Comparison of serological and molecular methods for differentiation between genotype A and genotype B strains of small ruminant lentiviruses

Small ruminant lentiviruses (SRLV) are a group of viruses of the Retroviridae family causing multisystemic, degenerative and chronic disease in sheep and goats. The viruses are divided into five genotypes (A, B, C, D and E) and at least 34 subtypes (A1–A27, B1–B5 and E1 and E2) (31). This group of viruses includes maedi-visna virus (MVV), typically infecting sheep, and caprine arthritis encephalitis virus (CAEV). Respectively these two viruses define genotype A (MVV-like strains) and B (CAEV-like strains). The A and B genotypes are the most widespread in goat and sheep populations worldwide, whereas the other three genotypes are restricted to limited geographical areas (12, 16). Strains belonging to genotype A were detected in goats and genotype B strains in sheep, confirming the possibility of interspecies transmission of SRLV (23, 28–31). Genotype B is considered the most virulent genotype for goats, while genotype A seems to be more pathogenic in sheep (2, 6, 24). In Poland, SRLV occur quite commonly, with herd-level seroprevalence of 61% and 33.3% in goats and sheep, respectively (19, 33). Genetic studies identified two genotypes (A and B) and 12 subtypes (A1, A5, A12, A13, A16, A17, A18, A23, A24, A27, B1 and B2) circulating in Poland. Updated knowledge of the different circulating genotypes could raise the effectiveness of control programmes, it being difficult at present to completely eliminate the viruses.

The main target cells for SRLV are monocytes and macrophages. Following the transformation of its RNA to DNA by the viral reverse transcriptase, the SRLV genome integrates as a provirus into the monocyte and macrophage genomes. The SRLV genome is comprised of three structural genes and three accessory genes. The structure is formed by gag (encoding capsid, matrix and nucleocapsid proteins), pol (encoding reverse transcriptase, protease and integrase) and env (encoding the surface and transmembrane protein), and information for the synthesis of proteins that regulate viral replication is carried by vif, rev and vpr-like. Additionally, non-coding long terminal repeats (LTRs) are located at both ends of the integrated (proviral) linear DNA (24).

Infection with SRLV persists throughout life and there is no treatment for it or vaccine against it. Therefore, the use of control programmes is the only way to avoid the spread of SRLV infection. These programmes rely on diagnostic tools to identify positive animals so that they may be eliminated. Therefore, the use of an appropriate test is crucial for the effective prevention and control of SRLV infections. Detection of SRLV infections is most commonly achieved using serological methods that detect antibodies to SRLV, while detection of the integrated provirus in infected monocytes and macrophages can be accomplished using PCR strategies. The two most commonly used tests for detecting specific antibodies against SRLV are agar gel immunodiffusion and enzyme-linked immunosorbent assays (ELISA). Agar gel immunodiffusion is highly specific, but its sensitivity is low, so it is increasingly being replaced by ELISA because of the immunosorbent assay’s good sensitivity, objectivity and ability to be automated. In addition, it is easy to perform and results are obtained quickly. Most diagnostic ELISAs use recombinant capsid and/or transmembrane proteins as antigens (27). More recently, real-time PCR assays have been developed for quantitative, sensitive, rapid and large-scale detection of SRLV (3, 5, 9, 10, 18, 20). However, despite advances in diagnostic techniques for detecting SRLV, there is no test capable of detecting all strains. This is mainly due to the high genetic and antigenic heterogeneity of these viruses. Many of the currently available diagnostic tests are still based on a monostrain format (genotype A or B), as it was assumed that the antigens of either strain can detect antibodies against both MVV and CAEV in infected animals’ sera (14, 38). However, it has been observed that ELISAs are more sensitive and specific when homologous antigens are used rather than heterologous ones (21, 37). The low cross-reactivity between genotype-mismatched SRLV antigen and antibody pairs starkly limits the diagnostic performance of monostrain ELISAs in a population where animals are infected with an SRLV genotype different from the one used in the test (6, 21, 37, 39). The high heterogeneity of the SRLV genome also hinders the usefulness of PCR to detect all SRLV strains, so it is suggested that PCRs should rather be developed based on strains circulating in a given area. Therefore, information on circulating genotypes would be helpful in selecting appropriate tests, especially in areas where genetic testing has not been conducted.

Current tools for SRLV characterisation include partial region sequencing and heteroduplex mobility assays, but these methods are quite complicated, take a long time and are not suitable for routine diagnosis (11, 28). Recently, the initial classification of A and B SRLV genotypes has become achievable by ELISA. To date, two such ELISAs have been developed, one based on the matrix and the other on the capsid epitope (17, 25). In addition, ELISAs based on the variable SU5 protein can be used as serotyping tools to provide information on SRLV subtypes, because the SU5 epitope is considered subtype specific (7, 25). Unfortunately, almost all of these ELISA tests have been developed in-house by non-commercial organisations and are not available worldwide. There is only one commercial test (the Eradikit SRLV Genotyping ELISA offered by In3 Diagnostic, Turin, Italy) that can distinguish between the A, B and E SRLV genotypes. However, its serotyping efficiency has not been well defined. In addition to serological tests, PCR protocols have been recently developed to detect the A and B genotypes and distinguish between them. This method uses genotype-specific primers and probes to detect genotype-specific nucleotide sequences (1, 10, 20, 40).

With the intention of simplifying the serotyping of SRLV field isolates and thereby improving the effectiveness of control programmes, this study investigated the serotyping efficiency of the Eradikit SRLV Genotyping ELISA test and the molecular typing efficiency of the newly developed real-time PCR targeting the LTR-gag region using samples originating from animals infected with known SRLV subtypes circulating in Poland.

It has been shown that the amino acid sequences of one immunodominant epitope of the capsid antigen are not conserved and have a variable region specific for each SRLV genotype. Therefore, ELISAs based on this region can be tools for classifying SRLV genotypes circulating in the field (13, 17). In this study, we evaluated the serotyping efficiency of the Eradikit SRLV Genotyping ELISA kit, the only commercial test based on capsid antigens able to distinguish between the A, B and E SRLV genotypes. For this purpose, a panel of 97 serum samples from animals infected in Poland with known SRLV isolates representing subtypes A1, A5, A12, A13, A16, A17, A18, A23, A24, A27, B1 and B2 was tested.

Our results revealed that the percentage of samples with their genotype assignment by sequencing in agreement with the assignment by serology was 42.3. This value is higher than the one obtained by Acevedo Jiménez et al. (1), who obtained only 26% of results consistent between sequencing and serological typing using the Eradikit SRLV Genotyping kit. However, the present value was much lower than that achieved by Nogarol et al. (26), who showed that 98.23% of genotyped sera were correctly classified using capsid antigens derived from genotypes A, B, C and E. The research performed by Acevedo Jiménez et al. (1) and Nogarol et al. (26) revealed that ELISAs based on the capsid antigens were unable to serotype 40.2% and 30.0% of analysed seropositive samples, respectively. The unserotypeable samples gave OD values lower than 0.4 for all antigens. We found that 19.6% of tested sera had low (OD < 0.4) reactivity with the antigens used in the Eradikit SRLV Genotyping kit, and their results were considered inconclusive. This can be explained by the basis of the kit on a single epitope representing one specific subtype; therefore, the weaker reactivity of sera from animals infected with different subtypes is not surprising. Even single amino acid substitutions can significantly affect the antigenicity and thus the genotypic specificity of the serologic response. The low reactivity of positive sera against immunodominant capsid epitopes may also suggest a late stage of infection, when antibodies to the p25 SRLV core protein are not always detectable. It is known that antibodies against capsid antigens are detectable early after infection and tend to decrease later, while antibodies against the envelope predominate at later stages of infection (20–33 weeks after infection) (4, 8, 21, 22, 39). Therefore, in many ELISAs antigens from the capsid and the envelope are used together to identify seropositive animals at all stages of infection. Furthermore, it should also be noted that the tested samples were archival and had been thawed and refrozen several times, which could have affected the stability of the antibodies.

It was observed that 20.6% of the tested sera gave indeterminate genotype results because they contained antibodies which reacted highly with all test antigens, and the reaction strengths with individual antigens differed too little to permit discrimination. This result was similar to that obtained by Nogarol et al. (26), who showed that the genotype of 17.07% of analysed positive sera was indeterminate using capsid antigens. This outcome may indicate the carriage of more than one SRLV subtype in a given animal; however, the real-time PCR did not confirm any co-infection. Higher reactivity of sera with both genotype A and B antigens may indicate co-infection with MVV and CAEV, and such a phenomenon was observed in sheep and goats in Poland earlier (28–31). However, it is difficult to explain the high reactivity of sera to all three antigens (A, B and E) observed in 60% of the indeterminately genotyped samples from this study, especially since genotype E has never been detected in Poland. Highly heterogeneous genotype E was identified only in Roccaverano goats in mainland Italy and Sarda goats on Sardinia (36). Therefore, these positive reactions are more likely to indicate the presence of nonspecific reactions or the occurrence of cross reactions. Despite the C-terminal epitopes of the P25–B3 subunit of genotype B (RRNNPPPP), genotype A (VRQNPPGP) and genotype E (MRQNPQPP) being different, all three genotypes’ N-terminal epitopes are similar (LNE/KEAER/TW); therefore, cross-reactions are possible (17, 26). Such cross-reactivity was also observed when matrix and SU5 antigens were used (6, 15, 40). We also observed that 75% of indeterminate sera in this study originated from goats, which is in line with previous reports indicating that the spectrum of antibody reactivity is wider in goats than in sheep (17).

Nogarol et al. (26) found that only 1.5% of samples were incorrectly classified using capsid antigens derived from genotypes A, B, C and E. Acevedo Jiménez et al. (1) noted 7.6% of samples to be misclassified using the Eradikit SRLV Genotyping kit. Our results included 17.5% of sera incorrectly classified by the Eradikit ELISA. These differences may be due to the sera sample panels having been different in these studies. Nogarol et al. (26) tested samples from animals infected with the A1, A8, A9, B1 and B2 subtypes. Acevedo Jimenez et al. (1) tested sera from animals infected with the A1, A2 and B1 subtypes. In our study, sera from animals infected with SRLV subtypes A1, A5, A12, A13, A16, A17, A18, A23, A24, A27, B1 and B2 were analysed, and most of these subtypes (A12, A13, A17, A18, A23, A24 and A27) are found only in Poland. We noted that most of the misclassified samples (82.3%) were genotype A-infected sera which reacted most strongly with an antigen derived from a genotype B strain. These observations differ from those of Lacerenza et al. (21), who suggested that genotype B strains may induce an antibody response against more conserved epitopes of the Gag protein. This phenomenon is not fully explained and requires further research. A further impediment to the specificity of the Eradikit ELISA worth mentioning is that the kit’s recombinant proteins are fused to glutathione S-transferase, and some animals have antibodies that react with the GST portion of the fusion protein, which can lead to false-positive reactions (8).

The differences in the sequence of the gag gene also allowed the development of PCR tests differentiating genotypes A and B. Acevedo Jiménez et al. (1) developed a nested PCR based on the gag gene–encoded capsid protein which attained 55.0% sensitivity. Kuhar et al. (20) developed a real-time PCR based on the gag gene–encoded matrix protein which showed 79.0% sensitivity, while Michiels et al. (2018) developed a real-time PCR also based on the gag gene which achieved 83.3% sensitivity in goats and 88.0% in sheep. The primers used in the nested real-time PCR performed in this study allow amplification of the LTR-gag fragment in order to detect either genotype A or genotype B SRLV. This real-time PCR was developed by Schaer et al. (40), who estimated that its sensitivity was 75.5%. Our study resolved the sensitivity of this test to be 80.2%. The less than 100% sensitivity may be due to the storage of the archival samples tested in this study for a quite a long time in the freezer, which may have reduced the amount of proviral DNA. However, the MVV and CAEV assays were highly sensitive, with a detection limit of five plasmid copies per reaction, being more sensitive than the real-time PCR designed by Kuhar et al. (20), which detected 100 copies per reaction. In addition, the samples used in the study were from a different pool of samples than those used for sequencing, so the negative results may be due to SRLV compartmentalisation. It can be assumed that the reason is not SRLV diversity, since the negative samples came from animals infected with subtypes A12, A13, A17 and A27, of which more than 75.0% were detected. Only subtype A1 was poorly detected, 25.0% of such samples having been identified, which may indicate the weaker ability of this method to detect this subtype.

In addition, Schaer et al. (40) indicated that all real-time PCR–positive samples were correctly classified as MVV or CAEV. Our results confirmed these findings. The MVV assay was able to detect various SRLV strains belonging to genotype A (subtypes A1, A5, A12, A13, A16, A17, A18, A23, A24 and A27), and the CAEV assay was able to detect subtypes B1 and B2. When genotype A strains were tested with CAEV-specific primers and probe, and genotype B strains were tested with MVV-specific primers and probe, no amplification was observed. Thus, the nested real-time PCR used in this study showed 100% specificity. When Kuhar et al. (20) used the real-time PCR developed by themselves, similarly no cross-reactivity was observed when samples from animals infected with the B1, B2, A1, A3, A4, A5, A14 and A15 subtypes were tested. These results indicate that these real-time PCR methods are reliable tools for accurately differentiating genotype A (MVV) from genotype B (CAEV). However, in order to validate the PCR applied here for international use, samples representing all other subtypes should be tested.

Our results showed that the Eradikit SRLV genotyping kit is not a reliable method for predicting SRLV genotype. Cross-reactivity was noted, many samples could not be serotyped, and some were misclassified. Unlike this ELISA, the nested real-time PCR based on the LTR-gag region proved to be a very good tool for error-free differentiation of the A and B genotypes. This preliminary classification of SRLV genotypes can help in the selection of ELISAs or the design of new PCR assays to be used in a given area for detecting SRLV.