Exploration of immunoglobulin transcriptomes from mice immunized with three-finger toxins and phospholipases A₂ from the Central American coral snake, Micrurus nigrocinctus

Introduction

Snakebite envenomings represent a major public health concern in tropical regions of the world (Williams et al., 2011). Despite emerging discoveries that may one day pave the way for novel biotechnology-based antivenoms (reviewed by Laustsen et al., 2016a; Laustsen et al., submitted), animal serum-derived antivenoms remain the cornerstone of snakebite envenoming treatment (Gutiérrez et al., 2011). Production of antivenom is challenged by a large variation in immunogenicity of many key snake venom toxins resulting in unpredictable immune responses in production animals (Cook et al., 2010; Guidolin et al., 2010). It has been shown that many of the immunogenic venom components are in fact not important for toxicity (Antúnez et al., 2010; Gutiérrez et al., 2009; Laustsen et al., 2015), and conversely, that some highly toxic venom components, such as α-neurotoxins (both short and long neurotoxins), phospholipases A₂, and P-I snake venom metalloproteinases may be poorly immunogenic (Schottler, 1951; Gutiérrez et al., 2009; Chotwiwatthanakun et al., 2001; Ownby & Colberg, 1990; Judge et al., 2006; Wong, Tan & Tan, 2016; Tan et al., 2016; Leong et al., 2015; Tan et al., 2015). Combined, this creates a challenge for antivenom production, since the goal of obtaining an antivenom with a strong, yet balanced response against all the medically relevant toxins becomes a complex endeavor.

Coral snakes (genera Micrurus, Leptomicrurus, and Micruroides) are, together with the sea snake Hydrophis (Pelamis) platura, the representatives of the snake family Elapidae in the Americas, comprising approximately 85 species (Campbell & Lamar, 2004; the Reptile Database—www.reptile-database.org). Although Micrurus species are only responsible for about 1-2% of snakebite cases in this continent, roughly corresponding to 750 to 1000 cases per year, envenomings by these snakes can be fatal if not treated properly and timely (Warrell, 2004; Gutiérrez et al., 2016; Bucaretchi et al., 2016). Envenomings resulting from coral snakebites are predominantly associated with descending neuromuscular paralysis, which may end in respiratory arrest (Warrell, 2004; Bucaretchi et al., 2016).

Production of antivenoms against Micrurus snakes is particularly challenging, as (a) it is very difficult to maintain coral snakes in captivity (Chacón et al., 2012); (b) the majority of Micrurus species provide a very low yield of venom, implying that the collection of the quantities of venom required for horse immunization and quality control testing demands the ‘milking’ of many specimens (Chacón et al., 2012; Bolaños, 1972); and (c) there is a variable extent of immunological cross-recognition between venoms from coral snakes of different species; hence, antivenoms raised against some species are not always effective in the neutralization of venoms of other species (Bolaños, Cerdas & Abalos, 1978; Tanaka et al., 2016). As a result, only a few laboratories manufacture Micrurus antivenoms, and several countries where these snakes inhabit completely lack this therapeutic resource, e.g., Venezuela, Ecuador, Peru, Bolivia, the Guyanas, and Paraguay, which severely limits the clinical management of these accidents.

Knowledge on the composition of the venoms of Micrurus species has increased steadily over the last years, as a consequence of proteomic characterizations (reviewed by Lomonte et al., 2016b). Two main venom phenotype patterns have been identified, i.e. venoms rich in neurotoxins of the three-finger toxin (3FTx) family, and venoms rich in phospholipases A₂ (PLA₂s) (Fernández et al., 2015). In addition to these two main protein families, other minor components of these venoms include L-amino acid oxidases, serine proteinases, metalloproteinases, nerve growth factor, C-type lectin-like proteins, Kunitz-type inhibitors, among others (Fernández et al., 2011; Fernández et al., 2015; Corrêa-Netto et al., 2011; Lomonte et al., 2016a; Sanz et al., 2016; Rey-Suárez et al., 2011; Rey-Suárez et al., 2016). In some cases, the toxins playing the main role in overall toxicity have been identified, these being 3FTxs and PLA₂s (Rey-Suárez et al., 2012; Vergara et al., 2014; Fernández et al., 2015; Castro et al., 2015; Ramos et al., 2016).

The limited immunogenicity of the highly toxic PLA₂s and 3FTxs (Fernández et al., 2011; Rosso et al., 1996; Alape-Girón et al., 1996) represents another difficulty in production of Micrurus antivenom, since it thwarts the goal of raising a balanced immune response against these medically relevant toxins. In order to further explore how these toxins interact with the mammalian immune system, we chose a mouse model and employed an NGS approach using the AbSeq™ technology developed by AbVitro (now Juno Therapeutics, https://www.junotherapeutics.com), based on Illumina sequencing (Fig. 1). The methodology was utilized to sequence immunoglobulin (Ig) encoding mRNA transcripts from splenic B-lymphocytes in mice subjected to immunization with either a 3FTx or a PLA₂ toxin from the venom of M. nigrocinctus (Central American coral snake). By this approach, the transcription levels of different immunoglobulin isotypes and dominant clones of B-lymphocytes with a particular usage of V (variable) and J (joining) gene segments can be determined for Ig heavy chain transcripts. This methodology has previously been employed for investigating B-cell populations in autoimmune (Stern et al., 2014) or infectious diseases (Tsioris et al., 2015; Di Niro et al., 2015), for example. By employing the AbSeq™ high-throughput approach, we explore, for the first time, the Ig transcriptome including VJ usage patterns in individual animals subjected to immunization with two relevant toxin classes of elapid snakes. Although exploratory in its nature and somewhat limited by a small sample size, this study thus provides novel insight into the humoral response of mice immunized with 3FTx or PLA₂ toxins and highlights important challenges of raising antibodies against poorly immunogenic toxins.

Materials and Methods

Results and Discussion

Three-finger toxins (3FTx) and phospholipase A₂s (PLA₂) are the two most abundant toxin families in the venom of M. nigrocinctus (Fernández et al., 2011), and generally they are the two snake toxin families which have been most investigated (Laustsen et al., 2016a). In the venom of M. nigrocinctus these toxins cause neuromuscular paralysis, owing to a combination of pre- and post-synaptic actions, and myotoxicity, providing the venom with its high toxicity (Rosso et al., 1996; Alape-Girón et al., 1996). In previous studies it was observed that 3FTxs and PLA₂s were recognized more weakly than larger proteins from this venom, by a therapeutic equine antivenom (Fernández et al., 2011). Despite their low immunogenicity, it was possible to raise an antibody response against both toxins in four out of five mice, although high variation in the antibody titer was observed (Fig. 2). Mice immunized with PLA₂ had a higher antibody titer than mice immunized with the 3FTx, in agreement with the higher molecular mass of the former.

Assessment of the mRNA from harvested mouse splenocytes indicated that it was of sufficient quality to proceed to sequencing (RIN^e scores between 5.2 and 6.4). A high-throughput sequencing approach (AbSeq™) was employed to investigate transcription levels of Ig isotypes and the usage of V and J gene segments for heavy chain assembly in mice that were immunized with a 3FTx or a PLA₂. Investigation of the 50 most frequent VJ combinations for the immunized mice did not, however, result in identification of a dominant combination, as the VJ usage was found to be similar across all samples (Fig. 3). This finding suggests that the generated antibody responses might be diverse and that multiple specific antibodies with low abundance are generated in each mouse.

Looking at the sequences of all mRNA transcripts encoding heavy chain variable domain (V_H) clones across each sample, we were able to find shared V_H clones with similar relative abundances in either the PLA₂-immunized or the 3FTx-immunized mice (Fig. 4). In comparison, almost no V_H clones were shared between mice immunized with different toxins (Fig. 5). This implies that the immunization procedure did indeed elicit specific, but different responses dependent on whether PLA₂s or 3FTxs were employed for immunization. The relatively high number of V_H clones found in both of the PLA₂-immunized mice (Fig. 4A) compared to lower number of V_H clones found across the three 3FTx-immunized mice (Figs. 4B–4D) further indicate that immunization with PLA₂s is more prone to give rise to antibodies transcribed in similar quantities. Also, an intermediate number of similar V_H clones was found in both the PLA₂-1 and 3FTx-3 samples (Fig. 5E), even though the correlation in relative abundance was not equally pronounced. This is likely explained by the fact that the majority of V_H clones found in both PLA₂-immunized mice are not expected to be specific towards the toxins, but instead are likely to be directed against other (environmental) antigens that the mice have encountered throughout their life.

The AbSeq™ antibody sequencing methodology is capable of determining the Ig isotype of the identified V_H clones. The coloring of the V_H clones in Figs. 4 and 5 reveals that a large number of the most abundant V_H clones present in the mice are of the IgA isotype, which was confirmed by further investigation of all mRNA transcripts from the splenocytes (Fig. 6). This is surprising, as IgG is known to be the dominant immunoglobulin class in mouse blood after the response to T-dependent protein antigens. The observation could possibly be explained by differences in expression levels due to different translation rates and half-lives of mRNA transcripts encoding different immunoglobulin isotypes. All approved antibody-based therapies on the market are based on IgGs (Walsh, 2014), which are also the desired isotype for antivenoms. In immunized horses for antivenom production, two isotypes of IgG are largely responsible for the neutralization of toxic effects in the case of viperid snake venoms (Fernandes et al., 2000). However, analysis of the transcripts obtained from the immunized mice revealed only a low percentage of IgG transcripts, as compared to the transcripts for other Igs (Figs. 4–6). With the acknowledgement of the small sample size employed in this study, this finding may indicate a difficulty in raising potent IgG antibodies against both 3FTxs and PLA₂s in rodents. If similar difficulty is present in horses, this may therefore have implications for antivenom production. Our results may further suggest that the immune response is slightly lower against 3FTxs than for PLA₂s based on the lower abundance of IgG transcripts in mice immunized with 3FTx (Fig. 6). Taken together with the results from the ELISA assay (Fig. 2) and the observation that immunization with PLA₂s is more prone to give rise to similar Ig transcripts (Fig. 4A vs. Figs. 4B–4D), we suggest that the PLA₂ toxins may possibly be slightly more immunogenic than the 3FTx, although neither toxin seems to have high immunogenicity. The underlying reason for this could possibly be due to the smaller molecular size of 3FTx compared to PLA₂s, or that PLA₂s may contain distinct epitopes better capable of eliciting an adaptive immune response than 3FTxs. Predictions of possible B-cell epitopes using the protein sequences and BepiPred 1.0 (Larsen, Lund & Nielsen, 2006) or DiscoTope 2.0 (Kringelum et al., 2012) with structural models of the investigated toxins as input do not indicate a major difference in the bare number of possible epitopes. However, the suggested difference in immunogenicity is further indicated by the fact that only two IgG-encoding mRNA transcripts are found in the top 20 most abundant Ig-encoding mRNA transcripts for only one out of three of the 3FTx-immunized mice. In comparison, six and nine of the top 20 mRNA transcripts for mice immunized with PLA₂s encode the IgG isotype (Fig. 7). It would be interesting to assess whether the immune response of horses against these elapid venom toxins is also characterized by a low proportion of IgG—a finding that would have evident implications for antivenom manufacture, as IgG has been shown to be the antibody isotype of therapeutic value (Fernandes et al., 2000). However, this is beyond the scope of this exploratory study.

Concluding Remarks and Outlook

In addition to demonstrating the utility of high-throughput sequencing technology, AbSeq™, for investigation of immune responses in animals immunized with snake venom toxins, the exploratory findings presented here may indicate possible difficulties in obtaining an IgG response against the medically important toxins of the 3FTx and PLA₂ families from M. nigrocinctus. Given that these proteins play key toxic roles in envenomings by elapid snakes, this underlines a drawback of current antivenom production based on immunized animal serum, since IgG has been shown to be the antibody isotype of therapeutic value (Fernandes et al., 2000). Although based on a small exploratory sample size, these findings therefore contribute to the understanding of snake toxin immunogenicity and indicate the possible difficulty in obtaining balanced immune responses in animals during the immunization process.