Comparative analysis of codon usage bias and host adaptation across avian metapneumovirus genotypes

Abstract

Avian metapneumovirus (aMPV) is a significant pathogen affecting poultry worldwide, causing respiratory disease and economic losses. This study investigated the genetic and evolutionary differences among aMPV genotypes through codon usage bias analysis. Using whole-genome and F gene sequences, we assessed phylogenetic relationships, codon usage patterns, evolutionary pressures, and host adaptation. Our results indicate clear genotype differentiation in the phylogenetic tree, with Group C identified as the earliest diverging lineage of aMPV. The F gene exhibits independent evolutionary trajectories, reflecting distinct selective pressures. Codon usage bias varies across genotypes and is primarily driven by selection pressure, with Groups B and C experiencing stronger selective constraints. The F gene, crucial for viral entry and adaptation, undergoes intense selection, optimising codon usage for host adaptation. Host adaptation analysis reveals that aMPV is most suited to chickens. Additionally, Group B exhibits the largest population size; however, recent declines, particularly in this genotype, suggest that vaccine-driven selection pressure may be influencing aMPV population dynamics. These findings provide critical insights into aMPV evolution, highlighting the role of codon usage bias and selection pressure in shaping viral adaptation. Understanding these evolutionary mechanisms may aid in vaccine development and disease control strategies.

Keywords: Avian metapneumovirus; Codon usage bias; Genotype; Host adaptation; Phylogenetic analysis.

Fig. 1

Phylogenetic and temporal evolutionary trajectory of the F gene in aMPV. (a) Phylogenetic analysis based on the whole genome. (b) Phylogenetic analysis based on the F gene. Sequences were aligned using multiple alignment with fast Fourier transform, and ambiguous regions were removed with trimAL. The best-fit model was selected via ModelFinder. Maximum likelihood phylogenetic trees were inferred using IQ-TREE with ultrafast bootstraps in PhyloSuite. (c) Temporal evolutionary trajectory. Bayesian molecular clock analysis of aMPV was performed using F gene sequences. The BEST Model Test package selected the optimal nucleotide substitution model. An optimised relaxed clock model accounted for substitution rate variation, while the coalescent Skygrid model inferred population size changes.

Fig. 2

Relative synonymous codon usage, correspondence analysis, and linear discriminant analysis. (a) Relative synonymous codon usage (RSCU) analysis based on the whole genome. RSCU values of >1.6 indicate codon preference, while values of <0.6 indicate underrepresentation. (b) Principal component analysis (PCA) based on the whole genome. PCA reduced the high-dimensional RSCU dataset into two principal components (Axis 1 and 2) for clustering visualisation. (c) RSCU analysis based on the F gene. (d) PCA based on the F gene.

Fig. 3

Differences in codon usage characteristics between genotypes. (a) Analysis based on the whole genome. (b) Analysis based on the F gene. A: adenine, G: guanine, C: cytosine, T: thymine, GC %: percentage of G and C in a gene or genome, GC3 %: GC content at the third codon position, GC3s: GC content at the third position of synonymous codons. ENC: effective number of codons, which quantifies codon usage bias, with lower values indicating stronger codon usage bias. Dinucleotide abundance was assessed using O/E ratios, with values of >1.25 indicating overrepresentation and <0.78 indicating underrepresentation. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 4

Evolutionary forces of codon usage bias in aMPV genotypes. (a) Whole genome. (b) F gene. PR2: Parity Rule 2, ENC: effective number of codons, F: Fusion. The regression line was generated using the least squares method, plotting GC content at the first and second codon positions (GC12 %) against GC content at the third codon position (GC3) to assess the balance between selection and mutational pressure. The selection pressure acting on each genotype was estimated using (1 − slope of regression line) × 100 %.

Fig. 5

Host adaptation and codon usage optimisation of aMPV genotypes. (a) Whole genome. (b) F gene. GG: chicken (Gallus gallus), MG: turkey (Meleagris gallopavo), AP: duck (Anas platyrhynchos), and AA: goose (Anser anser). CAI: codon adaptation index, RCDI: relative codon deoptimisation index, SiD: similarity index. Higher CAI and lower RCDI/ SiD indicate stronger adaptation to host codon usage and higher translational efficiency. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 6

Population expansion of each genotype in the context of codon usage. The shaded intervals represent the 95 % highest posterior density of the product of generation time and effective population size, while the middle line represents the inferred median effective population size. Bayesian skyline analysis was performed using BEAST 2.7 to infer historical population dynamics. The BEST Model Test selected the nucleotide substitution model, an optimised relaxed clock model accounted for rate variation, and a coalescent Bayesian skyline model was used as the tree prior. Markov chain Monte Carlo sampling (100M iterations) was conducted, with convergence assessed in Tracer (effective sample size > 200).