Comparative transcriptomics reveals divergent paths of chitinase evolution underlying dietary convergence in ant-eating mammals

Comparative transcriptomics reveals divergent paths of chitinase evolution underlying dietary convergence in ant-eating mammals
 

Rémi Allio^1,2,§,*, Sophie Teullet^1,§, Dave Lutgen^1,3,4,§, Amandine Magdeleine¹, Rachid Koual¹,  Marie-Ka Tilak¹, Benoit de Thoisy^5,6, Christopher A. Emerling^1,7, Tristan Lefébure⁸, and Frédéric Delsuc^1,*

¹ISEM, Univ. Montpellier, CNRS, IRD, Montpellier, France

²CBGP, INRAE, CIRAD, IRD, Montpellier SupAgro, Univ. Montpellier, Montpellier, France 

³Institute of Ecology and Evolution, University of Bern, Bern, Switzerland

⁴Swiss ornithological Institute, Sempach, Switzerland

⁵Institut Pasteur de la Guyane, Cayenne, French Guiana, France

⁶Kwata NGO, Cayenne, French Guiana, France

⁷Biology Department, Reedley College, Reedley, CA, USA

⁸Univ. Lyon, Université Claude Bernard Lyon 1, CNRS, ENTPE, UMR 5023 LEHNA, F-69622, Villeurbanne, France

^§Equal contribution

*Correspondence

Rémi Allio: remi.allio@inrae.fr

Frédéric Delsuc: frederic.delsuc@umontpellier.fr

Abstract

Ant-eating mammals represent a textbook example of convergent morphological evolution. Among them, anteaters and pangolins exhibit the most extreme convergent phenotypes with complete tooth loss, elongated skulls, protrusive tongues, and powerful claws to rip open ant and termite nests. Despite this remarkable convergence, comparative genomic analyses have shown that anteaters and pangolins differ in their chitinase gene (CHIA) repertoires, which potentially degrade the chitinous exoskeletons of ingested ants and termites. While the southern tamandua (Tamandua tetradactyla) harbors four functional CHIA paralogs (CHIA1, CHIA2, CHIA3, and CHIA4), Asian pangolins (Manis spp.) have only one functional paralog (CHIA5). These two placental mammal lineages also possess hypertrophied salivary glands producing large quantities of saliva to capture and potentially digest their social insect prey. We performed a comparative transcriptomic analysis of salivary glands in 23 representative species of placental mammals, including new ant-eating species and close relatives. Our results on chitinase gene expression suggest that salivary glands play a major role in adapting to an insect-based diet with myrmecophagous and insectivorous species highly expressing CHIA paralogs. Moreover, convergently-evolved pangolins and anteaters express different chitinases in their hypertrophied salivary glands and other additional digestive organs. CHIA5 is overexpressed in Malayan pangolin, whereas the southern tamandua exhibits high levels of CHIA3 and CHIA4 expression. Overall, our results demonstrate that divergent molecular mechanisms underlie convergent adaptation to the ant-eating diet in pangolins and anteaters. This work highlights the role of historical contingency and molecular tinkering of the chitin-digestive enzyme toolkit in this classical example of convergent evolution.

Figures & Tables

Figure 1: A. Mammalian chitinase gene family tree reconstructed using a maximum likelihood gene-tree/species-tree reconciliation approach on protein sequences. The nine chitinase paralogs are indicated on the outer circle. Scale bar represents the mean number of amino acid substitutions per site. B. Synteny of the nine chitinase paralogs in humans (Homo sapiens), tarsier (Carlito syrichta), nine-banded armadillo (Dasypus novemcinctus) and the two main focal convergent ant-eating species: the southern tamandua (Tamandua tetradactyla) and the Malayan pangolin (Manis javanica). Assembly names and accession numbers are indicated below species names. Arrows represent genes with scaffold/contig names and BLAST hit positions indicated below. Arrow direction indicates gene transcription direction as inferred in Genomicus v100.01 (Nguyen et al. 2022) for genes located on short contigs. Ψ symbols indicate pseudogenes as determined in Emerling et al. (2018). Genes with negative BLAST results were not represented and are probably not functional or absent.

Figure 2: Comparison of predicted ancestral sequences of the nine mammalian chitinase paralogs. A. Conserved residues of the canonical chitinolytic domain active site (DXXDXDXE). Arrows indicate paralogs in which changes occurred in the active site. B. Summary of the evolution of chitinase paralogs functionality. C. Conserved cysteine residues of the chitin-binding domain. The arrow indicates OVGP1 in which the last four cysteines have been replaced.

Figure 3: Expression of the nine chitinase paralogs in 28 mammalian salivary gland transcriptomes. Species are ordered taxonomically in the four major placental clades: AFR: Afrotheria, XEN: Xenarthra, EUA: Euarchontoglires, and LAU: Laurasiatheria. Expression level is represented as log10 (Normalized Counts + 1).

Figure 4: Comparative expression of CHIA1-5, CHIT1, CHI3L1-2, and OVGP1 in 64 transcriptomes from different organs in three mammalian species: the nine-banded armadillo (Dasypus novemcinctus), the Malayan pangolin (Manis javanica), and the southern tamandua (Tamandua tetradactyla). Non-functional pseudogenes are symbolized by the Ψ symbol and  horizontal bars indicate the digestive organs on the right side of the different graphs. Expression level is represented as log10 (Normalized Counts + 1).

Table 1: Detailed information on the tissues sequenced or retrieved from public databases for the project.

Zenodo supplementary files

CHIAs_OG_tree.zip contains raw CHIA sequences from the OrthoFinder orthogroups and the corresponding ML tree.  

Chitinases_ancestral_sequences.zip contains the alignment of the ancestral sequence reconstruction inferred by RAxML-NG. 

Chitinases_gene_tree.zip contains input and output files corresponding to the chitinase gene tree presented in Figure 1: 

- mammalina_species_tree_input_Generax.newick = species tree used for the reconciliation with Generax

- chitinase_gene_alignment_renamed_input_Generax.fasta = chitinase gene alignment with the sequence names renamed for Generax

- chitinase_gene_alignment_not_renamed.fasta = chitinase gene alignment with the original sequence names (for information)

- chitinases_gene_tree_sequences_renamed_input_Generax.newick = chitinase gene tree inferred with RAxML-NG and reconciled using the TreeRecs algorithm to find the optimal rooting scheme; this tree was used for Generax

- reconciled_chitinase_genes_tree_output_Generax.newick = reconciled chitinase gene tree inferred by Generax and presented in Figure 1

Chitinases_gene_expression.zip contains all orthogroup gene expressions plus chitinase gene expressions.

Kallisto_abundances.zip contains the abundances estimated with kallisto for each organ of each species.

Supplementary table figure 1B - BLAST results.xlsx contains BLAST results supporting sentence inferences. 

Transcriptome_assemblies.tar.gz contains the transcriptome assemblies obtain for each organ and species with Trinity.