is in resting e.coli

8/7/2019 IS in resting e.coli

1/10

Copyright 0 1994 by the G enetics Society of America

Insertion Sequence-Related Genetic Variationin Resting Escherichia coli K-12Thierry Naas,* Michel Blot,* WalterM. Fitcht and Werner Arber*

*Abteilung Mikrobiologie, Biozentrum der Universitat Basel, CH-4056 Basel, Switzerland, and tDepartmen t of Ecology andEvolutionary Biology, University of California, Zruine, California 92717Manuscript received September 13, 1993Accepted for publication November 10, 1993

ABSTRACTBacterial subclones recovered from an old stab culture of Escherichia coliK-12 revealed a high degreeof genetic diversity, which occurred in spite ofa very reduced rate of propagation during storage. Thisconclusion is based ona pronounced restriction fragment length polymorphism(RFLP) detected uponhybridization with internal fragments of eight resident insertion sequences ( I S ) . Genetic diversity wasdependent on the I S considered and, in many cases, a clear consequence of I S transposition.I S 5 wasparticularly active in the generation of variation. All subclones in whichIS30 had been active testify toa burst of IS30 transposition. This was correlated with a loss of prototrophy and a reduced growthon rich media.Apedigree of the entire clone couldbe drawn fromthe RFLPpatterns of the subclones.

Out of 118 subclones analyzed,68 different patterns were found but the putative ancestral populationhad disappeared. A few patterns were each represented by several subclones displaying improvedfitness. These results offer insightsinto the role of IS elements in the plasticityof the E . coli genome,and they further document that enzyme-mediatedD N A rearrangements do occur in resting bacterialcultures.

STAB cultures kept in tightly sealed glass tubes atroomtemperature offer a simple, convenientmethod of storage of bacterial strains for very long pe-riods. However, it is well known, to most scientists usingstabs, that these do no tguarantee geneticstability ofthestored bacterial strains. This is somewhat surprisingsince after the short periodof exponential growth fol-lowing inoculation, on e expects no fur the rgrowth-or atmost only a very slow residual growth-to occur duringstorage. Transposition of residential mobile genetic el-ements is a most common cause of mutation inbacteria(RODRIGUEZet u l. 1992). It was postulated thatunder thesteady conditions in stab cultures, transposition of in-sertion sequence(IS) elements might contributesignifi-cantly to genetic instability. This was based on the ob-servations (a) that ISmediated transposition does notrequire chromosomal replication [see GALASand CHAN-DLER (1989) for transposition mechanisms] ; (b ) that le-thal mutations accumulated inthe P1 prophage instabswere oftenrelatedto IS 2 transposition ( A R B E R et al.1978); (c) that the copy number of IS5 in Escher i chiacoli increases upon prolonged storage in stabs (GREENe t a l. 1984); (d) thatthe transposition frequency ofsome mobile elements seems to increase under starvingconditions (HALL 1988; SHAPIROand HIGGINS1989;MIT-TLER and LENSKI1990). While much is known on theDNA sequences of IS elements an d ontheir molecularbiology, lesssolid knowledge is availableon mechanismsgoverning their persistence in populations [see AJIOKAan d HARTL (1989) for a review] and on their effectiveevolutionary role [see CAMPBELL (1981) and HARTLet al.(1984) for discussion].Genetics 136 721-730 (March, 1994)

Eight different IS elements are identified as residenton the chromosome of E . coli K-12: IS2 (OHTSUBOan dOHTSUBO1978); IS2 (GHOSALand SAEDLER1979); IS3(TIMMERMANand Tu 1985); IS4 ( K L A E Ret a l. 1981); IS5(ENGLERand BREE1981); IS30(DALRYMPLEet al. 1984);1985; SENGSTAGet al. 1986). Several transposable ele-ments, most often transposons, have been also intro-duced into E . coli strains as tools for mutagenesis butwere not involved in our study. The total length of allknown IS copies in E . coli K-12 is about 50 kb (ca. 1%of the chromosome), andtheir distributionon thechro-mosome is no t fully at random,since some regions, e . g . ,near thereplication origin, are freeof elements (BIRKEN-BIHL an d VIELMETTER1989). The copy number of studiedIS elements and their distribution on the chromosomevary greatly among natural isolates of E . col i (BRAHMAetal. 1982; GREENet al . 1984; H A R T L an d SAWYER1988a,b;SAWYERand HARTL1986). Nevertheless their locationscould be mapped o n the chromosome of the strainW3110 (BIRKENBIHLand VIELMETTER1989, 1991; UMEDAand OHTSUBO1989, 1990a,b, 1991).

In this repor t we compare the restriction fragmentlength polymorphism (RFLP) of subclones made fromindividual viable cells recovered from a stab culture ofthe E . coli K-12 strain W3110, which had been storedinou r collection for the last 30 years. The hybridizationprobes were internal sequences of various IS elementsresident on its chromosome. The E . col i strain W3110 ismost likely the genetically best known strain for bothitsphysical (KOHARAet al. 1987) and genetic (BACHMANN1990) maps. We show here that this strain has suffered

IS250 (SCHWARTZ et Ul . 1988), IS186 (KOTHARYet U l .


2/10

722 T. Naas et al .intense genetic alterations during prolongedstorage. Itsresidential IS elements, particularly IS5 and IS30, but toa lesser extent also others, have caused an impressivegenetic variance as revealed in the restriction patterns.Indeed, all tested subclones which were recovered fromthe stab differed from the input strain by several muta-tional steps, most of which must be transpositional andthus enzyme-mediated and this under conditions of seri-ously reduced metabolic activities. Environmentaland ev elutionary implicationsof these findings are discussed.

MATERIALSANDMETHODSStrain, storageandsampling: Strain W3110 isa derivative ofE. coli K-12 (BACHMANN1987). In the late 1950s a culture ofW3110 came from the California Institute ofTechnology, Pasa-dena, to the University of Geneva, where it was used in ex-periments involving hgal transduction (ARBER 1958). In thosedays laboratory cultures for routine work were usuallykept onslants stored in the refrigerator and renewed every few weeks.In the spring of 1960 a stab was prepared with a cultureof thisstrain. Stabs are small glass vials of about 4 ml volume con-taining 2 mlof growth medium (nutrient broth, Difco, 8g/liter; Bacto agar, Difco, 7.6 g/l iter ). The stab is inoculatedby stabbing a loopful of the bacterial culture all the way downto the bottom of the solid medium. It is then covered with acotton plug and incubated at 37" until bacterial growth is vis-ible. Growth is most intense on top of the trace at the surfaceof the medium. T he tube is then sealed with a paraffined cork.Depending on thequality of the cork, stabs kept atroom tem-perature are protected from drying out, and bacteria keeptheir viability for many years. Ou r stabs are stored in a closedcupboard at a temperature of 20-25".The stab prepared in 1960 with W3110 wasonly occasionally

opened in order to take a small sample of the culture andresealed immediately. At two occasions, in 1965 and in 1972this stab culture was renewed as follows. The old stab wasopened, an agar plug was taken from the surface area with aplatine loop and introduced into a fresh stab. Thus the stabculture which we have now analyzedwas from a third genera-tion stab. I t was taken from the collection in 1990, i.e., aftera total storage under stab conditions for 30 years. For thispurpose an agar plug was removed from the stab, resuspendedin Luria broth (LB) and aliquots were immediately plated onLBagar. After overnight incubation, 118 colonies were ran-domly picked regardless of the colony size and grown over-night in 5 ml LB, from which 0.5 ml was stored in 15%glycerolat -70" for the record and 4.5 ml were analyzed (see D N Apreparations, blotting and hybridization).In growth experiments, a culture of W3110 used routinelyin our laboratory served as reference strain. To estimate thevariation that occurred duringgrowth inside the stab culturebefore sealing, a single colony of this reference strain was in-oculated in a stab culture tube. After 1 week of storage theoffsprings were analyzedby the same method as for the elderstab subclones.Prototrophyand growth of subclones: Prototrophy of thebacteria was tested on solid M9 minimal medium (SAMBROOK

et al. 1989) containing 1g/liter glucose. Kinetics of growthinnutrient broth,Difco, 8 g/liter, which corresponds to the me-dium in stabs, were determined by diluting an overnight cul-ture 1:50, incubating at 37" for 4 h and measuring the opticaldensity (600 nm) of aliquots at intervals. The growth rate ofeach subclone was compared to that of an arbitrarily chosenreference strain W3110 that is routinely used in our laboratory.Selection coefficientswere calculated after a methodoriginally

designed for chemostat competitions ( D~W IUIZ EN andHARTL1983),while inour experiments the two strains compared weregrown in separate cultures.DNA preparations,blotting and hybridization: Bacteria(4.5 ml) grown as described above were centrifuged and thepellet was resuspended in 0.7 ml TE-glucose(25 mM Tris-HC1,

10 mM EDTA, 50 mM glucose, pH 8.0) and 0.1 ml of a mixturecontaining lysozyme, proteinase K and RNase A to a final con-centration of 5,0.05 and 0.5 pg/ml, respectively. After 30 minof incubation at 37", 0.1 ml SDS (20% ) was added and incu-bation was pursued for another 30 min. DNA was sheared bypassing twice through a syringe (0.8 X 40-mm needle), andextracted once with phenol, twice withphenolchloroform andtwice withchloroform. DNAwas precipitated at -70" by adding2.5 volumes of ethanol (100%) and 1/10(v:v) of Na-acetate(3M ) .After centrifugation, theDNA pellet was resuspendedovernight in 0.2 ml TE buffer (Tris-HC110 mM, EDTA 1mM,pH 7.6).DNA concentra tion was dete rmined by absorbanceat 260 nm.Genomic DNA (10pg) was digested withEcoRV (GAT'ATC)for 6hr at37" and theresulting, about lo', different fragmentswere separated by electrophoresis on a 0.6% agarose gel inTBE buffer. DNA was then transferred on a nylon membrane(Pal, Biodyne A)by vacuum transfer (Pharmacia).The South-ern blot hybridizations were performed following standardprocedures and visualized on Fuji medical X-ray film. Internalfragments of eight IS elements having no EcoRV site were usedas probes for successive hybridizations (Figure 1).An internalHpaI-BglII 1.4k b fragment of IS50 and an internal 845-bpEcoRI fragment of yS [which is also known asTn 1000 (GUYER1978)] were also used to verify the absence of these two ele-ments from the strain W3110. Furthermore, a1.1-kb BglI-Ssp1internal sequence of the l ac2 gene derived from plasmidpNM480 (MINTON1984) was used as a hybridization proberepresenting a unique chromosomal gene.Probes were labeled with [a-"PIdATP (Random LabelingKit, Bic-Rad) , and then purified through gel filtration (support G50, Pharmacia) in a 1-ml spun column. For deprobing,the radioactive membranes were washedthree times inboilingdistilled water for 15min before a new hybridization with an-other probe was performed. All DNA methods were after (SAM-BROOK et al. 1989).Coding of the RFPL Each specific fragment that hybrid-ized to a particular IS probe under stringent conditions wasscored as one copy of the IS element and its electrophoreticmobility was measured. For the coding, only the presence orthe absence of a given band was considered, and not the rela-tive intensities, even in a few cases where particular bands a ppeared with double intensities as compared to the rest. All ofthe 118 subclones were compared and all of the observed elec-trophoretic mobilities were used to define possible hybridiza-tion fragments. In individual patterns the scores were 1 or 0depending on presence or absence, respectively, of a frag-ment. This resulted ina matrix (fragment X individuals)which was used for establishing the pedigree of the culture.Pedigree of the culture: IS elements are relatively dynamicin the genome and therefore may serve as a useful, rapidlyevolving genetic marker [ see SAWYERand HARTL (1986) fordiscussion]. Since IS elements are inherited, the subclonesthat share a common ancestor should also tend to share sitesat which IScopies are inserted in the chromosome. This p roperty was used to calculate a most parsimonious tree describingthe variation of hybridization patterns after prolonged storage,starting with a uniqueunknown pattern. The tree analysis wasdone by the procedure of FITCH (1971) which tries to find thattree that accounts for the descent of the sequences in the few-est number of changes. Gains and losses of fragments were


3/10

Variation inResting E. coli 723

b0HindIII 1327Hind111I

C

0 1258PstI &oRV

d I ,I 10 1426WnI E a R IIe

0PstI

1195PstI

f I ,I I~. IS30I 1 I ,

0 1221BsfXf E a R V

0SmaI

1443HindIIIIh I IS186

0 13'38FIGURE1.-Hybridization probes of IS elements. The restric-tion fragments of eight ISelements used asprobes forhybrid-ization to E. coli genomic DNA are shaded. Numbers indicatethe nucleotide coordinates in the published sequence of theIS element (see MATERIALS AN D METHODS for references). (a)740-bp HindIIETthlllI fragment of IS1 from plasmidpAW929 (unpublished, our collection); (b) 730-bp A d - H p a Ifragment of IS2 from plasmid pRAB2::IS2(MEet al. 1988);(c) 916bpHindIII fragmentof IS3 from plasmid pRAB2::IS3

(RAABE el al . 1988); (d ) 525-bp PstI-EcoRV fragment of IS 4from plasmid pFDX934K61 (unpublished, giftof B. RAK);(e )940-bp BglII-EcoRI fragment of IS5 from plasmid pRAB2::IS5(UEet al. 1988); (0 1200-bp PstI fragment of IS30 fromplasmid pAW304(STALDER and ARBER 1989); (g) 1141-bpBslXI-EcoRV fragment of IS150 from plasmidpFDX450(SCHWARTZet al. 1988); (h) 906 bpHindIII-SmaI fragment ofIS186 including 12-bp vector DNA from the plasmidpAW4O:IS186 (SENCSTAGet al. 1986).given equal weight. The program (ANCESTOR) gives strictlybifurcating trees but branches of zero length can be removedto give the polychotomies shown. Thus isolates 1, 13, 14, 25,

35 and 62 (see Figure 3) are all different from the same im-mediate ancestor (identical to isolate 10) by the change of asingle but different site. There are945 ways that these isolatesmight be related to each other in a strictly bifurcating tree, allof them equally parsimonious. Such multipliers from otherpolychotomies yield 1.8 X 10" trees of the same length (174changes) out of the 10'' possible pedigrees. The ambiguityimplied by that large number is already contained within thetree by the unresolved branch order andthus, although thereare many equally good trees, they are all subsumed under thetree shown.

RESULTSSince IS elements are inherited, individual colonies

that share a recent common ancestor should also tendto share most of the sites at which IS elements are pre-sent in the genome. These shared sites would give riseto restriction fragments of identical size hybridizingwiththe IS probe on Southernblots. A band shift for a par-ticular individual may result from a point mutation in arestriction site, an insertion, aninversion, a duplication,a transposition or a deletion affecting a fragment. In-crease of the copy number of an IS element may reflecta transposition of the IS element or a duplication of aDN A sequence containing the element.Large RFLP in a stab culture: Of the colonies recov-

ered from a 30 years old stab culture of strain W3110,118 wereanalyzedindividually. DN A fragments pre-pared from the chromosomewith EcoRVwereseparatedby electrophoresis and hybridized to probes from inter-nal sequences of IS elements. Chromosomal DN Ashowed hybridization to eight probes of resident IS el-ements, but not to the IS50and yS probes (only 60 in-dividuals were tested to ensure that these two elementswere absent from the strain).N o polymorphism was de-tected after hybridization to an internal l a d probe.Most individuals contained fewer than 15chromosomalcopies of a given ISso that direct countingof the num-ber of hybridization bands on gelswas reliable. Multipledigestion of the same genomic DN A from some of thestudied subclones with other restriction enzymes con-firmed that the countingof the IS copy number by thismethod was reliable (data notshown). In most cases, theintensities of hybridization were homogeneous for agiven DNA extract indicating a 1:l relation between afragment and anIS copy.Three exceptions were found:(i) with ISI, five fragments showed strong signals and twofragments (9.8 and 1.9 kb) weaker signals that likely cor-respond to the isoforms ISIF and ISIR, respectively, de-scribed for strain W3110 (UMEDAand OHTSUBO1991);(ii)with IS30a weak signal at 1.9 kb is due to a truncated copyof the element (UMEDAand OHTSUBO1990a);and (iii) withIS30 some subclones showeda fragment of 5.Gkb lengthwith a double intensity. This fi-agmentwas counted as oneIS element, although it carries a dimer of IS30 (seebelow)which could have arisen by a single DNA rearrangement

A striking polymorphism was displayed among thesubclones, indicating that a high genetic diversity arose


4/10

724 T.Naas ef al.IS7@6 IS5 IS30

3-

FIGURE2. P -u th er n hybridization of EcoRVdigested gen om ic DNA from 20 subclones (representing individuals #59-78) toprobesof IS IR6 , IS5 and IS30.No polymorphism wasobselved with IS I86w hile the hybridizationpatternsofelectrophoreticbandsobselved with IS5 were highly variable. With the IS30 probe, two types of patterns are noticed: on e with four fragments (threestrong and on e very we ak), and o n e with in addition several new bands including on e (marked *) that has a double intensityaccording to the a utoradiograp hy scanning. kb gives the fragm ent size in k ilobase (1-kb ladder, BRL).

in the studied strain. As illustrated in Figure 2, the con-tributions of the different IS probes to the revealedRFLPwere unequal, suggesting an ISdependentgeneticvariation. After coding, all data were compiled into a(fragmentsX individuals) matrix. Becauseof itssize, thismatrix is not shown here but it can be provided uponrequest to the authors.

In a control experiment,60individuals from a l-week-old stab cultureof a fresh subclone were also analyzedwith the same method. No RFLP was detected, indicat-ing that the preparation of the stab and its short timestorage did not induce any significant variance.

Natureofpolymorphism: Genetic variation acquiredupon prolonged storage was first measured with thecopy number p,$ iubclone for each IS element (Table1). For the eight IS probes used, the total copy numbervaried from 40 to 55 among individual subclones. Con-sidering an average length of an IS to be about 1 kb, thisrepresents a 15-kb difference for the chromosome sizewithin the same clone of bacteria. Although this is nota large proportion (cu . 0.5% of thechromosomelength), it indicates a fairly good amount of DNA rear-rangements The degreeof variation in the number ofcopies dependedonthe IS element: while IS1 wasstrictly constant in this study, IS186, IS150 and IS4showed little variation, but IS2. IS3, IS30 and particu-larly IS5 displayed a larger degree of variability. Thenumber of patterns per IS reflected a similar variationindicating a large polymorphism within the same cul-ture. Again ISZ, IS4, IS186 and IS150 had one majorpattern and noneor fewvariants,while the hybridizationpatterns of IS2, IS3, IS5and IS30 revealed a higher de-gree of polymorphism. In that respect, IS5 displayed a

remarkably high polymorphism with 46 different pat-ternscontaining from 9 to 19 copies. Such an ISdependent distribution of the genetic variation indi-cates that Isindependent mutations, e.g., pointmutations in restriction sites and random deletions, arenot mainly responsible for thevariation observed amongindividual colonies. Therefore, simple transposition orotherISmediated DNA rearrangements weremostlikely the cause of the observed genetic variation.

The frequency of occurrence of each fragmentamong individual colonies is also a measure of geneticvariation (Table 2). For a given IS element some frag-ments were always present while other fragments wereseen only in a fraction of the subclones. Again, the dif-ferent IS elements didnot contribute equally to thedif-ferentiationamong subclones: ISZ, IS4, IS150 andIS186 patterns showed no or little variation while IS2,IS3, IS30 and IS5 patterns were very variable.

The existing maps of IS elements present in E. coliK-12 strain W3110 had been constructed from gene li-braries obtained from single clones (BIRKENBIHLandVIELMETTER1989) and offer a reference to analyze thevariation among our subclones (Tables 1 and 2). Thegeneral patterns observed with the probes of ISZ, IS4,IS150 and IS186 are in accordance with these maps,except that one extra copy of IS1 is resident in oursamples. The distribution of the major and stable frag-ments with the probes of IS2, IS3, IS5 and IS30 are alsoin accordance with the published data. However, dif-ferences exist for thecopy numbers and the distributionof fragments for these highly polymorphic markers. ForIS2, seven fragments were seen in all individuals exceptone, three fragments in 40-60% of the individuals and


5/10

Variation in Resting E . coli 725TABLE 1

Distribution of IS elements among 118 individual colonies recovered froman old stab culture of strain W3110IS 1 IS 2 IS 3 IS 4 IS 5 IS30 IS150 IS186 All 8 IS elements

No. of IS copies per chromosome: aStandard strain W3110 6 14 6 1 >23 5 1 3 >59Range in our 118 subclones 7 8-11 4-7 1-2 9-19 4-10 1-2 3 40 -55Subclone #57 79 5 1 11 4 1 3 41Subclone #42 7 8 5 1 12 4 1 3 41Number of patterns 1 8 8 2 46 15 3 2 68 'Unique patterns 0 3 3 1 34 7 2 1 51

Hybridization patterns:

a The range of copynumber for each ISelement is compared to published data ofa standard strain after BIRKENBIHLand VIELME-ITER(1989,1990.'The number of different patterns illustrates the variability related to each IS element.'The total number of patterns differs from the sum of patterns per IS because some individuals showedmodifications with more than one IS

Subclones #42 and #57 are the closest to the putative ancestor.

probe.Unique patterns are those observed in only one of the 118 subclones.three otherfragments were rare. For IS3, four fragmentswere seen in all individuals except one, threefragmentsoccurred in about half of the individuals and one frag-ment was rare. For IS30, two categories of patterns wereobserved: individuals which possessed the three majorbands of a size similar to that described in the literature(BIRKENBIHLand VIELMETTER1989) and individuals thatcarried in addition adiversity ofother fragments includ-ing a 5.Gkb fragment showing a double hybridizationintensity which is likelyto contain two copies of the IS30element (Fig.2).The occurrence of IS30 as a dimerhasbeen identified to cause transpositional DNA rearrange-ments at high rates in E . coli (OLASZet al. 1993). Thestriking increase in the IS30 copy number in all sub-clones showing this 5.Gkb fragment is likely to be relatedto the presence of an IS30 dimer as a source for aburstof transposition. Inaccordance with the results of(GREENet al., 1984), IS5 showed a high mutator activityin the stab conditions. After 30 years of storage, thetransposition frequency of IS 5 can be estimated at 0.013per copy per year which is in accordance with the valueof 0.008 2 0.002 found previously (GREENe t al . 1984).

Pedigreeof the old stab culture: Sixty eight differentpatterns (57%) were found after hybridization of theeight IS probes to theDNA of 118individuals (Table 1).Fifty one different patterns were represented by oneunique individual subclone in the sample surveyed in-dicating that thepolymorphism arose under almost non-replicating conditions. Based on theparsimony method,a pedigree was designed (Figure 3) which required atleast 174 mutational steps to explain the observed dis-tribution of patterns. The average number of changesper observed pattern was 12.1 ? 2.6. The tree was un-rooted but contains a major cluster that allows one toassume a location for an ancestor: all individuals shownin the lower part of Figure 3 have had massive modifi-cations of the IS30 patterns linked to a doublecopy (seeabove), while in the upper par t individuals displayedunmodified IS30 patterns. Because nearly half of allthepatterns carried the same doublet leading to a burst of

IS30, it was assumed that thecritical event occurred be-fore most events at other IS elements happened. Thiswas used to determine a putative ancestor, located be-tween #57 and the groupof #42,58 and83 (Figure 3).It is interesting that theassumed genetic structure of theoriginal clone is no more observed at a high density.However this in line with the calculated average numberof 12 changes per observed pattern. As an illustration forthe variability acquired during storage, individual #72,on the otherside of the pedigree, required aminimumof 23 mutational steps from the putative ancestor to beexplained.

The high occurrenceof unique patternsindicates thatmany ofthe observed variations have their origin dur ingthe time of storage in the stab culture rather than pre-viously to the inoculation of the stab. Recent experi-ments with stabs prepared from subclones support thisinterpretation because the numberof mutants accumu-lates linearily with the time of storage in stabs (T. NAAS,M. BLOT,W. FITCHand W. ARBER, to be published). Itisassumed that in stab cultures, a residual growth is pos-sible on the expense of lysed bacteria. Such a crypticgrowth should not lead to more than oneeffective gen-eration for the entire cuItureand thus should notaffectan estimate of the mutation frequency based on thenumber of changes per pattern per time unit. For 30years (2.6 X lo5h), this frequency is about changesper pattern and per hour, and thus corresponds to ahigh level of mutagenesis. This value does nottake intoaccount lethal mutations, due for instance to transpo-sition or deletion in vital genes (REIF and SAEDLER1975),neither mutations not affecting the distribution of ISelements. Moreover two bottlenecks occurred when thestab was renewed in 1965 and 1972 that may have re-duced the genetic diversity within the stab. Therefore,this mutation frequency might be underestimated com-pared to the actual mutation rate. It also might be thatwe observed a biased sample of thepopulation, ifsome mutants had beenselected for duringstorage andled to a larger progeny by differential cryptic growth.


6/10

726 T. Naas e t al.TABLE 2

Size distribution of EcoRV fragments containing IS elementsIS 4 IS150 IS186 IS1 IS3 IS 2 IS30 IS5"

kb" % ' kb % kb % kb % kb % kb % kb % kb %7.8 1 7.0 100 7.4 100 9.8 100 9.37.7 l O O d 4.4 54 10.01 4.8 99 8.2 100 8.4 42 8.81.8 1 4.5 1 6.5 100 7.741 7.92.6 100 5.4 100 5.2 100 7.84.3 100 4.8 3 7.62.2 100 3.0 99 7.31.9 100 2.8 100 7.01.9 100 6.56.05.34.13.02.9

10010010039100406410010019937

9.99.88.78.07.67.57.26.96.65.6e5.24.34.03.02.72.52.32.0

98210021001134294225331111114100

12.912.110.810.09.18.58.17. 77.67.47. 16.96.66.46.15.95.75.65.55.25.04.64.44.14.03.93.83.63.53.43. 12.82.52.2

4.8

11001518158228142232242241343811009832244100266323944941620100

a IS elements are sorted by increasing polymorphism.'Size of EcoRV fragments hybrizing with the different IS probes.'Frequency of a given fragment in the population of subclones.Fragments observed in all subclones (100%) are in boldface.The 5.&kb fragment which hybridizes to the IS30 probe has a double intensity.

Thus these lineages should predominate in the cultureand appear as large groups of individuals in one cluster.In fact, most patterns were unique or had one or twosiblings, except five subsets of patterns (marked I-V inFigure 3) that contained between 5 and 12 individuals.These groups belonged to differentclusters in thepedi-gree, indicating that they werenot relics of the originalpopulation but new genetic combinations that appearedin the course of time. Their existence is more likely dueto reproduction than to independentidentical mutationevents. For instance, a mutation that improved fitnessmight have occurred before group I and I1 diverged byanother neutral ISrelated rearrangement.

Physiological studies with genetic variants: On richbroth medium, a great variety in colony sizes at 37" wasobserved among theindividuals. The occurrence of ben-eficialnew genetic combinations during storage wasthus addressed by comparing the growth rates, whiledetrimental mutations were analyzed by testing the dif-

ferent subclones for prototrophy (Figure 3 ) . With 56subclones (#62-118) the growth rate was determined at37" in a nutrient broth medium identical to the stabmedium except agar.A selection coefficient (s ) could becalculated by reference toa culture of W3110used rou-tinely in the laboratory, and its significancewas studiedby means of analysis ofvariance. Nearly half ofthe testedsubclones did not differ significantly from the referenceand their selection coefficient was then zero. This wasmainly the case for patterns represented by only onesubclone. Positive selection coefficients, indicating anadvantage, were found for several subclones in the u pper cluster and especially for the subsets I, I1 and N .Negative selection coeffkients were found for about onethird of the subclones and only in the lower cluster.Thesemutants were auxotroph and in most casesshowed a long lag phase after dilution as well as a slowgrowth. With some of these subclones viability could notbe maintained for more thantwo daysin liquid cultures.


7/10

Variation in Resting E . coli

coeffimtr trophyr c l e c t i a l pmcduuro

727

r-It52

0 5 10 15Number of changes

2025

04.004

+O.W00

+0.0060

+ O . W000+O.W

0

0

-0.007-0.006000

-0.0070-0.005

-0.005

-0.003-0.003-0.006-0.00600

0

PP

PPP

PP

PPPPP

P

P

1

1111

11

1

1

a11111

P

FIGURE3.-Pedigree of the studied bacterial population after prolonged storageasderived from the observed RFLP. The treeis unrooted butsubclones at the left (#57,42 ,58 and 83)are likely to bethe closest to the putative ancestor (see text for discussion).Each point represents a defined RFLP pattern after hybridization to IS elements, and individual subclonesare indicated by theirnumbers. Clusters labeled I-V are specific patterns represented by more than five subclones. The horizontal axis represents theminimum number of changes. Subclones testedfor their growth rate and prototrophy are underlined and the results are providedon the right: average selective coefficients(s ) as a measure of fitness and prototrophy ( p) orauxotrophy (a). In a crowded partof the figure, the relevant data for subclones #95, #97 and #l o3 are given in this order.


8/10

728 T. Naas et al .Possibly the burstof IS30 transposition shown by thesesubclones could have caused the growth defects. Thefact that the subset V contains nine siblings in the stabculture butshows a lower growth rate at37 is interestingand indicates that another component of fitness is en-hanced forthese individuals. We have not yet been ableto identify such a component which could perhaps im-prove viability of resting bacteria.

It is remarkable that all of the subclones in which noIS30 transposition had occurred (upper cluster of Fig-ure 3) remained prototrophic. In contrast most of thesubclones which underwent IS30 transposition (lowercluster of Figure 3) had become auxotrophic. Subclone#96 is an exception to this rule. Therefore, Itis not theburst of IS30 copies per se which caused the loss ingrowth performance but morelikely the loci affected bythe DNA rearrangements.

DISCUSSIONNature and dynamics of DNA rearrangements The

data obtained in the studies reported here confirm, ex-tendand generalize previous observations that I Smediated transposition contributes importantly to ge-netic plasticity and that this mutagenic activity is notlimited to growing cells but occurs also in resting bac-teria. The RFLP method employed is simple, but effi-cient and reliable. It involves only o ne restriction en-zyme,EcoRV, used to fragment chromosomal DNA andabout 10 hybridization probes includingall known resi-dent IS elements of the E. coli K-12 strain W3110. Al-though the strategy followed cannot reveal all sponta-neous mutations occurring to abacterial population northe specific mutagenesis mechanism, it is likely to revealmost transpositional DNA rearrangements, any otherDNA rearrangements as long as they affect DNA frag-ments carrying an IS element and thepresumably ratherrare cases in which a mutation destroys or creates anEcoRV cleavage site. A strong variation of the IS copynumber points to transposition as a major contributionto the detected structural genome alterations.

The high degreeof polymorphism which was revealedupon the analysis of subclones isolated from an old stabculture came as a surprise to us. Such a big variance isnot seen with bacterial cultures after prolonged propa-gation, during which periodic selection is known to re-duce geneticdiversity in the cell population (NowCKandSZILARD1950). It is thus most likely that the variancestrongly increased during the 30 years ofstorage of thestab at room temperature.Unfortunately, the nature ofthe culture of W3110 which was used in 1960 for theinoculation of the stab remains unknown. Possibly, ithad been a routinelaboratory culture rather than afreshsubclone. If so, some polymorphism might have alreadybeen present at that time. In particular, it remains un-known if the dimeric form of IS30, which has been de-tected in about 40% of the subclones isolated, was al-

ready present before the stab was inoculated. But thedevelopment of many different unique patterns of re-striction fragments must be largely attributed to trans-position upon storage of the stab. This hypothesis is sup-ported by recent studies involving stabs whichhad beeninoculated with fresh clones of bacteria. In particular, itis shown that mutants accumulate linearily with time ofincubation and the observed patterns aredifferent froma stab to another (T. NAAS,M. BLOT,W. FITCHand W.ARB E R, unpublished results).

Our knowledge of bacterial physiology in stab culturesis verypoor. There is at most a very slow,residual growthwhich could occur at the expense of dying cells. Thisargues for selection conditions differing from those ex-erted on fast growing bacterial cultures. This situationcould explain both thelarge number of different uniquegenome structures present and the fact that a few spe-cific structures are represented by several isolates. Thelatter are likely to have found afitness advantage in thestab culture and might thus have succeeded to grow tosome extent. A majority of such clusters analyzed haveindeed shown an increased fitness even in exponentialgrowth.

Among the eight residentIS elements of W3110, IS5proved to be themost mobile in the studied stab culture.Most other IS elements also showed transposition ac-tivities to a degree dependingon the IS considered. Toour surprise, ISI , although presentin sevencopies in thechromosome, was absolutely stable. This might be re-lated to the particular physiological condition in the stabkept at room temperature. Itis known that at low tem-peratures ISI mainly undergoesdeletion formation(REIF and SAEDLER1975; ARBER et a l . 1978). Such mu-tagenesis is likely to be often lethal which could explainthat we could not detect any ISI activity.

DNA rearrangements affected in multiple ways all vi-able members of the bacterial population in the stab. Inthe average, each subclone showed about 12 changes inthe RFLP patterns from that of the putative ancestralgenome, although individual deviations are consider-able from this average. In a crudeanalysis we could notfind any evidence for interdependence of DNA rear-rangements involving different IS elements that mightcorrespond to a generalresponse to starvation. Rather,the observed variation should be attributedto a numberof independently occurring consecutive events. Thisconclusion is supported by recent studies withstabsstored for various periods of time (0-20 years) indicat-ing that the numberof different ISrelated mutantsde-pend on thetime of storage. Moreover, the preparationof a stab and its storage for one week does not result inany measurable genetic variance (T . NAAS,M. BLOT,W.FITCHand W. ARB E R, unpublished results). However, itis possible that at particular times particular cells mightundergomore efficient transposition. Such activephases might depend on a particular physiological state


9/10

VariationinResting E . coli 729and it might thusalso depend on theenvironment. Tem-perature for instance was shown to induce transpositionof IS elements in Halobacterium (PFEIFERand BLASEIO1990). The methods used in this investigation could of-fer help in the elucidation of these questions. A goodexample of increased transposition activities is givenbyIS30. It was already known that this element can bringabout dimeric forms of IS30 and that such forms arehighl? active for transpositional DNA rearrangements(OLASZet al. 1993). The bursts of IS30 transpositiondocumented in this study showthat such transpositionalactivities can affect the bacterial chromosome indeed.

It may appear needless to state that transposition isenzyme-mediated and that these enzymes must be ex-pressed in stabs to explain their activity. Life in stabsmight representa particular conditionof stress.It is thusinteresting to note that many IS elements can undergotransposition under these conditionsof life.This may berelated by their postulated role as evolutionarily relevantvariation generators. The documentation of a high de-gree of variance to be attributed totransposition in stabsand thus to essentially resting cells can represent s u pport for anevolutionary role of mobile genetic elements.

Evolutionary biologyof bacteriaand IS elements over30 years: One fundamental question concerning mo-bile genetic elements addresses their long term main-tenance in genomes. Theoretical approaches have o pposed a parasitic style of life (DOOLITTLEand SAPIENZA1980) us. an adaptive role provided to the host (CAMP-BELL 1981).Because IS elements only carry genes relatedto theirtransposition function, they cannot express ben-eficial traits that might directly enhance the fitness oftheir host, as shown for some transposons (BLOTet al.1993). Rather IS elements may play a role in bacterialpopulations in the long term by creating a diversity ofgenetic combinations, with occasionallyone of them be-ing morefit [see~ E R(1991) for discussion]. From thispoint of view the pedigree of the stab culture and thefitness measured are very interesting. Less than half ofthe different mutants seemed neutral at least for theirgrowth rate and they were represented most often byone individual. Nearly another halfwas disfavored beingauxotroph mutantswith a lower growth rate, althoughone of these mutants existed as a group of nine indi-viduals with identical patterns, suggesting that it mighthave an advantage during storage. Three mutants a ppeared to have a better growth rate than the other in-dividuals and one patternshowing similar properties hasnot been tested for growth rate. Thus after 30 years,maybe five new advantageous genetic combinations oc-curred in the studied sample as a consequence of 1%related rearrangements.If similar kinds of favorable mu-tations occur at intervals, this might be sufficient toensure the maintenance of IS elements upon each re-placing an ancestor by a better adapted mutant. Evi-dence for rapid changes in the genetic structure of

population during stationary phase (ZAMBRANO et al.1993) illustrates what might have happened at intervalsduring 30 years of storage.

The observed increase of genetic variation in restingbacteria has its implications for environmental micro-biology. Bacteria spend most of their time in a restingstate rather than in exponential growth. However, mostpublished data involving quantitative experimentationto explore mutagenesis are based on work done withpropagating bacteria. As we suggest here, in part differ-en t laws may apply to spontaneous mutagenesis of rest-ing bacteria. The strategies applied in this study mayhelp to understand the role of physiology on the gen-eration of genetic plasticity in bacterial populations.Such investigationsare now inprocess in our laboratory.

We wish to thank B.LEVTN who stimulated the initiation of this study,B. RAKfor providing some IS probes an d A. ARINI fo r valuable discussions. This work was supported by grant 31.30040-90 from the SwissNational Science Foundation.

LITERATURE CITEDAJIOKA,J., and H A R ~ ,D., 1989 Popula tion dynamics of transposableelements, pp. 939-958 in Mobile DNA,edited by D. BERGand M.M. HOW. American Society for Microbiology, Washington D.C.ARBER,W., 1958 Transduction des charact6res g al par le bactkric-phage Lambda. Arch. Sci. 11: 259-338.ARBER,W., 1991 Elements in microbial evolution.J. Mol. Evol. 33:ARBER,W., S. IIDA, H. JURE,P.CASPERS, J. MEYERand C. H A N N I ,1978 Rearrangements of genetic material in Escherichia coli asobserved on the bacteriophage P1 plasmid. Cold Spring HarborSymp. Quant. Biol. 43: 1197-1208.BACHMANN,B. J., 1987 Derivations and genotypes of some mutant

derivatives of Escherichia coli-K12, pp. 1190-1219 in Escherichiacoli andSalmonella typhimur ium, edited by F. C. NEIDHARDT.American Society for Microbiology, WashingtonD.C.BACHMANN,B. J., 1990 Linkage map of the Escherichia coli K-12, Edi-tion 8. Microbiol. Rev. 54: 130-197.BIRKENBIHL,R. P., and W. V I E L M E ~ R ,1989 Complete maps of ISI,IS,?, IS3, IS4, IS5, IS30, and IS150 locations in Escherichia coliK12. Mol. Gen. Genet. 220: 147-153.BIRKENBIHL,R.P., and W. VIELMEITER, 1991Completion of the ISmapin Escherichia coli: IS186 positions on the Escherichia coli K12chromosome. Mol. Gen. Genet. 226: 318-320.BLOT, M.,J. HEITMANand W. ARBER, 1993 Tn5-mediated bleomycinresistance in Escherichia coli requires host genes. Mol. Microbiol.8 1017-1024.BRAHMA, N.,A. SCHUMACHER,J. CULLUMand H. SAEDLER, 1982 Distri-bution of the Escherichia coli K12 insertion sequences ISl, IS,?

and IS3 among other bacterial species.J. Gen. Microbiol. 128:CAMPBELL, A , 1981 Some questions about movable elements and theirimplications. Cold SpringHarbor Symp. Quant. Biol. 4 5 1-9.DM.RWPLE,B. , P. CASPERS and W. ~ E R ,1984 Nucleotide sequenceof theprokaryotic mobile genetic element IS30. EMBOJ. 3:2145-2149.DOOLITTLE,W. F., and C. SAPIENZA, 1980 Selfishgenes, the phenotypeparadigm and genome evolution. Nature 284: 601-603.DWUIZEN,D.E., and D.L. HARTL,1983 Selection in chemostats.Microbiol. Rev. 47: 150-168.ENGLER,J. A., and M.P.BREE, 1981 The nucleotide sequence andproteinxoding capability of the transposable element IS5. Gene14: 155-163.FITCH,W.M., 1971 Toward defining the course of evolution: mini-mum change for specific tree topology. Syst. 2001. 2 0 406-416.GALAS,D. J., and M. CHANDLER, 1989 Bacterialinsertion sequences,

pp. 109-162 in Mobile DNAedited by D. BERGand M. M. HOWE.American Societyfor Microbiology, Washington D.C.

4-12.

2229-2234.


10/10

730 T. Naas et al.GHOSAL, D.,and H. SAEDLER,1979 Nucleotide sequence of the trans-posable DNA-element IS2. Nucleic Acids Res. 6 1111-1122.GREEN,L., R. MILLER,D. E. DYKHUIZENan d D. L. HARTL,1984 Distri-bution of DNA insertion elements IS 5 in naturalisolates of Esch-erichia coli. Proc. Natl. Acad. Sci. USA 81: 4500-4504.GUMR,M. S., 1978 The gammade lta sequence of F is an insertionsequence.J. Mol. Biol. 126: 347-365.HALL,B. G.,1988 Adaptive evolution that requi res multiple sponta-

neous mutations I. Mutations involving an insertion sequence.Genetics 120: 887.HARTL,D. , an d S. SAWYER,1988a Multiple correlations among inser-tion sequences in the genome of natural isolates of Escherichiacoli, pp. 91-106 in Transposit ion, edited by A.J. KINGSMAN and S.M. KINGSMAN. Cambridge University Press, Cambridge.~ T L ,D. L., and S. A. SAWYER,1988a Why do unrelated sequencesoccur together in the genomeof Escherichia coli? Genetics 118:537-541.HARTL,D., D. E. DWUISEN andD. BERG,1984 Accessory DNAs in thebacterial gene pool: playground for coevolution. Ciba Symp. 102:233-245.KLAER, R., S. KUEHN,E. TILLMANNand P.STARLINGER,1981 The se-quence of IS4. Mol. Gen. Genet. 181: 169-175.KOHARA,Y., K. AKlyAMA and K. ISONO,1987 The physical map of thewhole Escherichia coli chromosome: application of a new strategy

for rapid analysis and sort ing ona large genomic library. Cell 50:495-508.KOTHARY, R.K., D. JONES and E. P. M. CANDI DO, 1985 IS186 an Esch-erichia coli insertion element isolated from a cDNA library. J.Bacteriol. 164: 957-959.MINTON,N. P., 1984 Improved plasmid vectors for the isolation oftranslational la c gene fusion. Gene 31: 269-273.MITTLER,J. E., and R. E. LENSKI,1990 Excisions of Mu from E. coliMCS2 are not directed mutations. Nature 344: 173-175.NOVICK,A., and L. SZILARD,1950 Experiments with the chemostat onspontaneous mutations of bacteria. Proc.Natl. Acad. Sci. USA 36:708-719.OHTSUBO,H., and E. OHTSUBO,1978 Nucleotide sequence of an in-sertion sequence, 1.51. Proc. Natl . Acad. Sci. USA 75: 615-619.OLASZ,F., R. STALDERand W. ARBER, 1993 Formation of the tandemrepeat (IS?O), and its role inIS30-mediated transpositional DNArearrangenments. Mol. Gen. Genet. 239: 177-187.PFEIFER, F.,and U. BMEIO, 1990 Transposition burst of the ISHF27Insertion elem ent family in Halobacter ium halobium.Nucleic Ac-ids Res. 18:6921-6925.RAABE, T., E. JENwand J. MEMR,1988 A selection cartridge for rapiddetection and analysis of spontaneous mutations including in-

sertions of transposable elements in Enterobacteriaceae. Mol.Gen. Genet. 21 5 176-180.REIF,H.J., an d H. SAEDLER,1975 IS 1 is involved in deletion formationin the g al region of Escherichia coli K12. Mol. Gen. Genet. 137:RODRIGUEZ,H., E. T. SNOW,U. BHATand E.L. LOECHLER,1992 AnEscherichia coli Plasmid-Based, mutational system in which supFmutants are selectable-insertion elements dominate the sponta-neous spectra. Mutat. Res. 270: 219-231.SAMBROOK,J., T. MANIATISand E. F. FRITSCH,1989 Molecular Cloning:A Laboratory Manual. Cold Spring Harbor Laboratory, ColdSpring Harbor, N.Y.SAWYER,S., and D. L. HARTL,1986 Distribution of transposable ele-ments in prokaryotes. Theor. Popul. Biol. 30: 1-16.SCHWARTZ,E. , M. KROEGER and B. RAK, 1988 1.5150: distribution,nucleotide sequence and phylogenetic relationships of a new E .coli insertion e lement . Nucleic Acids Res. 16: 6789-6802.SENGSTAG,C ., S. IIDA,R. HIESTANDNAUERand W. A R B E R , 1986 Termi-nal inverted repeats of prokaryotic transposable e lement IS186which can generateduplications of variable length at anidenticaltarget sequence. Gene 49: 153-156.SHAPIRO,J. A., and N. P. HIGGINS,1989 Differential activity of a transposable element in Escherichia coli colonies. J. Bacteriol. 171:STALDER,R., and W. A R B E R , 1989 Characterization of i n v i t r o con-structed IS30-flanked transposons. Gene 76: 187-193.TIMMERMAN,K. P., and C.-P. Tu, 1985 Complete sequence of IS3.Nucleic Acids Res. 13: 2127-2139.UMEDA,M., and E. OHTSUBO,1989 Mapping of insertion elementsIS1, IS2 and IS3 on theEscherichia coli K-12 chromosome. J.Mol.

Biol. 208: 601-614.UMEDA,M., and E. OHTSUBO,1990a Mapping of insertion elementIS3Oin the Escherichia coli K-12 Chromosome. Mol. Gen. Genet .UMEDA,M., and E. OHTSUBO,1990b Mapping of insertion elementIS5 in the Escherichia coli K-12 chromosome. J. Mol. Biol. 213:UMEDA,M., and E. OHTSUBO,1991 Four types of IS1 with differencesin nucleotide sequence reside in the Escherichia coli K-12 chrc-mosome. Gene 98: 1-5.ZAMBRANO,M. M., D. A. SIEGELE,M. A L M I R O N , A. TORMOand R. KOLTER,1993 Microbial competition: Escherichia coli mutants that takeover stationary phase cultures. Science 259: 1757-1760.

17-28.

5975-5986.

222: 317-322.

229-237.

Communicating editor: R. M A U E R

is in resting e.coli

Documents