In São Paulo state, rabies caused by the domestic dog genetic lineage (AgV2) has not been detected in dogs and cats since 1999 (Kotait et al, 2001), but these animals continue to become infected, albeit sporadically, with lineages circulating in bats, mainly the Desmodus rotundus/Artibeus lituratus, (compatible with AgV3) (Castilho et al. 2017; 2018), which is endemic in São Paulo State, and probably throughout Brazil.
Desmodus rotundus are also responsible for the absolute majority of rabies cases in herbivores, which are infected by hematophagous bats with rabies at the time they feed on these animals.
Unlike herbivores, dogs and cats, due to their predatory habits, can contract rabies from bats other than hematophagous bats, including the frugivorous bat Artibeus lituratus, which is also a reservoir of the RABV strain D. rotundus/A. lituratus, and to date, it is not possible to infer, through genetic sequencing of these isolates, which of these bats acted as the most likely source of infection of these infections (Castilho et al., 2018).
Research has already demonstrated a possible genetic differentiation between RABV isolates from D. rotundus and A. lituratus but has not discussed whether these groups would be specific lineages or sub lineages within the D. rotundus/A. lituratus lineage, nor demonstrated the use of these results in rabies epidemiological surveillance in dogs and cats infected with this lineage (Kobayashi et al., 2007, Fahl et al., 2012).
Based on the results of research studying different Brazilian RABV strains using the N gene, it is observed that to date, isolates considered to be from the same lineages generally do not vary more than 5% in nucleotide identity (Carnieli et al., 2013; Souza et al., 2017, Oliveira et al., 2010; Macedo et al., 2010), and this value of gene identity based on the N gene is also suggested by Nadin-Davis (2020) for the separation of RABV lineage.
For all concatenated RABV genes, according to Table 2, a maximum nucleotide identity variation of 4.8% was found, while for the N gene, the maximum variation found was 4.4%. Thus, if we use the gene identity values based on the N gene, there would be no sense in classifying the clusters found in this research in distinct lineages of RABV as suggested by Fahl et al. (2012), because the variation between them based on the N gene had a maximum value of 4.4%, thus belonging to the same lineage of RABV.
As for the analysis of the calculations of genetic identities of nucleotides in relation to the samples sequenced in this study, the following result is observed in increasing order of genetic conservation in mean values: G > L > M > N > P. In absolute values, the degree of conservation of genes would decline in the following order L > N = G > M > P (Table 1).
As for the analysis of amino acids in mean values we have the following order of variability of proteins N = L < M < G < P. In absolute values, the degree of variation of proteins would be in the following order L > N > M > G > P.
Oliveira (2014) conducted a phylogenetic study of RABV based on complete genomes using a very significant sample of RABV strains distributed worldwide, where it was obtained the following result regarding the genetic identities related to the five RABV genes: N > M > L > P > G. In the same research, the author demonstrated that if RABV is divided into two large groups, in which the author called aerial and terrestrial cycles of RABV, these values varied according to the group analyzed. In this case the result of the mean identities obtained for the aerial cycle was N > M > L > P > G, while for the terrestrial cycle was N > M > L > G > P. This shows that depending on the way the data is analyzed or categorized, different values of genetic identity can be obtained.
Although the N gene was shown to be the third most divergent gene on average in the Desmodus rotundus/Artibeus lituratus lineage, in terms of amino acids, the N protein along with the L protein were shown to be on average the most conserved proteins among the RABVs of this lineage. Protein N is considered the most conserved protein in RABV species (Wunner & Conzelmann, 2020), but this is the first study to demonstrate that protein L may be as conserved as protein N, again showing that some features of nucleotide and peptide identities considered to be the consensus for RABV as a whole may vary in more specific analyses.
The functional and organizational constraints imposed on RABV proteins make the same to be under intense purifying selection (Wunner & Conzelmann, 2020; Streicker & Biek, 2020; Oliveira, 2014; Kuzmina et al., 2013), which causes most of their nucleotide mutations to be synonymous, i.e. mutations encoding the same amino acid. Thus, it is possible that the third most variable gene on average in the strain studied in this research (gene N) encodes the most conserved protein (Tables 1 and 2).
Although purifying selection is the dominant type of selection in RABV proteins, studies show that there is a higher tolerance for amino acid changes in P and G proteins compared to other RABV proteins, causing nonsynonymous mutations to be fixed in larger numbers in these genes (Streicker & Biek, 2020). This could explain the fact that in this research the G gene proved on average to be the most conserved gene in the Desmodus rotundus/Artibeus lituratus lineage while corresponding to the second most divergent protein in this lineage.
In the recombination analysis, no evidence of this phenomenon was found in the isolates studied here, agreeing with results obtained by other authors (Kuzmin et al., 2012; Oliveira et al., 2020). Recombination between different RABV lineages is a rare event, few studies have presented events of this nature, all in China involving specific dog lineages (Liu et al., 2011; He et al., 2012; Ding et al., 2017).
As for topology, the Maximum Cluster Credibility phylogenetic tree demonstrated that the isolates sequenced in this study clustered into four clusters (Fig. 1), which were arbitrarily named according to the order of divergence in the phylogenetic tree and the main host found in each cluster. It can be suggested, according to the results found, that in São Paulo State, the Desmodus rotundus/Artibeus lituratus lineage is currently subdivided into at least four sub clusters named Desmodus 1, Desmodus 2, Artibeus 1 and Artibeus 2 (Fig. 1).
An interesting fact observed in the phylogeny obtained in Fig. 1 were the two independent and not directly related divergences, which gave rise to the two sub lineages associated with the species Artibeus lituratus. This shows that these two sublines diverged at different times and from different ancestral viruses, and according to the data of this study, the Artibeus 1 sub lineage seems to have been the first to diverge in the state of São Paulo, being perhaps the most recent common ancestor of the other 3 sub lineages circulating in the state of São Paulo. These two differentiation events can be observed in a subtle way in the research of Fahl et al. (2012), however the authors of this paper did not observe or discuss this fact.
As for the remaining groupings, it is noted that the Desmodus 1 sub lineage was probably the second to diverge, and interestingly, it is noted that the Desmodus 2 and Artibeus 2 lineages were the last to diverge directly from the same ancestral RABV.
In the analysis of each cluster, it was observed that in Desmodus 1 sub lineage only isolates associated with the hematophagous bat Desmodus rotundus were found.
In cluster Artibeus 1 there was a predominance of isolates from Artibeus lituratus. However, we found an isolate from a feline from Jaguariúna, an isolate from Desmodus rotundus from the municipality of Ubatuba and an isolate from a bovine from the municipality of São Carlos (Fig. 1), which can be considered very likely spillover events Artibeus lituratus → D.rotundus (Fahl et al, 2012, Oliveira et al, 2010, Menozzi et al, 2017).
In Desmodus cluster 2 isolates were found from Desmodus rotundus, equines and bovines, without the occurrence of probable spillover events.
In the Artibeus 2 cluster, as observed in the Artibeus 1 cluster, an isolate from a bovine was clustered that was also most likely due to an Artibeus lituratus → D.rotundus spillover event.
Although the results regarding the differentiation of the four sub lineages within the Desmodus rotundus/Artibeus lituratus RABV lineage circulating in the state of São Paulo found in this research are quite robust based on the high posterior and bootstrap values of the source nodes of each cluster, these results should not be extrapolated to other regions of the country due to the fact that little is known about the characterization of RABV isolates from the frugivorous bat Artibeus lituratus in these regions (Kobayashi et al, 2007). Furthermore, a preliminary phylogenetic analysis (data not shown) using 1281nt of the N gene from the samples in this study together with 191 other isolates from Brazil and Latin America show that several RABV isolates from D. rotundus and herbivores from other regions of Brazil cluster in the Artibeus 1 and Artibeus 2 sub lineages, and to infer whether this is due to spillover events or circulation of these sub lineages in D. rotundus in these regions, further research is needed.
In other Neotropical countries, there are no data on molecular characterization of RABV isolates from the genus Artibeus, however, serological studies demonstrate the existence of neutralizing antibodies to RABV in these bats, which may be indicative of the maintenance of rabies epidemiological cycles in these animals (Salmón-Mulanovich et al., 2009; Zieger et al., 2017, Seetahal et al., 2020). As such, genetic sequencing of RABV isolates from this genus of bats is critical for the characterization of lineages or sublineages associated with this genus of bats in these countries.
The divergence time of the most recent common ancestor (ACMR) of the Desmodus rotundus/Artibeus lituratus RABV lineage, according to the data of this research, was around 170 years, ranging from 107 to 237 years with 95% confidence interval, however this value is underestimated, since all the evolutionary signal of ancestry relationships of other sub lineages within this lineage, as well as of the different RABV lineages maintained in bats is not present in this analysis (Oliveira et al, 2020).
Oliveira et al. (2020) demonstrated that the five RABV genes evolve at different rates, a phenomenon known as heterotachy. Thus, using different RABV genes for evolutionary analysis may lead to different results regarding the rate of nucleotide substitution per year and time of most recent common ancestor.
Oliveira et al. (2020) using the concatenated G and L genes and a fairly representative sampling of the worldwide RABV distribution, arrived at an average divergence time of approximately 260 years for the D. rotundus/A lituratus lineage. Kuzmina et al. (2013) using the G gene, obtained the result of 309 years.
It is worth noting that the evolution of new genus or species-specific RABV lineages or the adaptation of the same lineage in different bat genera from spillover events depends on several factors, including the frequent contact of these animals and the phylogenetic proximity existing between the transmitting bat and the host (Streicker et al., 2010; Streicker et al., 2012; Mollentze et al., 2014). The bats Desmodus rotundus and Artibeus lituratus, besides belonging to the same family (Phyllostomidae) also share the same shelter types, a fact that corroborate the adaptation of the same lineage of RABV in these two genera (Greenhall et al., 1983; Arellano-Sota, 1988).
Thus, it can be inferred that currently in São Paulo state, the RABV lineage D. rotundus/A. lituratus is subdivided into 4 sub lineages, two of them associated with the hematophagous bat D. rotundus and the other two associated with the frugivorous bat A. lituratus. The use of a representative sample from a large geographic region, together with the phylogenetic signal provided by the analysis of all concatenated RABV genes provide subsidies for these findings.
In the phylogenetic analysis performed from a 799 nucleotides fragment of the N gene of RABV, the same one used in the routine molecular surveillance of rabies in the state of São Paulo, it is demonstrated the recovery of the same 4 clusters relative to the sub lineages described above, with high bootstrap values in their origin nodes, which demonstrates the possibility of differentiating these clusters without the need for modification of the RABV sequencing protocols currently used.
Interestingly, the two RABV isolated from cats present in this research clustered in the Artibeus 1 and 2 lineages, which, according to the results presented here, may suggest that the source of infection in these cases was most likely the frugivorous bat A. lituratus, demonstrating the great usefulness of this research for rabies molecular surveillance in São Paulo State.