Abstract

1. Introduction

Canine distemper virus (CDV) (Canine morbillivirus) belongs to the genus Morbillivirus, family Paramyxoviridae and is a pathogen distributed worldwide. CDV infection is often associated with severe disease in domestic dogs, but it has also been identified in wild species, such as carnivore and marine mammals, and non-human primates (Martella et al., 2008; ICTV, 2019). CDV infections may compromise several organs and systems, leading to respiratory, dermatological, gastrointestinal and central nervous system (CNS) signs, the latter frequently resulting in neurological sequelae (Rendon-Marin et al., 2019).

The CDV genome is composed by a single-stranded, negative-sense RNA molecule of approximately 15,700 nucleotides (nt) in length. The viral genome contains six genes, which encode eight viral proteins (Sidhu et al., 1993; Nguyen et al., 2016). The nucleocapsid (N) and matrix (M) genes encode proteins with structural functions. The phosphoprotein (P) gene encodes the P protein, which together with the large (L) protein (L gene), is involved in transcription and replication of the viral genome. The P gene also encodes V and C proteins, which participate in pathogenesis, virulence and immune evasion mechanisms. The hemagglutinin (H) and fusion (F) genes encode the glycoproteins located in the viral envelope, with roles in receptor binding and membrane fusion, respectively (Lamb and Parks, 2013).

Glycoprotein H is the main target for neutralizing antibodies and, as such, undergoes strong selective pressure from the host immune response, displaying the highest sequence variability among the CDV genes (Harder and Osterhaus, 1997; Mochizuki et al., 1999). This variability has encouraged the use of H as a target for phylogenetic studies, which, in turn, may lead to classification of CDV into distinct lineages (or genotypes) (Nguyen et al., 2016; Jo et al., 2019; Piewbang et al., 2020). Hence, CDV isolates/strains sharing less than 95% amino acid (aa) identity in the H protein have been classified into distinct lineages (Mochizuki et al., 1999; Loots et al., 2018; Duque-Valencia et al., 2019a). Furthermore, it is suggested that sequences with at least 98% nt identity may be clustered into the same subgenotype (Budaszewski et al., 2014; Costa et al., 2021).

Several CDV lineages have been described based on phylogenetic analysis of the H gene, considering nt or aa sequences. In general, lineages are named according to their region of identification and/or distribution, e.g., America (or North America) 1, Asia 1, Europe 1 (or South America 1), Arctic Like and Caspian (Jo et al., 2019; Duque-Valencia et al., 2020; Piewbang et al., 2020; Costa et al., 2021). When more than one lineage is identified in the same region, they are distinguished by cardinal numbers, e.g., Asia 1 and Asia 2 (Lan et al., 2007). Furthermore, viruses of the same lineage identified in different regions are named after both regions, e.g., Europe 1/South America 1 (Duque-Valencia et al., 2020).

Considering the role of CDV H glycoprotein in viral tropism and interactions with the immune system, nt- and aa-based analyses of this gene/protein may be also useful for studies of antigenic variability, vaccine protection/failure, and for cross-species spillover discussions (Iwatsuki et al., 2000; Martella et al., 2008; Duque-Valencia et al., 2019b). However, regarding the use of H for CDV classification, it is important to highlight that the evolutionary history of a gene may be different from that of the whole genome (Gregory, 2008). Classifications of natural recombinant viruses, for instance, may be biased by the analysis of single genes (Piewbang et al., 2019).

Herein, we performed an in silico analysis of all CDV CGs available in the GenBank database in order to investigate the suitability of H gene/protein for classification of CDV in lineages/genotypes. In addition, we analyzed the other viral genes and discussed which would be the most suitable target for CDV classification.

2. Material and methods

3. Results

3.1. Phylogenetic classification of CDV complete genomes

Complete CDV genome sequences were initially collected from GenBank (n = 196) and, after the exclusion criteria, we obtained a dataset of 116 CGs. Some of these sequences have received different classifications in CDV phylogeny studies. Thus, to cover all available classifications and to facilitate the contextualization of our results, we assigned compound names to the lineage of these sequences: Africa/Africa 1/Africa 2 and Africa/South Africa/Africa 1.

Herein, CDV CGs were distributed into lineages Asia 1 (39 sequences), Europe 1/South America 1 (12 sequences), Asia 4/Thailand (7 sequences), Africa/Africa 1/Africa 2 (7 sequences), Africa/South Africa/Africa 1 (2 sequences), Arctic Like (4 sequences), Asia 2 (14 sequences), America 2 (6 sequences), Caspian (3 sequences), Vaccine/America 1/Asia 3 (19 sequences; 3 and 16 clustered in group a and b, respectively). Two sequences (MT448054 and KF640687) were grouped into a yet unidentified lineage, termed “Undetermined 1″; and an additional sequence (MH316137), which did not have a defined lineage, was named “Undetermined 2″ (Fig. 1A and File S2). “Undetermined” classifications were based on ML analysis and previous reports of CDV evolution (Jo et al., 2019; Peserico et al., 2019). Furthermore, MT905031, MW876862 and MK037459, which were not previously classified (MT905031 and MW876862) or defined as Asia 5 based on the H gene (MK037459), were classified as Africa/Africa 1/Africa 2 considering their distribution in the ML and B-MCMC analyses (Fig. 1A and File S2). Details (metadata) of the sequences analyzed in the study are available in File S3.

Open in a separate window

Fig. 1

Phylogenetic trees based on nucleotide sequence analysis of the CDV complete genome and individual genes. One hundred and sixteen CDV complete genomes (CGs) were collected from the GenBank database. Phylogenetic analyses were performed with the Maximum Likelihood method and bootstrap of 1000 replicates, using MEGA-X software (version 10.2.5) (see Table 1). Analyses were performed considering the untranslated (UTR) and coding (CDS) regions of each gene. A: CG; B: hemagglutinin gene; C: C gene. In H tree, the America 1/Vaccine/Asia 3 sequences (groups a and b) were grouped in a single cluster (B). To facilitate their locations, sequences with discordant classification between H and CG were identified with the lineage color given by the CG (see Fig. A and B). Phylogenetic analyses of the other CDV genes are showed in File S4.

3.2. Lineage classification: complete genome versus individual genes

3.2.1. Nucleotide sequence

The phylogenetic classification of CDV based on the H gene showed disagreement with the CG analysis (both in ML and B-MCMC): in the CG phylogenetic tree, the America 1/Vaccine/Asia 3 sequences were distributed in two clusters (here termed groups a and b), whereas the H analysis grouped them in a single cluster (Fig. 1B and File S2). No disagreement was identified in the analysis of the C gene, i.e., it fully reproduced the CG classification by ML and B-MCMC analyses (Fig. 1C and File S2).

The phylogenetic trees of N, M, F, P and L genes showed at least one discrepancy from the CG by ML (File S4). The sequence MF926599 was identified as Asia 1 by the CG, but could not be classified by the N analysis. The sequences AY649446 and AY443350 (America 2 lineage by CG) were clustered with Vaccine/America 1/Asia 3 (group a) in the M phylogenetic analysis. In addition, MF926599 (identified as Asia 1 by CG) was classified as Vaccine/America 1/Asia 3 (group b) by the M gene. The sequence AY443350 (America 2 by CG) was classified as Vaccine/America 1/Asia 3 (group a) by the F analysis. In the L phylogenetic tree, the America 2 and Vaccine/America 1/Asia 3 (group a) sequences were grouped in the same cluster. The sequence MH316137, classified as Undetermined 2 by the CG, was grouped with the America 2 lineage in the P analysis. These disagreements were also observed when the CG vs. genes comparisons were performed by B-MCMC, except for the P gene, which reproduced integrally the CG classification obtained by Bayesian inference (File S2).

3.2.2. Amino acid sequence

Phylogenetic analysis based on the aa sequence of the H gene also showed some discrepancies with the CG classification. America 1/Vaccine/Asia 3 group a and b sequences were grouped in the same cluster in the H analysis (similar to that observed in the nt analysis). In addition, both America 2 and Undetermined 1 lineages were ungrouped after H phylogenetic analysis (Fig. 2A). The best models for aa analysis defined here are not available in the BEAUti/BEAST v.1.10.4 software package. However, the Bayesian analysis of H performed with the LG+G, the closest model to the LG+G+F (File S1), also showed the distribution of the America 1/Vaccine/Asia 3 groups a and b in the same cluster (data not shown).

Open in a separate window

Fig. 2

Phylogenetic trees based on amino acid sequence analysis of CDV genes. One hundred and sixteen CDV complete genomes (CGs) were obtained from GenBank. Phylogenetic analyses were performed with the Maximum Likelihood method and bootstrap of 1000 replicates, using MEGA-X software (version 10.2.5) (see File S1). A: hemagglutinin gene; B: phosphoprotein gene; C: C gene. The America 1/Vaccine/Asia 3 sequences (groups a and b) were grouped in a single cluster in the H tree (A). Sequences with discordant classification between H and CG were represented with the lineage color given by the CG. Sequences that did not cluster in any lineage were identified with “?”. Phylogenetic analyses of the other CDV genes are in File S5.

For the P and C genes, the aa-based analysis fully reproduced the CG classification (Fig. 2B and ​and2C).2C). For the M gene, sequences AF164967 and KU666057 (both America 2 by CG) could not be classified in any lineage, and the America 1/Vaccine/Asia 3 (group b) and Caspian lineages were grouped into the same cluster. Regarding the F phylogenetic tree, the sequence AY443350 (America 2 by CG) could not be classified in any lineage, and KF640687 (Undetermined 1 by CG) was clustered with Asia 4/Thailand. Phylogenetic analyses from the aa sequences of the N and L genes showed the same disagreements described for nt. Phylogenetic analyses of the aa sequences of N, M, F and L are showed in File S5.

4. Discussion

Hemagglutinin-based phylogenetic analyses may provide important insights on CDV evolution, potential vaccine failure, and virus jumping between host species (Iwatsuki et al., 2000; Duque-Valencia et al., 2019b). Interestingly, H is also the most frequently used gene for CDV phylogenetic classification (Martella et al., 2008; Nguyen et al., 2016; Duque-Valencia et al., 2019b; Piewbang et al., 2020). Herein, we first investigated whether the H gene/protein would be the best target for CDV phylogenetic classification. Subsequently, we explored the other CDV genes for their potential use in CDV classification.

Initially, we compared the phylogenetic classification of 116 CDV CGs available in GenBank with that obtained by analyzing their respective H. The aa and nt sequences of H were analyzed, as both have been used for CDV classification (Bolt et al., 1997; Mochizuki et al., 1999; Espinal et al., 2014; Nguyen et al., 2016). Interestingly, both H analyses (nt and aa) showed the same disagreement when compared to the CG-based phylogeny: the America 1/Vaccine/Asia 3 lineage was divided into groups a and b by the CG analysis, whereas it grouped in the same cluster in the H analyses. This finding was observed in the ML and B-MCMC analyses. Except for the H trees, groups a and b of the America 1/Vaccine/Asia 3 lineage were also separated in the trees from all other CDV genes based on nt or aa sequences, using ML or B-MCMC methods. Furthermore, the aa-based H analysis also showed additional disagreements with the CG, suggesting that the nt analysis may be the best choice when the H gene is used for CDV classification.

The difference between the H and CG trees (as well as the other genes) suggests that the clustering between groups a and b of the America 1/Vaccine/Asia 3 lineage in H is probably related to a recombination event involving this gene. This hypothesis is strongly supported by the recombination analysis of our dataset, in which the Snyder Hill vaccine strain (JN896987) (group b) was likely the major H donor for the three America 1/Vaccine/Asia 3 (group a) sequences (AY466011, AY542312 and AY445077) that clustered with the America 1/Vaccine/Asia 3 (group b). This result is also in line with that reported by Budaszewski et al. (2016), which suggests the Snyder Hill strain as a possible donor of the H gene for these sequences. Interestingly, the sequences AY466011, AY542312 and AY445077 are from viruses identified in free-living raccoons (Lednicky et al., 2004a; 2004b), whose ease circulation may contribute to co-infections, favoring recombination between different CDVs.

In addition, another 12 sequences showed probable recombination events in the H gene. Probable recombinations in CDV H have also been described in other studies (Han et al., 2008; Ke et al., 2015; Piewbang et al., 2019) and warn to the possibility that individual genes and whole organisms may have different evolutionary histories, which may lead to possible misunderstandings in single target-based classifications. Despite these findings, it should be emphasized that, although H seems not to be the most suitable target for phylogenetic classification, its analysis remains useful in studies of virus-host interactions and antigenic variability as well.

Due to the mismatch between H and CG-based classifications, we analyzed the other CDV genes to identify the one that would reproduce the CG classification. The F, L, M and N genes showed at least one CG discordance, both in the nt and aa analyses, observed in the ML or B-MCMC trees. Some CDV sequences also showed possible recombination events in these genes, especially in F - the gene with the highest number of probable recombinations described here.

Interestingly, the N and F genes have been previously used as targets in phylogenetic studies (Castilho et al., 2007; Sarute et al., 2013; Headley et al., 2015; Duque-Valencia et al., 2019a). The fusion protein signal-peptide (Fsp), an internal region of F gene, for example, was suggested as a target for phylogenetic studies due to its high variability, providing an easier differentiation between sequences (Sarute et al., 2013). This degree of variability, however, may be different from the CG, which may lead to discordant classifications when analyzing these two targets. Furthermore, the recombination evidence in the F gene described in our study, as well as those reported by other authors (Budaszewski et al., 2016), suggest careful use of this gene for phylogenetic classification purposes.

On the other hand, we observed that the classification obtained by the C gene analysis fully reproduced the CG results, both in the nt and aa sequences, in the ML or B-MCMC analysis. Furthermore, the P analysis (for aa in ML, or nt in B-MCMC) also reproduced the CG classification. Importantly, these were the only CDV genes that did not show strong recombination signals. Previous reports have described phylogenetic analyses based on P (Lednicky et al., 2004b; Lan et al., 2007), although some recombination events have been proposed for this genomic region (Budaszewski et al., 2016; Yuan et al., 2017). However, in our dataset, all recombination signals described for this gene were identified as possibly false positive recombination events by the RDP software. Overall, these findings indicate the C and P genes as suitable targets for CDV phylogenetic classification, especially when platforms for CG sequencing, considered the more reliable strategy for viral classification (Duque-Valencia et al., 2019b; Piewbang et al., 2019), are not available. Conversely, it is important to emphasize that C and P-based analyses may not allow for deep insights into CDV-host interactions, for which the H gene, as mentioned above, is probably the most suitable target.

In addition to lineage classification and recombination assessments, we calculated the geodesic distance between the phylogenetic tree of individual genes and that of the CG. This metric has been recently used to evaluate similarity between distinct phylogenetic trees considering both their topological differences and edge lengths (de Oliveira et al., 2021, 2022; Merchioratto et al., 2023). Here, the H gene was the fifth furthest from the CG (value 0.0680). In CDV, CG and H may have different nucleotide substitution rates (Piewbang et al., 2019). As the substitution rate is demonstrated by the edge length of the phylogenetic trees (ML), the difference between CG and H was likely reflected in the geodesic distance between these two targets. Interestingly, the geodesic distance between the P and CG trees found in our study was the smallest (0.0369).

Finally, considering the level of agreement between the phylogenetic analysis of the CDV CG with its respective genes, the geodesic distance between CG and genes, and the recombination findings, we observed that the H gene, especially its aa sequence, is not the most suitable target for CDV phylogenetic classification. Furthermore, considering the same criteria as above, we suggest the C and P genes are the best candidates for CDV classification. It is important to note that these results were obtained considering the CDV CGs available in GenBank to date. Therefore, future analyses – performed as more CDV CGs become available – are welcome to verify the limitations of H-based phylogenetic analyses and to consolidate the suitability of the C and P genes for CDV classification.

Funding

This work was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) - Finance code 001.

Author statement

All authors have contributed to, seen and approved the final and submitted version. We have no conflicts of interest to disclose.

Data statement

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

Footnotes

References