JOURNAL OF BACTERIOLOGY, Mar., 1965Copyright 0 1965 American Society for Microbiology

Vol. 89, No. 3Printed in U.S.A.

Mutations to Penicillin Resistance in theEnterobacteriaceae That Affect Sensitivityto Serum and Virulence for the Mouse1

ROBERT J. ROANTREE AND JOHN P. STEWARDDepartment of Medical Microbiology and the Max C. Fleischmann Laboratories of the Medical Sciences,

Stanford University School of Medicine, Stanford, California

Received for publication 7 October 1964

ABSTRACTROANTREE, ROBERT J. (Stanford University School of Medicine, Stanford, Calif.),

AND JOHN P. STEWARD. Mutations to penicillin resistance in the Enterobacteriaceaethat affect sensitivity to serum and virulence for the mouse. J. Bacteriol 89:630-639.1965.-Series of mutants resistant to benzylpenicillin or a-aminobenzylpenicillin werederived from serum-resistant strains of Escherichia coli and Salmonella by the gradient-plate technique. Serum-sensitive mutants were detected in series derived from 16 of the19 strains used, and these retained the parental 0 type. Most series were characterizedby a mutational step to a high degree of sensitivity to serum. Penicillin-resistant mu-tants of virulent S. typhimurium and S. enteritidis were less virulent than the parentstrains; those which were very sensitive to serum usually showed the greatest loss ofvirulence. One class of mutants from S. enteritidis was sensitive to human serum butvirulent for mice. We found that the mice lack bactericidal antibody against thisstrain and that immunization with it leads to a high degree of protection.

Strains of gram-negative enteric bacilli varygreatly in their sensitivity to the complement-dependent bactericidal system of normal mam-malian serum (Mackie and Finkelstein, 1931;Roantree and Rantz, 1960). However, for anyone strain, this quality is remarkably constant,and we have had no success in deriving serum-resistant strains from "naturally" sensitivestrains by use of active serum as the selectiveagent. Michael and Braun (1958) showed thatsensitive strains might be derived from resistantones. They isolated a series of penicillin-resistantmutants from strains of Shigella and Escherichiacoli, and each mutant showed greater resistanceto penicillin than its predecessor. A progressiveshift toward sensitivity to serum was noted ineach series, and the derived strains possessed thesame somatic antigens as the parent strains.Our interest in this technique was twofold.

First, it would be convenient to be able to obtainserum-sensitive strains of any particular 0 typeof E. coli or Salmonella so that antibodies tothese strains could be measured by the verysensitive bactericidal method. Second, the com-parison of the virulence of serum-resistant strains

I Presented in part at the 63rd Annual Meetingof the American Society for Microbiology, Cleve-land, Ohio, 5-9 May 1963.

of Salmonella with their serum-sensitive mutantsmight show whether resistance to serum is anecessary quality for virulence.The first part of the present study was devoted

to discovering whether serum-sensitive strainswhich still retain the parental 0 type could bederived with regularity from serum-resistantstrains of E. coli of various 0 types. The secondportion was concerned with an investigation ofthe virulence for the mouse of the penicillin-resistant mutant strains derived from virulent,serum-resistant strains of S. typhimurium and S.enteritidis.

MATERIALS AND METHODSBacterial strains. Strains of E. coli were obtained

from K. L. Vosti and L. A. Rantz of the Divisionof Infectious Diseases, Department of Medicine,Stanford University School of Medicine. Eachhad been isolated by selecting a single colony froma culture of human urine. E. Randall typed themby the tube agglutination method with antiseraobtained from the Communicable Disease Center,Atlanta, Ga. S. typhimurium 7 was obtained sim-ilarly, and its identification was confirmed by theCalifornia Department of Public Health. S.typhimurium 173 was provided by D. Rowley,University of Adelaide, Adelaide, Australia. It isvirulent for the mouse and was classified as S-2/444. Slide agglutinations with antisera specific for

630

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PENICILLIN RESISTANCE IN ENTEROBACTERIACEAE

the 0 antigenic factors showed it to have theantigenic formula 1,4,5,12. We obtained theseantisera from B. A. D. Stocker, Lister Institute,London, England; they had originally been ob-tained from Joan Taylor of the Salmonella Refer-ence Laboratory, Colindale, England. The antiserawere diluted 1:10 or 1:20 in distilled water. Theresults can be considered only qualitative innature, but they were specific in the positive testsbecause anti-04 and 05 did not agglutinate or-ganisms of the S. enteritidis series, and anti-09did not agglutinate those of the S. typhimuriumseries. An undiluted anti-019 serum agglutinatedmembers of neither series. S. enteritidis 203 wasreceived from the Division of Laboratories, Cali-fornia Department of Public Health. It possesses0 antigens 9 and 12 and is virulent for mice.Mice. Mature CF-1 mice of both sexes, obtained

from Carworth Farms, New York, N.Y., wereused for the determinations of LD5o and to supplyserum.

Cutlture media. Brain Heart Infusion (Difco)broth and nutrient agar (Difco) were the basicculture media used throughout. Eosin methyleneblue agar (EMB Agar, Difco) was employed whenappropriate for colonial differentiation.Serum. The serum for the bactericidal tests was

obtained from healthy human donors. Blood wascollected aseptically in sterile centrifuge bottles,allowed to stand at room temperature for about 1hr, and then centrifuged for about 30 min. Serumwas stored in convenient volumes at -14 C untiljust before use.Mouse serum used as medium for growth curves

was collected by pooling blood from at least 50mice. The blood was treated as described above,and the serum was not inactivated because it isnaturally inactive in bactericidal tests (Marcus,Esplin, and Donaldson, 1954).

Derivation of mutants resistant to penicillin. A0.1-ml amount of an overnight broth culture ofeach of the selected strains was spread on a gra-dient plate (Szybalski and Bryson, 1952) contain-ing benzylpenif illin or ca-aminobenzylpenicillin(Penbritin). After 48 or 72 hr, an isolated colonyin the zone of inhibition was selected for the inocu-lation of a tube of broth. Samples of the resultingovernight cultures were frozen at -14 C andsaved for future testing; 0.1 ml was plated on agradient plate containing the next higher concen-tration of antibiotic so that the next-step mutantmight be selected. This procedure was repeatedso that a series of penicillin-resistant mutants wasderived from each parent strain. The temperatureof incubation for all cultures was 37 C.

Testing of resistance to penicillin. E. coli strainsA27 and A48 and their penicillin-resistant mutantswere tested by inoculating 106 organisms from anovernight broth culture into 2-ml quantities ofbroth containing varying concentrations of ben-zylpenicillin. The greatest concentration of drugpermitting visible growth at 48 hr was taken as thevalue for degree of resistance.Salmonella strains 173 and 203 and their mutants

resistant to a-aminobenzylpenicillin were testedby inoculating nutrient agar plates containingvarious concentrations of a-aminobenzylpenicillinwith 50 to 100 organisms of each strain. The great-est concentration of antibiotic allowing the con-sistent appearance of colonies was taken as thevalue for the degree of resistance. When parentand mutant strains within any one series werecompared, they were plated on the same batchesof agar on the same day. Typical results are shownin Table 4.

Bactericidal tests. Samples of the strains to betested were thawed, and a loopful from eachsample was used to inoculate tubes containing 4.5ml of broth. After growth overnight, each culturewas diluted decimally to 10-6 in 0.85% saline. Thenumber of viable bacteria contained in 0.1 ml ofthe 10-6 dilution was determined by the pour-platemethod; from these data, the sizes of the inoculawere determined. To 0.1-ml portions of the 10-2and 10-5 dilutions, 0.4-ml amounts of active humanserum were added, and the suspensions were in-cubated at 37 C. The number of bacteria survivingin 0.1-ml portions of the suspensions was deter-mined by colonial counts from platings made at1 and 2 hr. For any given series of peni,illin-resistant mutants derived from one strain, thetests were done on the same day with serum fromthe same donor.

Determination of virulence of bacteria for themouse. Ten mature CF-1 mice housed togetherwere each inoculated intraperitoneally with thesame volume of one serial 10-fold dilution in salineof an overnight culture in broth. Sufficient 10-folddilutions were used to permit the calculation ofthe LD5o for 30 days (Reed and Muench, 1938).The LD5o, when greater than 100, is reported tothe closest whole power of 10.

Determination of rates of growth of selected strainsof Salmonella in mouse serum. Inocula of the orderof 103 organisms in 0.2 ml of saline were added to1.0-ml amounts of undiluted mouse serum. Plat-ings to determine numbers of viable bacteria weremade at zero-time and at hourly intervals up to 4,5, or 6 hr. The tubes were shaken by hand im-mediately before the cultures were taken.

RESULTSDerivation of serum-sensitive strains. The parent

strains were resistant to serum. All but oneshowed 70% or more survival after the exposureof an inoculum of 102 organisms to human serum;the exception, E. coli C20, showed 40% survival.The results from experiments in which E. colistrains A27 and A48 were used are shown inTable 1; these results are typical of the otherlines of mutants which became sensitive to serumduring the derivation of mutants resistant tobenzylpenicillin. A few showed some indicationof a step-wise increase in sensitivity to serum, asdid the A48 series, but most, such as the A27series, did not. The most noteworthy finding was

631VOL. 89, 1965

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

632ROANTREE AND STEWARD

the occurrence, in most series, of a mutation topenicillin resistance associated with a high degreeof sensitivity to serum, as seen in the fourth-stepmutants of A27 and A48. It was only through sucha step that strains were obtained which were

sufficiently sensitive to serum for use in measur-ing anti-O antibody. The detection of this step ina virulent strain seemed our best hope to assessthe effect of sensitivity to serum on virulence.Thus, the remainder of the early part of the studywas directed toward determining whether thisstep could be detected in most strains of E. coliand Salmonella. The criterion for occurrence ofthis step was the survival of fewer than 0.1% of105 organisms of the mutant strain in activehuman serum after 2 hr at 37 C. The mutantsisolated immediately prior in the series to thoseshowing a high degree of sensitivity to serumshowed survival of at least 5% of 102 organisms,and usually showed a much higher rate of sur-vival. The only exception was in the Salmonella173 line, as noted below.The results of the survey to detect this muta-

tion in a number of strains are shown in Table 2.Six of seven strains of E. coli of differing 0 typesshowed the one-step conversion. The place of themutational step in the series and the concentra-tion of penicillin in the plate on which it wasdetected varied greatly. One of two Salmonellastrains showed the same mode of conversion.

Similar experiments with a-aminobenzyl-penicillin used for the selection of mutants werecarried out (Table 3). In these experiments, iftwo or three different colonial types appeared on

TABLE 1. Conversion of two serum-resistant Escher-ichia coli strains into strains sensitive to

serum by deriving a series ofpenicillin-resistant mutants

Per cent survival afterexposure to serumt of an Resistance

Strain inoculum of topenicillin

106/0.1 ml 102/0.1 ml

units/m"

A27 TMTC* 87 100P127 TMTC 100 120P227 TMTC 100 140P327 TMTC 100 180P427 <0.01 0 240

A48 TMTC 100 10P148 TMTC 42 10P248 TMTC 22 20P348 TMTC 15 40P448 <0.01 0 80

* Too many to count.

TABLE 2. Conversion of serum-resistant Escherichiacoli and Salmonella strains into strains sensi-

tive to serum by deriving a series of mutantsresistant to benzyl penicillin

Con-Strain 0 antigen ver- Step Penicillin

sion

units/miE. coliA27 Ox8 + 4 800A48 6 + 4 800A62 25 + 1 200A76 12 + 5 1,000A94 7 + 5 600102 1 0 5 1,000108 Untypable + 4 400

Salmonella7 (typhimur- 1,4,5,12 0 7 1,000ium)

173 (typhimur- 1,4,5,12 + 2 50ium)

* Indicates mutational step at which conver-sion occurred or, if no conversion, last-stepmutant tested. Last column contains concentra-tions of benzylpenicillin in the gradient platesupon which the mutants in the preceding columnwere isolated.

the gradient plates, new sublines of mutants werebegun from each to determine whether the step toa high degree of sensitivity to serum took placein each subline, and whether its occurrence couldbe predicted by colonial appearance. Such sub-lines are designated a, b, c, or d under the ap-propriate parent strain in Table 3.

Eight of nine E. coli strains of types 04, 06,and 075 were successfully converted into strainshighly sensitive to serum in one major step. Inthe sublines of CIO and C14, the early occurrenceof this step in one subline did not predict itsoccurrence in the other. Some conversions, asin the sublines of ClI and C27, took place atwidely separated mutational steps and weredetected at very different concentrations of anti-biotic. Some mutants from S. typhimurium 173and S. enteritidis 203 derived on a-aminobenzyl-penicillin were very sensitive to serum, as dis-cussed below.

If one considers all the series of mutantsderived on both kinds of penicillin, it is note-worthy that in five instances mutants highlysensitive to serum were detected on the firstmutational step to penicillin resistance. Mutantsderived from S. typhimurium 7 on either mediumshowed no such conversion to sensitivity toserum, despite the large number of mutantstested. Even the 12th-step mutant from this

632 J. BACTERIOL.

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

VrOL. 89, 1965 PENICILLIN RESISTANCE IN ENTEROBACTERIACEAE

TABLE 3. Conversion of serum-resistant Escherichiacoli and Salmonella strains into strains sensitiveto serum by deriving a series of mutants resist-

ant to a-aminobenzylpenicillinPeIcli

Strain

E. coliClO

C1l

C12C13C14

C15C20C21

C27

Salmonella203 (enteri-

tidis)173 (typhi-murium)

7 (typhimur-ium)

0 antigen Line

6

4

6756

47575

75

9,12

1,4,5,12

1,4,5,12

ababc

aba

abab

ababcabcd

Con-ver-sion

0++++0++0++++++

0++000

Step*

952567114244351

4126612777

Penicillinconcn

pg/ml

1608020160240

2,0002020

4002080804016020

* Indicates mutational step at which conver-sion occurred or, if no conversion, last-stepmutant tested. Last column contains the concen-trations of a-aminobenzylpenicillin in the gradientplates upon which the mutants in the precedingcolumn were isolated.

strain showed only slightly more sensitivity toserum than did the original parent strain.

There seemed little difference between a-aminobenzylpenicillin and benzylpenicillin intheir roles as selective agents for the mutationto sensitivity to serum. Mutants to resistance tobenzylpenicillin gained in resistance to a-amino-benzylpenicillin, and vice versa.The serum-sensitive mutants from the strains

of E. coli which could be typed initially showed,in every case, the same 0 type as that of theparent. Most were as easily typable as the parentstrains, but two showed qualitatively weakeragglutination, and one strain developed a tend-ency to clump which made identification moredifficult.

Alterations in virulence for the mouse. Series ofmutants from S. typhimurium 173 and S. enter-itidis 203 were suitable for studies of virulence forthe mouse. The methods of derivation of themutant strains and the data concerning theirsensitivity to serum, resistance to a-aminobenzyl-penicillin, mode of growth in broth, and virulencefor mice are depicted in Fig. 1, 2, and 3.

Figure 1 shows the series of mutants obtainedby serial passage of strain 173 upon gradientplates containing benzylpenicillin in the lowerlayer at the concentrations indicated. The plateupon which the fourth-step mutant, P4173, wasinoculated yielded large (L) and small (S) typesof colonies. One of each was picked to start theS and L sublines, but these did not differ greatlyin sensitivity to serum. The first-step mutant,P'173 was nearly as resistant to serum as theparent strain. Its virulence for the mouse wassignificantly less than that of the parent, but itwas still highly virulent. The second-step mutantstrain, P2173, showed a great increase in sensitiv-

FIG. 1. Series of mutants obtained by passage of Salmonella typhimurium upon gradient plates containingincreasing concentrations of benzylpenicillin. Tubes contain broth; crosshatching indicates growth in sus-

pension. Numbers in brackets show resistance to a-aminobenzylpenicillin in micrograms per milliliter;numbers in boxes are per cent survivors in serum after 2 hr of inocula of 105 (upper box) and 102 (lowerbox) bacteria per 0.1 ml.

633

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

634ROANTREE AND STEWARD

ity to serum, and its LD50 for mice was of the orderof 104 as compared with 102 for strain P1173.Strains P3173, P6173S, and P6173L showed little,if any, further decrease in virulence. Strain P2173gave rise to smooth colonies on nutrient agar andgrew in suspension in broth, but it was onlyweakly agglutinated by antisera directed againstthe 0-antigenic factors 1, 4, 5, and 12.

Figure 2 shows the derivation of three sublinesfrom the same 173 parent strain by passage ona-aminobenzylpenicillin. Three different colonialtypes were evident on the first gradient plate,and the three sublines, a, b, and c, were startedfrom one colony of each type. Lines b and ccontain no mutant which showed a marked in-crease in sensitivity to serum, even though theirsixth-6tep mutants showed a greater resistance toa-aminobenzylpenicillin than does that of line a.A more accurate method should be used, or moreobservations should be made, to determinewhether the sixth-step mutants in the b and clines were more sensitive to serum than the parentstrain. The second-step mutant in line a, Br2173a,showed a major step toward sensitivity to serum,even though it did not quite meet the criterion ofless than 0.1 % survival of 105 organisms. Thethird-stage mutant showed a characteristic

growth in the bottom of the tube, and subsequentmutants in the series also showed this growth.Determinations of virulence for the sixth-stepmutants only were attempted. The LD50 for line awas 107, the highest of any of the Salmonellamutants we have tested. The mutants represent-ing the b and c lines were much less virulent thanthe parent strain and had LD5o values of 105 ascompared with 102.

Figure 3 depicts a series of mutants derivedfrom S. enteritidis 203. In this group of experi-ments, a broth culture of strain 203 was startedfrom a single colony. Two colonial types werenoted on the first gradient plate containing a-aminobenzylpenicillin, and, from a single colonyof each of these, sublines a and b were initiated.Subline a consists of three mutants, none of whichbecame highly sensitive to serum. The second-step mutant, Br2203a, still retained most of thevirulence characteristic of the parent strain. Thefirst mutant in the b line showed a great increasein sensitivity to serum, a characteristic growth inthe bottom of the tube, and a marked loss ofvirulence. Because this represented the firstone-step mutant from a virulent strain whichshowed the conversion to sensitivity to serum,and because cumulative mutations to penicillin

(20)FIG. 2. Derivation of three sublines from a Salmonella typhimurium strain by passage on gradient plates

containing increasing concentrations of a-aminobenzylpenicillin. Tubes contain broth; crosshatching indi-cates growth in suspension, and sediment indicates growth in bottom of tube. Numbers in brackets shou'resistance to a-aminobenzylpenicillin in micrograms per milliliter; numbers in boxes are per cent survivorsin serum after 2 hr of inocula of 10O (upper box) and 102 (lower box) bacteria per 0.1 ml.

634 J. BACTERIOL.

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PENICILLIN RESISTANCE IN ENTEROBACTERIACEAE

FIG. 3. Series of mutants derived from Salmonella enteritidis by passage on gradient plates containingincreasing concentrations of a-aminobenzylpenicillin. Two methods are shown for the selection of back-mutants of Br120Sb. Tubes contain broth; crosshatching indicates growth in suspension, and sediment indi-cates growth in bottom of tube. NA, nutrient agar; Ser, human serum; S, organism sensitive to human serum;R, organism resistant to human serum. Numbers in brackets show resistance to a-aminobenzylpenicillin inmicrograms per milliliter; numbers in boxes are per cent survivors in serum after 2 hr of inocula of 105(upper box) and 102 (lower box) bacteria per 0.1 ml.

resistance were not involved, it seemed an ap-propriate candidate for more intensive study.Two methods for the selection of a back-mutant

were employed (Fig. 3). A 0.1-ml amount of a10-1 dilution of an overnight culture of strainBr'203b was subjected to the action of 1.0 ml ofserum for 4 or 5 hr; this resulted in the reductionof an inoculum of about 106 per 0.1 ml to lessthan 50 organisms. One of these colonies wasselected and grown in broth overnight. Theprocedure was then repeated with 0.1 ml of a10-1 dilution of that culture. Although we havesince employed more efficient methods, thistechnique demonstrated the stability of themutant strain. The 10th strain isolated in thismanner still had the same mode of growth inbroth, resistance to a-aminobenzylpenicillin,sensitivity to serum, and low degree of virulenceas the Br'203b strain. The 11th passage resultedin the selection of a strain which showed 10%survival of 102 organisms in serum; this strain washighly virulent. We had found earlier thatBr'203b could be differentiated from the parentstrain, 203, because it formed rougher and moretranslucent colonies on EMB Agar than did thelatter. When this method was used to analyze

the composition of the broth culture derived fromthe colony from the 11th passage, it was foundthat bacteria forming smooth colonies like thoseof the original 203 strain made up about 15%of the total population. These bacteria were thesurvivors of the test for sensitivity to serum;they grew in suspension in broth and, presumably,were responsible for virulence, because they werethe type surviving in the mice dying from chal-lenge with the culture as a whole. The remaining80 to 85% of the culture formed rough-appearingcolonies on EMB (like those of strain Br1203b),grew in the bottom when cultured in broth, andaccounted for the degree of resistance to a-aminobenzylpenicillin of the culture as a whole.The data used to determine the resistance to

a-aminobenzylpenicillin of the important strainsin the 203 series are shown in Table 4. The straindesignated Br'203b ser 11-3 is one picked from asmooth colony on EMB Agar after the culturefrom the 11th passage had been streaked on it.This strain and the one isolated after the 12thpassage through serum were indistinguishablefrom the original parent, 203, in resistance toserum, sensitivity to penicillin, and virulence forthe mouse.

VOL. 89, 1965 635

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ROANTREE AND STEWARD

TABLE 4. Degree of resistance to ce-aminobenzyl-penicillin of Salmonella enteritidis 203 and

mutants derived from it

Penicillin 203 o .uo o

NA 49t 70 76 50 60 65 77 73 92 87 110 108NA 53 53 ---0.5 57 56 53 56 77 66 54 64 72 53 112 1041.0 31 78 73 43 56 55 40 35 41 18 61 791.4 0 66 39 119 33 44 O O O 0 46 491.8 0 65 30 38 45 46 0 0 0 0 0 02.0 0 90 47 25 58 57 0 0 0 2 0 02.5 0 0 0 0 0 0 0 0 0 0 0 03.0 0 0 0 0 0 0 0 00 0 0 0

* Inocula were plated on nutrient agar (NA) andon agar containing the indicated concentrations ofa-aminobenzylpenicillin.

t Number of colonies appearing.

Strain Br'203b grew in broth at the bottom ofthe tube, leaving an optically clear supernatantfluid. The second method used to select a back-mutant was one in which a tiny loopful of brothfrom the surface of such a culture was inoculatedinto another tube of broth. After overnight incu-bation of this tube, the step was repeated. Eachstep in this series showed an increase in turbidityof the supernatant fluid and a decrease in thedeposit at the bottom of the tube. The fourthstep resulted in a culture which showed almost aslittle deposit as did strain 203. After one morepassage, a single colony was selected and culturedin broth. The resulting strain is the one designatedBr1203b6 in Fig. 3. This strain could not bedistinguished from strain 203 by its growth inbroth or its colonial form on EMB Agar, but itwas like strain Br'203b in its sensitivity to serum.The value for its resistance to a-aminobenzyl-penicillin was intermediate to those for 203 andBr1203b. Since it was sensitive to serum, it wassurprising to find it nearly as virulent for mice asthe 203 strain. Both its sensitivity to serum andits virulence have been repeatedly confirmed.These experiments, then, yielded two virulent

strains; one was a serum-resistant back-mutant,probably identical with the original parent strain,and the other resembled the original strain in itsmode of growth in broth but was sensitive toserum. For strains to have the degree of virulencepossessed by these, considerable multiplicationmust occur in the mouse (Hobson, 1957), whichshould therefore act as a selective agent by favor-ing the survival of the more virulent organisms.To determine what type of bacteria survived inthe mice, the spleens of animals in extremis or

very recently dead were homogenized and cul-tured on EMB Agar. The mice selected for thiswere those receiving the smaller inocula in theLD50 experiments. Colonies representing a diver-sity of colonial appearances were selected, andstrains from these were tested for their sensitivityto serum and their mode of growth in broth.When the culture derived from the colony from

the 11th passage through serum was used as thechallenge strain, the mice were effective selectiveagents for the virulent "smooth" type of bacteriawhich made up about 15% of the culture (Fig.3). This conclusion is plausible because a specialeffort was made to select the colonies giving theroughest appearance on EMB Agar. Isolatesfrom mice infected with strain Br1203b6 were notresistant to serum, so the virulence of this straincannot be attributed to a minority of serum-resistant organisms in the culture used for chal-lenge. The passages through mice of strainBr'203b and the seemingly identical derivativestrain obtained after 10 passages through serumresulted in the selection for strains which grew insuspension in broth, but not for serum-resistantorganisms. Among those growing in suspensionwere strains which are like Br1203b6 in theirvirulence for mice.Another group of experiments with S. enter-

itidis 203 was done to determine whether thefirst-step mutation to penicillin resistance,sensitivity to serum, and bottom-growth in brothcould be detected frequently. A single colony ofstrain 203 was selected and grown in broth over-night. After this culture had been suitably dilutedand plated on agar, each of three isolated colonieswas used to initiate an overnight culture in broth.Each of the three cultures was plated, as usual,upon gradient plates containing 5 jig/ml of a-aminobenzylpenicillin at the concentrated end.From the zones of inhibition of growth on theseplates, colonies of various morphologies werepicked and tested for mode of growth in broth.From one such culture, seven of nine coloniesgave rise to cultures which were sensitive to serumand grew at the bottom in broth. Bottom-growingmutants were not discovered among 13 coloniesselected from the other two cultures, but a strainsimilar to Br1203b6 was detected. It was verysensitive to serum, virulent for the mouse, andresistant to a-aminobenzylpenicillin at the 1.6jAg/ml concentration.

Partial characterization of the Salmonellamutants. It seemed likely that some of the muta-tions associated with sensitivity to serum amongthe Salmonella strains were similar or identicalto mutations to roughness, because of the mode ofgrowth in broth exhibited by Br3173a and

636 J. BACTERIOL.

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

VOL. 89, 1965 PENICILLIN RESISTANCE IN ENTEROBACTERIACEAE

Br'203b, and because their autoagglutinabilityor natural clumping did not allow identificationby specific antisera. Furthermore, the P2173strain was weakly agglutinable by antiseradirected against the 1, 4, 5, and 12 antigenicfactors even though P1173 and, oddly, P3173were easily and specifically agglutinated.

B. W. Nelson in this laboratory is working withlipopolysaccharides of the cell walls of parentstrains and derived serum-sensitive mutants.Acid hydrolysates of lipopolysaccharides obtainedby the phenol-water method of Westphal,Luderitz, and Bister (1952) have been analyzedby thin-layer chromatography. S. typhimuriumstrains 173 and P2173 and S. enteritidis strains203, Br'203b, and Br'203b6 have been studied inthis manner; the mutants show all the sugarscharacteristic of the parents, although the lipo-polysaccharide of strain Br1203b, especially,seems to contain less of the terminal sugarsrhamnose and tyvelose. Current quantitativestudies will be the subject of a later paper.The mode of growth of Brf173a and Br1203b

has been observed by dark-field microscopy, byuse of a microdroplet of broth containing one ortwo organisms under oil. These small inoculamultiply to form tortuous chains and clumps;these only occasionally show motility, althoughthe parent strain and the Br'203b' are motile.

Despite the tendency of strain Br1203b togrow in chains, its rate of growth in mouse serumclosely paralleled that of the 203 strain. In onetest, its generation time over the 2- to 4-hr periodwas 46 min, as compared with 52 min for theparent strain; in a second run over the 2 to 5-hrperiod, 51 min for strain Br1203b6 comparedwith 46 min for the parent strain. Final numbersat the end of the 4th and 5th hr were nearly thesame for the two strains.

Curves representing the growth of S. typhi-murium strain P3173 and its 173 parent showed amore pronounced lag period between the 0- and2-hr counts for the P3173 strain in three of fourcurves constructed. However, generation timesduring the 2- to 4-hr periods were comparable.Final counts done in six separate tests at 6 hrshowed no difference in numbers between the twostrains representing more than one generationand favored each in three instances.

DISCUSSIONThe finding of Michael and Braun (1958) that

mutants to penicillin resistance among theEnterobactericeae become sensitive to serumwithout losing their characteristic 0 antigenswas confirmed. We found that most serum-resistant strains of E. coli and Salmonella could be

converted to sensitivity to serum by this method,and that a one-step change to a high degree ofsuch sensitivity could be recognized in mostseries.As discussed below, it is likely that the muta-

tions to penicillin resistance with which we aredealing depend upon an intrinsic change in thebacterium unrelated to the production of penicil-linases. The production of penicillinases by themutants was not investigated. It is possible thatcolonies producing penicillinase masked thepresence of those showing the mutation to sen-sitivity to serum in series in which no suchmutation was detected. It seems unlikely thatthe small, but consistently demonstrable, differ-ences in penicillin resistance existing between theparents and mutants in the S. typhimurium 173lines and the S. enteritidi8 203 line were causedby penicillinase production. The inocula used inthese determinations were so small that, in effect,the penicillinase from a single organism wouldhave to neutralize all the penicillin which coulddiffuse to its location through the agar.

Since the development of a high degree ofsensitivity to serum occurred on the first-stepmutation to penicillin resistance in series ofmutants from five different strains, the former isnot necessarily the result of cumulative mutationsto penicillin resistance. That a single mutationalevent confers both resistance to penicillin andsensitivity to serum is further demonstrated bythe isolation of a back-mutant in the S. enteritidis203 series having properties identical to those ofthe original parent. Hotchkiss (1951), using thepneumococcus, and Banic (1959), using Sal-monella, showed that the small step-wise muta-tions to penicillin resistance which occur whenmutants are derived in vitro represent separatemutational events occurring at different locationson the chromosome. The finding that one muta-tion or one class of mutations to penicillin re-sistance is characterized by a high degree ofsensitivity to serum indicates that some of thesteps to penicillin resistance can be thus defined.Thus, one should be able to map the locus forany particular strain by the appropriate geneticexperiments. Among the mutations to penicillinresistance observed in the S. enteritidis 203 series,two are characteristic. One of these resulted in aserum-resistant strain which gave rise to mucoidcolonies. This strain was not agglutinable inspecific antisera. The other mutation resulted inan inability to form colonies on deoxycholate-lactose-agar.

It is tempting to postulate, as did Michael andBraun (1958), that the mutations to penicillinresistance resulting in increases in sensitivity to

637

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

68ROANTREE AND STEWARD

serum are ones modifying the structure of thecell wall. The evidence that penicillin inhibits thesynthesis of the mucopeptide layer is convincing(Park and Strominger, 1957; Rogers and Mandel-stam, 1962). The action of the complement-dependent bactericidal system on sensitive strainsresults in the formation of spheroplasts (Muschel,Carey, and Baron, 1959; Michael and Braun,1959), indicating disruption of the mucopeptidelayer. The formation of spheroplasts by serummay be dependent upon the action of lysozyme(Inoue et al., 1959), but the presence of thisenzyme is evidently unnecessary to the bac-tericidal effect (Michael and Braun, 1964).

It may be, however, that a change in the over-lying lipopolysaccharide layer of the cell wall(Weidel, Frank, and Martin, 1960) is necessaryto allow the penetration of the macromoleculesof the complement-dependent bactericidal systemto the inner rigid layer. It has long been knownthat the rough mutants of virulent Salmonellastrains are avirulent (Lingelsheim, 1913) andsensitive to serum (Thj0tta and Waaler, 1932;Rowley, 1956). Although our E. coli mutantsretained the parental 0 type, studies utilizingquantitative absorptions of bactericidal antibodyby the parent strain A27 and its P427 mutanthave indicated that the two strains are not anti-genically identical (Steward, Collis, and Roan-tree, 1964).

Certain mutants of the Salmonella strains usedin the present studies showed some characteristicsresembling those of rough strains, such as thegrowth at the bottom in broth as illustrated bystrains Br3173a and Br1203b, and the weakagglutinability of strain P2173. But analyses ofthe lipopolysaccharide of the cell walls from S.typhimurium P2173 and S. enteritidis Br1203bby B. W. Nelson show that the lipopolysac-charides from these mutants possess the fullrange of sugars characteristic of the two species.Thus, these mutants are not rough by presentdefinition (Liuderitz et al., 1960); they fall intoneither the rough I nor rough II category (Beck-mann, Subbaiah, and Stocker, 1964). S. typhi-murium P3173 is an example of a mutant which issensitive to serum and avirulent but retains allthe characteristics of a smooth strain. Its genera-tion time in mouse serum differs little from thatof its virulent parent strain, so that in this in-stance whatever is responsible for its sensitivityto human serum may be responsible for itsavirulence.We concluded from the results of the ex-

periments with the series of mutants of S.typhimurium 173 derived on media containingbenzylpenicillin or a-aminobenzylpenicillin thatmutations to penicillin resistance were likely to

render the strain avirulent whetheror not the mu-tations involve any great increase in sensitivity toserum. However, those mutations associated withlarge increases in sensitivity to serum were char-acterized by the most striking losses of virulence.Mice are unique in lacking a demonstrably

active complement-dependent bactericidal system(Marcus et al., 1954), and strains of Enterobac-teriaceae sensitive to serum of other animalsmultiply when placed in millipore chambers inthe peritoneal cavities of mice (Steward andRoantree, 1961). However, there is evidence thatsuch serum-resistant strains are more virulentfor the mouse than are serum-sensitive ones(Maal0e, 1948; Rowley, 1954). Rowley postulatedthat one requirement of a virulent strain wasresistance to serum. One way of explaining theparadox that resistance to serum is necessary forvirulence in animals lacking a humoral bac-tericidal system would be to postulate that serum-sensitive organisms are more susceptible to in-tracellular killing after attachment of antibodyand whatever components of complement attachto bacteria after exposure to mouse serum.The results from the early experiments with the

S. enteritidis 203 strain seemed compatible withthe hypothesis that resistance to serum is ofimportance for virulence. The very serum-sensi-tive mutant, Br'203b, was of decreased virulence,and the serum-resistant Br2203a, although ofgreater resistance to penicillin than Br'203b,retained most of the virulence of the parentstrain. The detection of a virulent back-mutantof Br'203b by selecting a serum-resistant strainsupported the belief that the two qualities areclosely linked. However, the discovery of theserum-sensitive virulent mutant, Br1203b6, de-rived from Br'203b with repeated subculture fromthe very top of broth cultures as the selectivemeans, seemed to contradict the idea that re-sistance to serum was a necessary quality of avirulent strain. Screening for determination ofsensitivity to serum was done with human sera.When the sera of mice were investigated forbactericidal antibody against this strain in testswith fetal calf serum as a complement source,they were found to contain none. Immunizationwith the Br1203b6 leads to the appearance ofbactericidal antibodies and a high degree ofprotection for the mouse against challenge withthe Br1203b6. These findings will be presentedmore extensively later; however, it should benoted here that the existence of such strains doesnot prove that resistance to serum is an unim-portant quality for virulent strains, but only thatit is unnecessary for virulence if the test animalslack bactericidal antibody or perhaps some othercomponent to complete the bactericidal system

638 J. BACTERIOL.

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

PENICILLIN RESISTANCE IN ENTEROBACTERIACEAE

at the time of challenge. It will be of interest toascertain whether immunization is as effective inprotecting mice against the serum-resistantparent as it is against the serum-sensitive mutant.

ACKNOWLEDGMENTThis investigation was supported by Public

Health Service grant AI 02755 from the NationalInstitute of Allergy and Infectious Diseases.

LITERATURE CITEDBANIC, S. 1959. Transduction to penicillin andchloramphenicol resistance in Salmonella typhi-murium. Genetics 44:449-455.

BECKMANN, I., T. V. SUBBAIAH, AND B. A. D.STOCKER. 1964. Rough mutants of Salmonellatyphimurium. II. Serological and chemicalinvestigations. Nature 201:1299-1301.

HoBSON, D. 1957. The behaviour of a mutantstrain of Salmonella typhimurium in experi-mental mouse typhoid. J. Hyg. 55:322-333.

HOTCHKISS, R. D. 1951. Transfer of penicillinresistance in pneumococei by the desoxyribo-nucleate derived from resistant cultures. ColdSpring Harbor Symp. Quant. Biol. 16:457-461.

INOUE, K., Y. TANIGAWA, M. TAKUBO, M. SATANI,AND T. AMANO. 1959. Quantitative studies onimmune bacteriolysis. Biken's J. 2:1-20.

LINGELSHEIM, V. 1913. Zur Frage der Variation derTyphusbacillen und verwandter Gruppen.Zentr. Bakteriol. Parasitenk. Abt. I Orig. 68:577-582.

LUDER1TZ, O., F. KAUFFMANN, H. STIERLIN, AND0. WESTPHAL. 1960. Zur Immunochemie derO-Antigenevon Enterobacteriaceae. II Vergleichder Zuckerbausteine von Salmonella S-, R-, undT-Formen. Zentr. Bakteriol. Parasitenk. Abt.I Orig. 179:180 -186.

MAAL0E, 0. 1948. Pathogenic-apathogenic trans-formation of Salmonella typhimurium. ActaPathol. Microbiol. Scand. 25:414-430.

MACKIE, T. J., AND M. H. FINKELSTEIN. 1931.Natural bactericidal antibodies: observationson the bactericidal mechanism of normal serum.J. Hyg. 31:35-55.

MARCUS, S., D. W. ESPLIN, AND D. M. DONALDSON.1954. Lack of bactericidal effect of mouse serumon a number of common microorganisms. Science119:877.

MICHAEL, J. G., AND W. BRAUN. 1958. Relation-ships between bacterial resistance to serum andpenicillin. Proc. Soc. Exp. Biol. Med. 97:104-108.

MICHAEL, J. G., AND W. BRAUN. 1959. Serumspheroplasts of Shigella dysenteriae. Proc. Soc.Exp. Biol. Med. 100:422-425.

MICHAEL, J. G., AND W. BRAUN. 1964. Analysis ofsequential stages in serum bactericidal reac-tions. J. Bacteriol. 87:1067-1073.

MUSCHEL, L. H., W. F. CAREY, AND L. S. BARON.1959. Formation of bacterial protoplasts byserum components. J. Immunol. 82:38-42.

PARK, J. T., AND J. L. STROMINGER. 1957. Mode ofaction of penicillin. Biochemical basis for themechanism of action of penicillin and for itsselective toxicity. Science 125:99-101.

REED, L. J., AND H. MUENCH. 1938. A simplemethod of estimating fiftv per cent endpoints.Amer. J. Hyg. 27:493-497.

ROANTREE, R. J., AND L. A. RANTZ. 1960. A studyof the relationship of the normal bactericidalactivity of human serum to bacterial infection.J. Clin. Invest. 39:72-81.

ROGERS, H. J., AND J. MANDELSTAM. 1962. Inhibi-tion of cell-wall-mucopeptide formation inEscherichia coli by benzyl penicillin and 6-[D (-)-a-aminophenglacetamido] penicillamicacid (ampicillin). Biochem. J. 84:299-303.

ROWLLY, D. 1954. The virulence of strains ofBacterium coli for mice. Brit. J. Exp. Pathol.35:528-538.

ROWLEY, D. 1956. Rapidly induced changes in thelevel of nonspecific immunity in laboratoryanimals. Brit. J. Exp. Pathol. 37:223-234.

STEWARD, J. P., L. R. COLLIS, AND R. J. ROAN-TREE. 1964. Effects of active immunization andof total body x-irradiation upon the humoralbactericidal system of the guinea pig as meas-ured with strains of enteric bacilli. J. Immunol.92:616-625.

STEWARD, J. P., AND R. J. ROANTREE. 1961. Effectof mouse peritoneal fluid on strains of entericbacilli. Proc. Soc. Exp. Biol. Med. 108:654-658.

SZYBALSKI, W., AND V. BRYSON. 1952. Geneticstudies on microbial cross resistance to toxicagents. I. Cross resistance of Escherichia coli tofifteen antibiotics. J. Bacteriol. 64:489-499.

THJ0TTA, T., AND E. WAALER. 1932. Dissociationand sensitiveness to normal serum in dysenterybacilli of type III. J. Bacteriol. 24:301-316.

WEIDEL, W., H. FRANK, AND H. H. MARTIN. 1960.The rigid layer of the cell wall of Escherichiacoli strain B. J. Gen. Microbiol. 22:158-166.

WESTPHAL, O., 0. LUDERITZ, AND F. BISTER. 1952.Uber die Extraktion von Bakterien mit Phenol/Wasser. Z. Naturforsch. 76:148-155.

639VOL. 89, 1965

on Decem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from