Copyright 0 1988 by the Genetics Society of America Adaptive Evolution That Requires Multiple Spontaneous Mutations. I. Mutations Involving an Insertion Sequence Barry G. Hall Molecular and Cell Biology U-44, University of Connecticut, Storrs, Connecticut 06268 Manuscript receivedJuly 18, 1988 Revised copy accepted September 7, 1988 ABSTRACT Escherichia coli K12 strainx342LD requires two mutations in the bgl (@-glucosidase)operon, bglRo + bglR+ and excisionof IS103 from within bglF, in order to utilize salicin. In growing cells the two mutations occurat rates of 4 X lo-* per cell division and <2 X 10”‘ per cell division, respectively. In 2-3-week-old colonies on MacConkey salicin plates the double mutants occur at frequencies of lo-’ per cell, yet the rate of an unselected mutation, resistance to valine, is unaffected. The two mutations occur sequentially. Colonies that are 8-12 days old contain from 1% to about 10% IS103 excision mutants, from which the Sal+ secondary bglRo + bglR+ mutants arise. It is shown that the excision mutants are notadvantageous within colonies; thus, they must result from a burst of independent excisions late in the life of the colony. Excision of IS103 occurs only onmedium containing salicin, despite the fact that the excision itself confers no detectable selective advantage and serves only to create the potential for a secondary selectively advantageous mutation. DAPTIVE evolution is an active, dynamic proc- 1986). This study focuses on adaptive mutations that A ess that is presumed to be subject to biological result fromthe movement of mobile genetic elements. constraints, and it is usually assumed that evolutionary Molecular evolutionists and molecular systematists processes can ultimately be understood in terms of make certain assumptions about the ultimate source thoseconstraints. The initial event in evolutionary of genetic variation. First, it is assumed that sponta- processes is thegeneration of geneticvariation by neous mutations arise randomly. They may be fixed mutation;thus, an understanding of evolutionary into populations by selection or by chance events, but processes depends upon an accurate understanding of their occurrence is presumed to be random and to be the natureof spontaneous mutations and spontaneous relatively unaffected by normal environmental con- mutation rates. ditions. Base pair substitutions have been studiedvery thor- Because mutations arerare events, it is usually oughly, but there is still relatively little precise infor- assumed that multiplemutations arethe resultof mation onthe ratesat which theseevents occur. temporallyseparate, andtherefore independent, Studies on the molecular basis of substitutions have events. This assumption underlies most methods for focused on errors that occur during DNA synthesis constructing phylogenies on the basis of molecular or during repair of damaged DNA. That focus, in data. Therefore,if some phenotype requires multiple part, has led to an infrequently stated generalassump- mutations in order tobe exhibited, then each of these tion that replication errors are theprimary source of mutations is expected to besequentially fixed into the spontaneousmutations and virtually all studiesof population either by drift or by selection. This study spontaneousmutation rates in bacteriahave been provides evidence suggesting that both of these as- conducted on exponentially growing cultures under sumptions need to be critically reexamined. conditions that arealmost never found in nature. Microorganisms have proven to be useful model Although we usually think of mutations in termsof systems for studying evolutionary processes (reviewed base pair substitutions, mutations include any changes in several articles in MORTLOCK1984; HALL 1983; in the structureof the genome. It is becoming increas- HARTL1986), and one of the most powerfulap- ingly clear that mobile genetic elements play impor- proaches has been that of “directed” or “experimen- tant roles in creating genetic variation in a wide vari- tal” evolution (HALL 1982a, 1983; CLARKE 1984). In ety of organisms (PFEIFERet al. 1983; FEDEROFF1983; a variety of studies I have selected spontaneousE. coli MCCLINTOCK1965; CIAMPI,SCHMID and ROTH 1982; mutants that have evolved the ability to utilize novel SCORDILIS,REE and LFSSIE 1987; PRENTKIet al. 1986; carbonand energy sources(HALL 1982b,1983; BARSOMIANand LESSIE 1986; SAEDLERet al. 1980; KRICKERand HALL 1984; PARKER and HALL1988). MILLERet al. 1984; MILLER,DYKHUIZEN and HARTL In most ofthose studies Escherichia coli cells were 1988; BRECLIANOand KIDWELL1983; HARTLet al. streaked onto thesurface of MacConkey agar indica- Genetics 120 887-897 (December, 1988) 888 B. G. Hall tor plates that contained a novel sugar which the cells IS103 were incapable of utilizing. On MacConkey plates the cells form colonies by utilizing small peptides in the medium, and the colonies reach a maximum size of T-7 I I I about 10’ cells in two to three days. (Colony size is bglR I bglG I bglF I bg lB presumably limited by local exhaustion of the primary growth resource, peptides, from the medium.) These colonies are white because they are unable to ferment FIGURE 1.-Structure of the bgl operon in strain ~342.IS103 is the novel sugar, but after a periodof seven to 30 days inserted into the bglF gene. (depending upon the specific new function being se- 1987). We recently reported that the x342 line of E. lected), outgrowths or papillae appear on the surface coli carries a 1.4-kb foreign DNA fragment inserted ofmany colonies. The papillae are red, indicating into the bgEF gene (PARKER,BETTS and HALL 1988) sugar fermentation, and it is easy to streak out their (Figure 1). Recent results (B. G. HALL,unpublished cells on identical medium to obtain single red colonies, data) show that this foreign DNA is a newly discovered and thus to isolate pure cultures of cells that have insertion sequencenow designated IS 103. Since strain evolved the ability to utilize a novel sugar. x342 carries the wild-type bglR’ allele (PARKER,BETTS This approach is particularly relevant to the ques- and HALL1988), two events (excision of IS103 from tion of the nature of spontaneous mutations because within bglF and activation of bglR) are required to it has permitted the isolation of mutants that cannot permit the strain to grow on salicin. Since we had be obtained by direct selection on minimal medium. measured the spontaneous rate of bglR’ + bglR+ as One of the most thoroughlystudied experimental 2-6 X lo-’ per cell division, and had shown that the evolution systems is the evolved &galactosidase (EBG) rate of excision of IS103 from within bgZF was <2 X system in which E. coli strains bearing a deletion for 10”’ per celldivision (PARKER,BETTS and HALL the major portion of the lac2 gene were forced to 1988), it seemed unlikely that strain x342 colonies evolve a new system, the ebg operon, for catabolism consisting of about lo9 cells would generate salicin- of lactose and other 0-galactoside sugars (reviewed in utilizing (Sal+)papillae. HALL1982a, 1983).Several years ago we showed that In this paper I show that after less than 2 weeks of all of the ebg mutants, isolated as single-step papillae incubation, 60% of the x342 colonies on MacConkey on MacConkey lactose medium,had to carry two salicin medium generate Salf papillae. The source of spontaneousmutations, one in thestructural gene these double mutants is determined, and the evolu- ebgA and one in the regulatory gene ebgR, in order to tionary implications of the results are discussed. grow on lactose (HALLand CLARKE1977). I later pointed out that these spontaneous double mutants MATERIALS AND METHODS occurred some 108-fold morefrequently than ex- pected on the basis of independent mutations (HALL Genetic nomenclaturefor the bgl operon: Until recently “bglC” was used to designate the gene specifying the @- 1982a). glucoside transport system and “bgls” todesignate the gene Some recent observations with another system have specifying a positive regulatory locus (PRASADand SCHAE- again suggested that double mutants may occur sev- FLER 1974). Two recent papers (MAHADEVAN,REYNOLDS eral orders of magnitude more frequently than ex- and WRIGHT1987; SCHNETZ,TOLOZYKI and RAK 1987) reversed those definitions. In order to avoid further confu- pected. sion, BARBARABACHMANN, curator of the E. coli Genetic The bgl operon is cryptic in E. coli and cannot be Stock Center, has renamed these loci (B. BACHMANN,per- expressed unless a mutation in a site or short region sonal communication). In this paper we use the new desig- designated bglR activates theoperon (PRASADand nations as they will appear in the next edition of the linkage SCHAEFLER1974). When it is activated, the operon is map of E. coli. bglF now designates the @-glucoside-specific enzyme I1 of the phosphotransferase system, ie., the trans- inducible and permits utilization of the P-glucoside port protein. bglG now designates the gene encoding the sugars salicin and arbutin. The bglF gene specifies a positive regulator of expression of the bgl operon. bglB, @-glucosidetransport protein that both transports and whose definition is unchanged, specifies phospho-o-glucosi- phosphorylates its substrates, and the bgZB gene spec- dase B. E. coli strains: Strain HfrC (HfrC metBl relA SPOT) ifies a phospho-P-glucosidase (PRASADand SCHAEFLER K12 and strain x342 (HfrCproC metBl relA SPOTX- bglF::IS103) 1974). The wild-type allele of bglR is designated bglRo were obtained from the E. coli Genetic Stock Center.