ΦX174 Attenuation by Whole-Genome Codon Deoptimization

Abstract

Natural selection acting on synonymous mutations in protein-coding genes influences genome composition and evolution. In viruses, introducing synonymous mutations in genes encoding structural proteins can drastically reduce viral growth, providing a means to generate potent, live-attenuated vaccine candidates. However, an improved understanding of what compositional features are under selection and how combinations of synonymous mutations affect viral growth is needed to predictably attenuate viruses and make them resistant to reversion. We systematically recoded all nonoverlapping genes of the bacteriophage ΦX174 with codons rarely used in its Escherichia coli host. The fitness of recombinant viruses decreases as additional deoptimizing mutations are made to the genome, although not always linearly, and not consistently across genes. Combining deoptimizing mutations may reduce viral fitness more or less than expected from the effect size of the constituent mutations and we point out difficulties in untangling correlated compositional features. We test our model by optimizing the same genes and find that the relationship between codon usage and fitness does not hold for optimization, suggesting that wild-type ΦX174 is at a fitness optimum. This work highlights the need to better understand how selection acts on patterns of synonymous codon usage across the genome and provides a convenient system to investigate the genetic determinants of virulence.

Keywords: bacteriophage; codon bias; epistasis; fitness landscape; live-attenuated vaccine; synthetic biology.

Fig. 1

Invented data illustrate that epistasis affects how a desired level of attenuation is achieved. When a substantial amount of attenuation is desired (the “targeted attenuation range” is at a low fitness level), the amount of attenuation (e.g., number of deleterious mutations) will be harder to achieve if mutational effects combine synergistically (negative epistasis) because fitness declines at an increasing rate. In this case, the targeted amount of attenuation will be easier to achieve if mutations combine antagonistically (positive epistasis) because a larger range in the number of deleterious mutations results in the same level of fitness effects. Notice that this pattern is reversed if a slight level of attenuation is desired (near y = 0). If sign epistasis, or even irregular magnitude epistasis is observed, then the underlying nature of interactions is more difficult to predict and generalize.

Fig. 2

ΦX174 genome organization and capacity for deoptimization relative to host genes. (a) Genes on the ΦX174 genome are labeled at the top and shown as white boxes. Recoded regions are shown as filled blue boxes. These fragments are named consecutively (e.g., A1, A2, A3, A4, F1, F2, F3, … ). Transcript expression levels are shown as filled gray bands. The band heights are proportional to the relative number of transcripts by RT–qPCR (Zhao et al. 2012). (b) Codon adaptation index (CAI) of Escherichia coli, wild-type ΦX174, and recoded ΦX174 genes. Genes highly expressed in E. coli are enumerated on the secondary y axis. The white space between recoded fragments (the BsmBI sites) was enlarged for visualization. Lines connect genes that were fully recoded. Genes containing only some recoded fragments (e.g., deoptimized A1) do not have black boarders.

Fig. 3

Fitness effects of deoptimizing ΦX174 genes. (a) The fitness of wild-type and deoptimized ΦX174 strains containing recoded genes are shown in replicate (colored dots). Means and standard error bars are shown in black. Fitness values that are significantly different from wild type are indicated with asterisks (ANOVA, P < 0.01). (b) Fitness is plotted against measures that potentially explain fitness decreases. Fitness is the number of doublings per hour (log2 of the ratio of the phage concentration at 60 min divided by the phage concentration at time zero). Gene expression levels are from Logel and Jaschke (2020) and are normalized to gene A. Structural proteins are shown as empty circles. Deoptimizing gene G yielded no viable phage. The total number of independent fitness measurements is provided in Supplementary Material online. At least three replicates were performed for every strain. y axis are the same in panels (a) and (b). The percent of codons changed and change in CAI are highly correlated (R² = 0.92).

Fig. 4

Fitness of ΦX174 when deoptimized gene fragments are combinatorially joined. The fitness of variants containing one deoptimized fragment (a) and all possible within-gene variants (b) was measured and compared with wild type (gray horizontal line). Significant differences (ANOVA, P < 0.01) are indicated with asterisks. In (b), fragment lengths are drawn to scale. Filled colors indicate the deoptimized fragments whereas unfilled blocks indicate wild-type fragments. In both (a) and (b), fitness is shown as log2-fold increase in the number of phage per hour. At least three replicates were performed for every strain (see mentary Material online). y axis are the same in panels (a) and (b).

Fig. 5

Effect of recoded fragment on all possible backgrounds. (a) Regressions of each recoded fragment’s fitness effect under the additive model against background fitness for gene A. Fitness effect is the difference between the background and the background plus the recoded fragment (e.g., A2 -> A1+A2, A2+A3 -> A1+A2+A3). Horizontal lines indicate a perfect fit to the additive model with no residual effects of background. Sloped regression lines indicate antagonistic/synergistic epistasis. Solid regression lines indicate that the additive model can be rejected (linear model, P < 0.05). The overall fit of epistatic models for each gene is shown in table 2. In (b), fragment fitness effects are shown against the change in CAI. Slight point jitter was used for visualization. Linear regressions are shown with P and R² values. Change in CAI (Xia 2007 method) is proportional (CAI of recoded gene over CAI of background). AIC values of alternative models are shown in supplementary table S2, Supplementary Material online.

Fig. 6

Fitness of recoded viruses correlates with codon usage bias and mRNA folding stability. (a) Codon usage bias (CAI) compared with fitness for viruses optimized and deoptimized in genes A, F, G, and H. Fitness and CAI of wild type are indicated with gray horizontal and vertical lines. Points to the right of these lines are optimized. Points to the left are deoptimized. (b) Viral fitness compared with mRNA folding stability (mfold). Wild-type values are indicated with gray horizontal and vertical lines. Less-stable transcripts (mostly deoptimized genes) have less negative values and are right of wild type. More stable transcripts (most optimized genes) have more negative values. R² and P values shown are from individual (for each gene) linear regressions. A complete model comparing all indices with adjusted P values for multiple comparisons is shown in supplementary table S2, Supplementary Material online. Codon optimized viruses are shown with empty circles. Those significantly different from wild type are labeled (ANOVA, P < 0.05).