Accepted May This article has been cited by other articles in PMC. Abstract Many bacteria are highly sexual, but the reasons for their promiscuity remain obscure. Did bacterial sex evolve to maximize diversity and facilitate adaptation in a changing world, or does it instead help to retain the bacterial functions that work right now?
In other words, is bacterial sex innovative or conservative? Our aim in this review is to integrate experimental, bioinformatic and theoretical studies to critically evaluate these alternatives, with a main focus on natural genetic transformation, the bacterial equivalent of eukaryotic sexual reproduction.
First, we provide a general overview of several hypotheses that have been put forward to explain the evolution of transformation. Next, we synthesize a large body of evidence highlighting the numerous passive and active barriers to transformation that have evolved to protect bacteria from foreign DNA, thereby increasing the likelihood that transformation takes place among clonemates.
Our critical review of the existing literature provides support for the view that bacterial transformation is maintained as a means of genomic conservation that provides direct benefits to both individual bacterial cells and to transformable bacterial populations. We examine the generality of this view across bacteria and contrast this explanation with the different evolutionary roles proposed to maintain sex in eukaryotes. Introduction Bacteria have long been appreciated as genomic shape-shifters that gain and lose genes with great regularity [ 1 ].
This fluidity is intuitively encapsulated by the idea of the core genome, or the fraction of genes shared by all or most strains of a given species. Although there are wide margins of error on these estimates, owing to difficulties of discerning recent from distant gene-transfer events or HGT occurring within or across species, it is clear that HGT plays a key role in shaping bacterial genomes.
Equally, the genes subject to HGT seem to have an outsized role on bacterial ecology and evolution, influencing where bacteria are found, what they can consume or degrade, their susceptibility to antibiotics and their virulence as pathogens, among others [ 3 ].
Because of these conspicuous benefits, it is easy to draw the conclusion that HGT is uniformly positive, and that bacteria engage in promiscuous sex in order to enhance their adaptability, much like meiotic sex enhances eukaryotic adaptation. However, this conclusion is too narrow and potentially misguided. First, it is important to bear in mind that there is a perception bias with respect to the benefits of HGT because the only instances of HGT that are observed in bacterial genomes are those that have passed the filter of natural selection [ 6 ].
Second, just like in eukaryotes, recombination in prokaryotes is associated with several potential costs that limit when, where and how recombination can occur [ 7 , 8 ]. Finally, while there are superficial similarities between eukaryotic sex and recombination in bacteria, the mechanisms underlying these processes are vastly diverged, as are their potential costs and benefits [ 6 ]. Thus, although it is tempting to look to eukaryotes to help understand the benefits of sex in prokaryotes, this should be done with caution.
The regulation and mechanisms underlying bacterial sex appear to have evolved independently across prokaryotic groups, although some of the core genes associated with recombination are broadly conserved.
Also, the frequency of recombination can be highly variable even within a species [ 6 , 9 , 10 ]. In spite of this, we believe a unifying benefit to bacterial sex can still be found. Here, we argue that the benefits of bacterial sex, and transformation in particular, lie in its genomic conservatism and not in the opportunities transformation can provide for genomic innovation.
Bacteria utilize three primary mechanisms to acquire exogenous DNA as a substrate for recombination: Although all three of these mechanisms of recombination contribute to HGT, only natural transformation is exclusively encoded by genes present on the bacterial chromosome. For that reason, transformation is the only process that may have evolved as a form of bacterial sex [ 6 , 11 ]. Accordingly, and as opposed to genes for conjugation and transduction that are predominantly carried by accessory infectious elements, the costs and benefits of transformation are borne solely and directly by the competent bacteria themselves.
In addition to the three classic mechanisms of DNA exchange, it has become clear in recent years that DNA may also be transferred between bacteria by other mechanisms, including nanotubes [ 12 ], micro-vesicles [ 13 ] or gene-transfer agents [ 14 ]. However, the prevalence and impact of these agents still remains to be established and we therefore focus on transformation for the remainder of this review.
We first discuss the diverse evolutionary costs and benefits of natural competence. Next, we outline the numerous strategies bacteria use to increase the likelihood that transformed DNA is derived from within the same species. Finally, we consider experimental and theoretical support for the idea of conservative sex and conclude with suggestions for further study. In addition to the discussion below, we also refer interested readers to several excellent reviews that provide more mechanistic or species-specific details of competence induction and transformation [ 15 — 17 ].
Costs and benefits of natural transformation a Physiological costs of transformation Natural transformation is coordinated by a large and complex molecular machinery dedicated to the uptake of DNA from the environment, its intracellular processing and, potentially, its genetic incorporation through recombination. Although there are similarities in the mechanisms of DNA binding, uptake and incorporation across species, distinct patterns of competence regulation together with the sporadic distribution of competence across bacteria imply that transformation has had multiple independent origins [ 18 , 19 ].
Equally, the rates of natural transformation can vary markedly within a single species [ 9 , 10 ], suggesting that transformation is evolutionarily labile and that bacteria face a trade-off between its costs and benefits. Potential costs are numerous and diverse. The very act of transformation is energetically costly, sometimes involving the transcription of more than genes, only a fraction of which are required for recombination [ 18 ].
This cost of entering this persister-like state was proposed to explain why only a fraction of cells in Bacillus subtilis become competent [ 21 ]. On the other hand, under conditions where rapidly dividing bacteria experience high levels of mortality e. Another cost arises through recombination itself because recombination is by default a DNA damaging process; evidence suggests that chromosomal integration of DNA that is successfully taken up by the cell can ultimately lead to double-stranded breaks that are lethal unless repaired [ 22 , 23 ].
Finally, there are the potentially considerable costs associated with competence-induced cell lysis in, for example, Streptococcus pneumoniae, where competent cells actively lyse non-competent members of the same population [ 24 ]. Were cells induced to become competent by starvation, DNA could in principle provide sufficient energy for continued replication or repair.
Consistent with this possibility, some species, like Haemophilus influenzae, require nutritional down-shifts to induce competence and purine depletion activates the competence activator sxy [ 26 ]. However, several factors argue against this hypothesis as a general explanation for the maintenance of transformation: In addition to these concerns, it remains uncertain, on energetic grounds, whether the costs of DNA transport are sufficiently offset by any metabolic savings provided by exogenous DNA [ 11 ].
Thus, despite the intuitive appeal of this idea, the evidence in its favour is currently limited. Early experimental evidence indicated an immediate benefit of DNA uptake on transformant survival relative to the remainder of the population in B.
However, these earlier results were countered in the same species by evidence that genotoxic stress did not induce competence [ 33 ], as predicted by the original idea.
This idea is supported by the fact that some, but not all [ 35 ], naturally transformable human pathogens such as S. Current evidence suggests, however, that the benefits of competence induction in these species may be unlinked from the effects of transformation per se i.
DNA uptake and integration [ 38 , 39 ]. Thus, although these reports reveal that at least some forms of stress can induce competence, they do not always provide evidence that this response is adaptive, nor provide clear indications of the mechanisms underlying these benefits. Furthermore, they do not take into account the abundant evidence showing that competence in many species is induced by quorum sensing, irrespective of exogenous stress, and that other unambiguous forms of stress, like temperature and pH, can even repress natural transformation [ 40 — 43 ].
This gives rise to a suite of potential benefits and costs of transformation associated with genetic recombination that are largely identical to those studied for meiotic sex in eukaryotes. Below, we briefly review two of the principles—epistasis and Hill—Robertson interference—through which transformation may be favoured, in both cases through either helping to purge deleterious or fix beneficial mutations.
Selection with epistasis, i. At the most basic level, recombination can have a detrimental effect as it breaks up co-adapted gene complexes e. This was shown theoretically to also provide a benefit to natural transformation in bacterial populations [ 49 , 50 ]. However, empirical work has shown that while negative epistasis sometimes exists, it is far from pervasive and there are many systems in which positive or no epistasis was reported reviewed in [ 51 , 52 ].
For this reason, the deterministic mutation hypothesis has been broadly disregarded as a main contender to explain the ubiquity of sex, including natural transformation. A second class of explanations for why recombination can be beneficial relies on the Hill—Robertson effect [ 53 , 54 ]. Here, an interaction between natural selection and stochastic effects in finite populations through random genetic drift or mutation produces genetic associations negative linkage disequilibria that reduce genetic variance for fitness.
By breaking up these associations, recombination can increase the efficacy of natural selection and genes that increase the recombination rate can be indirectly selected for. Advantages of recombination stemming from the Hill—Robertson effect come in different forms, and may involve both beneficial and deleterious mutations e. A number of mathematical and simulation models have been developed to specifically investigate variants of the Hill—Robertson effect in the context of bacterial sex, confirming that natural transformation can be favoured both in populations subject to recurring deleterious mutations [ 21 , 60 ] and in adapting bacterial populations [ 21 , 61 , 62 ].
Experimental work testing these ideas has focused on the Fisher—Muller model, the central prediction of which is that recombination accelerates the adaptation rate of sexual relative to asexual variants [ 57 , 58 ].
Evolution experiments with the Gram-negative bacterium Escherichia coli [ 63 ] undergoing plasmid-mediated recombination and the yeast Saccharomyces cerevisiae [ 64 ] support this hypothesis, while both reports strongly suggest the crucial role of recombination in relieving the effects of clonal interference.
Reduced clonal interference has also been proposed as a key genetic mechanism underlying the evolutionary maintenance of natural transformation [ 46 ]. For example, transformable H. By contrast, a study using the highly transformable species Acinetobacter baylyi failed to detect any consistent evolutionary advantage of natural transformation during generations of experimental evolution [ 66 ].
It was later demonstrated that the benefits of natural transformation in A. Similar context-dependent benefits of transformation were shown in the human pathogen S.
Here, competence for natural transformation was disadvantageous when populations evolved in benign experimental conditions, but not when they evolved in the presence of periodic mild stress sub-inhibitory concentrations of kanamycin [ 68 ].
In addition, evolving competent S. Taken together, the few experimental tests of Fisher—Muller advantages to bacterial transformation provide a mixed picture: By contrast, natural transformation can also affect allele frequencies within a population because of its inherent asymmetry: This was first recognized by Redfield [ 49 ], who showed with a mathematical model that if bacteria carrying deleterious alleles are more likely to die and thereby release their DNA into the environment than wild-type bacteria, this creates a bias towards taking up deleterious alleles and thus a distinctive cost of natural transformation that is independent from any effects derived from gene shuffling.
More recently, models of well-mixed and spatially structured populations were developed that explicitly incorporated a pool of free eDNA subject to decay [ 62 , 69 ]. These models recovered a similar detrimental effect of transformation in populations adapting to new environmental conditions because the eDNA may build up an over-representation of old, non-beneficial alleles.
This hypothesis is based on the observation that successful transformation only requires homology between short stretches of DNA at the ends of the transformed fragment and the host chromosome, while the regions in the middle can freely vary [ 72 , 73 ].
As a consequence, transformation can lead to either the incorporation of new genetic material as is usually emphasized or its expulsion, if, for example the transforming DNA lacks a MGE that is present in the transformed recipient.
But under what conditions would this work? They argue that as a consequence of this bias transformed cells will be more likely to incorporate short rather than long DNA stretches, in turn preferentially leading to the loss of MGE.
However, generalizing their idea, we suspect that such a bias may not be strictly necessary. Although transformation with long DNA fragments can lead to MGE acquisition, it can also—and potentially at the same rate—lead to their loss if transformed fragments cover the insertion site of the MGE.
Thus, as long as there is any natural transformation with DNA fragments that are shorter than the MGE in question, there should be an automatic bias towards exclusion of the element. The existence of this bias should then be independent of the size distribution of incorporated DNA, even though its magnitude may still be affected by the size distribution.
This more general version of the defence hypothesis remains to be specifically tested. In addition to corroborating their hypothesis by means of a mathematical model, Croucher et al. In a small preliminary analysis for this review, we screened the whole genomes of 10 strains of S. Consistent with the predictions of the Croucher et al. Although no strong conclusions can be drawn from these limited tests, we believe this approach is likely to be very powerful for further tests of these ideas, as it can tie experimentally validated differences in transformation rates to the number of mobile elements carried in a given genome.
The potential for consensus Despite the numerous studies of competence across a broad swath of competent bacterial species, there still remains considerable uncertainty about its function because no single explanation seems to capture the particulars of different species growing in different contexts.
And this may in fact be the answer: However, a consensus option may lie in the fact that not all transformable DNA is of equal appeal or value to each bacterial species. But it goes further than this. As we detail below, all competent bacterial species go to extreme lengths to ensure that the DNA they take up or recombine is either from the same species or from the same strain.