The Constructive Black Queen hypothesis: new functions can evolve under conditions favouring gene loss

Abstract

Duplication is a major route for the emergence of new gene functions. However, the emergence of new gene functions via this route may be reduced in prokaryotes, as redundant genes are often rapidly purged. In lineages with compact, streamlined genomes, it thus appears challenging for novel function to emerge via duplication and divergence. A further pressure contributing to gene loss occurs under Black Queen dynamics, as cheaters that lose the capacity to produce a public good can instead acquire it from neighbouring producers. We propose that Black Queen dynamics can favour the emergence of new function because, under an emerging Black Queen dynamic, there is high gene redundancy spread across a community of interacting cells. Using computational modelling, we demonstrate that new gene functions can emerge under Black Queen dynamics. This result holds even if there is deletion bias due to low duplication rates and selection against redundant gene copies resulting from the high cost associated with carrying a locus. However, when the public good production costs are high, Black Queen dynamics impede the fixation of new functions. Our results expand the mechanisms by which new gene functions can emerge in prokaryotic systems.

Keywords: Black Queen hypothesis; constructive neutral evolution; gene duplication; horizontal gene transfer; modelling; neofunctionalization; public goods.

Figure 1

Neofunctionalization may occur via duplication and divergence or a cBQ; (A) a simple depiction of neofunctionalization via duplication; intragenomic duplication of a locus followed by mutation can lead to that locus being lost (loss pathway), else it may diverge, yielding some new function (divergence pathway, pink locus); (B) neofunctionalization is hypothesized to be possible under a Black Queen dynamic; the starting conditions are that all cells in a population carry a locus (blue) that codes for production of a good (blue dots); this public good cannot be monopolized (extracellular blue dots), which creates the conditions for the evolution of cheaters that do not contribute to public good production, yet derive benefit from its presence in the environment; cheaters may completely lose the public good locus, rendering them obligately dependent on producers (loss pathway); alternatively, it may be possible for the locus to diverge, giving rise to a new function (divergence pathway, pink locus); created with BioRender.com.

Figure 2

Multiple genotypes may be present in a population of cheaters during a Black Queen dynamic; a producer (Black Queen locus, black) can evolve new function (neofunctional locus, orange) because not all members of a community are required to produce a public good; direct mutation from one state to the other is possible in principle but unlikely; however, under a Black Queen dynamic, the population of cells diverges into producers and cheaters, with cheaters requiring proximity to producers to access the public good (cell population, top left); at the genetic level, multiple states may be present among cheaters (loci, right hand side) and may have varying fitness; locus deletion may be most selectively advantageous (no protein production, no locus maintenance, three upward arrows) but will not lead to neofunctionalization; however, three broad cheater genotype classes are, in principle, available for neofunctionalization; promoter inactivation retains the locus without expression; in this case, the cost of protein production is saved (two upward arrows), but the locus remains as raw material for neofunctionalization; a weak cheater may emerge through a promoter mutation that reduces but does not eliminate public good production; the cheater may nevertheless be selectively advantageous (one upward arrow) if it derives a greater portion of the public good from neighbouring cells than it contributes; neofunctionalization would render it an obligate cheater; a cheater phenotype may also arise through loss of activity; in this case, the mutation is selectively neutral; it is tolerated because loss of public good production has no detrimental effect, but the genotype is still able to produce (nonfunctional) protein; the expressed locus is under relaxed selection, so it is free to acquire new function; created with BioRender.com.

Figure 3

Black Queen dynamics can accelerate neofunctionalization; plots show the mean time (timesteps) for a new gene function to be fixed (present in >99% in total population) for a model with gene duplication alone (Ohno, orange) versus gene duplication plus a Black Queen dynamic (cBQ, black); duplication is implemented via “gene transfer” events (see Materials and methods); this enables the novel gene to also be acquired by producers in the cBQ model; error bars: 95% CI (100 replicate simulations and bootstrap with ); (A) Black Queen dynamics can accelerate neofunctionalization under deletion bias; when duplication rate is lower than gene deletion rate (), a cBQ substantially accelerates the evolution of new gene function relative to the Ohno model; at higher duplication rates, where deletion bias is eliminated, the difference between the two models is diminished; the cost of carrying a gene () is zero, the cost of good production () is 0.5, and the benefit of neofunctional genes () is 0.07; for the cBQ model, the size of public-good sharing neighbourhood () is 3; (B) the acceleration of evolution by a cBQ dynamic diminishes as deletion rate increases; this contrasts with the result, shown in (A), that the relative acceleration increases as duplication rate decreases (see main text for explanation); duplication rate () is , , , and, for the cBQ model, ; (C) in both models (Ohno, cBQ), the time taken to reach fixation increases as the cost of carrying genes increases, but this increase is much more rapid for the Ohno model; as the cost of carrying a gene increases, so does the strength of selection against gene redundancy; consequently, the time to fixation becomes very high under the Ohno model as intragenomic redundancy is selected against; under the cBQ model, this cost has less effect on the time to fixation (see main text for explanation); , , , , and for the cBQ model, ; (D) emergence of a cBQ depends on the cost of public good production; at moderate public good cost, a Black Queen dynamic accelerates fixation of a new gene, but when the cost of public good production is high, a Black Queen dynamic impedes fixation of a new gene; , , , , and for the cBQ model, .

Figure 4

Genetic redundancy is robust against gene cost in cBQ model; (A) plot shows the average age of cheater lineages as a function of time in the cBQ model; age count starts when a cheater arises from a producer through mutation or deletion and increases by one when a cheater reproduces (see Materials and methods); cost per gene () is set to , and the other parameters are the same as in Fig. 3C, except that the benefit of neofunctional genes () is set to zero so that the age of lineages represents a situation before the evolution of a new gene function; results from three replicate simulations are shown; the plot shows that the mean age of lineages is ~3000 generations; if cost for carrying a gene is denoted by , selection induced by this cost has a timescale of approximately generations; thus, the plot indicates that the timescale of selection induced by cost for carrying a gene is slower than the turnover of cheater lineages if ; (B) the mean total number of “mutable” loci across both producer and cheater populations (circles, solid line) or only the cheater population (triangles, broken line), as a function of cost for carrying a locus in the Ohno model; (C) the mean total number of “mutable” loci across both producer and cheater populations (circles, solid line) or only the cheater population (triangles, broken line), as a function of cost for carrying a locus in the cBQ model; in panels (B) and (C), mutable loci are defined as those that can mutate without reducing the fitness of the carriers; the number of mutable loci quantifies the degree of genetic redundancy; the plots show that the number of mutable loci decreases as the cost increases in both models, but this decrease is slower in the cBQ model, indicating that selection against genetic redundancy is buffered; the parameters are the same as in (A); each simulation was run for time steps, and the number of loci was averaged over time with the first time steps discarded; error bars: 95% CI (100 replicate simulations and bootstrap with ).

Figure 5

Emergence of a cBQ depends on the population size and lineage longevity of cheaters; (A) the plot shows the population sizes of producers (circles, solid line) and cheaters (triangles, broken line) averaged over time as a function of public good production cost for the cBQ model; the population size of cheaters is high at moderate public good cost, but low when this cost is low or high; error bars: 95% CI (100 replicate simulations and bootstrap with ); the parameters are the same as in Fig. 3D, except that the benefit of neofunctional genes () is set to zero so that the results represent a situation before the evolution of a new gene function (the same parameters are used in panels (B)–(E); (B) the plot shows the average age of parasite lineages as a function of public good production cost; at intermediate public good production cost (), the lifespan of cheaters is high generations), leaving ample time for the exploration of sequence space and thus neofunctionalization is expedited; at high public good production cost (), the lifespan of cheaters is low ( generations); cheaters are thus rapidly turned over, reducing the opportunity for mutation events to yield new function; lower panels: snapshots of simulations depicting spatial distributions of cheaters (red) and producers (black) on the model grid for different public good costs (denoted by ); (C) at low cost (), cheaters are rare while producers are abundant; this reduces the number of loci at which a novel gene function may emerge; consequently, the Black Queen dynamic is only very weakly constructive; (D) at intermediate cost (), cheaters become abundant, and there are thus many more opportunities for the evolution of a new gene function; (E) at high cost (), cheaters induce travelling wave patterns, which slows down the evolution of novel functions.