The role of APOBEC3-induced mutations in the differential evolution of monkeypox virus

Abstract

Recent studies show that newly sampled monkeypox virus (MPXV) genomes exhibit mutations consistent with Apolipoprotein B mRNA Editing Catalytic Polypeptide-like3 (APOBEC3)-mediated editing, compared to MPXV genomes collected earlier. It is unclear whether these single nucleotide polymorphisms (SNPs) result from APOBEC3-induced editing or are a consequence of genetic drift within one or more MPXV animal reservoirs. We develop a simple method based on a generalization of the General-Time-Reversible (GTR) model to show that the observed SNPs are likely the result of APOBEC3-induced editing. The statistical features allow us to extract lineage information and estimate evolutionary events.

Figure 1:

Schematic of the stochastic evolution model. (a) There are different potential synonymous mutation sites along the genome. Some potential mutations are APOBEC3-induced (red), some are reverse-APOBEC3-induced (blue), and some are neither (green). The transitions (from open circles to hash-filled circles) are indicated by arrows. Transitions of each type occur independently at their type-specific transition rate. (b) An example of two lineages $(i)$ and $(j)$ arising from a common ancestor that shared a common evolutionary path until time $t^{(ij)}$. Each black dot represents a mutation. An open circle represents the time a lineage is sampled. The waiting time between two successive mutations are independent and randomly distributed with the same total rate $N\bar{v}$. The specific type of mutation is also randomly chosen according to the relative rates of the different mutation types. (c) Counts of mutations using the common ancestor as a reference. (d) Counting the numbers of mutations of lineage $(i)$ at time $t^{(i)}$ using the lineage $(j)$ at time $t^{(j)}$ as a reference. Since sequences share part of their evolutionary paths starting with the common ancestor, mutations ${\alpha }_1$ and ${\alpha }_2$ are not identified by using lineage $(j)$ as a reference. The subsequent mutations of lineage $(j)$ count towards $n^{(i\mid j)}$. For completeness, we included all mutations instead of only synonymous ones in the schematic sample sequences in (c) and (d).

Figure 2:

Simulations of different scenarios. (a-b) Four representative trajectories of an iid substitution process of the JC69 limit. A total of 10⁶ nucleotides are used for each simulation and both TC → TT and A → C mutations are linearly correlated with the total number of point mutations.

Figure 3:

Simulations of scenarios that include mutation and selection following Eqs. 11.13. We used ${\beta }_0=1$, ${\mu }_0=0.5$ and assumed that with each birth, one daughter has equal probability 0.1 of acquiring each type of mutation. The total initial numbers of possible mutations $N_i$ for synonymous non-APOBEC3 mutations, synonymous APOBEC3 mutations, and hidden nonsynonymous mutations is set to 10⁵,10³, and 10³, respectively. (a) Numbers of synonymous APOBEC3-relevant mutations $n_{\mathrm{a}}$ associated with the numbers of synonymous non-APOBEC3-relevant mutations $n-n_{\mathrm{a}}$ from samples of simulation trajectories. Here, no selection for hidden mutations $\left({\sigma }_{\mathrm{h}}=0\right)$ is present. (b) Each diamond symbol represents the slope of an independently simulated selection-mutation process, as shown on (a). Slopes for different ${\sigma }_{\mathrm{a}}$ and 25 uniformly sampled values of $\mathrm{l}\mathrm{o}\mathrm{g}K$ are shown. (a) and (b) show that large population size $K$ and strong selection coefficient ${\sigma }_{\mathrm{a}}$ can lead to a larger effective mutation rate of APOBEC3-driven mutations. By design, under weak selection strength ${\sigma }_{\mathrm{a}}$, the effective mutation rate of APOBEC3-driven mutations is identical to the non-APOBEC3-driven mutations. (c) Values of $n_{\mathrm{a}}$ associated with $n-n_{\mathrm{a}}$ under different strengths ${\sigma }_{\mathrm{h}}$ of selection of hidden beneficial mutations. Strong selection of hidden beneficial mutations can restore the increased effective mutation rate of APOBEC3-driven mutations to the same level as the pure mutation processes. (d) For large ${\sigma }_{\mathrm{h}}>{\sigma }_{\mathrm{a}}$, the expansion of $n_{\mathrm{a}}$ is limited (red diamonds).

Figure 4:

Relative molecular clocks for different types of mutations with respect to an arbitrarily chosen reference genome, instead of time, exhibit linearity. (a) The number of synonymous APOBEC3-relevant mutations (TC → TT) is plotted against the number of synonymous mutations for the genomes collected before 2016. (b) The number of synonymous APOBEC3-relevant mutations (TC → TT) is plotted against the number of synonymous mutations for the genomes collected after 2016.

Figure 5:

(a) The number of observed APOBEC3-relevant mutations $n_{\mathrm{a}}$ of all genomes plotted against the number of observed synonymous mutations $n$, using the genome KP849470 as a reference. The two genomes KJ642617 and MK783029 are highlighted since they are the closest genomes to the intersection of the two lines corresponding to the pre-2016 and post-2016 genomes. (b) The number of observed APOBEC3-relevant mutations $n_{\mathrm{a}}$, using genome KJ642617 as the reference, plotted against the number of observed synonymous mutations $n$. The open circle represents the location of the hypothesized most recent common ancestor $(MRCA)$ of the post-2016 genomes in the animal reservoir. The inset shows that the line of the post-2016 group does not intersect the origin, suggesting that KJ642617 is not a common ancestor of the post-2016 group. (c) The number of observed APOBEC3-relevant mutations $n_{\text{a}}$ plotted against the number of observed synonymous mutations $n$, using MK783029 as reference. The inset shows that the line of the pre-2016 group does not pass through the origin, suggesting that MK783029 has already undergone some APOBEC3-editing after its ancestor’s transmission to human.

Figure 6:

The number of synonymous APOBEC3-induced mutations plotted against the collection date of the post-2016 genomes, using KJ642617 as the reference. The red diamond represents the genomes within the lineage with the longest branch length, starting with MK783028 and ending with ON694329.

Figure 7:

Different evolutionary environments shape the shared features of pre- and post-2016 groups of MPXV genomes. Pre-2016 cross-species transmission did not persist in human hosts. The blue dots represent the shared features due to evolution in the animal reservoir. The red dots represent the shared features due to evolution in the human hosts. We hypothesize an unobserved stage of continuous and persistent evolution in the human hosts between the 2017 cross-species transmission and the 2022 global outbreak.