes one of the most important causes of variability in the clinical presentation of the disease.

Recent studies based on whole genome sequencing have defined 16 haplogroups assorted into 6 major clades (Su et al., 2012; Lorenzi et al., 2016). Toxoplasma gondii strains have a clonal structure, particularly in Europe and North America, where the archetypal lineages type I, II and III comprise the majority of isolated strains (Su et al., 2012). In Europe, type II and, to a lesser extent, type III strains, are the dominating populations, both in domestic and wild environments (Fernández-Escobar et al., 2022), while in North America strains belonging to haplogroup XII also predominate in wild animals (Khan et al., 2011). By contrast, African and Asian T. gondii populations remain poorly explored (Chaichan et al., 2017; Galal et al., 2018), and isolates characterized from South America show a greater genetic diversity compared to the archetypal strains, being referred to as non-canonical or atypical strains (Lorenzi et al., 2016). These atypical strains usually exhibit greater virulence and have been associated with more severe clinical symptoms and long-lasting sequelae in humans (Su et al., 2012; Shwab et al., 2016; Hamilton et al., 2019).

Therefore, it is crucial to determine the strain causing infection to better understand the epidemiology of toxoplasmosis, predict its possible clinical consequences and design appropriate control and treatment strategies. To this end, molecular genotyping techniques such as whole genome sequencing (WGS), restriction fragment length polymorphism (RFLP) and microsatellites (MS) have been traditionally applied (Fernández-Escobar et al., 2022), as well as a recently described hi-res next-generation sequencing-based method (Joeres et al., 2024). Although these methods provide high discriminatory power, they possess important inherent limitations. For example, biological samples with enough parasite DNA are required, which usually are only present in tissues with high parasite burdens from clinical or severe cases. In addition, it is necessary to use invasive procedures such as biopsies or lumbar puncture, which are laborious, costly and risky to obtain, and the necessary equipment and procedures required for genotyping are often unaffordable for small laboratories. On the other hand, serotyping methods offer a cost-effective, rapid, sensitive and non-invasive alternative to DNA-dependent techniques, allowing the analysis of subclinical cases (Kong et al., 2003; Sousa et al., 2010; Maksimov et al., 2012a; Arranz-Solís et al., 2019, Arranz-Solís et al., 2021). This technique is based on the detection of antibodies in sera that react against certain segments of antigenic proteins presenting strain-specific polymorphic variations (Dard et al., 2016). Traditionally, these methods have been able to distinguish type II from non-type II strains in humans, mice, cats, sheep, pigs, chicken and turkeys (Kong et al., 2003; Peyron et al., 2006; Sousa et al., 2010; Maksimov et al., 2012b, Maksimov et al., 2013, Maksimov et al., 2018). However, the accuracy of prediction in several of these studies has been low, with many of them relying solely on samples from naturally infected patients or animals where the strain causing infection was unknown. Over the last decade a few studies have tested new peptides using samples from experimental infections performed in mice, cats, rabbits, chicken and turkeys with type I, II and III strains, considerably improving previous results (Maksimov et al., 2013, Maksimov et al., 2018; Arranz-Solís et al., 2019, Arranz-Solís et al., 2021). Of note, by combining different dense granule (GRA) pairs of homologous peptides, Arranz-Solís et al., (2019, 2021) we were able to further differentiate the three clonal lineages from each other and from atypical strains in mouse serum samples. Thanks to this approach, a characteristic and specific fingerprint was obtained for each type of strain, in a similar way to that from genotyping techniques.

Despite all these advancements, several limitations hinder the broad application of serotyping in large-scale epidemiological studies. For example, sequence polymorphisms in antigenic peptides used for serotyping do not always translate into the expected strain-specific reactions (Arranz-Solís et al., 2021). Secondly, it seems that each host species has different reactivity patterns against the same peptides, resulting in a need for individual standardization and characterization of peptides in each host species (McLeod et al., 2012; Maksimov et al., 2013, Maksimov et al., 2018; Arranz-Solís et al., 2021). Finally, there is a critical lack of bona fide and reference serum samples, which hampers a thorough characterization of peptides and reliable comparisons among studies (Arranz-Solís et al., 2019, Arranz-Solís et al., 2019; Sousa et al., 2023). In this sense, beyond mice, T. gondii experimental infections in other animal species are rare on account of growing ethical concerns, high costs and scarcity of appropriate facilities to allocate and handle livestock (Stelzer et al., 2019). In such experiments, frequently only type II or type III strains, the most prevalent strain types in Europe and North America, are used. In addition, serum samples from natural infections where the causative strain has been isolated and/or genotyped are scarce, and this is especially true for infections caused by strains other than type II or III, such as type I or atypical strains. This hinders serotyping studies attempting to differentiate archetypical and atypical strains or simply describing the specific signature for each strain type in other host species. Indeed, most serotyping studies have focused on analyzing samples from experimentally infected mice or naturally infected humans, and the few studies that have evaluated sera from other animals have used only a very limited number of peptides, samples and/or strain types, with inconsistent results (Kong et al., 2003; Sousa et al., 2010; McLeod et al., 2012; Maksimov et al., 2012a, Maksimov et al., 2013, Maksimov et al., 2018; Arranz-Solís et al., 2019, Arranz-Solís et al., 2021).

Considering the influence of the strains in the epidemiology of toxoplasmosis from both an animal and a public health point of view, in the present study we sought to investigate the usefulness of the panel of peptides previously characterized by Arranz-Solís et al. (2021) in sheep and pigs. To this end, we made use of valuable and well-characterized serum samples obtained from previous experimental infections with different type II and III strains, as well as from natural infections in sheep from which the causing strains were isolated and genotyped. Our results may lay the grounds for future large-scale serotyping studies in these and other animal species.

Serotyping can play an important role in determining the most prevalent T. gondii strains in abortion outbreaks, especially in small ruminants, or in animals with meat products destined for human consumption. Therefore, and considering their relevance for the epidemiology of toxoplasmosis, in the present study we aimed at investigating the usefulness of a panel of peptides previously shown to predict the strain type in mouse and human serum samples, now in domestic sheep and pigs (Arranz-Solís et al., 2021). To this end, we used serum samples from both experimental and natural infections where the causative strain was known, employed a wide panel of peptides and performed a peptide ratio strategy to compare homologous versions, which has been shown to mitigate the inherent individual variability in the reactivity of peptides (Arranz-Solís et al., 2021). Our promising results allowed us to suggest practical recommendations for future large-scale serotyping assays in these species that can be used to predict the strain type causing the infection.

Apart from humans, T. gondii serotyping has been previously investigated in other animals, including mice (Kong et al., 2003; Arranz-Solís et al., 2019, Arranz-Solís et al., 2021), cats (Maksimov et al., 2013), chickens (Sousa et al., 2010; Maksimov et al., 2018) and turkeys (Maksimov et al., 2018). To our knowledge, the only attempt in pigs and sheep was performed in 2010 by Sousa et al. In this work, a panel of 11 chicken and 15 pig serum samples from type II and III natural infections where the causative strain was isolated (Dubey et al., 2006; Sousa et al., 2006), as well as field samples from 35 chickens, 29 pigs and 50 sheep for which the strain causing the infection was unknown, were assessed with a panel of 4 GRA6 and GRA7 recombinant peptides consisting of three repetitions of selected polymorphic sequences (Sousa et al., 2008, Sousa et al., 2009). The results obtained with these peptides allowed room for significant improvement as less than 15% of the pig serotypes matched the genotype of the strain, with most of the samples being non-reactive. However, in this work only 4 peptides were used and only the individual OD values were considered for determining the serotype (Sousa et al., 2010). A good serotyping peptide should be reactive enough and have homologous versions for one or more different strains, allowing to compare their reactivity rather than considering each one independently. By relativizing the OD values with that from a whole tachyzoite lysate antigen, a RIPC ratio can be obtained and so determine the predominant peptide in each pair (Arranz-Solís et al., 2021). Based on this criterion, in our study the best peptide pairs to discriminate infections caused by different strains in sheep and pigs were derived from GRA6-213 and GRA6-44 peptides, together with GRA3-38 in sheep and GRA7-224 in pigs.

Although many T. gondii proteins have been used in the serodiagnosis of the disease, only a few have been tested for serotyping purposes. In total, to our knowledge, peptides derived from 62 different T. gondii proteins have been tested up to date for serotyping (details can be found in Arranz-Solís et al., 2019). Among them, polymorphic regions from GRA, ROP, SRS and other diverse proteins such as OWP, Toxofilin and yet-to-be-described proteins (currently annotated as hypothetical proteins) have been assessed. However, GRA-derived peptides have been shown to provide the most promising results in previous studies where peptides from GRA, MIC, SRS and ROP were evaluated (Kong et al., 2003; Maksimov et al., 2013; Arranz-Solís et al., 2021). Hence, it does not come as a surprise that in our initial assessment in sheep samples of the peptides previously characterized in mice and human sera (Arranz-Solís et al., 2021) the best peptides were derived from GRA3, 5, 6 and 7 proteins. After comparing homologous peptides, the GRA6-II-44 vs. GRA6-III-44 ratio was very effective in differentiating infections caused by type II and III strains (especially in sheep), in contrast to what was observed in mice (not antigenic) and humans (not efficient) (Arranz-Solís et al., 2021), confirming that each species might have different reactivity patterns against the same peptides. The GRA3-I/III-43 vs. GRA3-II-43 and GRA6-I/III-213 vs. GRA6-II-214 ratios resulted highly specific in detecting type II infections in sheep, as all the samples with ratios below 1 belonged exclusively to this group, while reactions above 1 were present in type III infections but some also appeared in type II infections. By contrast, GRA7-III-224 vs. GRA7-II-224 ratio showed poor predictive abilities in sheep, especially for type III strains, possibly due to the low reactivity of these 2 peptides with most samples, a fact that was previously reported in mice and human samples (Arranz-Solís et al., 2021). By contrast, this GRA7 ratio was very efficient in predicting the strain type in pig infections. Regardless of the individual reactivity of each peptide, with the combination of the 5 GRA ratios an accurate prediction was achieved in 94% and 90% of the experimentally infected sheep and pig samples, respectively, which confirms that with this strategy the relevance of the individual variability in each animal and sample can be attenuated, as it considers the global vision of the different ratios between homologous versions of peptides.

To evaluate the possible variability depending on the time after infection, we compared the reactivity in several samples from the same animal collected at different time points, ranging from 2 to 9 weeks pi in sheep, and between 3 and 6 weeks pi in pigs. Except for a few samples, particularly in pigs, no substantial differences were found in the global prediction when considering all peptide ratios. However, variations in the reactivity of individual peptides were detected for some animals. This suggests that the time after infection, perhaps related to the antibody levels, could influence the reactivity of these peptides producing a slightly different pattern that in turn may affect the ratio of peptide pairs. Maksimov et al. (2018) reported a gradual reduction in the reactivity of several peptides in experimentally infected turkeys and chickens over time until 9 weeks pi. This pattern was not reproduced in our study. These differences might be explained by the individual variability of the host species regarding antigen recognition, or the intrinsic differences among peptides used in each study. Nonetheless, in our study, and despite the individual differences, the global strain type prediction when considering the 5 peptide pair ratios in the same animal at different time points was not affected, highlighting the importance of using a wide range of homologous peptide pairs for this purpose. Furthermore, since all the sheep and pig serum samples used in our study were obtained after oral infections with oocysts, we could not study the influence of the route of infection or parasite stages in the reactivity and prediction abilities of the tested peptides. As for the influence of the dose of infection, all of our sheep samples belonged to animals infected with 10 oocysts (Sánchez-Sánchez et al., 2019; Vallejo et al., 2023), except for a few that were infected with 1000 oocysts of the TgShSp1 strain (Sánchez-Sánchez et al., 2023). Because the number of the latter samples was low (8), we could not draw any conclusions; however, there did not seem to be a difference in the peptides reactivity when comparing both infection doses. This fact was also observed in the study mentioned above in turkey and chicken samples (Maksimov et al., 2018). In pigs, serum samples were obtained from piglets infected with 1000 or sows infected with 400 oocysts, belonging to different strains. In general, although slightly better prediction rates were observed for piglets infected with 1000 oocysts, no significant differences were observed in the reactivity pattern between these groups. However, since different doses were not used within the same type of animals at the same age or under the same conditions, it is difficult to draw conclusions from this fact. Indeed, it is possible that the higher prediction rate observed in piglets was rather a consequence of their younger age, when a more controlled immunological status would be expected in comparison to adult pigs, which are more likely to have been exposed to other antigens that might elicit unspecific antibodies that react against the peptides.

When we took a closer look at the results obtained with each of the 3 different type II strain sheep serum samples, two different patterns, corresponding to each of the ToxoDB genotype numbers, were observed. Namely, the TgShSp1 and TgShSp16 strains belonging to the ToxoDB #3 genotype (Pru-like) showed very similar patterns, while the ME49 ToxoDB #1 genotype elicited a slightly different response, with general lower reactivity levels. Although Pru and ME49 share the same amino acid sequence in all the segments studied from GRA3, GRA5, GRA6 and GRA7 proteins (Arranz-Solís et al., 2021), the detected reactivity was not identical. It is therefore possible that the presence of immunogenic epitopes in other segments of the proteins might elicit different reactivity in strains that share the same sequence in the epitopes where the peptides are located (Arranz-Solís et al., 2019). However, because we only had 4 ME49 samples, this result needs to be confirmed in future analysis using a larger number of samples and a higher variety of type II strains. In pigs, on the other hand, no significant differences were observed when comparing the two different type II and two different type III strains beyond a slightly better prediction rate for GRA6-I/III-213 vs. GRA6-II-214 and GRA3-I/III-43 vs. GRA3-II-43 ratios for the ShSp1 type II and the ShSp24 type III infected piglets, respectively. Since the ShSp1 and CZ-Tiger belong to the same ToxoDB genotype #3, and the ShSp24 and CZ-Šimková belong to the same ToxoDB genotype #2 (López-Ureña et al., 2023), similar reactivity patterns were expected. Nonetheless, as stated above, it is possible that these slight differences in prediction rates can be explained by the type of animal infected rather than the isolate itself, as piglets, which showed the better prediction rates, were infected with the ShSp1 and ShSp24 strains (Largo-de la Torre et al., 2022).

Although not as high as experimental infections, the accuracy rate in predicting the strain type in sheep samples from natural infections caused by type II and III strains showed moderate levels (75%), with the GRA6-II-44 vs. GRA6-III-44 and GRA3-I/III-43 vs. GRA3-II-43 ratios being the most and the least reliable ones, respectively. Samples obtained from natural infections possess inherent limitations, as there exist uncontrolled parameters that might influence the reactivity against peptides, such as the immunological status, infection dose, time post-infection at sampling, cross-reactivity with other pathogens or multiple current or past infections with other T. gondii strains, among others. Hence, it is expected that these samples are challenging to assess, especially considering the absence of sera obtained from sheep infected with type I, or even atypical, strains, and the scarcity of sera obtained from the more common type II and III infections where the strain was isolated and/or genotyped (Sousa et al., 2010; Fernández-Escobar et al., 2022). Despite these limitations, and even though in our study only 18 type II and 2 type III samples were available, reactivity profiles could be determined, which would in turn be useful to perform future comparisons with other samples and infer whether those infections are caused by type II, III or other strains showing matching or different reactivity patterns.

In pigs, although no samples from naturally infected animals were studied, our serotyping results from experimental infections with 4 different type II or III strains looked promising, especially the combination of GRA6-44, GRA6-213/214 and GRA7-224 peptides, which showed to be very efficient in differentiating infections caused by type II and III strains, improving the single previous attempt performed by Sousa et al. (2010). Contrary to that observed in sheep samples, the GRA7-III-224 vs. GRA7-II-224 ratio showed good results and correctly predicted the strain in 35/40 samples. By contrast, pig samples were low reactive against GRA5-II-38, showing a poor prediction accuracy. In addition, since we used the peptides previously selected in sheep for the pig sera assessment, it is possible that some of the peptides we ruled out in sheep might have worked in pigs and could be worth including in future assays in these and other animals. For example, despite being lowly antigenic, ROP8-I/III(II)-305 peptides could be an interesting option, as our initial assessment revealed ratios that were low but correct in 3/4 sheep samples. Similarly, as mentioned above, GRA6-44 peptides showed much better results in sheep and pigs compared to mice and humans (Arranz-Solís et al., 2021), supporting the notion that peptides that do not work in some animal species might in others and vice-versa. It is therefore crucial to refine and standardize serotyping trials for each animal species, as there are inherent differences and reactivity patterns that might not match previously described ones. Moreover, our results highlight the need for including further samples from animals infected with other strains, including type I and other less common but more clinically and epidemiologically relevant atypical strains, so as to identify potential strain-specific signatures that might allow to define patterns in samples from field conditions.

Finally, two of the GRA5 and GRA6 epitopes in our peptides have a single different version for each of the archetypal strains, I, II and III, in contrast to the other peptides that have biallelic variations, usually I/III and II. Regarding GRA6-44, we only included in the final panel of peptides the type II and III versions, as only serum samples from animals infected with type II and III strains were available. Given the promising results obtained with this peptide ratio, it is tempting to speculate that samples from animals infected with type I strains would show a higher reactivity with the GRA6-I-44 version when compared to the type II and III versions. Thus, in future experiments it might be worth including all the possible peptide versions to compare the reactivity patterns obtained here for type II and III infections. As for the GRA5-38 peptides, only the I and II version were used, as the type III version could not be obtained after several attempts in the past where the lyophilized peptide could not be resuspended (Arranz-Solís et al., 2021). If new peptides batches are ordered in the future, new solvents, such as urea, could be tried to resuspend this peptide and compare its reactivity in sheep, pig, or other species serum samples.

Altogether, our findings support the results previously described by Arranz-Solís et al. (2021) regarding the usefulness of GRA peptides in the serotyping of infections caused by T. gondii. We have significantly advanced in the serotyping of two of the most important animal species for T. gondii infection epidemiology, sheep and pigs, showing a promising approach that might be useful for future epidemiological studies and to apply a similar strategy in other relevant animal species. In this sense, it is crucial to initially validate the reactivity and usefulness of peptides in each animal species, so as to select the most antigenic and strain-specific ones to be used. A practical recommendation for future serotyping epidemiological studies in sheep and pigs is to include at least GRA6-213, GRA6-44 and GRA3-43 peptide pairs for the former and GRA7-224, GRA6-213 and GRA6-44 for the latter, as well as testing at least 2 samples from the same animal to confirm the obtained results. Finally, further investigation could be conducted to assess the specificity of these peptides against samples from animals infected with other related apicomplexan parasites, such as Neospora caninum or Sarcocystis spp., or even other common pathogens in sheep and/or pigs, to investigate the possible cross-reactions that might interfere with the interpretation of the T. gondii strain type prediction.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The animal studies where these serum samples were collected were approved by Local and National Animal Welfare committees following proceedings described in EU legislation (Council Directive 2010/63/EU). These previous studies were conducted in accordance with the local legislation and institutional requirements under the following protocol numbers: 416-2016 and 1063/2021 for sheep, and PP 55/2016 and PROEX 293.7/20 for pigs.

DA-S: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing. LT: Investigation, Methodology, Writing – review & editing. ET: Investigation, Methodology, Writing – review & editing. NL-U: Investigation, Methodology, Writing – review & editing. BK: Resources, Writing – review & editing. MF: Supervision, Writing – review & editing. LO-M: Conceptualization, Funding acquisition, Resources, Supervision, Validation, Writing – review & editing. GÁ-G: Conceptualization, Funding acquisition, Resources, Supervision, Validation, Writing – review & editing.