Vol. 155, No. 2JOURNAL OF BACTERIOLOGY, Aug. 1983, p. 541-5480021-9193/83/080541-08S02.00/0Copyright 0 1983, American Society for Microbiology

Gene in the Major Cotransduction Gap of the Escherichia coliK-12 Linkage Map Required for the Expression ofChromosomal Resistance to Tetracycline and Other

AntibioticsANTHONY M. GEORGE1 AND STUART B. LEVY' 2*

Departments of Molecular Biology and Microbiology' and ofMedicine,2 Tufts University School of Medicine,Boston, Massachusetts 02111

Received 4 January 1983/Accepted 10 May 1983

In Escherichia coli K-12, amplifiable resistance to tetracycline, chlorampheni-col, and other unrelated antibiotics was mediated by at least four spatiallyseparated loci. Tetracycline-sensitive mutants were isolated by TnS insertionalinactivation of an amplified multiply resistant strain. One of these, studied indetail, showed coordinate loss of expression of all other resistance phenotypes.The Tn5 element in this mutant mapped to 34 min on the E. coli K-12 linkage map.We have designated the locus marA (multiple antibiotic resistance). Tetracycline-sensitive mutants containing marA::TnS regained all resistance phenotypes atfrequencies of 10-8 to 10- upon precise excision of TnS. Moreover, a newlydescribed tetracycline efflux system (A. M. George and S. B. Levy, J. Bacteriol.155:531-540, 1983) was inactivated in tetracycline-sensitive mutants, but recov-ered in tetracycline-resistant revertants. In merodiploids, F-prime marA+ ex-pressed partial or complete dominance over corresponding mutant chromosomalalleles. Dominance tests also established that a previously amplified host and amutant marA allele were preconditions for the expression of phenotypic resis-tances.

In the accompanying paper (9), we have de-scribed high-level resistance to tetracycline,chloramphenicol, and other antibiotics in plas-midless strains of Escherichia coli. Initially,spontaneous Tet and Cml mutants were isolatedby selection on complex agar medium containingeither drug at 5 ,ug/ml. These low-level resist-ance mutants were then "amplified" to high-level resistance (>100 ,ug/ml) either by stepwisetransfer of clones on plates containing incremen-tally higher concentrations of tetracycline orchloramphenicol or by growth for many genera-tions in liquid medium containing tetracycline orchloramphenicol at 5 ,ug/ml. Amplification ofresistance by tetracycline or chloramphenicolselection resulted in coincident resistance to anumber of other antibiotics (9). In the absence ofdrug selection, resistance reverted to low levelswithin 100 generations.We present evidence here for involvement of

four regions of the E. coli K-12 chromosome inthis resistance. The isolation of tetracycline-sensitive mutants by the insertion of TnS into asingle locus, which mapped in the major cotrans-duction gap of the E. coli K-12 chromosome,allowed identification of the marA locus (formultiple antibiotic resistance). marA is essential

for the expression of all the emergent chromo-somal resistance phenotypes and the tetracy-cline efflux system (9).

MATERIALS AND METHODSBacteril strai. All strains were E. coli K-12 (Table

1 or listed below).Media. MacConkey lactose, L, and minimal A liquid

or agar media supplemented with antibiotics, aminoacids, or other requirements were prepared as de-scribed elsewhere (9).Selction of Tet and Cml mutants and amplification of

resistan. The methods employed for the selectionand amplification of resistance phenotypes and mini-mum inhibitory concentrations (MICs) are describedin the accompanying paper (9).P1 transductions. The use of P1 vir for generalized

transduction has been described previously (6, 11). Inmapping experiments, cotransducibility was convert-ed to map distance in minutes by the formula of Wu(17), taking 2.3 min of E. coli chromosomal DNA asthe maximum amount packaged by P1 bacteriophage.When donor strains carried transposons, a correctionfactor was subtracted from 2.3 min to allow for thecontribution of transposon DNA to the transducingparticle (8).TnS mutagenesis. The amplified Tet mutant AG102

was infected with A b221 c1857 rex::TnS (obtainedfrom A. Wright) by the method of Shaw and Berg (15).

541

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

TABLE 1. E. coli K-12 strainsStraina Characteristicsb Source or reference

DO-23 uncB402 argE3 thi-l rpsL xyl mtl galK This laboratory, F. Gibson via J. DaviessupE44

AG100 As DO-23 but uncB+ and A (gal-uvrB)A- 9

AG102 As AG100, but expresses Tcl0r 9AG107 As AG100, but expresses Tc6OT 9AG406 As PA309, but expresses Tc5Or This paperAG426 As PA309, but expresses Cm50 This paperAG440 As GMS407, but recA This paper [Hfrll1(KL16recA) x

GMS407]AG446 As GMS407, but expresses Tc5OO This paperAG466 As GMS407, but expresses Cm5OT This paperAG1025 As AG102, but contains marA20::TnS This paper (X::TnS -- AG102)AG1027 As AG1025, but contains zdd-230::Tn9 This paper (Pl * PLK1253 -- AG1025)

and zde-234::TnlOAG1029 As PLK1253, but contains marA20::TnS This paper (Pl * AG1025 - PLK1253)AG4465 As AG446, but contains marA20::Tn5 This paper (Pl * AG1025 - AG446)AG4665 As AG466, but contains marA20::TnS This paper (P1 - AG1025 - AG466)PA309 gal-6 rpsL9 argHI his-i X- gyrA tonA2 1

thr-l mtl ara-13 xyl-7 malAl lacYl trpleu-6

GMS407 argE3 lacY) galK2 manA4 mtl-l tsx-29 12supE44 uidA

FS173 leuB6 argG6 metBI lacYI or lacZ4 xyl-7 16rpsLI04 ksgBI tonA2 tsx-i X- supE44

PLK1253 trpR trpA9605 his-29 ilv pro arg thyA 3deoB or deoC tsx Arac zdd-230::Tn9zde-234::TnlO

a This table does not contain many previously described and constructed strains that were used in singleexperiments. These appear in the appropriate sections of the text.

b Phenotypes given in the table only represent the levels of Tcr or Cmr selected. Each of these mutantsexpresses other antibiotic phenotypes at various levels (9).

After infection, washing, and aeration at 30°C for 90min, enrichment for tetracycline-sensitive cells by twocycles of D-cycloserine treatment (11) in the presenceof tetracycline (5 p.g/ml) was carried out. Sampleswere spread over MacConkey agar plates containingkanamycin at 30 ,g/ml, and the plates were incubatedfor 24 to 36 h at 37°C. Kanamycin-resistant cloneswere purified on master plates containing kanamycinand replica-plated to plates containing tetracycline orchloramphenicol at 5 or 10 ,g/ml. To distinguishbetween tetracycline-sensitive back-mutants and TnSinsertional tetracycline-sensitive mutants, we exam-ined the level of Tcr in AG102 after growth in theabsence of tetracycline for the same number of genera-tions as the transfected culture; 5,000 of 5,000 testedAG102 clones retained resistance to tetracycline at 10FLg/ml (Tc10f).Mating experiments. Time of entry and gradient of

transfer matings were performed between Hfr and F-strains as described previously (6, 10). Resistant do-nors and recipients were constructed by selection andamplification before matings. In the time of entrymatings, recombinants were selected on amino acid-supplemented minimal agar plates that were thenreplica-plated to antibiotic-supplemented agar plates.Hfr strains were obtained from the E. coli GeneticStock Center (B. Bachmann, Curator); their origins oftransfer are given in reference 2. Conjugal transfers ofplasmids with appropriate selection and counterselec-

tion were performed as described previously (6, 11).Tetracycline transport assays. Tetracycline transport

assays were performed as described in the accompa-nying paper (9).

Definitions. Amplification of resistance and pheno-typic designations are defined in the accompanyingpaper (9). Tet and Cml denote mutants selected bytetracycline and chloramphenicol, but these mutantselaborate cross-resistance to many antibiotics (9).

RESULTSMapping of resistance regions. To determine

whether selection and amplification of Tet andCml mutants and the attendant multiple resist-ance phenotypes were derived from chromo-somal mutation(s), we mapped Tcr and Cmr byHfr gradient of transfer and F' plasmid domi-nance.

(i) Transfer of resistance phenotypes. Twotypes of experiments were performed. In thefirst, Hfr donors (KL96, B8, and U7) wereamplified to high-level Tcr or Cmr and thenmated with PA309. In a second series of mat-ings, sensitive Hfr donors were crossed withtetracycline- or chloramphenicol-amplified re-cipients. In neither type of experiment was high-level Tcr or Cmr totally transferred or lost, even

542 GEORGE AND LEVY J. BACTERIOL.

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI GENE FOR CHROMOSOMAL DRUG RESISTANCES 543

after very long matings. Without exception, wecould only obtain transfer of about 25% of theTcr or Cmr levels originally expressed in thedonors. The transconjugant resistance pheno-types consistently mapped between trp and his,regardless of the origin of transfer of the donorHfr strain. In Cml mutants only, a second locusmediating Cmr mapped between gal and trp, andwe assumed this to be the cmlA mutation (13, 14).

In time of entry matings spanning the trp-hisregion, Tcr was not expressed in transconju-gants after mating a resistant donor (KL96 am-plified with tetracycline or chloramphenicol)with a sensitive recipient (PA309). This failure tolocate Tcr in the trp-his region was an unexpect-ed result since the gradient of transfer matingshad consistently mapped the gene for Tcr to thisregion. Our interpretation was that Tcr mightrequire the cooperation ofwell-dispersed alleles.

(ii) Mapping by F' plasmid dominance. Trans-acting wild-type episomes (F13, F112, F123,F126, F500, F506, and F621) expressed partialdominance to corresponding regions of tetracy-cline- or chloramphenicol-amplified mutantchromosomes (Table 2). These tests define Tcr

TABLE 2. Location of resistance loci by dominancetests% of recipient MIC retained by

Donor F' Recipient transcoijugantsbplasmid"a

Tetracycline ChloramphenicolF13 (7 to 14) AG406 90 40

AG426 90 50F112 (87 to 97) AG406 100 40

AG426 100 40F123 (27 to 31) AG406 100 100

AG426 100 50F126 (16 to 31) AG406 100 100

AG426 100 50F150 (41 to 45) AG406 100 100

AG426 100 100F152 (12 to 17) AG406 100 100

AG426 100 100F500 (34 to 45) AG446 80 80

AG466 85 90F506 (31 to 37) AG446 55 60

AG466 50 70F621 (30 to 36) AG446 50 50

AG466 40 60a The numbers within parentheses represent the

chromosomal region (in minutes) spanned by F' plas-mid DNA. F' plasmid strains were obtained from theE. coli Genetic Stock Center and are listed in reference2. F621 was obtained from R. M. Bitner (3).

b A minimum of 50 transconjugants from each mat-ing cross were screened for residual Tcr or Cmr byreplica-plating purified transconjugants to plates con-taining drugs. More than 95% of the transconjugantsfrom each mating produced the result listed in thetable. The percentages represent (transconjugant MIC/recipient MIC) x 100.

and Cmr in the leu-lac (min 7 to 14) and trp-his(min 30 to 37) regions and Cmr in the argH-leu(min 87 to 97) and gal-trp (min 16 to 31) regions.F150 and F152 did not express dominance.Although these mapping data do not localize

discrete resistance alleles, they do support a

multigenic model of chromosomal tetracyclineand chloramphenicol resistances.

Identfication of a resistance locus by insertion-al inactivation. AG102 (TclOr; a second-step Tetmutant) was mutagenized with TnS using a A::TnS vector, and Kmr clones were isolated. Tenthousand Kmr transductants were screened forTc5 and Cms. Of these, three were completelysensitive to tetracycline and choramphenicol,and 7 were less resistant to tetracycline or

choramphenicol or both. All 10 mutants elabo-rated mucoid phenotypes on complex agar at37°C. One of the tetracycline-sensitive mutants,designated AG1025, was selected for furtheranalysis. AG1025 expressed greater sensitivityto tetracycline and chloramphenicol than even

the original parent, AG100 (Table 3). Randomand directed Tn5 insertions into AG100 (strainsAG1004 and AG1005 in Table 3) did not alter theMICs for tetracycline, minocycline, and chlor-amphenicol or produce mucoidy.AG1025 did not exhibit the tetracycline efflux

system seen in the tetracycline-resistant parent,AG102 (9). Moreover, AG1025 showed an activeuptake of tetracycline that was nearly twice thatof the sensitive AG100 strain (data not shown).This result was consistent with the lower MIC oftetracycline in AG1025 compared with that inAG100 (Table 3). Strain AG1025 had regainedsensitivity to all other antibiotics (nalidixic acid,rifampicin, puromycin, penicillin G, ampicillin,cephalothin, and minocycline) toward whichAG102 showed resistance (9).A reversion test, designed to select Tet rever-

TABLE 3. MICs of tetracycline, minocycline, andchloramphenicol for AG100 derivatives

MIC (1&g/mlY'Strain

Tetracycline Minocycline Chloramphenicol

AG100 1.0 2.0 4.0AG102 10.0 5.8 14.5AG1004b 1.0 1.9 4.0AG1005b 1.0 1.9 3.8AG102SC 0.8 1.6 2.5

a Average of two determinations for which MICswere within 10o of each other. Antibiotic concentra-tions were in increments of 0.1 (to 2 pLg/ml), 0.5 (2 to 5Wg/ml), and 1.0 (5 to 20 pg/ml).b AG1004 contains a random Tn5 insertion (X::Tn5AG100); AG100S contains a directed Tn5 insertion

(P1 *AG1025 -S AG100).c Derivative of AG102 showing TnS insertional inac-

tivation of resistance.

VOL. 15;i 1983

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

544 GEORGE AND LEVY

tants of AG1025 after precise excision of thetransposon, was performed by plating largenumbers of late exponential phase cells on Mac-Conkey agar plates supplemented with 5 ,ug oftetracycline or chloramphenicol per ml. Pre-sumptive Tet clones were purified on the sameplate type and then screened on kanamycin-supplemented plates. AG1025 reverted to Cm5r(Kms) at a frequency of 4 x 10-8 (average ofthree experiments) and to Tc5r (Kms) at a fre-quency of 1 x 10-7 (average of three experi-ments). All Cmr revertants expressed Tc5r andvice versa; and all revertants, although nowKms, remained extremely mucoid on complexmedia at 37°C. That the reversion frequency ofAG1025 to Tc5r was nearly 100-fold higher thanthe spontaneous mutation rate from Tcs to Tcr(9) and was accompanied by loss of Kmr indicat-ed that AG1025 was not a back-mutant.

Inactivation of resistance by a P1 transducinglysate from AG1025. A Pl lysate of AG1025should transduce this TnS insert into other am-plified Tet and Cml strains. Kmr was transducedby P1 to a number of Tet and Cml mutants of E.coli K-12 at frequencies of 10-8 to 10-6, and 85to 100% of transductants selected for Kmr ex-pressed Tcs and Cms, indicating that recombina-tion in the region carrying the TnS insertion wasprecise. In each case, Kms Tet revertants wererecovered at frequencies of about 10-8. All KmrTcs transductants also lost all other resistancephenotypes, and all Kms Tet revertants recov-ered all resistance phenotypes.

In these reversion tests Kms Tet revertantswere recovered from strains that were amplifiedwith tetracycline or chloramphenicol selectionbefore transduction to Kmr. The mapping evi-dence (above) that Tcr and Cmr were expressedfrom several loci prompted us to examine thefeasibility of recovering low-level Tcr or Cmrafter transduction of sensitive strains withPl * AG1025. However, after PA309 andGMS407 were transduced to Kmr we could notrecover Kms Tet revertants even at a frequencyof 10-10. This result indicated that the TnS-inactivated locus alone could not produce Tcr(after loss of TnS) in otherwise sensitive strains.Mapping of TnS in AG1025. The approximate

position of the TnS insertion in strain AG1025was obtained from bacterial conjugation experi-ments. Hfr donors (KL96, AR101, and U7) wereamplified to Tcr and Cmr and then transduced toKmr by P1 * AG1025. The isolation of Kmr Tcstransductants indicated that the TnS elementwas inserted into the correct locus in the Hfrstrains. These donors were mated with PA309and recombinants (argH+, leu+, lac+, gal',trp+, and his+) were selected and scored by thegradient of transmission. Recombinant classesindicated close linkage of Kmr to his and trp in

all matings, and poor linkage to the more distantgal, lac, leu, and argH markers (for example, inthe mating of KL96 (Tets, Km9 with PA309, outof 176 selected Kmr recombinants, 133 weretrp+, 105 were his', 56 were gal', 6 were leu+,and 1 was argH+). More precise mapping byinterrupted matings located the TnS insertionnear the midpoint (34 to 35 min) of the trp-hisregion. Next, we attempted to determine P1cotransductional linkage of TnS to known mark-ers in the midrange of the trp-his region (Table4). TnS and manA were 1.4 to 1.7% cotransduci-ble, which represents a map distance of 1.6 minapart (17). TnS and ksgB were not cotransduci-ble. The ksgB locus has been recently reposi-tioned (7) to min 36.5 on the clockwise side ofmanA. This datum enabled us to position theTnS site on the counterclockwise side of manA,near relB at the edge of the major cotransductiongap.

Recently, the major cotransduction gap hasbeen spanned by the insertion and mapping ofvarious transposons in a "leapfrog" pattern (3,8). We utilized one of these new strains,PLK1253, which contains TnlO and Tn9 inser-tions near the TnS-inactivated locus. WithAG1025 as the P1 donor strain and PLK1253 asthe recipient, TnlO and Tn9 were crossed out atfrequencies of 55 and 18%, respectively. In thereciprocal cross with PLK1253 as the donor,TnS was crossed out of AG1025 at frequencies of53% by TnlO and 6% by Tn9. These data indicat-ed that TnS was much closer to TnlO at min 34.2than to Tn9 at min 33.3. The transductionsdescribed above were used to construct strainscontaining all three transposons; with a manArecipient (GMS407), a series of two-factor cross-es were performed (Table 4). TnS and TnlO werecotransduced at frequencies of 68 to 89o, de-pending upon the donor strains used and wheth-er Kmr or Tcr was the selected phenotype.When all of the crosses listed were considered,the data were consistent with the map order ofgenes and transposon insertions given in Fig. 1.The TnS insertion was placed at min 34.05 justinside the major cotransduction gap. We havedesignated the new locus marA (multiple antibi-otic resistance).The TnS-inactivating insertions of the other

two originally isolated tetracycline-sensitive mu-tants of AG102 also mapped in the marA locus,but we cannot determine whether these wereindependent insertion mutants.Dominance testing of the marA locus. None of

the amplified F' strains transferred expressibleTcr or Cmr to sensitive recipients. This failure ofexpression could not have been caused by domi-nance by the wild-type chromosomal marA+allele in the recipients because merodiploid anal-yses (Table 2) indicated only partial dominance

J. BACTERIOL.

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI GENE FOR CHROMOSOMAL DRUG RESISTANCES 545

TABLE 4. P1 transductional mapping of Tn5 in AG1025

Selected Unselected % Cotrans- DistanceDonor Recipient marker marker ducibilitya apart

(mm)

AG1025 GMS407 mr Man' 1.4 (5/352) 1.6Man' Kmr 1.7 (6/352) 1.6

AG1025 FS173 Kmr KsgS 0 (0/264) >1.8FS173 AG1025 Ksgr Kms 0 (0/362) >2.0AG1027 GMS407 Man' Kmr 1.6 (10/642) 1.4

TCr 3.3 (21/642) 1.3Cmr 0 (0/642) >1.6

K,,,mr IMan+ 1.6 (21/1,322) 1.4TCr 71 (938/1,322) 0.2Cmr 22 (293/1,322) 0.7

TCr Man+ 2.8 (24/848) 1.3Kmr 80 (674/848) 0.1Cmr 8.5 (72/848) 1.0

Cmr Man+ 0 (0/462) >1.6Kmr 21 (98/462) 0.75TCr 11 (49/462) 1.0

AG1029 GMS407 Man' Kmr 1.7 (4/240) 1.4TCr 2.5 (6/240) 1.3Cmr 0 (0/240) >1.6

K(mr Man' 1.4 (4/294) 1.4TCr 68 (200/294) 0.2Cmr 16 (48/294) 0.8

TCr Man+ 1.9 (3/160) 1.4IKmr 89 (143/160) 0.1Cmr 8.1 (13/160) 1.0

Cmr Man+ 0 (0/141) >1.5Kmr 17 (24/141) 0.8TCr 8.5 (12/141) 1.0

a (The frequency of unselected markers)/(selected markers) is given within parentheses.

by episomal marA+ alleles. When tetracycline-or chloramphenicol-amplified strains containingF506 or F621 were mated with a tetracycline-amplified, TnS-inactivated recipient (AG4465),50 to 60%o of the level ofTcr or Cmr expressed bythe F' donor was transferred and expressed inthe recipient (Table 5). Thus presumably mutantF' plasmid marA alleles were trans-active in arecipient that had been previously amplified, butcontained the TnS-inactivated chromosomalmarA allele. These results suggested that acooperativity existed between the marA locusand other mutant allele(s) in regions external tothat spanned by the F' plasmids. In matingsbetween wild-type strains containing F506 andF621 and AG4465, no Tcr or Cmr was expressedin the transconjugants, indicating that, indeed, amutant marA allele and a tetracycline-amplified,TnS-inactivated marA recipient were prerequi-sites for resistance.The F500 plasmid probably does not extend

much beyond the relB locus because very littleTcr from tetracycline- or chloramphenicol-am-plified strains containing F500 was expressed inthe AG4465 recipient (Table 5).marA lou affects the expression of highlevel

Tcr and Cmr. Representative K-12 strains were

amplified to high levels (50 to 100 ILg/ml) of Tcrand Cmr by selection in the presence of eitherdrug. These mutants were then transduced toKm' with P1 * AG1025 and the purified trans-ductants were scored for Tc' and Cm' (Table 6).Nontransduced, amplified strains were includedas controls to monitor maximum residual Tcrand Cmr. These were found to be reduced byless than 10%o of the amplified levels.When the selection and amplification was with

tetracycline, 93 to 99% of clones selected forKmr were sensitive to tetracycline and chloram-phenicol (Table 6), even though the recipientswere resistant to drug concentrations up to 10-fold higher than was the original strain, AG102,from which the Pl-inactivating lysate was de-rived. When the selection and amplification waswith chloramphenicol, two results were ob-tained. In some cases, 98 to 99%o of Kmr trans-ductants were sensitive to tetracycline andchloramphenicol, as seen with tetracycline se-lection. But in three examples (AG1315,AG2295, and AG3065), although 83 to 100lo ofthe Kmr transductants were Tet', the level ofCmr was still about 20%o of the original amplifiedlevels (as seen in the size of colonies on plateswith 10 and 20 ,ug of chloramphenicol per ml).

VOL. lSS, 1983

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

546 GEORGE AND LEVY

zde-234,th1Q(J") zdd-230-1Th mm Ma k gB

~~II I I

31 3 33 34 35 36 37

L19 2.6

FIG. 1. Segment of the E. coli K-12 linkage mapfrom min 31 to min 37, including relevant genes andtransposon insertions. This region includes the majorcotransduction gap from min 31.1 to min 34.4. Theclockwise direction of the circular map corresponds toa direction of left to right in the above figure. Theposition of marA20::Tn5 was deduced from the P1tranductional analysis in Table 4. The numbers on barsbelow the figure indicate average P1 cotransducibility(percent) between the connected markers, from thetwo-factor crosses in Table 4.

These anomalous chloramphenicol results mightsuggest that independent mutational events mayoccur at different levels of chloramphenicol se-lection in separate experiments. However, thedata in Table 6 clearly demonstrate a distinctionbetween high-level Tcr Cmr and low-level Cmrcomponents.The residual Cmr component could be due to

loci such as cmlA and cmlB that might not be

TABLE 5. Dominance testing of the marA locus

% of donor MICexpressed in

Donor F' Recipientb transconjugantscplasmida

Tetra- Chloram-

cyctine phenicol

F500 AG439 or AG440 <5 <5AG4465 5 <5

F506 AG439 or AG440 <5 <5AG4465 45-60 50

F621 AG439 or AG440 <5 <5AG4465 50-65 50-60

a Regions spanned by F' plasmids are given in Table2. Donor strains containing the indicated F' plasmidswere amplified with tetracycline or chloramphenicol to,resistance levels of 50 pLg/ml before being mated withthe recipients.

b AG439 is a Rif' derivative of GMS407 (Table 1);AG440 is a Rifr recA derivative ofGMS407; AG4465 isa Rifr derivative of GMS407 that was amplified toTc5Or then Tn5-inactivated with P1 * AG1025.

c Transconjugants were selected on minimal agarplates supplemented with mannose as the sole carbonsource (selection against the recipients) and rifampicinto 60 ,ug/ml (selection against the donors). A minimumof 50 transconjugants from each mating were screenedfor levels of resistance to tetracycline and chloram-phenicol, and more than 90%o of these produced theresult listed in the table. The percentages represent(transconjugant MIC/donor MIC) x 100.

TABLE 6. TnS insertional inactivation of high-levelTetr and Cml' in E. coli K-12 strains

No. of KmrStrain' transductants % Tetsc % Cmlsd

testedb

AGll15 319 93 93AG1315' 364 83 0AG2075f 264 99 99AG2295f 364 98 0AG24959 88 98 98AG2695' 188 99 99AG3065h 100 100 0AG3095h 174 98 98

a Strains were amplified on antibiotic-supplementedplates before transduction to Km' with P1 * AG1025.

b KMr transductants were purified on kanamycin-supplemented (30 jg/ml) agar plates and replica-platedto tetracycline- or chloramphenicol-supplemented (5,10, 20, or 30 ,ug/ml) agar plates.

c Tetracycline sensitivity was interpreted as an MICof <5 ,ug/ml.dChloramphenicol sensitivity was interpreted as an

MIC of <10 tig/ml.e Derived from AG100 via AG111 (Tc100') and

AG131 (Cm1009).f Derived from U7 via AG207 (Tc6O') and AG229

(Cm809.8 Derived from B8 via AG249 (Tc8O') and AG269

(Cm8OT.h Derived from AR101 via AG306 (Cm5Of) and

AG309 (Cm8Or).

under the control of the TnS insertional inactiva-tion and -would not be seen in tetracycline-amplified strains because the cmlA allele is notselected by tetracycline. To test this possibility,we mated Hfr donors AG2295 and AG3065 withPA309 and established linkage of residual Cmr togal (data not shown), consistent with the posi-tion of cmUA. In another experiment, RE103(cmlA) (14) and JF703 (cmlB) (5) were trans-duced to Kmr with Pl - AG1025; when the trans-ductants were screened for the low-level Cmrseen in the cml mutants, Cmr was found to beundiminished. These results suggested thatcmlA and cmlB were not affected by the TnSinsertional inactivation and that residual Cmr inchloramphenicol-amplified strains might be me-diated by these loci.

DISCUSSIONMerodiploid dominance tests indicated that

Tet and Cml phenotypes were determined fromleu-lac, gal-trp, trp-his, and argH-leu regions ofthe E. coli K-12 linkage map. Each region con-tributed Cmr or Tcr Cmr components, and wesurmised that an additive or synergistic cooper-ativity between loci may be a precondition forthe expression of high-level resistance. Thispremise was supported by the following obser-

J. BACTERIOL.

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI GENE FOR CHROMOSOMAL DRUG RESISTANCES 547

vations: (i) the consistent mapping of Tcr andCmr in the midrange of the trp-his region ingradient of transfer matings; (ii) the contrastingfailure to map transfer of Tcr in this region bytime of entry matings; (iii) the expression oftransferred mutant episomal alleles (in F' plas-mids sp the 34-mm region) required othermutant alleles in an amplified host (Table 5).

Several mutants derived-by Tn5 mutagenesisexpressed Tcs or reduced lrCr presumably byinsertional inactivation. A PI transducing lysatefrom one of these mutants, AG1025, caused theloss of all resistance phenotypes in more than90o of Kmr transductants. Pl transductionalmapping positioned TnS in AG1025 in the marAlocus at 34 min in the major cotransduction gapof the E. coli K-12 linkage map (Table 4 and Fig.1). It is difficult to invoke a single locus thatelaborates each of these phenotypes, and it islikely-and the merodiploid analyses and rever-sion tests support this conclusion-that two ormore loci are indicated. However, our datasuggest that a single locus-that at 34 min-iscritical for the expression of Tcr, Cmr and otherresistance phenotypes from itself and other al-leles.Our identification of this apparent multiple

resistance allele and pleiotropic locus in themajor cotransduction gap is interesting, particu-larly in view of recent suggestions (4) that thisregion is genetically "silent" either in lackinggene clusters or in containing more "exotic"genes with less detectable mutant phenotypes.The position of the marA locus shortens themajor cotransduction gap of 3.2 min by 0.4 min.The stability of TnS in this locus during Pltransduction suggests its usefulness as a newmarker in a relatively void region of the K-12chromosome.

In the accompanying paper (9) we describedan energy-dependent efflux system for tetracy-cline. This efflux was lost upon TnS insertionalinactivation, but was recovered upon reversionto Kmn Tet Cml (unpublished results). All low-level Tet mutants exhibited tetracycline effluxand therefore presumably contain marA, but alllow-level Cml mutants were not Tcr and did notexhibit tetracycline efflux; only first-step Cmlmutants that expressed Tcr and all second-stepCml mutants (which expressed coordinate Tcr)contain the marA mutation, as defined by thepresence of the efflux system (9).We propose that the expression of resistance

phenotypes requires a functional marA allele.The marA+ allele exerts partial (or complete)dominance in trans to the marA allele of the hostchromosome (Table 2), but marA+ is nonfunc-tional-in so far as resistance is concerned-inwild-type strains since the introduction of epi-somal marA+ alleles into amplified hosts inacti-

vated by marA20::Tn5 does not produce Tcr orCmr. However, subsequent amplification ofthese partial diploids proceeds with relative easedue to a marA+-to-marA transition in the F'plasmid while the host chromosome retains theinactivating TnS insertion.The mechanism of amplification of resistance

and its instability when the selective pressure isremoved are difficult to explain without addi-tional data. However, it could be speculated thatamplification occurs by duplication of one ormore mutant alleles as a consequence of tetracy-cline or chloramphenicol selection.The amplification process, the attendant mul-

tiplicity of resistance phenotypes, and the pres-ence of a tetracycline efflux system suggest twomodels for the operation of the marA locus. Inthe first, one might imagine a single regulatingallele that is functional in Tet and Cml mutantsand dysfunctional or nonfunctional in wild-typestrains or TnS-inactivated mutants. The allelealso regulates the function of other spatiallyseparated alleles involved in the expression ofresistance phenotypes. The second model con-tains the same features as the first, with theaddendum that marA may be one of two or moremembers of an operon. In this situation onewould expect that TnS in AG1025 was probablyinserted in the promoter of the operator-proxi-mal gene or in the structural region of a criticalregulatory or structural gene. In both models thepresence of more than one marA allele cannot beexcluded.

ACKNOWLEDGMENTSThese studies were supported by Public Health Service

grant A116756 from the National Institutes of Health.We thank A. Wright for helpful suggestions.

LITERATURE CITED1. Bachnann, B. J. 1972. Pedigrees of some mutant strains

of Escherichia coli K-12. Bacteriol. Rev. 36:525-557.2. R n, B. J., and K. B. Low. 1980. Linkage map of

Escherichia coli K-12, edition 6. Microbiol. Rev. 44:1-56.3. BItna, R. M., and P. L. Kuempel. 1981. P1 transduction

map spanning the replicaton terminus of Escherichia coliK-12. Mol. Gen. Genet. 184:208-212.

4. Bouche, J. P., J. P. Ge_gne, J. Lomrn, and J. M.Louam. 1982. Relationships between the physical andgenetic maps of a 470 x 103 base-pair region around theterminus of Escherichia coli K-12 DNA replication. J.Mol. Biol. 154:21-32.

5. Chal, T.-J., and J. Foulds. 1977. Escherichia coli K-12tolF mutants: alterations in protein composition of theouter membrane. J. Bacteriol. 130:781-786.

6. Cwrh, R., m. 1981. Gene transfer, p. 243-265. In P.Gerhardt (ed), Manual of methods for general bacteriolo-gy. American Society for Microbiology, Washington,D.C.

7. Fouts, K. E., and S. D. Barbour. 1981. Transductionalmapping of ksgB and a new Tn5-induced kasugamycinresistance gene, ksgD, in Escherichia coli K-12. J. Bacte-riol. 145:914-919.

VOL. lS5, 1983

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

548 GEORGE AND LEVY

8. Fouts, K. E., and S. D. Barbour. 1982. Insertion of tran-sposons through the major cotransduction gap of Esche-richia coli K-12. J. Bacteriol. 149:106-113,

9. George, A. M., and S. B. Levy. 1983. Amplifiable resist-ance to tetracycline, chloramphenicol, and other antibiot-ics in Escherichia coli: involvement of a non-plasmid-determined efflux of tetracycline. J. Bacteriol. 155:531-540.

10. Low, K. B. 1973. Rapid mapping of conditional and auxo-trophic mutations in Escherichia coli K-12. J. Bacteriol.113:798-812.

11. Miler, J. H. 1972. Experiments in molecular genetics.Cold Spring Harbor Laboratory, Cold Spring Harbor,N.Y.

12. Novd, M., and G. Novel. 1973. Mutants of E. coli K-12unable to grow on methyl-A-D-glucuronide: map location

of uidA. Locus of the structural gene of P-D-glucuroni-dase. Mol. Gen. Genet. 120:319-335.

13. Reeve, E. C. R. 1968. Genetic analysis of some mutationscausing resistance to tetracycline in Escherichia coli K-12. Genet. Res. 11:303-309.

14. Reeve, E. C. R., and D. R. Suttle. 1968. Chromosomallocation of a mutation causing chloramphenicol resistancein Escherichia coli K-12. Genet. Res. 11:97-104.

15. Shaw, K. J., and C. M. Berg. 1979. Escherichia coli K-12auxotrophs induced by insertion of the transposable ele-ment TnS. Genetics 92:741-747.

16. SparDg, P. F., Y. Ikeya, and D. Ellot. 1973. Two geneticloci for resistance to kasugamycin in Escherichia coli. J.Bacteriol. 113:704-710.

17. Wu, T. T. 1966. A model for three-point analysis ofrandom general transduction. Genetics S4:405-510.

J. BACTERIOL.

on August 21, 2020 by guest

http://jb.asm.org/

Dow

nloaded from