STUDIES ON ANTIBIOTIC SYNERGISM AND ANTAGONISM: THE EFFECTIN VITRO OF COMBINATIONS OF ANTIBIOTICS ON BACTERIA OF

VARYING RESISTANCE TO SINGLE ANTIBIOTICS1

J. B. GUNNISON, M. C. SHEVKY, J. A. BRUFF, V. R. COLEMAN, AND E. JAWETZWITH THE TECHNICAL ASSISTANCE OF D. GROOM AND P. Howm

Department of Microbiology, University of California School of Medicine, San Francisco, California

Received for publication January 12, 1953

Earlier studies on combined antibiotic actionfrom this laboratory have resulted in the formula-tion of a scheme for combined antibiotic action.Antibiotics have been divided tentatively intotwo groups: I. penicillin, streptomycin, bacitra-cin, and neomycin; II. chloramphenicol, aureo-mycin, terramycin (Jawetz and Gunnison,1952a,b; Jawetz et al., 1952). Mixtures of group Idrugs often resulted in synergism, never inantagonism. Mixtures of group II drugs showedno combined effects beyond simple addition.Tue mixture of an antibiotic of group I with adrug of group II resulted in either synergism orantagonism apparently depending in part uponthe sensitivity of the test bacteria to the groupI drug. Thus, if the bacteria were highly sensitiveto the group I antibiotic, the addition of a groupII drug often interfered with the early bactericidalaction of the first drug (antagonism). On theother hand, if the organisms were relativelyresistant to the group I drug, in some instancesthe addition of a group II drug resulted in syner-gism, i.e., in a marked increase of early bac-tericidal action and the killing of greater numbersof bacteria than could be expected from simplesummation of single drug effects. Most of theearlier tests were carried out with bacteriaisolated from random clinical infections.

Since the relative susceptibility of micro-organisms to group I drugs appeared to influencethe observed end result of group I + group IIcombined antibiotic action, it was deemed im-portant to compare the behavior of bacterialvariants of graded resistance to group I drugswhen exposed to group I + group II combina-tions. The present paper reports such a study.

1 Supported in part by grants from the ResearchCommittee of the University of California Schoolof Medicine, from Burroughs Wellcome and Co.,and a research grant (E-214) from the NationalInstitutes of Health, Public Health Service.

Bacteria were selected which were highly sensi-tive to group I drugs and which were killed morerapidly when exposed to a group I drug alonethan when a group II drug was added. Resistantvariants were obtained then from these culturesand subjected to tests with antibiotics singly andin combination. It was thought possible thatalthough a given pair of group I + group IIantibiotics acted antagonistically against theorinally sensitive culture they might actsynergistically against the more resistant variantsderived from it.

MODS AND MATERALS

Definition. "Synergism" in vitro has been de-fined as the ability of two drug to produce amarked increase in the bactericidal rate withinthe first 24 hours of exposure as compared withthe rate with either drug alone and the killingof greater numbers of bacteria than could beexpected from simple summation of single drugeffects. The term "synergism" has been reservedfor combined action unequivocally in exes ofsimple algebraic summnation, and the term "addi-tion" has been used for other instances of posi-tive summation. "Antagonism" in vitro has beendefined as a marked decrease in the early bac-tericidal rate as compared with that of the moreactive single drug even though the combinationafter prolonged incubation may kill more or-ganisms than either drug alone. These definitionshave been chosen because they seem to correlatewell with results in the treatment of experimentalinfections in mice and with certain clinicalevidence (Jawets and Gunnison, 1952a,b).

Antibiotic.. Stock solutions in sterile buffered0.85 per cent salt solution were stored in therefrigerator, and final dilutions were prepared inbroth immediately before use. Commercialcrystalline preparations of potassium penicillinG and streptomycin sulfate were used. Bacitracin

150

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ANTIBIOTIC SYNERGISM AND ANTAGONISM

(lot no. B-480420) was supplied by Dr. L. Smith;terramycin hydrochloride (lot no. WB507019), byDr. G. L. Hobby; and chloramphenicol (lot no.134006), by Dr. G. Rieveschl.Media. Proteose no. 3 agar (Difco) and broth

of similar composition were used. For Strepto-coccUs pyogenes, one per cent sheep blood wasadded to the agar.

Bacteria. The following organisms and variantsderived from them were studied: Micrococcuspyogenes var. aueus, strain H; Micrococcuspyogenes var. albus, strain M, isolated from acase of subacute bacterial endocarditis, andstrain L, isolated from a case of osteomyeltis;Strepococcus pyogenes, strain C203; and Klebsiellapneumoniae, strain A-D.Metods. A sensitive strain was obtained from

each culture by repeated selection of singlecolonies from drug-free agar. Resistant variantswere selected by serial transfer in graduallyincreasing concentrations of antibiotic in brothor in agar, by inoculation of large numbers (10°)on various concentrations of antibiotic in agar,or by the gradient plate method (Szybalski andBryson, 1952). Single colonies were isolated froma plate with a given concentration of antibioticand reisolated 3 to 6 times on antibiotic containingagar, picking a single colony each time. Allstrains were preserved by freezing in broth con-taining the highest concentration of antibioticthat would permit growth at a normal rate andwere stored at -20 C. For the final tests, thefrozen cultures were inoculated on agar contain-ing antibiotics in the highest amount permittinggrowth at a normal rate and from this into 100 mlof drug-free broth to be used as the inoculum.

Plate counts of the inoculum were made onagar containing antibiotic as well as on drug-freeagar to determine the proportion of resistant cells.In all instances, at least 80 per cent of the cellswere resistant to the selected concentration ofantibiotic.For the tests proper, broth containing

antibiotics was inoculated with the 18 hourculture to give a concentration of 106'6 to 107Oorganisms per ml in a total volume of 15 ml.Samples of 0.5 ml were removed at intervalsduring incubation at 37 C, and the number ofviable bacteria was estimated by plate counts.Details of the methods and their interpretationhave been reported previously (Gunnison et al.,1950b).

RESULTS

The first organism studied, M. pyogenes var.albus, strain M, was killed rapidly by bacitracinin a concentration of 10 ,g per ml. A variantstrain was obtained which was resistant to thisconcentration of bacitracin. The results ofcomparative tests of the resistant and sensitivestrains exposed to bacitracin and terramycinalone and in combination are shown in figures 1and 2. With the sensitive strain, the addition ofterramycin (1 jug per ml) to bacitracin (10 ,ug perml) resulted in antagonism; but with the re-sistant variant, the two drugs in these concen-trations had an additive effect. This suggestedthat antagonism might be converted into syner-gism as the resistance of the culture to the groupI drug increased. However, when a larger amountof bacitracin (100 ,ug per ml) was used which wasbactericidal for the more resistant variant, theterramycin again reduced the rate of killing justas it had done with the sensitive strain at thelower level of bacitracin.

Six additional bacitracin resistant variants ofthis micrococcus with different degrees of re-sistance were tested. With those variants whichwere inhibited but not killed by a given dose ofbacitracin, the combination of that dose withterramycin had an additive effect; with thesewhich were rapidly killed by that dose of baci-tracin, the combination had an antagonisticeffect. Each variant was tested with bacteriostaticand bactericidal amounts of bacitracin alone andcombined with terramycin. In each instance,terramycin showed addition when mixed withbacteriostatic doses of bacitracin but interferedwith actively bactericidal doses.

In view of these results, the sensitive strainwas tested further with a wide range of concen-trations of bacitracin. Again it was found thatstatic doses of bacitracin in a narrow rangeshowed addition when combined with terra-mycin, whereas this group II drug interfered withbactericidal doses of bacitracin over a much widerrange (figures 1 and 2). There was no essentialdifference in the pattern of behavior of the sen-sitive strain and the resistant variants derivedfrom the same original culture. With the resistantforms, the levels of bacitracin concentrations atwhich addition or antagonism occurred werehigher as the resistance increased, but the samesequence of events occurred. In no instance did adose of bacitracin that was completely ineffective

19531 151

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

J. B. GUNNISON ET AL.

against the strain tested show a joint effect, when acting alone; but whenever actively bac-either additive or antagonistic. tericidal dos of penicillin were used, the terra-The next organism studied was M. pyogenes mycin interfered. These findings applied not

var. aureus, strain H, against which terramycin only to a sensitive strain but also to the variantswas synergistic with bacitracin but antagonistic at higher levels of penicillin concentration, ex-

MICROCOCCUS PYOGENES vAr ALBU8 (M)Bacitracin- Sensitive Strain

9 'N

s . ~~~~~~~~~~~~~~~~~~~~~i7 l Brbmycin.n 1

45 rramyn tcr.n 10* +Terrgy.cin I

Bacitracin IAug. +Terramycin 1lp.

3~~~~~~~~~~~~~~Bctam\ IOO 09.

2

o 4 7 X24 4 7 24

q ~~~BAC;tracin- Resistant VAriant

x~~~~~rrwyi Terr_ac;ytcin lOOu9 . WO

% \ \_rramycGin1iSmainOAV

t5 xBocifracin 10* + \eaitrcinloo.85t Ssein 1^ t \>~~~~rrmyinAI

M \ B~~~~~~~~~~~~~~~acitr&cin100A1.

24 0 4 74OURS AFTER INOCULATION

24

Figure 1. Combined effect of bacitracin and terramycin on strains of Micrococcu8 pyogenes var. albus,strain M, sensitive or resistant to bacitracin. With the bacitracin sensitive strain, bacitracin 10 pg +terramycin 1 pAg result in antagonism. With the bacitracin resistant variant, this same drug combina-tion results in addition. However, when the drugs are tested over wide ranges, antagonism and additionare demonstrated with both strains.

to penicillin (Jawetz et al., 1952). Eight variantsat different levels of penicillin resistance wereselected for the tests summarized in table 1 andfigures 2 and 3. Again, addition was shown in anarrow range of penicillin concentrations ap-proximately equal to its bacteriostatic range

cept those against which penicillin was altogetherineffective. In one instance, with a culture re-sistant to 15 units of penicillin per ml no additiverange at all could be demonstrated (figure 3).A phenomenon rarely encountered in previous

tests was noted with certain of these resistant

152 [VOL. 66

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ANTIBIOTIC SYNERGISM AND ANTAGONISM

variants but not with the sensitive strain;namely, antagonism of penicillin to bactericidaldoses of terramycin. With rapidly bactericidaldoses of both drugs, the interference was mutual;i.e., the mixture was less effective than eitherdrug alone (table 1). However, this observationwas the exception rather than the rule.

bacitracin resistant variant was obtained andcomparative tests made as shown in table 2 andfigure 2. Again, as with the cultures of micrococci,an increase in resistance of the culture did notchange an antagonistic relationship of the groupI and II drugs into a synergistic one but merelyraised the level of the doses of bacitracin at which

Units of Sacitracin. Penicillin or Straptomycinper ml. added to Tervarnycin ( I,ui per ml )

Reslstance ofStrain to Group lAntiboftic Employed

0.2 units/mi. [lACITRACIN*1O.units/ml [

40O un'its I0nl.[PENICILLIN

2.0 units /ml. [

0.001 units/m. [BACITRACINas units/nil. [

0.5 units/mlPENICILLiN200.0 units/ml. [

0.1 units /mli. [STREPTOM4YCIN

30.0 units /nal.t

QOO 0.01 0.6a 0 0

I 10 600 1000* 0 . I

M.pyogne& var albus ('M)

M.pyoenes var aureus (N)

*.pyoqents (Co20)

K.pneumonisa (A-D)

1]Ef fect equal Terramydsn alone. mE:ffect Greater than eitherLJHJDlPFE.EEriCE di.drug alone. ADDITION

E Effect 1te0 I+han Oroup I alone.* Effect equals 4roupl drug lone.ANTAOONISM B INDIFFERENCE

Figure S. Spectra of combined effect of increasing concentrations of bacitracin, penicillin, or strepto-mycin (group I drugs) plus a constant amount of terramycin (group II drug) on strains of varying re-sistance to the group I drugs. The resistance to the group I drug used is expressed as that concentra-tion of antibiotic incorporated in agar which permitted growth of at least 80 per cent of the cells ofthe test culture. (Spectra suggested by those of Marshall and Hrenoff (1937) for single disinfectants.)

Another culture of M. pyogenes var. albus,strain L, and penicillin resistant variants derivedfrom it were exposed to mixtures of penicillinand chloramphenicol with essentially the sameresults as those described above.

Streptococcw pyogenes, strain C203, was knownto be highly sensitive to both penicillin and baci-tracin. Terramycin had been shown to interferewith the activity of these group I drugs (Gunnisonet al., 1950a; Speck et al., 1951). A relatively

antagonism was demonstrable about 10-fold.Mutual antagonism occasionally resulted whenbactericidal concentrations of both drugs wereused. Similar findings were obtained with arelatively penicillin resistant strain.

Klebsiella pneumoniae, strain A-D, was studied,and variants resistant to terramycin, penicillin,or streptomycin were isolated. It had been es-tablished that terramycin interfered with theactivity of both penicillin and streptomycin

19531 153

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

J. B. GUNNISON ET AL.

TABLE 1

Combined effect in vitro of penicillin and terramycin on a penicillin 8ensitive strain and a penicillinresistant variant of Micrococcus pyogenes var. aureus, strain H

EJYICT OFPENICILLIN(snrrs/) | Penicillinalone | Penicil ( +g teXr)ramycin | Penicllin + terrnycin(uNITs/ML) Penicillin alone Pncli+termcn(100 Avainl)

Penicillin sensitive strain

0.01 inactive = terramycin alone = terramycin alone0.02 static addition - terramycin alone0.05 slowly cidal addition - terramycin alone

0.1-100 rapidly cidal antagonism to penicillin = penicillin alone

Penicillin resistant variant

1.0 inactive = terramycin alone = terramycin alone5.0 static addition addition20.0 slowly cidal addition antagonism to terramycin100.0 rapidly cidal antagonism to penicillin mutual antagonism

Static = growth inhibited for 24 hours.Slowly cidal - 99% killed in 24 hours.Rapidly cidal - 99.99% killed in less than 24 hours.Terramycin 1 jig/ml alone = bacteriostatic.Terramycin 100 pg/ml alone - rapidly cidal.

MICROCOCCUS PYOGENES var. AUREUS (H4)

Resistance ofsirnam to Penicillin

Units of PENICILLIN per mladded to TERRAMYCIN ( ,1qper nl)

0.01 0.1 1 10 1oo 1000I I I S S S

0.0Z Uni;s/ml. :

2.0 lVno+s/ml.

5.0 Unit%n/mnl

15 Units/mnl.

20 Units/rni.

40 Units/ni.

Ir-I_I_

I mEffect equals Terramycin axoneINDIFFERENCE

Key j Effect grea+er than eiher drug aton

Effect Iess than Penicillin aloneANTAGONISM

Figure 8. Spectra of combined effect of increasing concentrations of penicillin plus a constant amountof terramycin on strains of Micrococcus pyogenes var. aurew, strain H, of varying resistance to peni-cillin. The resistance to penicillin is expressed as that concentration incorporated in agar which per-mitted growth of at least 80 per cent of the cells of the test culture. One strain resistant to 15 unitsof penicillin per ml showed no additive effect.

[voL. 66154

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ANTIBIOTIC SYNERGISM AND ANTAGONISM

against this organism (Gunnison et al., 1950a;Jawetz et al., 1951a). With the terramycin re-sistant variant, however, this drug no longerinterfered with the action of streptomycin exceptfor a slight, transitory antagonism when large,somewhat bacteriostatic doses of termmycin wereused (100 ;&g per ml). This was not surprisingbecause ineffective amounts of group II drugsnever interfered with group I drugs. When ter-ramycin was combined with penicillin, the resultswere less clear-cut as this terramycin resistant

ml showed an additive effect in the presence ofteramycin, whereas with the resistant strain200 to 400 units were required to give this effect.

Difficulty was encountered in obtaining strep-tomycin resistant strains which were not alsostreptomycin dependent, but a variant whichwas not dependent was isolated finally by thegradient plate method. This strain differed inbehavior from the penicillin or bacitracin re-sistant variants of the organism tested thus farin that no antagonism could be demonstrated

TABLE 2Combined effect in vitro of bacitracin and terramycin on a bacitracin sen8itive and a bacitracin

resistant variant of Streptococcus pyogenes, strain CSO0

EFFECT OFBACITEACIN(uNITs/M) Bacitmcin alone Bacitracin + terramycin Bacitracin + terramycin

(1 pg/mi) (10 pg/mi)

Bacitracin sensitive strain

0.002 inactive = terramycin alone = terramycin alone0.005 slowly cidal addition antagonism to terramycin0.01 rapidly cidal antagonism to bacitracin mutual antagonism0.05 rapidly cidal = bacitracin alone = bacitracin alone

Bacitracin resistant variant

0.01 inactive = terramycin alone = terramycin alone0.02 static addition = terramycin alone0.05 slowly cidal addition - terramycin alone0.2 rapidly cidal antagonism to bacitracin mutual antagonism0.5 rapidly cidal = bacitracin alone not tested

Static - growth inhibited for 24 hours.Slowly cidal - 99% killed in 24 hours.Rapidly cidal - 99.99% killed in less than 24 hours.Terramycin 1 pg alone - slowly cidal.Terramycin 10 pg alone = rapidly cidal.

variant multiplied slowly and was thereforesomewhat less sensitive to penicillin alone.

Test with a sensitive strain showed, as withthe cocci, that static or slowly bactericidAamounts of penicillin or of streptomycin gave anadditive effect in a narrow zone when combinedwith terramycin or chloramphenicol even thoughthe group II drugs were antagonistic to activelybactericidal amounts. The results with a peni-cillin resistant variant (figure 2) follow the same"spectrum" of activity as with the sensitivestrain; i.e., a narrow range of additive actionand a wide range of antagonistic effect as theamount of penicillin were increased. With thesensitive strain, 0.5 to 1.0 units of penicillin per

(figure 2). Here, the lower doses of streptomycinwhen combined with terramycin showed theusual additive effect. Above this range, however,there was a sharp end point where the strepto-mycin alone was rapidly bactericidal. There wasno antagonism whatever between terramycinand any concentration of streptomycin. Althoughhere, too, antagonism was not converted intosynergism as resistance to the group I drug de-veloped, in this instance the antagonism wascompletely obliterated; whereas with strainsresistant to penicillin or bacitracin it merelyoccurred at higher drug levels.

In view of the results reported here, all of theoriginal cultures previously studied (Jawets

19531 155

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

J. B. GUNNISON ET AL.

et al., 1952) were retested over a wider range ofconcentrations to see whether similar patterns ofbehavior were shown. In particular it was neces-

sary to determine whether any of the instanceswhich had been reported as "synergism" weremerely examples of the additive effect of bacterio-static amounts of group I drugs when combinedwith group II drugs which might be antagonisticwhen higher bactericidal doses of the group Idrug were used. With all pairs of drugs previouslydesignated as antagonistic when acting on a

given organism, there was a narrow zone ofconcentrations of the group I antibiotic in whichthe effect was additive, whereas the group II drugalways interfered with larger doses. On the otherhand, with all pairs that previously had beendesignated as synergistic against a given or-

ganism there was an additive effect with all con-

centrations of group I drug over a wide range,including bactericidal amounts, and no inter-ference at any level. Furthermore, in thesesynergistic systems the combination was more

effective than 2 to 10 times the dose of either drugacting alone.

DISCUSSION

The aims and results of the work reported inthis paper are restated in the following questionsand answers:

(1) Can both synergism and antagonism bedemonstrated with a given pair of drugs actingon a single strain of bacteria, depending upon therelative proportion of drugs in the mixture? Inearlier papers in this series it was stated thatsynergism could not be converted into antago-nism nor vice versa within a given test system bysimply altering the proportion of drugs present.The quantitative data of the present work sup-

port this conclusion. An additional fact is broughtout, however. When the whole range of concen-

trations of an "antagonistic" pair of drugs isstudied, a narrow range often can be discoveredwithin which the drugs are additive in theiraction. With resistant bacterial variants, theadditive zone tends to be somewhat wider thanwith the original sensitive strains. This additiverange occurs when both drugs are present inbacteriostatic quantities. The magnitude of addi-tive action within this range of concentrations isoften small so that the death rate of bacteriaexposed to the combination of drugs usually

does not equal, and rarely exceeds, the death

rate obtained by doubling the concentration ofasingle drug. While this effect does not fulfill thecriteria for "synergism" used in our work, itwell might account for the reports by others ofboth antagonistic and synergistic effects of agiven drug pair against a single strain of micro-organisms (Lankford and Lacy, 1949; Spicer,1950; Bliss et al., 1952).The spectrum of combined action within a

single "antagonistic" test system therefore mayinclude the following consecutive zones as theconcentration of the active, group I drug is raisedprogressively; (a) effect equal to that of theinterfering drug, (b) additive effect (usually anarrow zone), (c) effect of more active agentdiminished by interfering drug (zone of antago-nism), and (d) effect equal to that of the moreactive agent alone (demonstrated only infre-quently).

(2) Is antibiotic antagonism a unilateralphenomenon or can any drug interfere with anyother drug? Earlier work had suggested thatantagonism was unilateral in that group II drugswere capable of interfering with group I drugs,provided the microorganism was sensitive to thelatter, but not vice versa. The additional studiesreported in this paper indicate that, in general,these conclusions are justified. Under specialcircumstances, however, when high doses ofgroup I drugs act upon bacteria relatively re-sistant to these drugs but for which a group IIdrug is rapidiy lethal, the group I drug mayantagonize the group II agent. The circum-stances under which this event has been observedsuggest that the mode of action may be distinctfrom that seen in the much more commonantagonism of a group II agent to a group I drug.In rare instances, mutual antagonism was

observed when moderately bactericidal doses ofboth drugs were employed. This was noted par-ticularly with a drug sensitive hemolytic strep-tococcus. With this same organism, mutualantagonism had been seen as a chance observa-tion in vivo (Jawetz and Speck, 1950) but laterhad been discounted because of its extremerarity.

In contrast to the ordinarily unilateral natureof antibiotic antagonism, synergism is a mutualphenomenon and has never been changed toantagonism by changing the relative concentra-tions of drugs. Only one of the members of asynergistic drug pair needs to exhibit a biological

156 [voL. 66

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ANTIBIOTIC SYNERGISM AND ANTAGONISM

effect in the concentration entering into thecombination (Jawetz, 1952). The in vitro studyof the dosage relationships in such pairs of drugsis limited by the rapid killing effect of highdrug concentrations, making the demonstrationof additive phenomena impractical.

(3) To what extent does knowledge of thegroup I resistance of a microorganism permitprediction of the effect of a combination ofgroup I + group II drugs? It had been suggestedearlier that with a group I sensitive microor-ganism antagonism might result; with a group Iresistant microorganism, synergism. The materialpresented in this paper indicates that such astatement is undoubtedly an oversimplification ofthe situation, though perhaps of some practicalvalidity. The first requirement for the occurrenceof antagonism, as defined in our studies, is therapid bactericidal action of one drug which canbe reduced by the addition of a second agent. Inthe present paper it has been demonstrated thateven with moderately resistant bacterial variants,antagonism could occur, provided that sufficientlyhigh concentrations of the group I drug wereemployed to result in rapid bactericidal action.Antagonism is most likely to occur with

organisms highly susceptible to the group Idrug employed and, conversely, synergism wouldbe most likely with organisms resistant to thatdrug. As shown in this paper. however, therelative degree of resistance as such is not thedetermining factor in combined action. The basicfeatures which determine whether the action of agroup I + group II combination on a micro-organism will result in indifference, antagonism,or synergism must be more essential and un-alterable cellular characteristics than the merelevel of resistance to the group I drug.

(4) The in vitro demonstration of antibioticantagonism has given rise to concern about itsclinical occurrence although the phenomenonundoubtedly is rare in the treatment of patients(Lepper and Dowling, 1951). Can experimentalevidence explain this apparent discrepancy?Antibiotic antagonism can be demonstratedreadily in the treatment of experimental infectionsof animals. However, the phenomenon occursonly under special circumstances which rarelyprevail in clinical practice. It has been pointedout repeatedly that antagonism is limited bytime and dose relationships (Jawetz et al.,1951a,b; Speck et al., 1951; Ahern et al., 1952;

Ercoli and Crminati, 1952; Lepper et al., 1952).An excess of either of the participating drugsfrequently overcomes antagonism in vivo (Specket al., 1951; Ahern et al., 1952), and this has beendemonstrated in vitro in some instances. With thehighly drug sensitive organisms, against whichantagonism is most likely, the drug excess ob-tained with ordinary doses in therapy must belarge so that demonstrable antagonism would notbe expected. With drug resistant bacteria, onthe other hand, it is unlikely that sufficiently high,bactericidal levels could be reached in the tissuesto permit antagonism. These points add topreviously summarized evidence in minimizingthe likelihood of observable antibiotic antago-nism in clinical situations.

SUMMARY

Bacterial variants of graded antibiotic re-sistance were tested for their behavior towardantibiotic combinations in order to investigatethe role played by resistance toward one agentin the response to a mixture of drugs. It wasfound that antibiotic antagonism with a certaindrug pair acting on an antibiotic sensitivebacterial strain was not converted to synergismwhen the same drugs acted on an antibiotic re-sistant variant. The type of combined action,whether synergism, indifference, or antagonism,was essentially stable within a given test systemalthough wide variations in drug concentrationwere necessary to demonstrate this action againstbacteria of varying resistance.When "antagonistic" drug pairs were subjected

to a detailed study of dosage relationships, anarrow additive zone was found often. This oc-curred in the range of bacteriostatic concentra-tions of both drugs and never had the magnitudeof true synergism. Synergistic test systems didnot contain areas of antagonism. The relation-ship of these results to earlier work is discussed,and factors limiting the occurrence of antibioticantagonism in animal and human infections aresummarized.

REFERENCES

AHERN, J. J., BURNELL, J. M., AND KIRBY, W.M. M. 1952 Lack of interference of chlor-amphenicol with penicillin in a hemolyticstreptococcal infection in mice. Proc. Soc.Exptl. Biol. Med., 79, 568-571.

BLISS, E. A., WABRTH, P. T., AND LONG, P. H.

1953] 157

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

J. B. GUNNISON ET AL.

1952 Studies of combinations of antibioticsin vitro and in experimental infections in mice.Bull. Johns Hopkins Hosp., 90, 149-169.

ERCOLI, N., AND CARMINATI, G. M. 1952 Anti-spirochetal interference between antibioticsand arsenoxide. Science, 116, 579-582.

GUNNISON, J. B., COLEMAN, V. R., AND JAWETZ,E. 1950a Interference of aureomycin andof terramycin with action of penicillin invitro. Proc. Soc. Exptl. Biol. Med., 75, 549-552.

GUNNISON, J. B., JAWETZ, E., AND COLEMAN,V. R. 1950b The effect of combinations ofantibiotics on enterococci in vitro. J. Lab.Clin. Med., 36, 900-911.

JAWETZ, E. 1952 Antibiotic synergism and an-

tagonism: A review of experimental evidence.Arch. Internal Med., 90, 301-309.

JAWETZ, E., AND GUNNISON, J. B. 1952a Studieson antibiotic synergism and antagonism: Ascheme of combined antibiotic action. Anti-biotics and Chemotherapy, 2, 243-248.

JAWETZ, E., AND GUNNISON, J. B. 1952b Anexperimental basis of combined antibioticaction. J. Am. Med. Assoc., 150, 693-695.

JAWETZ, E., AND SPECK, R. S. 1950 Joint actionof penicillin with chloramphenicol on an ex-

perimental streptococcal infection of mice.Proc. Soc. Exptl. Biol. Med., 74, 93-96.

JAWETZ, E., GUNNISON, J. B., AND SPECK, R. S.1951a Studies on antibiotic synergism andantagonism: The interference of aureomycin,chloramphenicol, and terramycin with theaction of streptomycin. Am. J. Med. Sci.,222, 404-412.

JAwETZ, E., GUNNISON, J. B., SPECK, R. S., ANDCOLEMAN, V. R. 1951b Studies on anti-biotic synergism and antagonism. The inter-ference of chloramphenicol with the actionof penicillin. Am. Med. Assoc. Arch. Inter-nal Med., 87, 349-359.

JAWETZ, E., GUNNISON, J. B., BRIFF, J. A., ANDCOLEMAN, V. R. 1952 Studies on antibioticsynergism and antagonism. Synergismamong seven antibiotics against various bac-teria in vitro. J. Bact., 64, 29-39.

LANKFORD, C. E., AND LACy, H. 1949 In vitroresponse of staphylococcus to aureomycin,streptomycin, and penicillin. Tex. Repts.Biol. Med., 7, 111-124.

LEPPER, M. H., AND DOWLING, H. F. 1951Treatment of pneumococcic meningitis withpenicillin compared with penicillin plus aureo-mycin: Studies including observations on anapparent antagonism between penicillin andaureomycin. Am. Med. Assoc. Arch. Inter-nal Med., 88, 489-494.

LEPPER, M. H., JACKSON, G. G., AND DOWLING,H. F. 1952 Effectiveness of treatment ofpneumococcic infection of mice with a com-bination of multiple doses of aureomycinand penicillin administered over a three-dayperiod. J. Lab. Clin. Med., 40, 820-821 (Ab-stract).

MARSHALL, M. S., AND HRENOFF, A. K. 1937Bacteriostasis. J. Infectious Diseases, 61,42-54.

SPECK, R. S., JAWETZ, E., AND GUNNISON, J. B.1951 Studies on antibiotic synergism andantagonism. The interference of aureomycinor terramycin with the action of penicillinin infections in mice. Am. Med. Assoc. Arch.Internal Med., 88, 168-174.

SPICER, S. 1950 Bacteriologic studies of thenewer antibiotics: Effect of combined drugson microorganisms. J. Lab. Clin. Med., 36,183-191.

SZYBALSKI, W., AND BRYSON, V. 1952 Geneticstudies on microbial cross resistance to toxicagents. I. Cross resistance of Escherichiacoli to fifteen antibiotics. J. Bact., 64, 489-499.

158 [voL. 66

on Decem

ber 11, 2020 by guesthttp://jb.asm

.org/D

ownloaded from