Vol. 164, No. 1JOURNAL OF BACTERIOLOGY, OCt. 1985, p. 114-1220021-9193/85/100114-09$02.00/0Copyright © 1985, American Society for Microbiology

Resistance to Chloramphenicol in Proteus mirabilis by Expression ofa Chromosomal Gene for Chloramphenicol Acetyltransferase

I. G. CHARLES,' S. HARFORD,'t J. F. Y. BROOKFIELD,2 AND W. V. SHAW'*Departments ofBiochemistryi and Genetics,2 University of Leicester, Leicester LE] 7RH, United Kingdom

Received 19 March 1985/Accepted 22 June 1985

Proteus mirabilis PM13 is a well-characterized chloramphenicol-sensitive isolate which spontaneously givesrise to resistant colonies on solid media containing chloramphenicol (50 ,ug mlP) at a plating efficiency of 10"to 0-5 Such chloramphenicol-resistant colonies exhibit a novel phenotype with respect to chloramphenicolresistance. When a single colony grown on chloramphenicol agar is transferred to liquid medium and grownin the absence of antibiotic for 150 generations, a population of predominantly sensitive celis arises'. Thismutation-reversion phenomenon has been observed in other Proteus species and Providencia strains, whereinresistance has been shown to be mediated in each case by the enzyme chloramphenicol acetyltransferase. Thecat gene responsible for the phenomenon is chromosomal and can be cloned from P. mirabilis PM13 with DNAprepared from cells grown in the absence or the presence of chloramphenicol. Recombinant plasmids whichconfer resistance to chloramphenicol carry an 8.5-kilobase PstI fragment irrespective of the source of hostDNA. The location of the cat gene within the PstI fragment was determined by Southern blotting with a catconsensus oligonucleotide corresponding to the expected amino acid sequence of the active site region ofchloramphenicol acetyltransferase, and the direction of transcription was deduced from homology with thetype I cat variant.

The mechanisms by which microorganisms express resist-ance to antibiotics is of clinical and theoretical importance.This report describes a novel phenomenon whereby a pop-ulation of chloramphenicol-sensitive cells of Proteusmirabilis gives rise to resistant cells at a very high frequency(10-4 to 10-5 per cell per generation).Genes specifying the enzyme chloramphenicol acetyl-

transferase (CAT) occur in many genera of bacteria, whereinthey confer resistance to chloramphenicol. The regulation ofcat gene expression depends in the main on factors whichcorrelate with the taxon of the bacteria in which they reside.The cat genes of gram-negative bacteria are commonlyfound on large R-plasmids and produce CAT constitutively.The most commonly observed variant in this category is thetype I polypeptide encoded by Tn9 and naturally occurringplasmids such as NR1 (also called R100 or R222) and R6 orthe in vitro plas,mid constructs such as pACYC184 andpBR328 (reviewed by Shaw [35]). The nucleotide sequenceof the type I cat has been determined (1), and the gene hasbeen demonstrated to be subject to catabolite repressionmnediated by cyclic AMP and the catabolite gene activatorprotein at the level of transcription (14, 22). Circumstantialevidence suggesting transcriptional activation by the cyclicAMP-catabolite gene activator protein system has not beenobserved for the type II and type III genes found on non-Fplasmids in enteric bacteria (35).The synthesis of CAT in gram-positive bacteria is induced

by chloramphenicol and selected analogs (43), and theregulation appears to be one of posttranscriptional control,which may be regulated by the secondary structure of theCAT mRNA (9, 15).Any study of chloramphenicol resistance and CAT syn-

thesis in P. mirabilis is bound to be complicated by two

* Corresponding author.t Present address: Glaxo, Greenford, Middlesex, United King-

dom.

considerations. The first is the well-documented phenome-non of selective amplification of genes encoding resistancedeterminants carried by certain R plasmids in P. mirabilis.Rownd and Mickel (33) described this phenomenon astransitioning and have shown that the increase in CATsynthesis due to gene amplification is reversible in thatgrowth of transitioned cells in drug-free medium results inloss of resistance determinants with concomitant decrease inresistance. The second consideration has been the knowl-edge that strains of P. mirabilis lacking detectable plasmidsmay nevertheless harbor cat determinants which (i) areexpressed poorly, (ii) specify a polypeptide similar to that ofthe type I CAT (45), and (iii)'have been described as mutableto high-level expression (37).The present studies were undertaken to gain a better

understanding of the structure and function of the putativechromosomal cat gene of P. mirabilis, its relationship toanalogous genes carried on plasmids in gram-negative andgram-positive bacteria, and the mechanism underlying thehigh-level expression of CAT in selected strains of P.mirabilis.

MATERIALS AND METHODS

Bacteria, plasmids, and phage. The bacteria, plasmids, andphage used in this report are listed in Tables 1 and 2. Allbacteria were grown in Penassay nutrient broth (DifcoLaboratories) unless otherwise specifically stated.Chromosomal DNA preparations. P. mirabilis genomic

DNA was purified by the method of Chow et al. (11), with anadditional final cesium chloride buoyant density centrifuga-tion step.

Plasmid preparations. Plasmids were prepared by the rapidmethod of Birnboim and Doly (7) with the protocol describedby Maniatis et al. (25).

Conjugation and transformation of E. coli and P. mirabilis.Strains were transformed by a modified version (19) of the

114

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

CHROMOSOMAL cat OF PROTEUS MIRABILIS 115

TABLE 1. Bacterial strains and plasmids

Strain or Genotype or comment Reference orplasmid source

PlasmidspAT153 tet bla, cloning vehicle (41)pIC100 pAT153 with P. mirabilis This study

cat in PstI sitepIC101 pAT153 with P. mirabilis This study

cat in PstI siteM13mp9 Cloning vehicle (42)pIC025 M13mp9 with pBR328 cat in This study

AccI site

P. mirabilisPM13 Wild type, tet+ cat+ (13)PM2 Wild type, tet+ cat- NCTC 3177 and

this studyIC100 PM2 containing pIC100 This studyIC1o0 PM2 containing pIC101 This study

E. coliSk3430 thi aroD6 leuC S. Kushner via

A. R. HawkinsIC500 Sk3430 containing pIC100 This studyIC501 Sk3430 containing pIC101 This studyJM103 A(lac pro) thi rpsL supE (27)

endA sbcB hsdR F'traD36proA lacI AlacZM15

IC025 JM103 containing pIC025 This study

protocol of Kushner (21). Plate matings were carried out as

described by Barth et al. (2).Southern blotting and hybridization. Restriction endonu-

clease fragments were separated by gel electrophoresis andtransferred to nitrocellulose filters by the method of South-ern (39). The consensus active site cat 20-mer oligonucleo-tide (5'-CCATCACAGACGGCATGATG-3') was labeledwith [-y-32P]ATP and polynucleotide kinase by the method ofMaxam and Gilbert (26). Hybridization was carried out as

previously described (6). Other DNA samples were nicktranslated (31) and hybridized under the conditions outlinedby Maniatis et al. (25).

Purification of DNA fragments. DNA fragments generatedby endonuclease digestion were purified by separationthrough low-melting-point agarose and extracted by thefreeze-squeeze method of Tautz and Renz (40).

Construction of recombinant plasmids. Plasmids werecleaved with restriction endonucleases under incubationconditions recommended by the vendors. 5'-Phosphategroups were removed by inclusion of calf intestinal phospha-tase (0.5 U .g-1). Digested plasmid DNA was incubatedwith T4 phage DNA ligase and a threefold molar excess ofinsert DNA in a buffer containing 50 mM Tris (pH 7.6), 10mM MgCI2, 1 mM ATP, and 100 p.g of bovine serum albuminper ml in a volume of 10 to 30 p.l (8). Other reactionscontaining one-half and twice the amount of insert DNAwere set up to optimize ligation conditions, and a linearizedplasmid control without insert DNA was included to checkthe effectiveness of the phosphatase treatment.Enzyme purification. The affinity chromatography purifi-

cation of CAT from P. mirabilis cells grown in the presenceof chloramphenicol has been described (45). CAT was puri-fied from cultures of chloramphenicol-sensitive P. mirabilisby the same method, but with a 10-fold larger cell mass.Enzyme assays. CAT was measured by spectrophotomet-

ric and radiometric methods (34). Protein concentration wasdetermined by the method of Lowry et al. (23). Inductionexperiments were carried out as described previously (43)with a range of inducer concentrations from 0.1 to 10 ,uM ofeither 3-deoxychloramphenicol or the 3-fluoro analog ofchloramphenicol.

Expression of chloramphenicol resistance. Cultures weregrown in Penassay broth at 37°C to the midexponentialphase, and approximately 106 cells were spread on Penassayagar plates containing 50 p.g of chloramphenicol per ml andincubated at 37°C for 48 h. The resulting colonies containingresistant cells were transferred to plates containing 100 p.g ofchloramphenicol per ml and thence stepwise to 200 and 500p.g of chloramphenicol per ml.The fall-off of chloramphenicol resistance in P. mirabilis

cultures grown in the absence of selective pressure (asdetermined by the drop in efficiency of plating on agarcontaining 50 p.g of chloramphenicol per ml and by theconcomitant and parallel decrease in CAT levels) was as-sessed by direct plating experiments and CAT enzymeassays. Studies of the underlying basis of the resistance

TABLE 2. Strains of Proteus, Morganella, and Providenciatested for presence of chloramphenicol resistance and CAT

synthesis

Species NCTC Selectable CATno. synthesisa

Proteus mirabilis 10374 +9559 +6396 +6197 -

3177 -60 -

Proteus vulgaris 10376 -

10015 -4175 -

Morganella morganii 10375 +10041 -7381 +5845 +2818 +2815 +1707 -235 -232 +

Providencia rettgeri 10377 C8893 +7481 +7480 -7479 -7477 -7476 +7475 +

Providencia species 10318 C10286 -8113 -8056 -6345 -2481 -

+ and - indicate the presence and absence, respectively, of selectablechloramphenicol resistance mediated by CAT. Strains marked (+) have thesame phenotype as P. mirabilis PM13. C indicates constitutive CAT produc-tion. These strains may harbor R plasmids conferring resistance to multipleantibiotics (see Results).

VOL. 164, 1985

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

116 CHARLES ET AL.

phenotype and estimates of the spontaneous mutation ratewere carried out by means of the Luria-Delbruck fluctuationtest (24). A cell suspension prepared from a single colony ofP. mirabilis PM13 was used to inoculate 10 1-ml tubes ofPenassay broth as well as a single 10-ml sample. Aftergrowth of all cultures to the midexponential phase of growth(approximately 107 cells per ml) the large culture was dividedinto 10 aliquots, and an equal volume (0.1 ml) of each of the20 samples was plated on Penassay agar plates containingchloramphenicol (50 ,ug ml-'). After growth at 37°C for 24 h,the number of colonies on each plate was counted, and theresults with samples from independent cultures were com-pared with the number of resistant cells detected in the largerculture volume (see Results).

Definitions. In this paper we refer to enhancement as thephenomenon whereby there is an increase in the level ofCAT produced in P. mirabilis cells that have been grown inthe presence of chloramphenicol as compared with cells thathave not. No mechanism is implied by this term. Thephenotype of cells of P. mirabilis which have been grown inthe presence of chloramphenicol is referred to as CAT',whereas that of P. mirabilis cultured in the absence ofchloramphenicol is described as CAT-. The latter do containCAT activity that can be detected by the radiometric methodand purified by affinity chromatography, but it is dramati-cally lower than that in CAT' cells (see Results).

Sources of materials. All restriction endonucleases, DNApolymerase I, T4 DNA kinase, and low-melting-pointagarose were purchased from BRL Ltd.; T4 DNA ligase andcalf-intestinal phosphatase were purchased from BoehringerMannheim Biochemicals; Fujimex RX X-ray film was fromHannimex UK Ltd.; sheets of nitrocellulose were fromSartorius; agarose was from Miles Laboratories, Inc.;Penassay broth and solid media were from Difco Laborato-ries; salmon sperm DNA, polyvinyl pyrollidone, bovineserum albumin, Ficoll 400, and all antibiotics were fromSigma Chemical Co. All other reagents were of analyticalgrade.

RESULTSCharacterization of the CAT in P. mirabilis PM13. The

CAT from strain PM13 was purified to homogeneity bothfrom cells exhibiting the resistant (CAT+) state and thesensitive (CAT-) state, and various properties and kineticparameters were compared. They were eluted in identicalfashion from highly substituted chloramphenicol base-agarose at 1.0 M NaCl in 5 mM chloramphenicol (45) andgave lines of identity with one another by two-dimensionalgel diffusion in agar against a goat antiserum to the type I(Escherichia coli) variant of CAT (45). Both preparationswere observed to migrate at the same rate in polyacrylamidegel electrophoresis under nondenaturing conditions (18). Theapparent Km values for chloramphenicol were 16 and 14 ,uMfor enzyme from CAT- and CAT' cells, and the values were73 + 5 and 82 + 6,uM for acetyl coenzyme A, respectively.Although the CAT purified from cells grown in the absenceof chloramphenicol is present at low levels, it is indistin-guishable from the CAT purified from cells grown in thepresence of chloramphenicol (50,ug ml-'). For cells grown in0, 50, 100, 200, and 500 ,ug of chloramphenicol, the CATspecific activity was <1, 124, 150, 134, and 118 nmol min-1mg-1. (Extracts from cells grown in the absence of chloram-phenicol contained CAT which could be detected by theradiometric assay [34] and purified by affinity chromatogra-phy [45], but which was below the sensitivity of the spec-trophotometric assay [1 nmol-1 min-' mg-'] used in these

experiments.) The results support the view that the resist-ance phenotype is likely to be due to an increase in theamount of enzyme produced rather than to any alteration inthe properties of the protein or the synthesis of a secondCAT variant.Enhancement of CAT synthesis in P. mirabiis and origin of

the resistance phenotype through mutation. Experiments car-ried out under conditions which induce CAT synthesis inStaphylococcus aureus (43) and Bacillus pumilus (15) pro-duced no effect on the CAT levels in P. mirabilis PM13. Theincubation of resting cell suspensions with chloramphenicol,3-deoxychloramphenicol, or 3-fluorochloramphenicol wasnot accompanied by significant increases in the specificactivity of CAT (data not shown). Such results, althoughnegative, strongly suggest that induction by chloramphenicolor its analogs is unlikely to play a role in the enhancement ofchloramphenicol resistance observed in Proteus spp.The application of the Luria-Delbruck fluctuation test to

the efficiency of plating of chloramphenicol-resistant cellsrevealed a non-Gaussian variation among samples takenfrom independent cultures as compared with equal samplesfrom a single large culture. Since a mutation to chloramphen-icol resistance can occur in any generation before exposureto the selective agent, the large variance observed favors thehypothesis of a mutational event rather than induction ofCAT by exposure of cells to the antibiotic. In a typicalexperiment (see Materials and Methods) comparing 10 smallcultures with a single 10-fold larger culture volume theresults were as follows. Six of the small cultures yielded nochloramphenicol-resistant cells of P. mirabilis PM13 in sam-ples containing approximately 107 cells per ml, whereas theother four cultures yielded 5, 12, 61, and 267 cells resistant tochloramphenicol from comparable populations. By way ofcontrast, 10 equivalent samples from the larger culture eachyielded from 167 to 233 resistant cells (mean 207 ± 14).

Single step mutants resistant to 50 pug of chloramphenicolper ml after growth for 48 h (see Materials and Methods)were transferred sequentially to plates containing higherconcentrations of chloramphenicol (100, 200, and 500 ,ugml-1). Individual colonies were taken from each plate andgrown for transfer to liquid media containing the sameconcentration of chloramphenicol used in the selection plate,and cells were harvested at the midexponential phase ofgrowth. The specific activity of CAT in crude cell extractswas below the limits of detection by the routine spectropho-tometric assay in extracts prepared from bacteria grown inthe absence of chloramphenicol (Table 3), whereas thespecific activity of CAT from cells grown in the presence ofchloramphenicol was high and essentially independent of theconcentration of antibiotic in the growth medium. Themutation to the CAT' state, (i.e., high efficiency of platingon chloramphenicol agar) seems to be of an all-or-nothingnature in that all cells which will grow on 50,g ml-' will alsogrow on 500 ,ug ml-'. Once selected, CAT' cells aremaintained in the population so long as there is selectivepressure favoring their persistence. Figure 1 epitomizes theresults obtained with P. mirabilis PM13 (as well as otherstrains of Proteus spp. and related genera) when CAT' cellsare grown in the absence of the antibiotic. The exponentialdecline with time of the efficiency of plating of cells on 50 ,ugof chloramphenicol per ml and the parallel fall in the specificactivity of CAT for the population as a whole constitute thephenomenon which is analyzed in greater detail below (seeDiscussion).

P. mirabilis PM13 was tested for enhanced levels ofresistance to other antibiotics after selection on 50 jxg of

J. BACTERIOL.

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

CHROMOSOMAL cat OF PROTEUS MIRABILIS 117

0 30 60 90 120GENERATIONS

FIG. 1. Disappearance of chloramphenicol-resistant cells of P.mirabilis PM13 from a uniformly resistant population as a functionof time. Cells were collected by centrifugation after growth inPenassay broth containing chloramphenicol (50 ,ug ml-') and, afterwashing, were suspended in antibiotic-free medium. Samples takenat appropriate intervals were divided and taken for simultaneousplating on Penassay agar containing 50 ,ug of chloramphenicol per mland on plates lacking the antibiotic and for lysis before measurementof the specific activity of CAT. The experimental points giving line(a) represent the efficiency of plating (E.O.P.) of cells on chloram-phenicol, and line (b) depicts the parallel and exponential fall in CATfor the population. Growth in the absence of chloramphenicol was

followed for 120 generations by means of repeated dilutions (10-2) oflate-exponential-phase cells into fresh medium. The shaded area

represents an envelope containing a range of theoretical curves forthe fall in efficiency of plating as a function of growth, such plotsbeing generated by solutions to the equation (see Discussion) whichrelates the efficiency of plating to parameters for both selection andmutation. For the case wherein the disappearance of resistant cellsfrom the population is solely due to selection against their persis-tence, the initial rate of reduction in the efficiency of plating shouldbe low. For the other extreme, wherein the rate of back-mutation to

sensitivity is the sole determinant of disappearance of resistant cells,an immediate and exponential fall in the efficiency of plating ispredicted. The position of the experimental points between the twoextremes described by the boundaries of the shaded area suggeststhat both back-mutation to sensitivity and selection against resistantcells are likely to play a role.

chloramphenicol per ml but remained uniformly sensitive toampicillin, kanamycin, nalidixic acid, spectinomycin, andstreptomycin (data not shown).How widespread is the phenomenon? Additional represen-

tatives of Proteus spp., Morganella spp., and Providenciaspp. were screened for the occurrence of the enhancement

phenomenon described above, i.e., the appearance of chlor-amphenicol-resistant cells arising from a predominantlychloramphenicol-sensitive population at high frequency (ap-proximately 10' per cell per generation). Thirteen of 32strains from the National Collection of Type Cultures(United Kingdom) showed the phenomenon (Table 2),whereas only two produced CAT at high levels without priorexposure to chloramphenicol. The latter were not studiedfurther because they displayed resistance to multiple antibi-otics, suggesting the presence of plasmid-mediated resist-ance (data not shown).

Absence of a plasmid in P. mirabilis PM13. P. mirabilisPM13 was screened for the presence of plasmids by twomethods (7, 20), neither of which yielded evidence of plas-mids when the DNA from cell lysates was subjected toagarose gel electrophoresis. Positive controls in such exper-iments included E. coli and P. mirabilis strains harboringwell-characterized plasmids (data not shown). Attempts totransfer chloramphenicol resistance from strain PM13 tosuitable E. coli recipients by conjugation were also unsuc-cessful. Subsequent experiments (see below) were consis-tent with the view that the cat gene is a chromosomal one instrain PM13.

Cloning of the P. mirabilis cat gene from the CAT' andCAT- states. Preliminary studies revealed that the CAT fromP. mirabilis was likely to be similar to the type I variant instructural and kinetic properties (45). Also, the P. mirabilisCAT bound the dye crystal violet (I. G. Charles, unpub-lished experiments), a property demonstrated by the type Ienzyme (30) but by none of the other CAT variants found ingram-negative bacteria (A. D. Bennett, Ph.D. thesis, Uni-versity of Leicester, 1984). The likely high degree of homol-ogy of the CAT of P. mirabilis with its type I counterpartwas exploited in the strategy used to clone the cat gene of P.mirabilis.

Plasmid pBR328 (38) contains a type I cat gene on a773-base-pair TaqI fragment. To facilitate the isolation ofthis fragment for hybridization experiments, the entire 773-base-pair fragment was cloned into phage M13 mp9 (42) asoutlined in Materials and Methods. One recombinant(pIC025) was isolated as the replicative form, and the entirestructural gene for cat was isolated by a double digest withPstI and BamHI.

Identification and isolation of the DNA fragments of the P.mirabilis chromosome carrying the cat gene was achieved byfirst using the type I cat gene (isolated from pIC025) to probevarious restriction endonuclease digests of chromosomalDNA from P. mirabilis. A PstI digest yielded a single bandon hybridization (data not shown), indicating that this frag-ment was likely to contain the entire P. mirabilis cat gene.PstI-digested chromosomal DNA samples (prepared from P.mirabilis in both the CAT' and CAT- states) were sepa-rately ligated with pAT153 which had been cut by PstI andtreated with alkaline phosphatase. After transformation, theCAT-containing cells, which were likely to harbor recombi-nant plasmids (selected on 10 ,ug of chloramphenicol per ml),were then screened (7) for insert DNA with positive resultsin each case for an insert of approximately 8.5 kilobases(kb). Plasmids from one each of the resulting transformants(designated pIC100 for the recombinant produced from DNAin the CAT' state and pIC101 for the recombinant producedfrom DNA in the CAT- state) were used for large-scaleplasmid preparations.

Identification and localization of the cat genes in plasmidspIC100 and pIC101. The synthetic 20-mer cat active siteconsensus probe was used to locate the cat gene within the

VOL. 164, 1985

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

118 CHARLES ET AL.

8.5-kb PstI fragment of plasmids pIC100 and pIC101. Hy-bridization of the 32P-labeled probe to separated restrictionfragments yielded results (Fig. 2) which were interpreted asevidence for the localization of the cat gene on a 500-base-pair fragment located near one end of the insert (Fig. 3). Thedirection of transcription of the gene was inferred from thishybridization data in relation to the restriction map of thetype I cat gene.Chromosomal organization of the P. mirabilis cat gene. P.

mirabilis can amplify cat genes carried on plasmids such asNR1, and the phenomenon of chromosomal cat gene ampli-fication has accounted for increased chloramphenicol resist-ance in E. coli and B. subtilis (28, 44). The P. mirabilisgenome was examined by Southern blotting for selectivegene amplification. A Clal-to-PstIl fragment of pIC100, en-compassing the entire cat gene, was isolated after electro-phoresis through low-melting-point agarose and labeled with[32P]dATP by nick translation. P. mirabilis chromosomalDNA samples prepared from late-exponential-phase culturesgrown in 0, 50, 100, 200, and 500 ,ug of chloramphenicol perml were each digested with PstI and, after electrophoresis,were probed with the labeled fragment. No significant dif-ference in hybridization signal intensity was detected in therestriction fragments arising from DNA prepared from cellsthat have been grown in chloramphenicol as comparedwhich those that had not (Fig. 4). Southern blotting was alsoused to detect a possible change in orientation of cat DNAwithin the chromosome of the CAT' state compared withthe CAT- state, at least within the 8.5 kb of DNA spanning

0 1 2 3 4 5 6 7 S 9 10

0.006 S S 0| § |kb

0 a c *e,o 0_o

DIRECTION OFTRANSCRIPTION

FIG. 3. Restriction map of pIC100 (and pIC101) with restrictionendonucleases ClaI, EcoRI, HindIlI, PstI, and Sall. The black boxbounded by sites for EcoRI and Clal represents the smallestrestriction fragment detected by hybridization to the consensusactive site oligonucleotide (see Fig. 2). The arrow indicates thedirection of transcription as inferred from the location of the EcoRIsite in the gene for the type I cat (1) and from subsequent nucleotidesequence analysis (10).

the cloned fragment. P. mirabilis chromosomal DNA wasprepared from late-exponential-phase cultures grown in ei-ther no antibiotic or 500 jig of chloramphenicol per ml. Bothpreparations were separately digested with PstI, HindIll,and EcoRI, subjected to agarose (0.7%) gel electrophoresis,and probed with the entire 8.5-kb 32P-labeled insert of

(b)size (a)in kbin

23.

sizein kb

23.0-

9.7-6.4-

4.5-

2.3-2.0-

0.7-

(a) (b) 9.76.4

4.5

2.3-2.0-

*v -. n0_ _ w

B C D E F G0.7-

0

_

*W

A B C D E F G ABC D E F G

FIG. 2. Agarose gel (0.7%) showing endonuclease digest patternof pIC100 after ethidium bromide staining. (a) Lanes after digestionwith the following enzymes: A, ClaI; B, ClaI and PstI; C, ClaI, PstI,and PvuII; D, ClaI, PstI, and HindIll; E, ClaI, PstI, and EcoRI; F,ClaI, PstI, and Sall; G, Clal, PstI, EcoRI, and Sall. (b) Autoradi-ograph of the DNA shown in part (a) after transfer to nitrocelluloseand hybridization with a y-32P-labeled consensus cat active siteoligonucleotide (See Fig. 3 and Materials and Methods).

FIG. 4. Agarose (0.7%) gel electrophoresis (a) of PstI digests of(A) pIC100 at high concentration for visualization with ethidiumbromide and (B) pIC100 at a concentration estimated to correspondto approximately one copy of cat per chromosome for comparisonwith subsequent (C through G) loadings of chromosomal DNA.Lanes C through F were loaded with Pstl digests of chromosomalDNA from P. mirabilis PM13 cells grown in the presence of:hloramphenicol at (C) 500, (D) 200, (E) 100, and (F) 50 ,ug ml-'.u-ane G was a PstI digest of chromosomal DNA from strain PM13grown in the absence of chloramphenicol. (b) Autoradiograph of theDNA in (a) after transfer to nitrocellulose by the method of Southern(39) and hybridization with a 32P-labeled cat probe comprising the1.7-kb PstI-ClaI fragment of pIC100 (see Materials and Methods andFig. 3). Efforts were made to load equal amounts of chromosomalDNA in tracks C through G. The higher fraction of high-molecular-weight DNA fragments in lanes C, D, and E and the weakly positivediscrete hybridization signals (marked with arrows) suggest that thecorresponding PstI digests may have been incomplete. Scanning ofthe autoradiograph with integration and summation of the major andminor peaks yielded the intensities of hybridization for comparisonwith the notional value of unity for the pIC100 control. The relativeintensities for lanes C, D, E, F, and G in relation to that of lane Bwere 2.5, 2.4, 2.6, 3.0, and 3.1, respectively.

J. BACTERIOL.

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

CHROMOSOMAL cat OF PROTEUS MIRABILIS 119

sizein kb

23.0-

9.7~~9 .

0.7-

A B C D E F G H B C D E F G H(a) (b)

FIG. 5. Use of the cloned fragment containing the cat of P. mirabilis to probe chromosomal DNA from cells of strain PM13 grown in thepresence (+) or absence (-) of chloramphenicol (500 p.g ml-') as indicated. (a) Agarose (0.7%) gel electrophoresis showing digests ofchromosomal DNA with EcoRI (C and D), HindIlI (E and F), and PstI (G and H). (b) Autoradiograph of the DNA shown in (a) after transferto nitrocellulose and hybridization with a 32P-labeled probe comprising the PstI fragment from pIC100 (see Fig. 3 and Materials and Methods).Lanes A and B contain PstI digests of pIC100. Only the lower concentration (B) was used for the hybridization control in (b), wherein anintense signal at approximately 8.5 kb may be seen to correspond with that in lanes G and H for the PstI digest of chromosomal DNA.

pIC100. Figure 5 shows that the pattern of hybridizationproduced for each pair of digests is indistinguishable, indi-cating that no detectable rearrangement has occurred withinthis 8.5-kb cat-containing fragment.

Expression of the P. mirabilis cat gene in E. coli and in a cat-strain of P. mirabilis. Plasmids pIC100 and pIC101 were bothtransferred into E. coli and into P. mirabilis PM2, the latterstrain having previously been demonstrated to be negativefor CAT by a sensitive radiometric assay (34) and to lackhomologous cat DNA by Southern blot experiments (datanot shown). The resulting transformants were then tested forenhanced expression of cat as observed in strain PM13(Table 3). Expression of the P. mirabilis cat gene wassufficiently poor in E. coli that the specific activity of theenzyme could not be measured by the spectrophotometric

TABLE 3. Synthesis of CAT from the cloned cat gene of P.mirabilis PM13 in chloramphenicol-sensitive (cat-) bacteria

Sp act of CAT (nmol min-' mg-')Bacterium Plasmid

No chloramphenicol Chloramphenicola

P. mirabilis PM2 None 0P. mirabilis PM2 pIC100 85 87P. mirabilis PM2 pIC101 88 85

E. coli Sk3430 None 0E. coli Sk3430 pIC100 <1E. coli Sk3430 pIC101 <1

a CAT specific activity by the spectrophotometric method in extracts ofcells grown in the absence or presence of chloramphenicol (10 pLg ml-'). Thezero value for P. mirabilis PM2 and for E. coli Sk3430 indicates that no CATwas detected by the more sensitive radiometric method. Although thepresence of pIC100 or pIC101 in E. coli promoted CAT synthesis as detectedby the latter method (Fig. 6), it was below the level of detection byspectrophotometry. Plasmid-free P. mirabilis PM2 and all three strains of E.coli failed to grow in 10 pug of chloramphenicol per ml.

method and required use of the radiometric technique (Fig.6). The minimum inhibitory concentration for chloramphen-icol with E. cQli Sk3430 carrying pIC100 or pIC101 is only 10p.g ml-', whereas the corresponding concentration for strainPM2 with either plasmid is 100 jig ml-' as compared with 10,ug rml-n without pIC100 or pIC101. The constitutive expres-sion of the cat gene from strain PM13 in the geneticbackground of different P. mirabilis strains is addressed inthe Discussion.

DISCUSSIONThe results presented here show that a number of strains

of the Proteus-Provideqtcia group are capable of expressinghigh levels of chloramphenicol resistance under appropriateconditions.The mechanism of the phenomenon appears to involve the

selection of mutants in the population. Preexisting mutantcells (which occur spontaneously at a frequency of approx-imately 10-5 per cell per generation) are selected by growthon chloramphenicol and are maintained in the population foras long as they are exposed to chloramphenicol selection.Cultures of P. mirabilis in which 100% of the cells arephenotypically resistant to chloramphenicol undergo a pop-ulation reversion to the chloramphenicol-sensitive state dur-ing growth in the absence of the antibiotic, the rate of thisprocess being of the order of 10-2 per cell per generation.The rate of decrease in the efficiency of plating of cells onantibiotic agar (Fig.1) is too low to correspond with a modelwherein a putative intracellular inducer is progressivelydiluted by cel; division. An alternative explanation, theprogressive reduction in the proportion of resistant cells inthe population, may be seen to operate by mutation orselection or some combination of both factors. Back-mutation to the sensitive phenotype at a very high rate wouldyield a concomitant reduction in the efficiency of plating ofthe population on antibiotic-containing media. Such a model

VOL. 164, 1985

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

120 CHARLES ET AL.

predicts that in any interval bt, a proportion of cells ,u8t(where ,u is the back mutation rate per generation) wouldrevert to sensitivity. If q is the frequency of resistant cells ina population after t generations, then the integrated expres-sion describing the phenomenon is

qt = qo e->' (1)where qo is the initial frequency of resistant cells in thepopulation and q, is their frequency after t generations. Aplot of ln(q,) or ln(efficiency of plating) against time shouldyield a straight line, its slope being equal to the back-mutation rate.A second alternative is selection against resistant cells in

the absence of the antibiotic. If s is the selective coefficientagainst resistant cells, then a proportion of the cells (sbt) isselectively eliminated in bt generations, then the change in qin St generations is

q = -sbt q(l - q) (2)

and, when integrated, yields the equation for the proportionof resistant cells remaining after t generations

=q qo (3)(1-qo) est + qo

Equation 3 describes a linear plot of ln(efficiency of plating)against time after an initial slow reduction in the efficiency ofplating, the duration of such a lag being dependent upon howclose qo is to 1. The plot becomes linear when nearly half ofthe cells in the population are sensitive, after which ln(ef-ficiency of plating) decreases with a slope of -s. Thecompound expression corresponding to equations 1 and 3 inthe case wherein both selection and back mutation areinvolved is

qt =[. q0(p. + s) (4)[tL + s(1 - qO)Ie(L+s)t + sqo

The latter reduces to produce equations 1 and 3 as specialcases when s and ,u, respectively, are zero.

Irrespective of the relative contributions of mutation andselection, as t becomes large, the slope of ln(efficiency ofplating) against t becomes -(,u + s). The amount of selec-tion, however, determines the duration of the lag before theslope becomes linear.The shaded region in Fig. 1 describes the graphical limits

of predictions from equation 4 when (,u + s) is equal to theexperimentally derived slope of -0.044. A comparison of thepossible range of curves calculated from theory for differentvalues of ,u and s with that actually determined experimen-tally suggests that mutation and selection are likely to beoperating (calculations not shown). In addition to such directevidence bearing on the phenomenon, there are reasons forbelieving that several conventional explanations are un-likely.Although there is a superficial resemblance between the

induction phenomenon observed for cat expression in gram-positive bacteria and the phenomenon observed with P.mirabilis, they differ in important respects. First, nondivid-ing P. mirabilis cells cannot be induced by chloramphenicolor congeners known to induce cat in staphylococci (e.g., the3-deoxy and 3-fluoro analogs). Second, the Luria-Delbruckfluctuation test produced marked variation in the frequencywith which resistant colonies appeared from cells taken fromindependent cultures, indicative of a mutation event ratherthan a response to environmental change. Third, the decay

1 2 3 4 5FIG. 6. Radiometric demonstration of CAT activity in crude cell

extracts from E. coli Sk3430 harboring pIC100 (lane 1), pIC101 (lane2), or pBR328 (lane 5). Lanes 3 and 4 contain ntgative controlswhich correspond, respectively, to extracts trom E. coli Sk3430without a plasmid and Methylophilus methylotrophus, an obligatemethylotroph which has been shown previously to lack both CATactivity and a cat gene detectable by hybridization (6). The assayinvolves ascending chromatography on silica gel in chloroform-methanol (85:15, vol/vol) of the products of acetylation of[14C]chloramphenicol by CAT in the presence of acetyl coenzymeA. The assay is quantitative only when CAT is limiting and bothsubstrates are present in excess (34). In this instance the prepon-derance of 1,3-diacetylchloramphenicol (top arrow) suggests thepresence of a large excess of CAT for each of the positive extracts.The middle arrow indicates the relative mobility of 3-acetyl-chloramphenicol, and the bottom arrow indicates the position ofunmodified chloramphenicol.

in the efficiency of plating with time is far too slow torepresent the dilution by cell division of a conventionalinducer or gene product present only at time zero.An apparent precedent for the proposed reversible muta-

tion event accounting for increased resistance in Proteusspp. is the so called transitioning phenomenon (reviewed byRownd [32]). The level of chloramphenicol resistance insuch a case is mediated by changes in cat gene dosage via avariation in plasmid copy number, the extent of tandemduplication of plasmid-borne genes, or by a combinationof both mechanisms. Two observations argue againsttransitioning as the explanation for the reversible enhance-ment of cat expression in P. mirabilis. First, in the case ofplasmid transitioning all cells appear to express cat consti-tutively and exhibit a high efficiency of plating on mediacontaining chloramphenicol before an increase in gene dos-age occurs. Second, and notwithstanding the limitations ofnegative experiments, all efforts to detect a plasmid in strainPM13 have been unsuccessful. Furthermore, the possibilityof cat amplification by tandem duplication, within thegenome of P. mirabilis was investigated by Southern hybrid-ization with negative results. No significant difference wasobserved in the extent of hybridization between the labeledP. mirabilis cat probe and its counterpart in restrictionendonuclease digests of total DNA prepared from a popula-tion of resistant cells as compared with DNA from cellsgrown in the absence of chlorampheriicol.

Recently it has become clear that gene expression inprocaryotes may be modulated by specific and reversibleDNA inversions, the best known example being phasevariation in Salmonella spp. (36). One of the consequencesof such inversions is that the restriction patterns of the DNA

J. BACTERIOL.

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

CHROMOSOMAL cat OF PROTEUS MIRABILIS 121

from the alternate states are different. Southern blottingprocedures designed to detect rearrangements within cat or

its flanking sequences in P. mirabilis PM13 failed to identifyany differences when DNA prepared from cells in the CAT'state was compared with DNA from CAT- cells. There iscircumstantial evidence, on the other hand, that the site ofaction of the mutation may be outside of the cat gene and 5'control regions. The efficiency of plating experiments showthat for P. mirabilis in the CAT' state virtually 100% of thecells are expressing the cat gene at a high rate; thus, thehighly transcribed gene must be represented at least once pergenome. Furthermore, the Clarke and Carbon equation (12)predicts that not more than 104 recombinants need bescreened to isolate such a genp in a cloning experiment. Inthe present study the cat gene was cloned from P. mirabilisDNA isolated both CAT' cells (pIC100) and from CAT-cells (pIC101). The frequency of isolation of such clones was

independent of the origin of the cells and was within thelimits of the Clarke-Carbon prediction, and the P. mirabilisDNA inserts from CAT' and CAT- cells were observed tobe identical by restriction endonuclease digestion. Takentogether, the results favor the view that the site of themutation which underlies the enhancement of chloramphen-icol resistance and cat expression (and their reversion) islikely to be outside the cat gene and its flanking sequences as

represented in pIC100 and pIC101. Expression studies em-

ploying these plasmids in backgrounds which lack cat (E.coli and P. mirabilis PM2) are consistent with the notion thatthe control of cat expression in P. mirabilis PM13 may resideoutside the cloned segment. Whereas the absence of high-level expression of the P. mirabilis cat in E. coli might berelated to the general problem of the anomalous expressionwhich has been observed reciprocally between genes in E.coli and P. mirabilis (3, 4), there are examples of P. mirabilisgenes being expressed at normal levels in E. coli back-grounds (5, 16, 17). Furthermore, it is more difficult to

account for low-level expression of cat in PM2, since theapparent single copy of the chromosomal cat gene in strainPM13 yields resistance to chloramphenicol concentrationsas high as 500 ,ug ml-'. The data available are compatiblewith the hypothesis that a host specific element, present inPM13 but not in PM2, is responsible for promoting thehigh-level cat expression which is characteristic of theCAT' state.Taken together, the results favor a mechanism wherein a

reversible mutation event occurs at high frequency. Suchmutations are characteristic of transposition or geneticswitch mechanisms (29). Although the data for the P.mirabilis cat system do not support a rearrangement of a

DNA segment within an 8.5-kb region which includes the 5'control region, such a switching event could occur at a more

distant locus as is the case in phase variation in S.typhimurium. The important predictions of such a model forthe CAT system in P. mirabilis would be (i) the absence of a

difference between the structural gene for cat in the CAT'and CAT- states, (ii) evidence of transcriptional control viaa trans-acting element (protein or RNA); and (iii) the directdemonstration (by mutation or its isolation and characteriza-tion) of a control locus which is external to the P. mirabilisDNA in pIC100 and pIC101, and (iv) that the activity of theputative switch region shows positional dependence.

ACKNOWLEDGMENTS

The CAT phenomenon in P. mirabilis has been a puzzle for morethan a decade. W.V.S. thanks J. N. Coetzee, N. Datta, T. J. Foster,and R. W. Hedges for their advice and for their collaboration in

preliminary (unpublished) studies which helped to shape our view ofthe problem. We are grateful to Dorothy Jones for access to hercollection of bacterial strains and for guidance in taxonomicquestions.

This work was supported by grants from the Medical ResearchCouncil and the Science and Engineering Research Council of theUnited Kingdom.

LITERATURE CITED1. Alton, N. K., and D. Vapnek. 1979. Nucleotide sequence anal-

ysis of the chloramphenicol resistance transposon Tn9. Nature(London) 282:864-869.

2. Barth, P. T., L. Tobin, and G. Sharp. 1981. Development ofbroad host host-range plasmid vectors, p. 439-448. In S. B.Levy, R. C. Clowes, and E. L. Koenig (ed.), Molecular biology,pathogenicity and ecology of bacterial plasmids. Plenum Pub-lishing Corp., New York.

3. Baumberg, S., G. Cornelis, M. Panagiotakopoulos, and M.Roberts. 1980. Expression of the lactose transposon Tn951 inEscherichia coli, Proteus, and Pseudomonas. J. Gen. Micro-biol. 119:257-262.

4. Baumberg, S., and S. Dennison. 1975. Variation in expression ofsex factor genes in the Proteus-Providencia group relative toEscherichia coli. J. Bacteriol. 123:278-286.

5. Beck, Y., and W. F. Coetzee. 1983. Molecular cloning of Proteusmorganii phenylalanine deaminase gene in Escherichia coli.Gene 21:25-31.

6. Beschle, H. G., I. G. Charles, and W. V. Shaw. 1984. Use of asynthetic oligonucleotide to identify a chromosomal gene forchloramphenicol acetyltransferase in a plasmid-bearing Flavo-bacterium. J. Gen. Microbiol. 130:3335-3338.

7. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extractionprocedure for screening recombinant plasmid DNA. NucleicAcids Res. 7:1513-1523.

8. Burns, D. M., and I. R. Beacham. 1983. A method for theligation of DNA following isolation from low melting tempera-ture agarose. Anal. Biochem. 135:48-51.

9. Byeon, W. H., and B. Weisblum. 1984. Post-transcriptionalregulation of chloramphenicol acetyl transferase. J. Bacteriol.158:543-550.

10. Charles, I. G., J. W. Keyte, and W. V. Shaw. 1985. Nucleotidesequence analysis of the cat gene of Proteus mirabilis: compar-ison with the type I (Tn9) cat gene. J. Bacteriol. 164:123-129.

11. Chow, L. T., R. Kahmann, and D. Kamp. 1977. Electronmicroscopic characterization of DNA's of non-defective dele-tion mutants of bacteriophage Mu. J. Mol. Biol. 113:591-609.

12. Clarke, L., and J. Carbon. 1976. A colony bank containingsynthetic ColEl hybrid plasmids representative of the entire E.coli genome. Cell 9:91-99.

13. Coetzee, J. N., and T. G. Sacks. 1960. Transduction of strepto-mycin resistance in Proteus mirabilis. J. Gen. Microbiol. 23:445-455.

14. De Crombrugghe, B., I. Pastan, W. V. Shaw, and J. L. Rosner.1973. Stimulation by cyclic AMP, and ppGpp of chloramphen-icol acetyl transferase synthesis. Nature (London) 241:237-239.

15. Duvall, E. J., D. M. Williams, P. S. Lovett, C. Rudolph, N.Vasantha, and M. Guyer. 1983. Chloramphenicol-inducible geneexpression in Bacillus subtilis. Gene 24:171-177.

16. Eitner, G., B. Adler, V. A. Lanzov, and J. Hofemeister. 1982.Interspecies recA protein substitution in Escherichia coli andProteus mirabilis. Mol. Gen. Genet. 185:481-486.

17. Eitner, G., A. S. Solonin, and V. I. Tanyashin. 1981. Cloning ofa recA-like gene of Proteus mirabilis. Gene 14:301-308.

18. Fitton, J. E., and W. V. Shaw. 1979. Comparison of chloram-phenicol acetyltransferase variants in Staphylococci. Purifica-tion, inhibitor studies and N-terminal sequences. Biochem. J.177:575-582.

19. Hawkins, A. R., A. J. F. Da Silva, and C. F. Roberts. 1985.Cloning and characterization of the three enzyme structuralgenes QUT B, QUT C, and QUT E from the quinic acidutilization gene cluster in Aspergillus nidulans. Curr. Genet.9:305-311.

VOL. 164, 1985

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

122 CHARLES ET AL.

20. Kado, C. I., and S. T. Liu. 1981. Rapid procedure for detectionand isolation of large and small plasmids. J. Bacteriol.145:1365-1373.

21. Kushner, S. k. 1978. An improved method for transformation ofEscherichia coli with ColEl-derived plasmids, p. 17-23. InH. B. Boyer and S. Nicosia (ed.), Genetic engineering.Elsevier/North-Holland Publishing Co., Amsterdam.

22. Le Grice, S. F. J,, and H. Matzura. 1981. Binding of RNApolymerase and the catabolite gene activator protein within thecat promoter in Escherichia coli. J. Mol. Biol. 150:185-196.

23. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J.Biol. Chem. 193:265-275.

24. Luria, S. E., and M. Delbruck. 1943. Mutations of bacteria fromvirus sensitivity to virus resistance. Genetics 28:491-511.

25. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning. A laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

26. Maxam, A. M., and W. Gilbert. 1980. Sequencing end labelledDNA with base specific chemical cleavages. Methods Enzymol.65:499-560.

27. Messing, J., R. Crea, and P. H. Seeburg. 1981. A system forshotgun DNA sequencing. Nucleic Acids Res. 9:309-321.

28. Meyer, J., and S. Iida. 1979. Amplification of chloramphenicolresistance transposons carried by phage PlCm in Escherichiacoli. Mol. Gen. Genet. 176:209-219.

29. Plasterk, R. H. A., and P. Van de Putte. 1984. Genetic switchesby DNA inversions in prokaryotes. Biochim. Biophys. Acta782:111-119.

30. Proctor, G. N., and R. H. Rownd. 1982. Rosanilins: indicatordyes for chloramphenicol-resistant enterobacteria containingchloramphenicol acetyltransferase. J. Bacteriol. 150:1375-1382.

31. Rigby, P. W. J., M. Dieckmann, C. Rhodes, and P. Berg. 1977.Labeling deoxyribonucleic acid to a high specific activity invitro by nick-translation with DNA polymerase I. J. Mol. Biol.113:237-251.

32. Rownd, R. H. 1982. R plasmid drug resistance gene amplifica-tion in bacteria, p. 273-279. In R. T. Schimke (ed.) Geneamplification. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.

33. Rownd, R. H., and S. Mickel. 1971. Dissociation and reas-sociation of RTF and r-determinants of the R factor NR1 inProteus mirabilis. Nature (London) New Biol. 234:40-43.

34. Shaw, W. V. 1975. Chloramphenicol acetyltransferase fromchloramphenicol resistant bacteria. Methods Enzymol. 43:737-755.

35. Shaw, W. V. 1983. Chloramphenicol acetyltransferase: enzy-mology and molecular biology. Crit. Rev. Biochem. 14:1-46.

36. Silverman, M., and M. Simon. 1983. Phase variation and relatedsystems, p. 537-557. In J. A. Shapiro (ed.) Mobile geneticelements. Academic Press, Inc., New York.

37. Smith, D. H., J. H. Harwood, and F. A. Rubin. 1972. Studies onthe regulation of the chloramphenicol acetyltransferase medi-ated by R factors, p. 319-335. In V. Krcmery, L. Rosival, andT. Watanabe (ed.), Bacterial plasmids and antibiotic resistance.Springer-Verlag, New York.

38. Soberon, X., L. Covarrubias, and F. Bolivar. 1980. Constructionand characterization of new cloning vehicles. Deletion deriva-tives of pBR322 and pBR325. Gene 9:287-305.

39. Southern, E. 1975. Detection of specific sequences among DNAfragments separated by gel electrophoresis. J. Mol. Biol. 98:503-517.

40. Tautz, D., and M. Renz. 1983. An optimized freeze-squeezemethod for the recovery of DNA fragments from agarose gels.Anal. Biochem. 132:14-19.

41. Twigg, A. J., and D. Sherratt. 1980. Trans-coniplementablecopy-number mutants of plasmid ColEl. Nature (London) 283:216-218.

42. Vieira, J., and J. Messing. 1982. The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencingwith synthetic universal primers. Gene 19:259-268.

43. Winshell, E., and W. V. Shaw. 1969. Kinetics of induction andpurification of chloramphenicol acetyltransferase from chloram-phenicol-resistant Staphylococcus aureus. J. Bacteriol. 98:1248-1257.

44. Young, M. 1984. Gene amplification in Bacillus subtilis. J. Gen.Microbiol. 130:1613-1621.

45. Zaidenzaig, Y., J. E. Fitton, L. C. Packman, and W. V. Shaw.1979. Characterization and comparison of chloramphenicol ace-tyltransferase variants. Eur. J. Biochem. 100:609-618.

J. BACTERIOL.

on August 8, 2019 by guest

http://jb.asm.org/

Dow

nloaded from