Research ArticleVIROLOGY

Competition between social cheater viruses is driven by mechanistically different cheating strategies

See allHide authors and affiliations

Science Advances  21 Aug 2020:
Vol. 6, no. 34, eabb7990
DOI: 10.1126/sciadv.abb7990
  • Fig. 1 High-frequency mutations detected by experimental evolution of MS2 and their genomic effects.

    (A) Trajectories of mutations that increased in frequency and attained a frequency of at least 10% in one of the replicate lines A or B. (B) An illustration of the MS2 genome with the genomic location and effect of the mutations in (A). Shapes represent mutation type, and colors represent specific mutations, as in (A). RdRP, RNA-dependent RNA polymerase. (C) The location of Δ1764 (encircled) with respect to the RNA secondary structure of the TR loop. (D) The effect of Δ1764 on the lysis protein results in two peptides of the protein created by the truncated lysis product and by the frameshifted replicase reading frame. The amino acid affected by Δ1764 is marked with a red arrow.

  • Fig. 2 Two major lineages of defective mutants.

    (A) Frequency of mutants in 20 plaques seeded from p15. Of the mutations that arose in the original experiment, the only one observed in the plaques was A535G. (B) Results of haplotype inference based on MinION sequencing for virus populations from p15 line A (left) and line B (right). Each mutation is represented by a circle (color-coded as in Fig. 1A) whose size is roughly proportional to its frequency at p15. Our results suggest that two major cheater lineages are present in both populations with additional different hitchhiking mutations in each line. The full list of inferred haplotypes is given in table S1. (C and D) Details as in Fig. 1A and p1 to p14 are the same as in Fig. 1A. Passaging was reinitiated from a frozen stock of p14. p15 was created at an MOI of 1, yet p16 was created at lower MOIs: MOI of 0.1 (C) and MOI of 0.01 (D), marked in green. All panels show a strong decrease in defective frequencies at lower MOIs.

  • Fig. 3 The mechanism endowing an advantage to Δ1764.

    (A) Populations from p5, p8, p10, p13, and p15 of line B were used to infect cells to test the intracellular replication of the different mutants. We tracked the intracellular frequency of the two major mutations found in the experiment versus WT or WT-like variants. Results show similar dynamics for WT/A535G and A1664G, as opposed to a marked decline in frequency of Δ1764 that is compensated after overnight growth (labeled as “on”). (B) In vitro binding assay between the MS2 coat protein and the RNA of the TR loop (fig. S6B) reveal higher affinity of Δ1764 RNA for the coat protein. The initial fluorescence data from the Monolith NT.115 pico instrument (NanoTemper Technologies) was used to calculate a Kd (dissociation constant) of 137.6 ± 20.4 μM for the WT RNA and a Kd of 48.8 ± 8.7 μM for the Δ1764 RNA [n = 4 independent measurements (fig. S6, C and D); error bars represent the SD; y axis is normalized to fraction bound].

  • Fig. 4 Game theory–based modeling of a single cheater and WT dynamics.

    (A) Theoretical fitness matrix of a cooperator-defector system, with notations, as described previously (47). (B) WT and single Δ1764 cheater model, empirical frequency measurements of line B (solid line) with an interval (band) derived from the posterior distribution. (C) Posterior distributions for inferred parameters (dark blue) versus prior distributions (gray). (D) Host cell lysis c3000 bacteria post infection with MS2 populations from p1, p10, and p15 from lines A and B. Notably, passaged populations with high proportions of Δ1764 (p15A, p10B, and p15B) display reduced cell lysis. On the other hand, p10A, which has a high proportion of A1664G, does not display reduced cell lysis.

  • Fig. 5 A second partial cheater lineage takes over the viral population.

    (A) Additional passages initiated from a p14 frozen stock reveal a decline in Δ1764 and increase in A1664G frequencies. Only mutations displayed in Fig. 1 are shown; fig. S1C displays a few additional minor mutations that arose. (B) The position of the A1664G synonymous mutation on the secondary RNA structure lysis hairpin, constructed by Mfold (70). The inferred RBS is boxed, and the start codon is marked in purple. The A1664G mutation creates a stronger stem-loop structure, as evident by an inferred ΔG of −11.6 kcal/mol for the WT RNA and −15.2 kcal/mol for A1664G RNA. (C) Synthetic constructs bearing GFP were generated, one with the RBS of WT virus and one with the A1664G mutant. (D) Fluorescence level comparison of the WT- and A1664G-bearing constructs suggest reduced protein levels for the A1664G RBS. The plasmid bearing the A1664G RBS construct had a GFP expression level much lower than the WT construct and equal to that of the bacteria bearing no plasmid. Fluorescence levels are calculated as the median GFP level of the population. Each construct was tested in three biological replicates (fig. S8), and the group means were compared using t tests with false discovery rate correction for multiple testing (P values of <0.01 are marked with **). ns, not significant; A.U., arbitrary units.

  • Fig. 6 Modeling the dynamics of cheater viruses.

    (A) Empirical frequency measurements of line B (solid line/s) with an interval (band) derived from simulations based on the posterior distribution. (B) Posterior distributions for inferred parameters (dark blue) versus prior distributions (gray). The dashed lines represent the median of the posterior distribution. (C) Relationships between pairs of parameters across joint posterior distributions. We conclude that the advantage of Δ1764 is greater than A1664G in the presence of the WT (left) yet that A1664G is able to replicate better in the presence of Δ1764 than vice versa (middle and right). Results are shown for line B; results for line A lead to similar conclusions (fig. S7, C to E). (D) Proposed mechanistic model for the antagonism between both cheaters. During coinfection of a cell with both cheaters, A1664G produces abundant coat products. Because of its efficient binding of coats, Δ1764 is packaged before its genome is replicated. Accordingly, more A1664G genomes will be released from the cell than Δ1764 genomes.

Supplementary Materials

  • Supplementary Materials

    Competition between social cheater viruses is driven by mechanistically different cheating strategies

    Moran Meir, Noam Harel, Danielle Miller, Maoz Gelbart, Avigdor Eldar, Uri Gophna, Adi Stern

    Download Supplement

    This PDF file includes:

    • Figs. S1 to S8
    • Tables S1 and S2
    • References

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article