The Critical Role of Codon Composition on the Translation Efficiency Robustness of the Hepatitis A Virus Capsid

Abstract

Hepatoviruses show an intriguing deviated codon usage, suggesting an evolutionary signature. Abundant and rare codons in the cellular genome are scarce in the human hepatitis A virus (HAV) genome, while intermediately abundant host codons are abundant in the virus. Genotype-phenotype maps, or fitness landscapes, are a means of representing a genotype position in sequence space and uncovering how genotype relates to phenotype and fitness. Using genotype-phenotype maps of the translation efficiency, we have shown the critical role of the HAV capsid codon composition in regulating translation and determining its robustness. Adaptation to an environmental perturbation such as the artificial induction of cellular shutoff-not naturally occurring in HAV infection-involved movements in the sequence space and dramatic changes of the translation efficiency. Capsid rare codons, including abundant and rare codons of the cellular genome, slowed down the translation efficiency in conditions of no cellular shutoff. In contrast, rare capsid codons that are abundant in the cellular genome were efficiently translated in conditions of shutoff. Capsid regions very rich in slowly translated codons adapt to shutoff through sequence space movements from positions with highly robust translation to others with diminished translation robustness. These movements paralleled decreases of the capsid physical and biological robustness, and resulted in the diversification of capsid phenotypes. The deviated codon usage of extant hepatoviruses compared with that of their hosts may suggest the occurrence of a virus ancestor with an optimized codon usage with respect to an unknown ancient host.

Keywords: codon usage; fitness landscape; genotype–phenotype maps; sequence space; translation efficiency.

Fig. 1.

—Relative Codon Deoptimization Index (RCDI), tRNA Adaptation Index (tAI), and translation elongation Rate (R_c) of the capsid of several picornaviruses. PV 1 (poliovirus 1) is depicted in turquoise, HRV 14 (human rhinovirus 14) in violet, HRV 89 (human rhinovirus 89) in green, and HAV (human hepatitis A virus) in brown. Analyses were performed in 100 codon length windows sliding every 15 codons. Doted horizontal lines depict the average from all the windows. Regarding RCDI and tAI, PV 1 significantly differed (P < 0.001) from the rest of viruses and HRV 14 also differed significantly (P < 0.001) from HAV. R_c was significantly higher (P < 0.001) in HAV, and in PV 1 was also significantly higher (P < 0.001) than in HRV 14 and HRV 89. The HAV fragments under study are indicated with black solid lines.

Fig. 2.

—Analyses of the mutant spectra. Genotype distribution of VP3 (A) and VP1 (B) in populations L0, F0.05LA, and F0.2 LA.

Fig. 3.

—Effect of mutations inducing changes in the codon frequencies in the translation rate. The actual translation efficiency of the VP1 clones is shown. Results are based on three different experiments, each including two replicas. (A) Conditions of no shutoff, (B) conditions of moderate shutoff, and (C) conditions of high shutoff. The dotted red line correspond to the average of a negative control corresponding to cells transfected with the digested vector alone. A new pdf file, corresponding to panels A, B and C, has been uploaded since the dotted red line in panel A is not visible. Statistically significant differences (P < 0.05) are depicted by different combinations of letters (ab=a, ab=b; bc=b, bc=c; a ≠ b, a ≠ c, b ≠ c; a ≠ d, b ≠ d, c ≠ d). The blue line plots represent the P-num, which predicts the potential occurrence of secondary structures in the RNA of the different genotypes; the lower the P-num value the higher the secondary structures in the RNA. (D) Information on the codon replacements present in each clone, compared with the most abundant genotype (VP1-1). Red and green background colors represent a change to a less and more frequent codon, respectively, and light gray a mutation not changing the codon frequency. ^aTheoretical translation elongation rate (R_c) of codons calculated using the data on human tRNA copy numbers available in http://gtrnadb.ucsc.edu/Hsapi19/Hsapi19-summary.html, last accessed July 22, 2019. ^bR_c values corrected based on tRNA copy numbers available in (Iben and Maraia 2014).

Fig. 4.

—Translation efficiency landscapes. (A), (B), and (C) panels show the VP3 landscape, and (D), (E), and (F) panels show the VP1 landscape. (A) and (D) panels include the landscape in absence of cellular shutoff (0.00 µg/ml of AMD), (B) and (E) panels include the landscape in conditions of moderate cellular shutoff (0.05 µg/ml of AMD, and [C] and [F] panels in conditions of high cellular shutoff [0.20 µg/ml AMD]). The landscapes were built by representing the relationship between three parameters: each genotype, their Relative FLuc/RLuc activity (phenotype) and its abundance (percentage). Black dots correspond to the L0 genotypes, red dots to the F0.05LA genotypes and blue dots to the F0.2LA genotypes.

Fig. 5.

—Estimation of the translation efficiency robustness through G → P maps. The number of different genotypes and phenotypes for each fragment and population, and their location in the G → P maps of the VP3 (A) and VP1 (B) fragments is depicted. L0 population: black circles. F0.05LA population: red circles. F0.2LA population: blue circles. Arrows connect one mutation away genotypes. Continuous arrows connect genotypes expressing the same phenotype. Discontinuous arrows connect genotypes expressing different phenotypes. Black, red, and blue arrows denote absence, moderate, and high cellular shutoff conditions, respectively. T_s and T_v are transition and transversion mutations, respectively.

Fig. 6.

—Translation efficiency robustness and phenotype accessibility, and its relationship with folding diversity. (A–F) Translation efficiency robustness (black plots) and phenotype accessibility (blue plots). Analyses of VP3 (A, B, and C) and VP1 (D, E, and F) fragments of populations L0 (A and D), F0.05LA (B and E), and F0.2LA (C and F) under conditions of no (0.0 µg/ml of AMD), moderate (0.05 µg/ml of AMD), and high cellular shutoff (0.2 µg/ml of AMD) are shown. Red circles represent the optimum conditions of each population. (G–I) Capsid behavior under extreme conditions of populations L0, F0.05LA, and F0.2LA grown in their optimum conditions and population F0.2LA grown in conditions of moderate cellular shutoff (F0.2LA-0.05). Boxplot diagrams show the fragility to 300 MPa (high hydrostatic pressure) for 1 min (G), the denaturability at pH 2 for 1 h (H), and the plasticity for RNA uncoating measured as the log₁₀ 50% uncoating time in minutes (I). The bottom and upper limits of the boxes represent the 25th and 75th percentiles, respectively. The bottom and upper whiskers represent the 5th and 95th percentiles, respectively. A solid black line represents the median.