INFECTION AND IMMUNITY, Sept. 1972, p. 308-315Copyright © 1972 American Socicty for Microbiology

Vol. 6, No. 3Printed int U.S.A.

Polynucleotide Sequence Relatedness Among ThreeGroups of Pathogenic Escherichia coli Strains

DON J. BRENNER, G. R. FANNING, A. G. STEIGERWALT, 1. ORSKOV, AND F. 0RSKOVDivision of Biocheemistry, Walter Reed Army Ilnstituite of Research, Washinlgtoni, D. C. 20012,and Initerntationial Escherichiia Cenitre (WHO), Statens Serluministitutt, Copenihageni, Detnmark

Received for publication 18 May 1972

Escherichia coli strains that cause dysentery-like disease, parenteral infection,and infantile diarrhea form specific groups based on mobility of 0 and K antigensin immunoelectrophoresis. Members from each of these groups were assayed forgross nucleotide sequence relatedness. The method used was interspecific deoxy-ribonucleic acid (DNA) reassociation reactions carried out free in solution. Reas-sociated DNA was separated from unreacted DNA by passage through hydroxy-apatite. DNA relatedness between these groups was approximately 80c,%.. Thegroups containing those strains causing parenteral infection and those responsiblefor dysentery-like disease showed preferentially high intragroup DNA relatedness.The group containing strains responsible for infantile diarrhea did not show pref-erentially high intragroup DNA relatedness with the reference strain employed.These strains, however, did exhibit preferentially high DNA relatedness to a secondreference strain.

The species Escherichia coli encompasses alarge number of strains that are extremely similarin morphology and metabolic patterns. Threeparameters of deoxyribonucleic acid (DNA)relatedness in a large number of E. coli strainswere assayed in a recent study (3). Data obtainedin this study indicated up to a 4%' difference inguanine plus cytosine content, and as much as5 X 108 dalton differences in genome size in E.coli strains. Furthermore, as much as one-fourthof the DNA from some typical E. coli strainshad diverged to a point where there were nodemonstrable nucleotide sequences held incommon. The frequency of relatedness betweenE. coli strains forms a Gaussian distribution.The above parameters can be used in the molecu-lar description of a species (3).With the increasing incidence or recognition

of human pathogenic strains, or both, it hasbecome increasingly important to determine theserotype of strains isolated from clinical sources.It is known that strains with certain specific 0and K antigens are closely associated with patho-logical conditions, whereas other serotypes arenormally nonpathogenic. In the pathogenicstrains, certain serotypes are predominantlyassociated with infantile diarrhea, other serotypesare associated with parenteral infection, andstill others with dysentery-like infections. Usingimmunoelectrophoretic (IE) patterns of simpleextracts containing the 0 lipopolysaccharide

antigen and a possible K acidic polysaccharideantigen, it was found that the serogroups thatare highly frequent in the normal intestinal canal(which are the same serogroups frequent inparenteral disease) give one IE pattern-1A.Strains associated with human infantile diarrheahad a similar cathodic 0 antigen mobility butno special K antigen line, and thus gave an IEpattern which was a variant of 1A, termed 1B.E. coli strains from dysentery-like disease gave athird IE pattern, 2, with no K antigen line and anegatively charged 0 antigen. Finally, it shouldbe stressed that the remaining large number ofE. coli strains, about which nothing is knownregarding possible relatedness to specific diseasesor special ecological situations, can be groupedin these three IE groups (8).

It appears, therefore, that a strong correlationexists between the electrophoretic mobility ofsurface antigens and the type of pathologicalcondition produced by E. coli strains. In thepresent study, we sought to determine whether therelationship between type of pathogenicity andelectrophoretic mobility is reflected in prefer-entially high polynucleotide sequence related-ness.

MATERIALS AND METHODS

Organisms and media. The strains used in this studyare listed in Table 1. All Statens Seruminstitut strainsare antigenic test strains of the respective E. coli 0

308

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

DNA RELATEDNESS IN PATHOGENIC E. COLI STRAINS

TABLE 1. Bacterial straints employed

Strain Sourcea

E. coli OIA CDCE. coli 02 SSI

E. coli 02A CDCE. coli 04 SSI

E. coli 04 CDCE. coli 06 SSI

E. coli 06 CDCE. coli 07 SSI

E. coli 07 CDCE. coli 08 F. J. SkermanE. coli 09 CDCE. coli 025 CDCE. coli 026 SSI

E. coli 028 SSI

E. coli 032 SSI

E. coli 068 SSI

E. coli 075 SSI

E. coli 075 CDCE. coli 0111B4 S. FalkowE. coli 0112 SSI

E. coli 0114 SSI

E. coli 0115 SSI

E. coli 0119 SSI

E. coli 0124 SSI

E. coli 0128 SSI

E. coli 0136 SSI

E. coli 0142 SSI

E. coli 0145 SSI

E. coli K-12 University ofWashington

Alkalescens-Dispar 01 CDCAlkalescens-Dispar 02 CDCAlkalescens-Dispar 03 WRAIRAlkalescens-Dispar 03 CDC

(ceylonensis)Alkalescens-Dispar 04 CDCAlkalescens-Dispar 05 CDCAlkalescens-Dispar 06 CDCAlkalescens-Dispar 07 CDCAlkalescens-Dispar 08 CDCShigella boydii 1 WRAIRS. boydii (etousa) 7 WRAIRS. dyseniteriae I WRAIRS. dyseniteriae II WRAIRS. dysenteriae III WRAIRS. flexneri 24570 S. FalkowS. flexeteri (newcastle) 6 WRAIRS. sonntei WRAIRS. sonnei virulent WRAIRS. sonnei avirulent WRAIR

a CDC, Center for Disease Control, Atlanta,Georgia. SSI, Statens Seruminstitut, Copenhagen,Denmark. WRAIR, Walter Reed Army Instituteof Research, Washington, D. C.

antigens. Bacteria were maintained on meat extractagar or nutrient slants. Unlabeled DNA was obtainedfrom organisms grown to stationary phase at 37 C in4.5 liters or more of Brain Heart Infusion broth. 32PO4-labeled DNA was obtained from cells grown to sta-

tionary phase in a glucose salts broth containing min-imal amounts of inorganic phosphates and 10 to 15mCi Of 32PO4 per liter (1).DNA preparation. The purification of both labeled

and unlabeled DNA, as well as the shearing ofDNA toan average single-stranded fragment size of 1.25 X 105daltons, has been described (1).DNA reassociation. Relatedness studies based on

relative percentage of DNA reassociation are depend-ent on both the specificity and the completeness ofDNA duplex formation. To determine nucleotide se-quence relatedness, DNA from one or more referenceorganisms is labeled with 32PO4 and then reacted freein solution with a series of unlabeled DNA prepara-tions. DNA reassociation is a concentration-dependentreaction. In these experiments, the concentrations oflabeled and unlabeled DNA were 0.1 ug/ml and 150,4g/ml, respectively. These concentrations were chosenso that, under the conditions of incubation, little orno reassociation occurs between labeled fragments,whereas maximal reassociation occurs between labeledDNA fragments and complementary unlabeled frag-ments. Specificity of reassociation is controlled by us-ing salt concentration and incubation temperatures atwhich sequences containing only randomly comple-mentary bases cannot form stable reassociation prod-ucts. Specifically, the labeled and unlabeled DNAspecies, in 0.28 M PB (phosphate buffer, an equimolarmixture of Na2HPO4 and NAH2PO4; pH 6.8), weredenatured in a boiling water bath for 3 to 4 min. Sam-ples were then cooled to 60 C or 75 C and incubatedat this temperature for 16 hr. In control reactions atthese criteria, 0.1 ,ug of labeled DNA per ml reassoci-ates at less than 1%,' whereas unlabeled DNA fromthe reference strains employed shows 80% or greaterreassociation.

After incubation the samples are diluted in deion-ized water to 0.14 M PB, and single-stranded DNA isseparated from duplexed DNA by chromatography onhydroxyapatite (HA, reference 2). HA is equilibratedwith 0.14 M PB + 0.4% sodium dodecyl sulfate (SDS)and held at 60 or 75 C. Single-stranded DNA does notbind to HA at this salt concentration, whereas reasso-ciated DNA is bound to HA. Four 15-ml washes with0.14 M PB + 0.4% SDS at the incubation tempera-ture serve to remove unreacted DNA from HA. Reas-sociated DNA iseluted eitherin four 15-ml washes with0.4 M PB or, when the thermal stability of the duplexedDNA is of interest, in a series of 15-ml washes with0.14 M PB at increasing 5 C temperature increments upto 100 C. Radioactivity is assayed by Cerenkovcounting.

RESULTSWith few exceptions, antigens from the strains

tested exhibited the following immunoelectropho-retic behavior: (i) group 1A, 0 antigen migrationtowards the anode and the cathode and a specialanodic K antigen (strains highly frequent in nor-mal human colon and causing parenteral infec-tion); (ii) group 1B, 0 antigen migration towardsthe cathode (strains associated with infantile diar-rhea); (iii) group 2, 0 antigens migrate towards

309VOL. 6, 1972

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BRENNER ET AL.

the anode (strains causing dysentery-like disease).Representative strains from each of these groupswere tested forDNA relatedness within and acrossgroup lines. Also included in the study were E. colistrains whose 0 antigen groups were known, butwhich had not been subjected to immunoelec-trophoretic analysis. However, previous experi-ence with many strains belonging to these 0groups indicate 100% correlation between 0group and IE pattern. These strains were as-signed to the same group as strains with identicalO antigens that had been assayed immunoelec-trophoretically. In addition, several Shigellastrains were tested.

E. coli 075 was the arbitrarily chosen DNAreference strain from group 1A. Labeled DNAfrom this strain was reacted with DNA from tenother strains belonging to group 1A, as well aswith DNA from seven group 1B strains and sixgroup 2 strains as shown in Table 2. Strain 075is closely related to members of all three im-munoelectrophoretic groups as shown by highrelative binding at 60 C and the negligible dropin relative binding obtained at 75 C as comparedto that at 60 C. Despite the high level of bindingof strain 075 to all strains tested, it is evidentthat preferentially high reaction occurred withmembers of group 1A (89%) as opposed tostrains from group 1B (77%), group 2 (81%), orstrain K-12 (81 %).

Strains 075, O1A, 02A, 04, 06, and 07 (strainsin parentheses in Table 2) were arbitrarily as-signed to groups IA and lB without being sub-jected to electrophoretic analysis. With the pos-sible exception of strain 02A (group 1A) theirrelatedness to 075 was similar to that shown bystrains placed in the group on the basis of im-munoelectrophoretic analysis of 0 antigens. Thedata from group IA suggest that some strainswith common 0 antigens are extremely similar(04, 06, 07), whereas other strains with similar 0antigens (02 and 075) appear to be significantlydifferent in terms of their total DNA.

Strain 0114 was chosen as the DNA referencestrain for group lB. This strain exhibits approxi-mately equal average relatedness to the threegroups of E. coli strains (Table 3). Strain 0114was also tested for relatedness to a group ofShigella strains. As expected, DNA reassocia-tion between the shigellae and 0114 was some-what lower than the average obtained between E.coli strains (3). From this data it appears thatthe grouping of pathogens based on electropho-retic mobility of 0 antigens in group 1B does notreflect preferentially high overall DNA related-ness among members of this group.

Results obtained using strain 0112 as the repre-sentative from group 2 are shown in Table 4.

TABLE 2. DNA relatedniess of E. coli 075 toE. coli strains

% Relative bindingSource of unlabeled DNA'

At 60 C At 75 C

Group IA E. coli strains075(075)(O1A)02(02A)04(04)06(06)07(07)

Average

Group iB,(09)0260680114011901420145

E. coli strains

Average

Group 2 E. coli strains0750320112011501240136

Average

E. coli K-12

lOOb90879280929293928584

89

77807279778076

77

757878887775

79

81

lOOb92858876899194938280

87

77827279788276

78

737673817471

75

78

a Strains in parentheses were not tested im-munoelectrophoretically.

b Not included in average.

Relatedness values obtained from optimal, 60 Creassociation reactions are highest between 0112and other strains from group 2. Data from similarreactions carried out at 75 C indicate that whilethe highest average reaction still occurs between0112 and members of group 2, it is more diffi-cult to distinguish between group 2 and group 1Breactions.

Strain 0115, also a member of group 2, ex-hibits preferential overall DNA sequence related-ness to strains from group 1B; in fact, in 60 Creactions, it showed least relatedness to strainsfrom group 2 (Table 5). A reexamination of thedata reveals that, in addition to 0115, there are

310 INFECT. IMMUNITY

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

DNA RELATEDNESS IN PATHOGENIC E. COLI STRAINS

TABLE 3. DNA relatedness of E. coli 0114 to E. coliand Shigella strains

% Relative bindingSource of unlabeled DNA'

At 6oC At 75 C

Group 1B E. coli strains0114(08)(09)(025)026068(011OB4)011901420145(128)(Alkalescens-Dispar 08)

Average

Group 1A E.rcoli strains(02A)02040607075(Alkalescens-Dispar 01)(Alkalescens-Dispar 03)(Alkalescens-Dispar 04)(Alkalescens-Dispar 05)

Average

Group 2 E. coli strains0280320112011501240136

Average

E. coli K-12Shigella strainsS. boydiiS. boydii (etousa)S. dyseiiteriae IS. dysenteriae IIS. dysenlteriae IIIS. flexneriS. flexneri (newcastle)S. sonneiS. sonnei virulentS. sonznei avirulent

Average

lOOb8686758174798878837979

81

86797978857879787679

80

747981888078

80

78

71767275727774787777

75

other strains in which the pattern of DNA re-latedness does not agree with strain groupingsbased on pathogenicity and 0 antigen immuno-electrophoretic mobility. These include 02A and025. Despite these exceptions, and the seeminginability to distinguish group 1B strains fromother strains on the basis of DNA reassociation,it appears that both group lA and group 2strains show preferentially high intragroup DNArelatedness.Data obtained from DNA reassociation ex-

periments were subjected to statistical analysisin order to assess the validity of grouping basedon DNA relatedness. When the data were plottedon probability paper, 075 (group lA) reactionsformed a Gaussian distribution with group IAstrains, but not with strains from groups 1B or 2.Similarly 0112 (group 2) reactions comprise aGaussian distribution with group 2 strains moreso than with strains from groups IA or lB.Strain 0114 (group iB) did not form a Gaussiandistribution in reactions with strains from anyof the groups. Strain 0115 (group 2) reactionsexhibited a Gaussian distribution only withgroup 2 strains, despite the fact that the reac-tions with group 1B strains were substantiallyhigher than reactions with group 2 strains. Thedata were also analyzed using the Kruskol-Wallis rank comparison test or X2 significancetest. There is a significant difference (P = <0.05)in the rank correlation of the different referencestrains to group 1A strains. The ranking is 075(group 1A) > 0115 (group 2) > 0112 (group 2) >0114 (group 1B). For group 1B strains, the rankcorrelation is 0115 (group 2) > 0112 (group 2) >0114 (group 1B) > 075 (group 1A), with P =

<0.01. For strains in group 2, the rank corre-lation is highly significant (P = <0.001) with0112 (group 2) > 0115 (group 2) > 0114 (group1B) > 075 (group IA).Genome size variations from 2.3 x 109 daltons

to 3.0 x 109 daltons exist among E. coli strainsas shown by nonreciprocal binding in reciprocalreassociation reactions and by differences ininitial rates of reassociation determined spectro-photometrically (3, 4). Differences in genomesize can account for what may appear to be ab-normally high or low binding. For example, inreciprocal reactions, 91% of E. coli K-12 DNAis equivalent to 74% of DNA from E. coli strain02A due to the larger genome size of 02A ascompared to that of K-12. It is, therefore, quitepossible that all or part of the differences in DNAbinding observed between the three groups ofpathogenic E. coli strains reflect genome sizedifferences, rather than comparably sized genomesthat contain differing amounts of related nucleo-tide sequences. Significant genome size differ-ences existing in the reference strains would be

a Strains in parentheses were not tested im-munoelectrophoretically.

b Not included in average.

311VOL. 6, 1972

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BRENNER ET AL.

TABLE 4. DNA relatedliess of E. coli strain 0112 toE. coli strains

% Relative bindingSource of unlabeled DNAa

At 60 C At 75 C

Group 2 E. coli strains0112028032011501240136

Average

Group IA E. coli strains(OlA)02(02A)04(04)06(06)07(07)075(075)(Alkalescens-Dispar 01)(Alkalescens-Dispar 02)(Alkalescens-Dispar 03)(Alkalescens-Dispar 03celonensis)

(Alkalescens-Dispar 04)(Alkalescens-Dispar 05)(Alkalescens-Dispar 06)(Alkalescens-Dispar 07)

Average

Group lB E. coli strains(08)(09)(025)026(1OllB4)0114011901420145(128)(Alkalescens-Dispar 08)

Average

E. coli K-12

a Strains in parentheses vmunoelectrophoretically.bNot included in average.

loot8684849390

87

818390817574757983817586857782

84807976

81

8288768380848585858283

83

83

were not tested

evident as nonreciprocal relatedness values atthese strains. The reciprocal binding datasented in Table 6 indicate that genomeof the reference strains are within 5% ol

7685666771747672696576777278

838178

another. Therefore, genome size differencescannot account for the differences in bindingobserved between 075 and group 1A strains(89%<0) and group 1B or group 2 strains (77 and79%, respectively). Similarly, the differences inbinding of 0112 and 0115 to members of thedifferent groups cannot be accounted for solelyon the basis of possible genome size differences.When relative relatedness data are plotted as

a frequency distribution, the mode for E. coliK-12 reaction with E. coli strains is 90 to 94%7,and the range of relatedness values around thismode approximates a Gaussian distribution(3). Frequency distribution graphs of related-ness for each of the four reference strains em-ployed in this study are shown in Fig. 1. Despitethe relatively small number of strains, especially

TABLE 5. DNA relatednless of E. coli strain 0115 toE. coli strains

% Relative bindingSource of unlabeled DNA'

At 60 C At 75 C

Group 2 E. coli strains0115028032011201240136

Average

71 Group IA E. coli strains(OIA)

74 02(02A)04

79 (04)74 0661 (06)80 0776 (07)81 07585 (075)737576

Average

77 Group I B E. coli strains(09)

76 026068

73 01140119

d im- 01420145

AverageIllUlgpre-sizes

f one

lOOb8183818280

81

8181858479838192858681

84

94938195958192

90

loob7982817880

80

7078797972757486758374

77

85857993927589

85

a Strains in parentheses were not tested im-munoelectrophoretically.

I Not included in average.

312 INFECT. IMMUNITY

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

DNA RELATEDNESS IN PATHOGENIC E. COLI STRAINS

TABLE 6. Reciprocal bintdinig as an indicationz ofrelative genlome size

%N Relative Maximum difference inReactiona binding at genome size based on

60 C reciprocal binding

075/0112 78 075 = 4'0 larger0112/075 81075/0114 79 0114 = 1'-0 larger

0114/075 78075/0115 88 0115 = 2%' larger0115/075 860112/0114 84 0114 = 4%C larger0114/0112 810112/0115 84 0115 = 4%C larger0115/0112 81 I0114/0115 88 0114 = 8%, larger0115/0114 95

a Number preceding slash line indicates sourceof labeled DNA.

in group 2, the relative relatedness data for eachreference strain with each group of strains ap-proximate a Gaussian distribution. With theexception of 0114, the group to which the refer-ence strain belongs can be separated from theother two groups on the basis of frequency dis-tribution. A great many more strains should betested with one or more reference strains fromeach group in order to determine whether eachgroup shows a distinct relatedness pattern withinthe overall frequency distribution of the speciesE. coli.

DISCUSSIONIn a recent study (3), a large number of E. coli

strains have been investigated by interstrainDNA-DNA reassociation, genome size, andguanine plus cytosine determinations in order toassess the degree of difference or divergence ingenetic material that is tolerable within a species.From this sampling, it is apparent that E. colistrains vary by at least 4% in guanine plus cytosinecontent and 23%o in genome size (enough DNAto specify some 600 average-sized genes). Asmuch as 25% divergence occurred between E. colistrains, based on their ability to form stable DNAduplexes between strains at 75 C. It is, therefore,evident that many E. coli strains display sub-stantial differences in their DNA. Within thedistribution of E. coli strains, one can identifyseveral subgroups. For instance, the Alkalescens-Dispar strains form a tight subgroup with 88 to95%0 relatedness to E. coli K-12, whereas Shigellastrains form a subgroup on the low end of theE. coli frequency distribution (3).

In the present study we sought to determinewhether pathogenic E. coli strains grouped onthe basis of immunoelectrophoretic mobilityshow preferential DNA relatedness. The data

indicate preferential relatedness in group lAstrains and in group 2 strains. There are excep-tions including strain 02A, and strain 0115, amember of group 2 that shows preferential re-latedness to group lB strains. The referencestrain from group 1B, 0114, reacted to approxi-mately the same extent with strains from allthree groups. Statistical treatments and fre-quency distribution graphs of these data sub-stantiate the conclusions drawn from mean rela-tive relatedness values.A variety of functions have been identified as

essential to the pathogenic capacity of differentbacteria. These include enzymes allowing in-vasiveness, toxin production, coagulase produc-tion, resistance to phagocytosis, and suscepti-bility to specific lysogenic converting phage.There are, however, no data on exactly how muchDNA in a given organism is concerned with theability to produce disease. Since both pathogenicstrains and nonpathogenic strains comprise thespecies E. coli, data from the present study are ofinterest in this regard.

Average relatedness among the three groups ofstrains is approximately 80%. Thus, preferentialrelatedness of strains within a given group musteither be due to sequences held in common inthe remaining 20%-O of DNA, or due to significantdifferences in genome size. Although it is truethat genome size differences as large as 25% doexist between certain E. coli strains, reciprocalbinding reactions do not indicate significantgenome size differences among the referencestrains employed in the present study. These dataare therefore consistent with an interpretationin which about one-half of the DNA not commonto all three groups is conserved among group IAstrains (where mean relative relatedness is 89%).Similarly, one-third of the DNA not common toall three groups is conserved among group 2strains. It follows that the "additional" DNA con-served in each of these groups is either a differentset of sequences, or a set of sequences originallypresent in an ancestor common to both groupsthat has diverged to a point where it is disimilarin group 1A and group 2.Data obtained from group 1B strains are more

difficult to interpret. The reference strain fromthis group, 0114, exhibits the same averagerelatedness to members of all three groups. Thisresult is consistent with an 80% core of DNA re-latedness between all three groups where the re-maining 20% differs among group lB strains tothe same extent that it differs between group lBstrains and strains from the other two groups. Itis also possible that 0114 is not representativeof DNA relationships among group 1B strains.Strain 0115, the atypical group 2 strain, showspreferential relatedness to group 1B strains. If

VOL. 6, 1972 313

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BRENNER ET AL. INFECT. IMMUNITY

0114 REACTIONS

I/\- GROUP IA075 REACTIONS

JP IA

12

L6l 0112 REACTIONS 0115 REACTIONS

E A_GROUP IA

GROUP lBGROUP IA

8-

6- GROUP 2

4 -GROUP lB

2 GROUP 2

60-69 70-79 80-89 90-99 100 60-69 70-79 80-89 90-99 100

% RELATIVE BINDING

FIG. 1. Frequency distribution of relatedness among E. coli strains. The values shown are obtained from 60 CDNA reassociation reactions in which E. coli 075, 0112, 0114, and 0115 were the referenice strains.

0115 reactions are considered typical for group lBrelatedness, we see preferentially high related-ness among these strains to at least the sameextent as is present in groups 1A and 2. It will beworthwhile to isolate the 20% of DNA not com-mon to the three groups and rereact it with eachgroup.There is no way to determine how much of the

DNA that is preferentially conserved amongpathogenic strains of E. coll is specifically asso-ciated with pathogenesis. The upper limit ofDNA specifically concerned with the pathogenicprocess is approximately 10%, enough to specifysome 250 to 300 genes of average size. Whereasthis figure provides an upper limit, it is possiblethat the bulk of preferentially shared sequencesare not concerned with pathogenicity. In fact,one may argue that pathogenicity is largelydetermined by single mutations.

Preferentially high nucleic acid relatedness

among pathogens occurs in several groups ofbacteria. Brucella species are virtually identical(5), although this result may be due to taxonomicsplitting of virtually identical strains with dif-ferent host specificity. Studies on the genusNeisseria that contains both pathogens and non-pathogens also indicate preferentially high re-latedness among pathogens. Strains of N.gonorrhoeae and N. meningitidis share at least80% of their DNA, whereas relatedness betweenseven nonpathogenic species varied from 15 to75% (6, 7). Another case in point is relatednessin shigellae compared to relatedness amongEnterobacter species. Strains of S. flexneri, S.boydii, S. dysentereae, and S. sonnei exhibit 80%or greater relatedness, whereasDNA species fromthe normally nonpathogenic species E. aerogenes,E. cloacae, E. liquifaciens, and E. hafniae havediverged in the range of 50 to 80% (Brenner,unpublished observations).

314

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

DNA RELATEDNESS IN PATHOGENIC E. COLI STRAINS

How does one account for the apparent con-

servation of DNA in pathogenic bacteria? Itis usually argued that pathogens have evolvedinto an ecological niche that is specialized enoughto prohibit gross divergence. In other words, thehost-parasite relationship is in delicate balanceso that significant divergence will be lethal or

prevent a bacterium from competing success-

fully in its environment. Alternatively, non-

pathogenic or rarely pathogenic bacteria havemuch more varied ecological niches and thereforedivergence is successful at a greater frequency.

ACKNOWLEDGMENTS

The authors are grateful to Helen C. Sing who performed thestatistical analyses. We wish to thank W. H. Ewing and G. J.Hermann for providing cultures of several of the E. coli strainsused in this study.

LITERATURE CITED

l. Brenner, D. J., G. R. Fanning, K. E. Johnson, R. V. Citarella,and S. Falkow. 1969. Polynucleotide sequence relation-

ships among members of the Entterobacteriaceae. J. Bacteriol.98:637-650.

2. Brenner, D. J., G. R. Fanning, A. Rake, and K. E. Johnson.1969. A batch procedure for thermal elution of DNA fromhydroxyapatite. Anal. Biochem. 28:447-459.

3. Brenner, D. J., G. R. Fanning, F. J. Skerman, and S. Falkow.1972. Polynucleotide sequence divergence among strains ofEscherichia coli and closely related organisms. J. Bacteriol.109:953-965.

4. Gillis, M., J. DeLey, and M. DeCleene. 1970. The determina-tion of molecular weight of bacterial genome DNA fromrenaturation rates. Eur. J. Biochem. 12:143-153.

5. Hoyer, B. H., and N. B. McCullough. 1968. Polynucleotidehomologies of Brucella deoxyribonucleic acids. J. Bacteriol.95:444-448.

6. Kingsbury, D. T. 1967. Deoxyribonucleic acid homologiesamong species of the genus Neisseria. J. Bacteriol. 94:870-874.

7. Kingsbury, D. T., G. R. Fanning, K. E. Johnson, and D. J.Brenner. 1969. Thermal stability of interspecies NeisseriaDNA duplexes. J. Gen. Microbiol. 55:201-208.

8. Orskov, F., I. Orskov, B. Jann, and K. Jann. 1971. Immuno-electrophoretic patterns of extracts from all Escherichia coliO and K antigen test strains: correlation with pathogenicity.Acta Pathol. Microbiol. Scand. Sect. B 79:142-152.

VOL. 6, 1972 315

on February 17, 2020 by guest

http://iai.asm.org/

Dow

nloaded from