PERSPECTIVES

und ihre Anwendung auf die Abderhalden’schen (Cold Spring Harbor Laboratory Press, New York, 1998). Abwehrfermentreaktion. Hoppe-Seyler’s Zs für 13. Posner, G. L. & Ware, J. Mengele. The Complete Story. vived. Even if harmful exchange events were physiologische Chemie 277, 222–232 (1943). (McGraw–Hill, New York, 1986). 100-fold more common than beneficial ones, 12. Müller-Hill, B. Tödliche Wissenschaft. Die Aussonderung von Juden, Zigeunern und Geisteskranken. (Rowohlt, Acknowledgements we would only see the latter in genomes today. Reinbek, 1984). English translation: Murderous Science. I thank the Präsidentenkommission of the Max-Planck- So, finding transferred genes in modern Elimination by Scientific Selection of Jews, Gypsies and Gesellschaft (MPG) for giving me access to the letters of Others in Germany 1933–1945 (Afterword by J. Watson) Butenandt and the Archive of the MPG for their excellent help. genomes shows that some transfers, like some mutations, are adaptive, but this finding does not address the larger issue of the average costs and benefits of exchange. Filtering by natural selection is like the fil- OPINION tering of lottery outcomes by the media. On the basis of what we read in the newspapers, we would expect everyone who buys a lottery Do have sex? ticket to be a winner. Of course, most scien- tists know better than to buy lottery tickets, Rosemary J. Redfield but many have failed to apply the same logic to the processes that generate genetic diversi- Do bacteria have genes for genetic CONJUGATION or TRANSFORMATION, and can be ty. Research papers do not explicitly claim that exchange? The idea that the bacterial physically recombined into their chromo- genetic exchange must be adaptive because processes that cause genetic exchange somes by various cytoplasmic proteins. The we see its benefits and not its harmful conse- exist because of natural selection for this many sequenced bacterial genomes contain quences — if this were done, the error would process is shared by almost all abundant examples of genes that were unam- be obvious. Nevertheless, the error is probably microbiologists and population geneticists. biguously acquired by horizontal transfer. responsible for much of the complacency However, this assumption has been For example, most of the physiologically with which most biologists view the evolution perpetuated by generations of biology, important differences between Escherichia of genetic exchange. microbiology and genetics textbooks coli and Salmonella typhimurium result from without ever being critically examined. recombination: genes for lactose, citrate and Rigorous approaches to sex propanediol use, and indole production, have Until recently, scientific approaches to the Terms such as sex and recombination have all been acquired in this way4. problem of the evolution of sex have almost different meanings in different contexts. exclusively been the domain of theoretical Here, I use recombination to mean the How not to study selection for sex population genetics. The formulation of breaking and joining of DNA strands; genetic Because the ability to create new genetic com- explicit mathematical statements is a rigor- exchange, gene transfer or HORIZONTAL TRANS- binations affects FITNESS only indirectly and ous tool for evolutionary analysis, but this FER to refer to processes that produce new because the outcomes are intrinsically unpre- rigour often demands a corresponding sacri- genetic combinations; and meiotic sex to dictable, selection for the creation of new fice of relevance. Mathematical modelling mean the cyclical alternation between hap- genetic combinations is much harder to inves- can show how selection on the genes that loid and diploid stages in eukaryotes. Sex tigate than selection on processes that con- cause genetic exchange might act, but only refers to any process selected by the benefits tribute directly to survival or reproduction. under hypothetical and unrealistic assump- of genetic exchange. One reason why so much misunderstanding tions about the processes and their conse- Understanding the evolutionary causes of surrounds the evolution of sex is that the least quences, which are required if the equations genetic exchange in bacteria has important rigorous and most misleading evidence has are to be solvable. Computer simulations are implications for our understanding of the had the greatest influence, whereas the more versatile, as equations need not be evolution of meiotic sex in eukaryotes. The strongest has been mostly overlooked. solved but only applied repeatedly, but the primary function of meiotic sex seems to be The large number of transferred genes we assumptions that underlie the programming to produce new combinations of chromoso- find in modern bacterial genomes has misled must still be simple. For example, both theo- mal genes, but extensive work by population many researchers about the benefits of genetic retical and computer models of the evolution geneticists has been unable to show why this exchange. Many of the transferred genes are of meiotic sex often assume that all muta- would be beneficial1–3. If bacteria do have obviously beneficial to their new hosts and this tions have identical effects on fitness. As a genes that have evolved for genetic exchange, is frequently interpreted as conclusive evidence consequence, although both mathematical then they provide much-needed indepen- that gene transfer must be adaptive. The for- and computer modelling have been useful for dent systems in which to study how sex can eign origin of many of these genes is firmly showing that some explanations for meiotic evolve. If they do not, then meiotic sex must established, but the bacterial genomes that they sex are possible whereas others are not, they have evolved to provide eukaryotes with are found in are unfortunately a very biased have failed to produce solid answers2,3,5,6. benefits that are not needed by bacteria and record of evolutionary processes. The problem, The power of long-term selection experi- its evolutionary causes must be sought of course, is natural selection. Because natural ments on microbial cultures (‘experimental among eukaryote-specific phenomena. selection eliminates almost all deleterious evolution’) has only recently been appreciated7. Bacteria have several well-studied process- changes, the genomes of modern organisms These experiments tell us how selection can act es that can transfer genes, and the analysis of are the result of several billion years of evolu- under laboratory conditions, by testing experi- genome sequences has revealed that these tionary success stories, with not a single failure mentally whether a given set of conditions processes have made important contribu- represented. In a way, the sequences we see are leads to a change in the frequencies of certain tions to bacterial evolution. DNA can be a type of anecdotal evidence — each represents genotypes in a population. Because many transferred between cells by , a unique event that has, against the odds, sur- thousands of generations can be followed, the

634 | AUGUST 2001 | VOLUME 2 www.nature.com/reviews/genetics © 2001 Macmillan Magazines Ltd PERSPECTIVES experiments can detect relatively weak selective Most bacterial recombination is ‘homolo- processes. However, a considerable limitation “The large number of gous’; that is, the two recombining segments to laboratory selection experiments is that cul- have identical or near-identical sequences ture conditions inevitably fail to reflect the evo- transferred genes we find in and base pairing between their strands lutionarily relevant conditions that are experi- modern bacterial genomes replaces one with the other. The REC PROTEINS enced by bacteria in their natural RecA and RecBCD have key roles in this environments. The problem is not that experi- has misled many researchers process, along with other proteins (RecE, menters do not wish to use natural conditions, about the benefits of genetic RecF, RecG, RecJ, RecN, RecO, RecQ, but that it is usually impossible to determine exchange.” RuvABC, Ssb, PolA, DNA and DNA which features of the natural environments of gyrase A and B). Because mutations in the microorganisms are most important. genes that specify these proteins disrupt Neither theoretical nor experimental mod- recombination, the genes were often given els of evolution have shed much light on genet- What causes genetic exchange? ‘rec ’ names when first discovered. They were ic exchange in bacteria. Two models found that Bacterial genetic exchange is not like meiot- thought to function mainly in the ‘recombi- exchange could be beneficial only under condi- ic sex. Whereas meiotic sex regularly mixes nation pathways’ that were believed to have tions that were generally more restrictive than two complete sets of genes and randomly evolved to promote genetic exchange. Effects for sexual recombination in eukaryotes8,9. reassorts the alleles into new individuals, on DNA repair and overall viability were Another found that the genes that cause genet- bacterial recombinants form by processes noted, but were usually considered to be sec- ic exchange interfere with selection on the that are non-reciprocal and fragmentary, ondary. Decades of genetic and functional genes that affect mutation rates10. One well- and that are not regular components of bac- analysis have shown that DNA replication controlled selection experiment that has terial life cycles. Any one recombination and repair are, in fact, the primary functions addressed this problem found that introducing event transfers a single fragment of the of these proteins, and that these functions genetic exchange into laboratory populations chromosome from one cell, called the are achieved by mechanisms that also of E. coli did increase their genetic variation, ‘donor’, to another, called the ‘recipient’. increase recombination12,13. For example, but that there was no concomitant increase in Three well-studied processes are responsible RecBCD, RecG and the RUV PROTEINS all con- the rate or extent of adaptation11. for most naturally occurring DNA transfer: tribute to restarting stalled replication Fortunately, the most powerful way to transduction by bacterial viruses, conjuga- forks14–16, and RecA carries out a process investigate the evolution of genetic exchange tion by bacterial plasmids and DNA uptake called RECOMBINATIONAL REPAIR and also does not depend on mathematical tractability, by naturally competent bacteria (transfor- regulates repair by sensing DNA damage12. nor on assumptions about selectively impor- mation) (FIG. 1). Once in the cytoplasm of A less-common process is non-homolo- tant components of the environment. the recipient, transferred DNA fragments gous or illegitimate recombination, in which Instead, the nature of the genes and the escape degradation only if they physically two unrelated sequences become connected processes responsible for genetic exchange recombine with the chromosome. This usu- either by incorrect rejoining of broken ends can reveal how selection has acted in shaping ally occurs by replacing a recipient sequence or by insertion of one DNA segment into them. Because such genes and regulatory with a very similar sequence from the another. The former is mediated by DNA lig- mechanisms evolved in the natural world donor, although unrelated donor sequences ase, which is essential for DNA replication over evolutionary time, they are more sensi- can sometimes be added to the recipient and repair, and the latter by TRANSPOSASES, tive and accurate indicators of selection than chromosome. which are encoded by, and essential for, the laboratory evolution experiments or evolu- replication of transposable genetic elements. tionary theory can ever be. This kind of Recombination. Without the breaking and So, both homologous and non-homolo- analysis does not depend on, or suffer from, joining of DNA strands, DNA transfer could gous recombination are carried out by pro- preconceptions about the evolutionary func- never lead to new genetic combinations. So, teins that have other important cellular tion of the process being studied. In fact, most to understand the causes of genetic exchange functions. But how do we know that the of the evidence has been produced by molec- we need to find out why cells have proteins recombination activities of these proteins ular biologists and bacterial geneticists, who that cause recombination. The strongest evi- have not also been selected? We can never had little concern for the evolutionary issues dence comes from the phenotypes of prove that they have not, but there is no jus- that their results are helping to clarify. Below, I mutants that lack these proteins and from tification for invoking such selection, first consider the processes that contribute to molecular analyses of the protein activities. because selection for each primary function genetic exchange, and what our current This evidence has shown that these proteins is so strong that it dwarfs any possible selec- understanding of their mechanisms and reg- exist to promote DNA replication and repair, tion for genetic exchange, and second ulation reveals about their evolution. not genetic exchange. because each protein seems to promote exchange only as a side effect of its other activity. For example, mutations in any of DNA Physical transfer recombination the ruvA, ruvB or ruvC genes cause a 50% reduction in cell viability, measured under • Transduction conditions in which no genetic recombina- • Conjugation 17 • Transformation tion is possible . This extreme decrease in Donor cell Recipient cell Recipient cell viability would certainly preclude success in Figure 1 | Genetic exchange in bacteria. Transduction, conjugation and transformation can transfer a the natural environment and is more than DNA fragment from a chromosome of a donor cell to a recipient cell. Physical recombination can then sufficient to account completely for the evo- integrate this DNA into the recipient chromosome. lution of these proteins. Furthermore, their

NATURE REVIEWS | GENETICS VOLUME 2 | AUGUST 2001 | 635 © 2001 Macmillan Magazines Ltd PERSPECTIVES a Collapsed replication fork Transduction and conjugation. Two of the tions usually arise as side effects of the activ- well-studied DNA-transfer processes — ities of other genetic parasites, most com- transduction and conjugation — depend on monly the short transposable elements

RuvC infectious agents that move DNA from cell called insertion sequences. So, conjugation, to cell. In both, gene transfer seems to be a like transduction, seems to transfer host simple side effect of the infectious activities genes by accident. Collapsed of these agents (FIG. 3). The strongest evi- new strands dence of how natural selection acts on these Competence and transformation. Our Old strands processes is the location and action of the improved understanding of conjugation genes responsible for them. Transduction is and transduction, and of the that the most common process18 (FIG. 3a). cause physical recombination of the trans- Transduction is also used for strain con- ferred DNA, consistently supports the struction in the laboratory19. It is caused by hypothesis that transfer and recombination the many bacterial phages (viruses) that of chromosomal genes are unselected side b Recombination intermediate occasionally package host DNA instead of effects of processes that have evolved for phage DNA into viral particles and then other functions. However, the evolutionary inject this DNA into new cells. All the genes function of a third DNA-transfer process RuvC involved in transduction are on phage remains controversial. This third process is genomes, not host chromosomes, which the development of a state called compe- indicates that there is selection for transfer of tence, in which bacteria can take up DNA phage DNA but not host DNA. By promot- fragments from their environment. A cell Donor strands Recipient strands ing production of infectious phages, these the chromosome of which recombines with genes strongly enhance their own evolution- such a fragment might change its genotype ary success. No host genes promote the and thus become ‘transformed’.Some bacte- packaging of DNA (host or phage), which ria cannot become naturally competent, at indicates that this packaging probably has no least under laboratory conditions (in E. coli, Figure 2 | The role of RuvC in DNA replication significant benefit to the host. In many ‘competence’ refers to artificially permeabi- and recombination. RuvC is a crossover phages, the gene product that is responsible lized cells), but many can22. junction , which cuts DNA for initiating DNA packaging recognizes a Unlike conjugation and transduction, at the circled positions to resolve the topologically sequence in the phage genome and trans- competence is not caused by infectious identical Holliday junctions that arise during duction depends on this protein mistaking a agents. The genes required are all chromo- a | DNA replication and b | recombination. host sequence for the phage sequence20. somal, which indicates that the benefit of There is no evidence that such host DNA uptake is to the recipient, not to another mode of action in replication fully accounts sequences have been modified to promote genetic element or parasite23. (DNA donors for their mode of action in recombination. packaging in phage particles. cannot benefit as they are already dead.) And, The Ruv proteins contribute to viability by The infectious agents responsible for unlike mutations in recombination genes, resolving four-stranded DNA structures conjugation are mostly plasmids — small, mutations in competence genes do not have called HOLLIDAY JUNCTIONS, which arise when circular DNA molecules that replicate inde- marked effects on viability; mutants that are replication forks are stalled, and which pre- pendently of the host chromosome (FIG. 3b); unable to take up DNA usually grow well vent further DNA replication and thus kill some transposons can also cause conjuga- under standard culture conditions. the cell if unresolved. The role of RuvC in tion. Both types of conjugative element Although competence has usually been recombination is the same as its role in cause their host cells to form a connection thought to exist to favour genetic exchange23–25, replication (FIG. 2). When DNA recombines to cells that lack the element and to pass a homologous DNA can, in principle, also be it forms Holliday junctions that are topo- copy of the DNA of the element to the new used as a template to repair otherwise-lethal logically identical to those at stalled replica- host cell. If chromosomal DNA is connected DNA damage26. Such a repair function could tion forks and viable recombinants are only to the element it too will be transferred21. be much more important to the cell than produced if the junctions are resolved. Again, we can infer selection for transfer of genetic exchange, because DNA damage is There is no indication that the activity of the conjugative element but not for the host more harmful and more common than delete- RuvC has been in any way modified by genes because all the genes specific to conju- rious mutation or an unreliable environment. selection for genetic exchange. Similar gation are on the element and there are no (Theory predicts that both of these factors can analyses can be done for other ‘recombina- host genes that specifically promote conju- select for genetic exchange under some cir- tion’ proteins — for example, RecA has gal DNA transfer. Some host proteins are cumstances.) Experiments that tested whether repair and recombination activities that are required for the conjugation process, such competent Bacillus subtilis cells could use essentially identical12. The presence of a as DNA polymerase, but these also make externally supplied DNA for repair were incon- damaged base in the substrate for repair, but direct contributions to host fitness, which clusive26–28 and had two serious weaknesses. not for recombination, is the only difference fully explain their roles in conjugation; they First, this type of experiment was not suffi- between these events. So, the genetic make no distinction between host and con- ciently sensitive to detect modest but potential- exchange produced by the various DNA- jugative-element substrates. Nor are there ly significant differences in survival. Second, transfer processes seems to depend on sequences or genes that physically connect the cells needed special treatment to become recombination that occurs as a side effect of host DNA to conjugative plasmids and competent, because they would not take up DNA repair and replication. cause their transfer. Instead, these connec- DNA under standard culture conditions.

636 | AUGUST 2001 | VOLUME 2 www.nature.com/reviews/genetics © 2001 Macmillan Magazines Ltd PERSPECTIVES a DNA transfer by transduction However, experiments in both B. subtilis in which competence has traditionally been and Haemophilus influenzae have shown no induced by transferring cells to a ‘starvation’ connection between the cellular machinery medium. Induction of competence genes for sensing DNA damage and that for absolutely requires an increase in cyclic AMP, inducing competence29. Because all known a signal produced when preferred energy DNA-repair mechanisms are induced by sources are depleted36–38. A more recent find- the presence of DNA damage, the failure of ing is that an essential feature of the DNA damage to contribute to the regula- H. influenzae starvation medium is its lack of Phage-infected Recipient donor tion of competence strongly indicates that purine nucleotides and nucleosides, the pres- DNA repair is not the main function ence of which prevents the transcription of b DNA transfer by conjugation of competence. competence genes39.So,for H. influenzae, the Common Surprisingly, the most obvious, immediate regulation of competence fits the predictions F and inevitable benefit of DNA uptake has of the nutrient hypothesis. generally been overlooked or, more recently, discounted23,24. Like other molecules that are Donor Recipient taken up by bacterial cells, DNA can be used 30 “Surprisingly, the most Rare as a nutrient . Some bacteria might break it down as a source of carbon and nitrogen, but obvious, immediate and its primary use is likely to be as a source of Hfr inevitable benefit of DNA nucleotides for DNA and RNA synthesis. This spares resources that would otherwise be uptake has generally been Donor Recipient needed for nucleotide synthesis, a very overlooked or, more c Gene transfer by competence ‘expensive’ cellular process31. Furthermore, many competent bacteria live in very DNA- recently, discounted. Like rich environments, so this uptake might make other molecules that are a substantial contribution to the energy bud- get of the cell. For example, H. influenzae, taken up by bacterial cells, Streptococcus pneumoniae and Neisseria DNA can be used as a Dead donor Competent recipient meningitidis live in respiratory tract mucus µ 32 nutrient.” Figure 3 | Methods of DNA transfer. (~300 g DNA per ml of mucus ); a | Transduction is the phage-mediated transfer Helicobacter pylori and Campylobacter jejuni of host genetic information. In a phage-infected live in gastrointestinal mucus (~200–400 µg bacterial cell, fragments of the host DNA are DNA secreted into the gastric lumen every Nutritional signals also have roles in reg- occasionally packaged into phage particles and 10 min (REF. 33)); and B. subtilis lives in soil ulating competence in other bacteria, can then be transferred to a recipient cell. (>10 µg DNA per g of soil34). In laboratory although interpretations have been ham- b | Conjugation is the transfer of DNA from a donor cell to a recipient that requires cell-to-cell cultures, competent bacteria degrade most of pered by the common assumption that contact. Genes on conjugative plasmids, such the DNA they take up and use the released genetic exchange must be more important as the F plasmid, encode products that are nucleotides mainly for DNA synthesis35. than food. In B. subtilis, competence is nor- necessary for this contact, and replication and Although Gram-positive bacteria directly mally induced by transfer to a nutrient-lim- transfer of the plasmid to the recipient. When, internalize only one DNA strand23, in nature, ited liquid medium or to solid media that on rare occasions, the F plasmid becomes the nucleotides released by hydrolysis of the lack a required amino acid or base40.Many of integrated into the host chromosome (Hfr), conjugation results in a partial transfer of the other strand will also be efficiently taken up the factors that regulate competence in donor chromosome. c | Cells that are competent and used. DNA is likely to be of more value as B. subtilis reflect nutrient availability; CodY can take up free DNA from their environment. a source of nucleotides than for DNA repair, senses nitrogen levels41 and PtsG controls For all three methods of DNA transfer, the because cells continuously need nucleotides CATABOLITE REPRESSION42. Regulatory mecha- donor chromosomal DNA will only be for DNA and RNA synthesis even in the nisms are also shared with known nutrient- permanently maintained and expressed in the absence of damage. acquisition processes, such as secretion recipient cell if it is integrated into the recipient 43 genome by physical recombination. The nutrient hypothesis, like the DNA- of degradative enzymes . Furthermore, repair hypothesis, can best be tested by look- although the comEA and comEC genes in ing at its regulation. If DNA is mainly a B. subtilis are essential for DNA uptake, the Natural selection for competence source of nutrients, competence should be comEB gene in the same competence-regu- The ability to take up DNA is tightly regu- induced by nutritional signals. The most lated OPERON encodes dCMP deaminase, an lated in most naturally competent bacteria. important signals are likely to be the deple- that is required for salvage of Although this regulation is a disadvantage tion of nucleic acid pools and of the energy dCMP44. This protein has no role in DNA for those doing selection experiments in the resources needed for nucleotide synthesis. uptake, but its presence in this operon might laboratory, it is an advantage for those Most research into the regulation of compe- reflect a role in processing the deoxynu- wishing to understand selection in the nat- tence has not been motivated by an interest in cleotides that DNA uptake provides. In ural environment. If competence evolved to its function but, nevertheless, evidence of Acinetobacter, competence genes are maxi- provide templates for DNA repair, the most nutritional regulation is accumulating. mally expressed when nutrients become effective form of regulation should induce The clearest evidence for nutritional regula- depleted and when growth ceases in late sta- competence when DNA is damaged. tion of competence comes from H. influenzae, tionary phase, although DNA cannot be

NATURE REVIEWS | GENETICS VOLUME 2 | AUGUST 2001 | 637 © 2001 Macmillan Magazines Ltd PERSPECTIVES taken up until the cells are transferred transfer that it causes might simply be a to fresh medium45. The involvement of “Why have no genes been form of transduction by a phage that can QUORUM-SENSING PEPTIDES in competence regu- no longer specify its own replication51. lation in various bacteria has been interpret- selected to cause genetic Cellular regulation of GTA-encoded genes ed as an adaptation for genetic exchange exchange, given that could therefore reflect selection to reduce by inducing DNA uptake when DNA from beneficial recombinants harmful effects on its host, rather than to CONSPECIFICS is likely to be available46,47. optimize genetic exchange52. The short, However quorum-sensing mechanisms have made so many repeated sequences called Chi are another often have well-established roles in nutrient contributions to modern example, the true functions of which were acquisition; many control secretion of not originally appreciated. Chi sequences degradative enzymes that release nutrients bacterial genomes? The were once thought to be abundant in the E. for the cell to take up48. Other quorum-sens- explanation … new genetic coli genome because they are needed to ing functions might act as early warning sig- produce genetic exchange by homologous nals of the nutrient shortages that are likely combinations are, like recombination, but are now known to ori- to result from a high population density. mutations, more often entate RecBCD-mediated repair at DNA Of course, many aspects of competence replication forks14,53. are not yet understood. N. meningitidis and harmful than beneficial …” H. influenzae have sequence-biased DNA- Conclusions uptake systems that cause them to preferen- The analysis presented here avoids specula- tially take up DNA from their own or close- either that DNA uptake causes DNA dam- tion about hypothetical conditions and con- ly related species23. In both H. influenzae age or that incoming single-stranded DNA straints, by drawing conclusions from genes and S. pneumoniae, competence-regulating sends a false signal of DNA damage. and mechanisms that have been produced by proteins seem to control the expression of However, these and similar puzzles natural selection. In effect, investigations into genes that have no obvious connection to might disappear once their causes are better regulatory systems ask the bacteria which fac- competence49,50. Only a small fraction of B. understood. For example, the ‘gene-transfer tors have been important to them in their subtilis cells become competent in laborato- agent’ (GTA) of Rhodobacter capsulatus was natural environment over evolutionary time. ry cultures, which indicate that an impor- originally thought to have evolved for For conjugation, transduction and the tant component of regulation has been genetic exchange. It packages 3–4-kb frag- enzymes that cause physical recombination, overlooked. The induction of DNA-repair ments of chromosomal DNA into protein the evidence is robust: genetic exchange enzymes, such as RecA, in some competent particles that can inject the DNA into new occurs as an unselected side effect of processes cells has been interpreted as an adaptation cells. However, we now know that GTA is that evolved for more immediate functions. for recombination. Instead, it could mean encoded by a defective PROPHAGE, so the Although questions remain to be answered about competence, the accumulating evi- Glossary dence for its nutritional regulation is shifting the burden of proof onto those who favour a CATABOLITE REPRESSION PROTISTS genetic exchange function. Transcriptional repression of a prokaryotic operon by the Single-celled eukaryotes. metabolic products of the enzymes that are encoded by Why have no genes been selected to cause the operon. QUORUM-SENSING PEPTIDES genetic exchange, given that beneficial recom- Peptides secreted and detected by cells. Cells respond to binants have made so many contributions to CONJUGATION extracellular peptide only when cell densities are modern bacterial genomes? The explanation In , transfer of DNA from a donor cell to a sufficiently high (the ‘quorum state’) that the extracellular recipient cell is mediated by direct cell–cell contact. concentration of the peptide exceeds a threshold. is probably the same as for the processes that create mutations: new genetic combinations CONSPECIFICS REC PROTEINS are, like mutations, more often harmful than Members of the same species. A general class of protein that participates in beneficial, and although the rare beneficial recombination. outcomes have been preserved, the processes FITNESS A measure of the capacity of an organism to survive and RECOMBINATIONAL REPAIR themselves have been selected against because reproduce. DNA repair made possible when a damaged DNA strand of their usually harmful outcomes. base-pairs with a complementary undamaged strand from Many factors have contributed to mis- HOLLIDAY JUNCTIONS a different molecule. conceptions about bacterial sex. One is ter- Cross-shaped junctions at which four strands of DNA minology. Molecular biologists and popula- meet and exchange partners, an important intermediate RUV PROTEINS of recombination. Proteins that translocate and resolve Holliday junctions. tion geneticists both use the term ‘recombination’,but the first group means HORIZONTAL TRANSFER TRANSDUCTION the machinery that breaks and joins DNA, Acquisition of genetic information from another cell. Virus- or phage-mediated introduction into a cell of a whereas the second means the new genetic DNA fragment derived from a different cell. OPERON combinations that this machinery can A genetic unit or cluster that consists of one or more TRANSFORMATION produce. Gene names can also be misleading genes that are transcribed as a unit and are expressed in a Change of the genotype of a cell brought about by uptake to non-specialists; the names of ‘rec’genes coordinated manner. of free DNA. reflect their laboratory discovery, not their

PROPHAGE TRANSPOSASE primary function. Another very misleading An inactive bacteriophage genome integrated into the An enzyme that carries out the site-specific DNA factor is the bias introduced by natural selec- host genome. recombination required for transposition. tion, which sweeps all the deleterious out-

638 | AUGUST 2001 | VOLUME 2 www.nature.com/reviews/genetics © 2001 Macmillan Magazines Ltd PERSPECTIVES comes under the carpet, giving the false 10,000-generation experiment with bacteria. Proc. Natl 34. Blum, S. A. E., Lorenz, M. G. & Wackernagel, W. Acad. Sci. USA 96, 3807–3812 (1999). Mechanism of retarded DNA degradation and impression that most exchange is adaptive. 8. Redfield, R. J. Evolution of bacterial transformation: is prokaryotic origin of DNases in nonsterile soils. Syst. Yet another is the erroneous belief that selec- sex with dead cells ever better than no sex at all? Appl. Microbiol. 20, 513–521 (1997). Genetics 119, 213–221 (1988). 35. Pifer, M. L. & Smith, H. O. Processing of donor DNA tion for ‘evolvability’ will override selection 9. Redfield, R., Schrag, M. & Dean, A. The evolution of during Haemophilus influenzae transformation: analysis for viability. Although it is true that evolv- bacterial transformation: sex with poor relations. using a model plasmid system. Proc. Natl Acad. Sci. Genetics 146, 27–38 (1997). USA 82, 3731–3735 (1985). ability — the ability to generate adaptive 10. Tenaillon, O., Le Nagard, H., Godelle, B. & Taddei, F. 36. Dorocicz, I., Williams, P. & Redfield, R. J. The genetic variation — will be indirectly Mutators and sex in bacteria: conflict between adaptive Haemophilus influenzae adenylate cyclase gene: strategies. Proc. Natl Acad. Sci. USA 97, cloning, sequence and essential role in competence. J. favoured by selection, this is usually much 10465–10470 (2000). Bacteriol. 175, 7142–7149 (1993). too weak to counteract direct selection on 11. Souza, V., Turner, P. E. & Lenski, R. E. Long-term 37. Macfadyen, L. P., Dorocicz, I. R., Reizer, J., Saier, M. H. experimental evolution in . 5. Effects of Jr & Redfield, R. J. Regulation of competence the maladaptive variation that is also gener- recombination with immigrant genotypes on the rate of development and sugar utilization in Haemophilus ated. This error underlies Weismann’s widely bacterial evolution. J. Evol. Biol. 10, 743–769 (1997). influenzae Rd by a phosphoenolpyruvate:fructose 12. Cox, M. M. Recombinational DNA repair in bacteria and phosphotransferase system. Mol. Microbiol. 21, accepted hypothesis that sexual reproduc- the RecA protein. Prog. Nucleic Acid Res. Mol. Biol. 63, 941–952 (1996). tion exists to prevent extinction by creating 311–366 (1999). 38. Macfadyen, L. P. Regulation of Intracellular cAMP 54 13. Cox, M. M. et al. The importance of repairing stalled Levels and Competence Development in Haemophilus genetic differences . replication forks. Nature 404, 37–41 (2000). influenzae by a Phosphoenolpyruvate:Fructose What does the study of bacterial genetic 14. Kuzminov, A. Collapse and repair of replication forks in Phosphotransferase System. Ph.D. thesis, Department Escherichia coli. Mol. Microbiol. 16, 373–384 (1995). of Zoology, University of British Columbia, British exchange processes indicate about the evolu- 15. Vincent, S. D., Mahdi, A. A. & Lloyd, R. G. The RecG Columbia, Canada (1999). tion of meiotic sex? Mutation and accidental protein of Escherichia coli dissociates 39. Macfadyen, L. P. et al. Competence development by R-loops. J. Mol. Biol. 264, 713–721 (1996). Haemophilus influenzae is regulated by the availability of genetic exchange provide bacteria with all 16. Seigneur, M., Bidnenko, V., Ehrlich, S. D. & Michel, B. nucleic acid precursors. Mol. Microbiol. 40, 700–707 the genetic variation they need, so perhaps RuvAB acts at arrested replication forks. Cell 95, (2001). 419–430 (1998). 40. Hauser, P. M. & Karamata, D. A rapid and simple eukaryotes evolved sexual reproduction 17. Lloyd, R. G. Conjugational recombination in resolvase- method for Bacillus subtilis transformation on solid because they get much less accidental deficient ruvC mutants of Escherichia coli K-12 media. Microbiology 140, 1613–1617 (1994). depends on recG. J. Bacteriol. 173, 5414–5418 (1991). 41. Serror, P. & Sonenshein, A. L. CodY is required for exchange than bacteria do, or because they 18. Milkman, R. et al. Molecular evolution of the Escherichia nutritional repression of Bacillus subtilis genetic need much more. Neither seems especially coli chromosome V. Recombination patterns among competence. J. Bacteriol. 178, 5910–5915 (1996). strains of diverse origin. Genetics 153, 539–554 (1999). 42. Frisby, D. & Zuber, P. Mutations in pts cause catabolite- PROTISTS likely. Meiotic sex arose in , not 19. Sternberg, N. L. & Maurer, R. Bacteriophage-mediated resistant sporulation and altered regulation of spo0H in plants or animals55. Both viral infections and generalized transduction in Escherichia coli and Bacillus subtilis. J. Bacteriol. 176, 2587–2595 (1994). Salmonella typhimurium. Methods Enzymol. 204, 43. Kunst, F., Msadek, T., Bignon, J. & Rapoport, G. The phagocytosis are likely to cause genetic 18–43 (1991). DegS/DegU and ComP/ComA two-component exchange in protists, and this exchange 20. Vogel, W. & Schmieger, H. Selection of bacterial pac systems are part of a network controlling degradative might easily be as common in protists as sites recognized by Salmonella phage P22. Mol. Gen. enzyme synthesis and competence in Bacillus subtilis. Genet. 205, 563–567 (1986). Res. Microbiol. 145, 393–402 (1994). exchange is in bacteria. Conversely, there is 21. Frost, L. S. : everybody’s doin’ it. 44. Inamine, G. S. & Dubnau, D. ComEA, a Bacillus subtilis no obvious reason why protists would need Can. J. Microbiol. 38, 1091–1096 (1992). integral membrane protein required for genetic 22. Solomon, J. M. & Grossman, A. D. Who’s competent transformation, is needed for both DNA binding and more genetic exchange than bacteria: the when: regulation of natural genetic competence in transport. J. Bacteriol. 177, 3045–3051 (1995). two groups have similar mutation rates and bacteria. Trends Genet. 12, 150–155 (1996). 45. Porstendorfer, D., Gohl, O., Mayer, F. & Averhoff, B. 23. Dubnau, D. DNA uptake in bacteria. Annu. Rev. ComP, a pilin-like protein essential for natural 56 overlapping genome sizes . We can only Microbiol. 53, 217–244 (1999). competence in Acinetobacter sp. strain BD413: hope that research into the molecular mech- 24. Levin, B. R. & Bergstrom, C. T. Bacteria are different: regulation, modification, and cellular localization. observations, interpretations, speculations, and J. Bacteriol. 182, 3673–3680 (2000). anisms and regulation that underlie meiotic opinions about the mechanisms of adaptive evolution in 46. Tortosa, P. & Dubnau, D. Competence for sex will provide new insights. prokaryotes. Proc. Natl Acad. Sci. USA 97, 6981–6985 transformation: a matter of taste. Curr. Opin. Microbiol. (2000). 2, 588–592 (1999). Rosemary J. Redfield is at the Department 25. Mortier-Barriere, I., Humbert, O., Martin, B., 47. Morrison, D. A. & Lee, M. S. Regulation of competence Prudhomme, M. & Claverys, J. P. Control of for genetic transformation in Streptococcus of Zoology, University of British Columbia, recombination rate during transformation of pneumoniae: a link between quorum sensing and DNA Vancouver, British Columbia, Canada V6T 1Z4. Streptococcus pneumoniae: an overview. Microb. processing genes. Res. Microbiol. 151, 445–451 e-mail: [email protected] Drug Resist. 3, 233–242 (1997). (2000). 26. Michod, R. E., Wojciechowski, M. & Hoelzer, M. DNA 48. Swift, S., Throup, J. P., Williams, P., Salmond, G. P. & Links repair and the evolution of transformation in the Stewart, G. S. Quorum sensing: a population-density bacterium Bacillus subtilis. Genetics 118, 31–39 component in the determination of bacterial phenotype. DATABASE LINKS RecA | Rec | RecE | RecF | (1988). Trends Biochem. Sci. 21, 214–219 (1996). RecG | RecJ | RecN | RecO | RecQ | RuvABC | 27. Wojciechowski, M. F., Hoelzer, M. A. & Michod, R. E. 49. Macfadyen, L. P. Regulation of competence DNA repair and the evolution of transformation in development in Haemophilus influenzae. J. Theor. Biol. SSB | PolA | DNA ligase | DNA gyrase | CodY | Bacillus subtilis. II. Role of inducible repair. Genetics 207, 349–359 (2000). PtsG | comEA | comEC | comEB 121, 411–422 (1989). 50. Rimini, R. et al. Global analysis of transcription kinetics 28. Hoelzer, M. A. & Michod, R. E. DNA repair and the during competence development in Streptococcus FURTHER INFORMATION Salmonella | evolution of transformation in Bacillus subtilis. III. Sex pneumoniae using high density DNA arrays. Mol. Escherichia coli | Redfield lab with damaged DNA. Genetics 128, 215–223 (1991). Microbiol. 36, 1279–1292 (2000). 29. Redfield, R. J. Evolution of natural transformation: 51. Lang, A. S. & Beatty, J. T. Genetic analysis of a bacterial testing the DNA repair hypothesis in Bacillus subtilis genetic exchange element: the gene transfer agent of and Haemophilus influenzae. Genetics 133, 755–761 Rhodobacter capsulatus. Proc. Natl Acad. Sci. USA 97, 1. Kondrashov, A. S. Classification of hypotheses on the (1993). 859–864 (2000). advantage of amphimixis. J. Hered. 84, 372–387 30. Redfield, R. J. Genes for breakfast: the have your cake 52. Lang, A. S. & Beatty, J. T. The gene transfer agent of (1993). and eat it too of transformation. J. Hered. 84, 400–404 Rhodobacter capsulatus and ‘constitutive transduction’ 2. Lenormand, T. & Otto, S. P. The evolution of (1993). in prokaryotes. Arch. Microbiol. 175, 241–249 (2001). recombination in a heterogeneous environment. 31. Stouthamer, A. H. The search for correlation between 53. Horiuchi, T. & Fujimura, Y. Recombinational rescue of Genetics 156, 423–438 (2000). theoretical and experimental growth yields. Int. Rev. the stalled DNA replication fork: a model based on 3. Keightley, P. D. & Eyre-Walker, A. Deleterious mutations Biochem. 21, 1–47 (1979). analysis of an Escherichia coli strain with a and the evolution of sex. Science 290, 331–333 (2000). 32. Matthews, L., Spector, S., Lemm, J. & Potter, J. chromosome region difficult to replicate. J. Bacteriol. 4. Lawrence, J. G. & Ochman, H. Molecular archaeology Studies on pulmonary secretions. 1. The overall 177, 783–791 (1995). of the Escherichia coli genome. Proc. Natl Acad. Sci. chemical composition of pulmonary secretions from 54. Weismann, A. Die Bedeutung der sexuellen USA 95, 9413–9417 (1998). patients with cystic fibrosis, bronchiectasis and Fortpflanzung fur die Selektiontheorie (Gustav Fischer, 5. Burt, A. Perspective: sex, recombination, and the laryngectomy. Am. Rev. Respir. Dis. 88, 199–204 Jena, 1886). efficacy of selection — was Weismann right? Evolution (1963). 55. Redfield, R. J. A truly pluralistic view of sex and 54, 337–351 (2000). 33. Hunt, J. N., Smith, J. L., Jiang, C. & Kessler, M. S. recombination. J. Evol. Biol. 12, 1043–1046 (1999). 6. West, S. A., Lively, C. M. & Read, A. F. A pluralist Effect of synthetic prostaglandin E1 analog on aspirin- 56. Drake, J. W. The distribution of rates of spontaneous approach to sex and recombination. J. Evol. Biol. 12, induced gastric bleeding and secretion. Dig. Dis. Sci. mutation over viruses, prokaryotes, and eukaryotes. 1003–1012 (1999). 28, 897–902 (1983). Ann. NY Acad. Sci. 870, 100–107 (1999). 7. Papadopoulos, D. et al. Genomic evolution during a

NATURE REVIEWS | GENETICS VOLUME 2 | AUGUST 2001 | 639 © 2001 Macmillan Magazines Ltd