Vol. 146, No. 2JOURNAL OF BACTERIOLOGY, May 1981, p. 621-6310021-9193/81/050621-11$02.00/0

A Family of r-Detenninants in Streptomyces spp. ThatSpecifies Inducible Resistance to Macrolide, Lincosamide, and

Streptogramin Type B AntibioticsYUKIO FUJISAWAt AND BERNARD WEISBLUM*

Department ofPharmacology, University of Wisconsin Medical School, Madison, Wisconsin 53706

Received 11 December 1980/Accepted 8 March 1981

Inducible resistance to macrolide, lincosamide, and streptogramin type Bantibiotics in Streptomyces spp. comprises a family of diverse phenotypes inwhich characteristic subsets of the macrolide-lincosamide-streptogramin anti-biotics induce resistance mediated by mono- or dimethylation of adenine, or both,in 23S ribosomal ribonucleic acid. In these studies, diverse patterns of inductionspecificity in Streptomyces and associated ribosomal ribonucleic acid changes aredescribed. In Streptomyces fradiae NRRL 2702 erythromycin induced resistanceto vernamycin B, whereas in Streptomyces hygroscopicus IFO 12995, the reversewas found: vernamycin B induced resistance to erythromycin. In a Streptomycesviridochromogenes (NRRL 2860) model system studied in detail, tylosin inducedresistance to erythromycin associated with N6-monomethylation of23S ribosomalribonucleic acid, whereas in Staphylococcus aureus, erythromycin induced re-sistance to tylosin mediated by N6-dimethylation of adenine. Inducible macrolide-lincosamide-streptogramin resistance was found in S. fradiae NRRL 2702 and S.hygroscopicus IFO 12995, which synthesize the macrolides tylosin and marido-mycin, respectively, as well as in the lincosamide producer Streptomyces lincol-nensis NRRL 2936 and the streptogramin type B producer Streptomyces dias-taticus NRRL 2560. A wide range of different macrolides including chalcomycin,tylosin, and cirramycin induced resistance when tested in an appropriate system.Lincomycin was active as inducer in S. lincolnensis, the organism by which it isproduced, and streptogramin type B antibiotics induced resistance in S. fradiae,S. hygroscopicus, and the streptogramin type B producer S. diastaticus. Patternsof adenine methylation found included (i) lincomycin-induced monomethylationin S. lincolnensis (and constitutive monomethylation in a mutant selected withmaridomycin), (ii) concurrent equixnolar levels of adenine mono- plus dimethy-lation in S. hygroscopicus, (iii) monomethylation in S. fradiae (and dimethylationin a mutant selected with erythromycin), and (iv) adenine dimethylation in S.diastaticus induced by ostreogrycin B.

Studies of inducible MLS (macrolide, lincos-amide, and streptogramin) resistance in Staph-ylococcus aureus have focused on the require-ments for induction by erythromycin, which ini-tially appeared to be the only MLS antibiotic(excluding oleandomycin) with demonstrable in-ducing activity (21, 23). Underlying these studieshas been the question, What property distin-guishes erythromycin from other MLS antibiot-ics and endows it with optimal inducing activity?A step in the right direction was taken by Pestkaet al. (17), who examined 53 erythromycin ana-logs and noted a correlation between inducingactivity and ribosome binding activity. As the

t Present address: Central Research Division, TakedaChemical Industries, Ltd., Yodogawa-ku, Osaka 532, Japan.

number and diversity of MLS-resistant orga-nisms studied have increased, however, it hasbecome clear that other MLS antibiotics couldalso induce. Dixon and Lipinski (7), Hyder andStreitfeld (11), and Malke (15) have reportedthat both lincomycin and erythromycin can in-duce Streptococcus pyogenes, and Allen (1)showed that the lincosamide celesticetin (butnot lincomycin) could induce MLS resistance inS. aureus. Moreover, Tanaka and Weisblum (19)reported that a mutant of S. aureus selectedwith carbomycin was inducible by both carbo-mycin and lincomycin, but not by erythromycin.Additional divergence from the S. aureus pat-tern was suggested by studies ofMLS resistancein Streptomyces spp. reported by Graham andWeisblum (9), who noted the presence of N6-

621

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

622 FUJISAWA AND WEISBLUM

monomethyladenine in Streptomyces strainsthat synthesize macrolide antibiotics, presum-ably associated with resistance to the antibioticsproduced by these organism.

The present studies were therefore under-taken to determine the extent ofdivergence fromthe "classical" case of S. aureus. As we shall

show, inducible MLS resistance actually com-

prises a family of distinct phenotypes character-ized by specific MLS subsets active as inducersof mono- or dimethylation or both, dependingon the organism, and that S. aureus constitutesone of many special cases. In the present studywe (i) determine the requirements for inductionof adenine monomethylation and establish itsassociation with MLS resistance in a Strepto-myces viridochromogenes model system; (ii)characterize the diversity of induction patternsby testing several additional strains of Strepto-myces for induction by a set ofMLS antibiotics;and (iii) propose a model to account for thediversity of induction specificities in Strepto-myces spp. based on the nucleotide sequence ofthe control region for MLS resistance in an S.aureus determinant.

MATERIALS AND METHODSBacterial strains. S. viridochromogenes NRRL

2860, provided by J. C. Ensign, who has described theproperties of this organism in a review (8), was grown

in enriched medium containing (per liter) 5 g of yeastextract (Difco) and 5 g of tryptone (Difco). Strepto-myces hygroscopicus IFO 12995, a producer of themacrolide antibiotic maridomycin, was obtained fromthe collection of the Takeda Co. The remaining Strep-tomyces strains were NRRL stock cultures used in ourprevious studies.

Determination of resistance on solid medium.Mycelial suspensions of cells grown in enriched me-

dium were disrupted by sonic oscillation for 10 toobtain a homogeneous suspension for preparation ofpour plates by the agar overlay method. The disksused to determine resistance contained 20 jig of testantibiotic.

Induction studies. To test induction in liquid me-dium, a sonicated cell suspension containing 106 col-ony-forning units per ml suspended in growth mediumwas incubated under the conditions of time and in-ducer concentration described. For S. viridochromo-genes, the induction medium was sampled, and suita-ble dilutions were plated onto medium containing 50,ug of tylosin per ml to determine the concentration ofinduced colony-forming units in the medium, and ontoantibiotic-free medium to determine the total viablecolony-forming units.

Selection ofmutants. For selecting mutants, MLSantibiotics producing large clear inhibition zones were

used. The rationale behind this choice is that MLSantibiotics showing small inhibition zones probablyinduce in situ, and selection using such antibioticswould have a high probability of inducing the culturerather than selecting desired constitutively resistantmutants.

Altered methylation of 23S rRNA. 23S rRNAwas labeled with either [2-3H]adenine or [methyl-"4C]methionine in the presence or absence of inducer,as required. Inocula were grown in enriched mediumsupplemented with 1 g of K2HPO4 and 2 g of glucoseper liter. For labeling experiments, actively growingmycelia from 100 ml of enriched medium were col-lected by centrifugation and suspended in 100 ml ofM-9 medium containing (per liter): Na2HPO4, 6 g;KH2PO4, 3 g; NaCl, 0.5 g; NH4Cl, 1 g; 0.1 M MgSO4, 10ml; 0.01 M CaCl2, 10 ml; and 50% glucose, 10 ml. TheM-9 medium was additionally supplemented with 2%Casamino Acids and [3H]adenine (specific activity,15,000 Ci/mol; final concentration, 4 ytCi/ml). Cellswere labeled by incubation overnight (generally 14 to18 h), collected, washed with 0.15 M NaCl solution,and stored at -20°C until used. Cells were disruptedby sonic oscillation, and labeled 23S rRNA was puri-fied from the 50S ribosome subunits (obtained bysucrose density gradient centrifugation) by extractionwith phenol, followed by depurination and separationof the resultant purines by column chromatographyon Dowex-50X8 and paper chromatography in an iso-propanol-ammonia-water (85:1.3:15) system as previ-ously described (13). To label cells with [methyl-'4C]methionine, enriched medium diluted fivefold withwater and supplemented with methionine (specificactivity, 50 Ci/mol, 0.1 ,uCi/ml) was used. methyl-'4C-labeled 23S rRNA was digested with RNase A, andthe resultant digest was fractionated with DEAE pa-per by the two-dimensional method of Brownlee et al.(3).

RESULTSWe first characterized the requirements for

induction by using S. viridochromogenes, basedon the fact that this organism is not known toproduce any MLS antibiotics, a factor which ifpresent might tend to mask induction by exog-enously added inducers. The disk sensitivitymethod was used initially for determination ofinduced-resistance specificity profiles.Patterns of induction in S. viridochro-

mogens The ability of tylosin to induce resist-ance to erythromycin and to carbomycin (Fig.1A) is manifested by the distorted erythromycinand carbomycin inhibition zones on the sidefacing the tylosin disk; the distorted inhibitionzone results from the fact that cells induced onsolid medium by tylosin, which diffuses from thedisk, are capable of growing closer to the eryth-romycin or carbomycin disk, whereas the mini-mal inhibition zone surrounding the tylosin diskreflects the effectiveness of tylosin as inducer ofresistance to itself.We have shown previously that MLS anti-

biotics lacking inducing activity can be used toselect constitutively resistant S. aureus mutants(23). As shown in Fig. 1B, a mutant selected forresistance to carbomycin shows coresistance toall remaining members of the test panel ofMLSantibiotics. When wild-type inducible S. aureus

J. BACTERIOL.

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDUCIBLE MLS RESISTANCE IN STREPTOMYCES 623

S xviridochromooenes NRRL 286t:)

try Car TylChal Cir KitMatv VrnB Cl n

FIG. 1. Specificity of induction and constitutive MLS resistance. Sonicated suspensions of S. viridochro-mogenes NRRL 2860 and a constitutive mutant derived from this strain were plated by the agar overlaymethod as indicated and tested for their response to antibiotic disks. (A) Tylosin (Tyl) induces resistance toerythromycin (Ery) and carbomycin (Car). (B) A spontaneous constitutive mutant selected for resistance tocarbomycin (10 pg/ml) shows increased resistance to a set ofMLS test antibiotics, as indicated in the figure.Abbreviations: Chal, chalcomycin; Cir, cirramycin; Kit, kitasamycin; Mar, maridomycin; VrnB, vernamycinB; CIn, clindamycin. Disks contained 20 pg of the named antibiotic.

was tested similarly for induction, it was notedthat erythromycin induced resistance to tylosin,whereas neither carbomycin nor tylosin had de-monstrable inducing activity (19). We concludethat both S. viridochromogenes and S. aureusbehave in a formally similar manner when testedfor induction, but with differences in inducerspecificity that remain to be explained.Requirements for induction in S. virido-

chromogenes The process of induction in S.viridochromogenes has both a time and inducerconcentration dependence. To obtain reproduc-ible quantitative data, it was necessary to pre-pare a homogeneous cell suspension. Baltz (2)has shown that a 100-fold-increased level in thenumber of colony-forming units is obtained if amycelial suspension of Streptomyces fradiae isdisrupted by sonication for 5 to 10 s beforeplating, and we employed this method withoutmodification. Under the conditions used, opti-mal induction, measured in terms of the appear-ance of colony-forming units, requires exposurefor 1 h to a concentration of tylosin in the rangeof 0.3 to 0.6 ug/ml (Fig. 2).Altered methylation of 23S rRNA in re-

sistant cells. [methyl-"4C]methionine was usedas the source of label, followed by digestion ofthe labeled 23S rRNA sample with RNase A,two-dimensional fractionation of the resultantdigest, and autoradiography (Fig. 3). Two addi-tional spots in the rRNA from induced cells,indicated with arrows, were seen: one distinctlylabeled fragment in the lower part of Fig. 3A,and a second, partially resolved spot in the upperpart of this figure. We have not attempted tomximie the resolution of the larger, partiallyresolved fragment from its neighbors; however,

the results suggest the possibility of at least twoinducibly methylatable sites in 23S rRNA. Thebase compositions of these oligomers were notinvestigated further, but from our previous stud-ies in S. aureus (12) we were able to locate themethylated adenine residue in a similar type offragment.

Quantitative data are obtainable by labelingcells uniformly with [2-3H]adenine, which allowsus to determine the average number of adenineresidues that have been methylated. We assumethat all adenine residues in 23S rRNA are la-beled to the same average specific activity, andthat the relative amount of each derivatizedform of adenine present in 23S rRNA is reflectedin the relative amount of radioactivity presentin that component. Since 23S rRNA containsapproximately 800 adenine residues, one meth-ylated adenine per 23S rRNA would comprise0.12% of the total adenine fraction. Results ofsuch an assay (Fig. 4, summarzd in Table 1)indicate the presence of 0.13 and 0.17% mono-methyl adenine relative to total adenine for theinducible and constitutively resistant strains, re-spectively. The apparent discrepancy betweenthe results shown in Fig. 3 and 4 remains to bereconciled.Reduced erythromycin binding by resist-

ant cells. 50S subunits from uninduced, in-duced, and constitutively resistant S. virido-chromogenes were tested for their ability to binderythromycin. The extent ofassociation betweenerythromycin and 50S ribosome subunits wasmeasured by adsorption of the complex to nitro-cellulose membrane filters. Results of the assay(data not shown) were similar to those obtainedin our previous studies of erythromycin binding

VOL. 146, 1981

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

624 FUJISAWA ANDWEISBLUMB

L

4._-

0C._-

0

0>,

A

Q

4J._-

c

.E0

0.0

105

104

0 20 40 60 80 100 120

Induction time (min)

0.1 0.5 1.0 1.5 2.0

Inducer concentration (pg tylosin/ml)

FIG. 2. Time course and inducer concentration requirements for induction. (A) Time course of induction.Tylosin ( pg/ml) was added to a sonicated suspension of actively growing cells. Samples were withdrawnand plated on antibiotic-free medium and on medium containing 50 pg of tylosin per ml. (B) Concentrationdependence of induction. Cells were induced with tylosin at concentrations between 0.1 and 2.0 pg/ml. Afterincubation for 60 min, portions were withdrawn and plated as described for Fig. 2.

by 50S subunits from uninduced, induced, andconstitutively resistant S. aureus (23). We con-clude that tylosin-inducible MLS resistance inS. viridochromogenes formally resembles eryth-romycin-inducible MLS resistance in S. aureusat the ribosome level.Variety ofMLS resistance phenotypes in

Streptomyces spp. We next surveyed a groupof selected Streptomyces strains to determinewhether additional specificities of inductionmight be present. Of several possible tests forinduction, we concentrated our effort on two: (i)the disk assay to determine specificity of induc-tion, and (ii) the [3H]adenine methylation assay,to determine the altered adenine methylationphenotype and to obtain an estimate of thenumber of altered sites. Results of the methyl-ation assay are presented graphically and intabular form. The strains chosen produce MLSantibiotics; however, in the case of S. virido-chromogenes (as well as other examples cited),MIS resistance is not restricted in distributionto MLS producers and seems to occur in manyStreptomyces strains that were tested. A widerange of phenotypes was found in our survey.We therefore present the data below in a formthat illustrates the diversity encountered, ratherthan exhaustively documenting requirements forinduction in all the samples.

Streptomyces lineolnensis NRRL 2936.E tion of the response of S. lincolnensisto antibiotic disks revealed the distorted inhibi-tion zones characteristic of induction (Fig. 5A).Ciramycin induced resistance to carbomycinand (more distinctly) to maridomycin. A morepronounced flattening of the carbomycin inhi-bition zone in response to cirramycin can be seenin Fig. 5B. To determine whether a commonmechanism is responsible for resistance to theseveral MLS antibiotics to which S. lincolnensisappears sensitive, a mutant selected with mari-domycin was tested (Fig. 5B). The mutant se-lected showed increased resistance to the set oftest antibiotics.We checked for the presence of methyl ade-

nine in S. lincolnensis grown in the absence ofantibiotic and in the presence of an inducingconcentration of lincomycin, and in the consti-tutively resistant strain selected with marido-mycin. Results parallel to those obtained in S.virodochromogenes were observed (Fig. 6 andTable 1), suggesting that adenine methylationsimilarly mediates lincomycin resistance in S.lincolnensis.S frdiae NRRL 2702. Erythromycin in-

duced resistance to lankamycin and to verna-mycin B (Fig. 7A) in S. fradiae, although forreasons stated below we felt that it should have

J. BACTERIOL.

66

1 05

1 04

103 103

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDUCIBLE MLS RESISTANCE IN STREPTOMYCES 625VOL. 146, 1981

A Binduced control

FIG. 3. Fingerprint analysis of 23S rRNA labeledwith [methyl-14CJmethionine. methyl-14C_labeled 23SrRNA was prepared as described and digested withRNase A, and the resulting oligonucleotides, frac-tionated by the method described by Brownlee et al.(3), were located by autoradiography. The arrowsindicate the position ofnew oligonucleotides presentin the RNA from induced cells: a larger partiallyresolved oligomer and a clearly resolved smaller one.(A) 23S rRNA from induced cells; (B) 23S rRNA fromuninduced controls.

relatively weaker inducing activity compared tomost of the antibiotics in our test panel. Exam-ination of the response of S. fradiae to our testgroup of antibiotics (Fig. 7B) showed resistanceto most of the members of this group with theexception of erythromycin and vernamycin B, towhich it appeared to be sensitive. We interpretthe absence of a significant inhibition zone forseven of the nine antibiotics tested to indicate ahigh level of inducing ability and a relativelylower level for the remaining two antibioticstested, erythromycin and vernamycin B. Wetherefore also used erythromycin to select forconstitutively resistant mutants and obtained astrain with higher levels of resistance to botherythromycin and vernamycin B. The use ofvernamycin B for this selection might have beena more logical choice.Examination of the methylation pattern in

23S rRNA revealed the presence ofmonomethyladenine in the uninduced control, as in our pre-vious study (9), in which we also noted thepresence of monomethyl adenine in 23S rRNAfrom Streptomyces cirratus ATCC 21731, anorganism that produces the macrolide antibioticcirramycin. In studying S. fradiae further, how-ever, we made the unexpected observation thatdimethyl adenine was the predominant methyl-ated forn present in the mutant selected witherythromycin (Fig. 8 and Table 1). The apparentshift from mono- to dimnethylation raises thequestion of whether one or two enzymes areresponsible for the observed rRNA methylationpatterns. The smaller zone diameter seen in theerythromycin-resistant mutant is reminiscent ofthe partially constitutive mutants of S. aureusfound in earlier studies (23), and we interpretthis pattern of mutation to reflect a nucleotidechange that only partially destabilizes the r-de-terminant control region, as discussed in furtherdetail below.

S. hygroscopicus IFO 12995. In testing thespecificity of induction we found that either ofthe two streptogramin B antibiotics, vernamycinB and ostreogrycin B, induced resistance toerythromycin (Fig. 9A). This observation isnoteworthy since it represents the first time wehave seen induction by a member of the strep-togramin B family. Erythromycin was used inan attempt to select constitutively resistant mu-tants, and the strain obtained in this way becamecoresistant to the remaining members of the testgroup as shown in Fig. OB.Equimolar concentrations of mono- and di-

methyl adenine were found in 23S rRNA fromboth the uninduced control and the constitu-tively resistant mutant (Fig. 10 and Table 1).Whereas rRNA from uninduced S. hygroscopi-cus conforms to the pattem of methylationfound for other macrolide producers, namely,the presence of monomethyl adenine, the simul-taneous presence of dimethyl adenine providesan additional variant methylation phenotype,possibly due to the presence of two methylatingenzymes. As discussed below, we feel thatmacrolide-producing strains may synthesizetheir own inducers and that this may accountfor the observed presence ofmethylated adeninein macrolide-producing strains apparently un-related to induction by exogenously added in-ducer.Streptomyces diastaticus NRRL 2560. Cir-

ramycin clearly induced resistance in S. diasta-ticus to kitasamycin and maridomycin (Fig.11A). The same disk assay did not show induc-tion by a streptogramin B-type antibiotic; thisfinding might not be expected from the fact that

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

626 FUJISAWA AND WEISBLUM

Io6

o

ato4%- 1 O4

E103

2a 10

1 o

10 I I I I I I I I I I .1 I I I I I I I I I I I I j10 20 30 40 10 20 30 40 10 20 30 40

FRACTION NUMBERFIG. 4. Separation ofmethylated adenines from S. viridochromogenes 23S rRNA bypaper chromatography.

Pooled fractions from 23S rRNA, labeled with [3HJadenine and fractionated through the Dowex-50X8 step,containing adenine plus methylated adenines, were lyophilized an(I further fractionated bypaper chromatog-raphy (with added carrier adenineplus methylated adenines) using a solvent system consisting ofisopropanol-ammonia-water (85:1.3:15, vol/vol). The chromatogram was cut into 1-cm-wide strips, which were counteddirectly in a liquid scintillation counter. (A) Control, uninduced; (B) induced by growth in medium containing1 ug of tylosin per ml; (C) constitutively resistant cells selected with carbomycin and grown in the absence ofadded antibiotic.

TABLE 1. Methylated adenine residues in 23S rRNAacpm in monomethyl ade- . .d

Species cpm in adenine nine (% relative to ade- cpm i dimethyl adeninenine)

S. viridochromogenes 826,000 276 (0.03) 243 (0.03)Tylosin induced 554,000 740 (0.13) 283 (0.05)Carbomycin resistant 581,000 984 (0.17) 208 (0.04)

S. lincolnensis 380,000 211 (0.05) 198 (0.05)Lincomycin induced 867,000 642 (0.07) 247 (0.03)Maridomycin resistant 426,000 844 (0.20) 168 (0.04)

S. fradiae 232,000 522 (0.23) 195 (0.08)Erytbromycin resistant 579,000 386 (0.06) 949 (0.16)

S. hygroscopiwus 575,000 994 (0.17) 1367 (0.24)Erythromycin resistant 153,000 513 (0.34) 494 (0.32)aAdenine and its methylated derivatives were isolated from 23S rRNA as described in Fig. 4. Radioactivity

(counts per minute, cpm) in each of the methylated adenine fractions was determined, and the results weretabulated; 0.12% relative to adenine corresponds to 1 residue per 23S. Boldface data indicate experimentalvalues corresponding to one or more methylated adenine residues per 23S' rRNA.

S. diastaticus synthesizes the virginiamycincomplex, a group of streptogramin antibiotics.However, in view of a relatively small inhibitionzone surrounding the vernamycin B disk (Fig.llB), we inferred that vernamycin B might nev-ertheless have inducing activity, and we used itin the assay of induced methylation. The results(Fig. 12) suggest the apparent induction of a lowlevel of dimethyl adenine formation which re-maIns to be optimized. This observation is of

interest since it provides an additional exampleof induction in a producing organism by anantibiotic produced by that organism. Moreover,these results further suggest the existence ofinducing activity in a streptogramin B-type an-tibiotic.Other Streptomyces spp. By means of the

disk method we have examined additional Strep-tomyces strains specifically chosen because oftheir role in the synthesis of antibiotics that do

J. BACTERIOL.

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDUCIBLE MLS RESISTANCE IN STREPTOMYCES 627

A Sincol nensiXs NRRL 2936

r~~~~~~~~~~~~~~~~~~~~~~~.r

B S.lincolnensisl'NRRL 2936

Ery Car TylChal Cir KitMar VrnmS Cln

FIG. 5. Inducible resistance in S. lincolnensis: disk sensitivity test. (A) Induction ofresistance to carbomycinand maridomycin by cirramycin; (B) resistance toMLS antibiotics in the wild-type strain and in a constitutivemutant selected with maridomycin at 10 pg/ml Drug abbreviations as described for Fig. 1.

106 AB C

S1lincolnensis, control S.lincolnensis, induced S.lincolnensis, constitutive¢t> 1O 62L|[|||1p|

105io36 66

i2 m6A m6A If6A

10 m aI I I I I I I I I I I I I I a I a I I I10 20 30 40 10 20 30 40 10 20 30 40

FRACTION NUMBERFIG. 6. Inducible resistance in S. lincolnensis: methylation of23S rRNA. (A) Wild type, uninduced controls;

(B) wild type, induced by growth in medium containing 10 pg oflincomycin per ml; (C) constitutively resistantmutant selected with maridomycin at 10 pg/ml.

not belong to the MLS families. Emination ofStreptomyces rimosus NRRL 2234 and Strepto-myces griseus NRRL B-1965, the organismsused to synthesize oxytetracycline and strepto-mycin, respectively, reveals that chalcomycininduces resistance to maridomycin in both thesestrains (data not shown).

DISCUSSIONIn the Streptomyces spp. we see inducible

MLS resistance in its most diverse and generalform, and our main task is to explain the molec-ular basis for this diversity. Our findings prompta redefinition of the MLS phenotype in terms ofthe fact that macrolides, lincosamides, and

streptogramin B-type antibiotics can all induceresistance, and that the induced ribosomal alter-ation involves either mono- or dimethylation orboth. This broader definition of the inducibleMIS resistance phenotype subsumes all the var-iant phenotypes found to date.The studies reported above were first under-

taken to determine whether a causal relationcould be shown between adenine monometh-ylation and MLS resistance. The use of an in-ducible strain, S. viridochromogenes, one notknown to produceMLS antibiotics, provided themodel system that enabled us to do this withequal facility as in S. aureus. To prove, however,that an organisim does not synthesize compounds

VOL. 146, 1981

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

628 FUJISAWA AND WEISBLUM

FIG. 7. Inducible resistance in S. fradiae: disk sensitivity test. (A) Induction of resistance to lankamycin(Lank) and vernamycin B by erythromycin; (B) resistance to MLS antibiotics in the wild-type strain and in aconstitutive mutant selected with erythromycin at 10 pg/mL. Drug abbreviations as described for Fig. 1.

with inducing activity may be difficult, so thatthe use of a strain not known to produce anyMLS antibiotics is the optimal choice at thistime.

S. viridochromogenes provided us with anadditional example of induction specificity inwhich erythromycin was not the most potentinducer. It was previously shown that erythro-mycin was not the sole inducer in staphylococciand streptococci (1, 7, 11, 15, 19). Althougherythromycin has maximal inducing activity incertain cases, this did not appear to us to be thecase universally after we examined the Strepto-myces spp.

In the course of our studies ofMLS resistancewe have been faced with the problem of explain-ing the basis for induction specificity. In studiesof a series of 53 erythromycin analogs Pestka etal. (17) first noted a correlation between ribo-some binding and inducing activity, and theyinferred that formation of a ribosome-erythro-mycin complex constituted a significant step ininduction. More recent studies of MLS resist-ance by Shivakumar et al. (18) and from ourlaboratory (10) have suggested that the processof induction requires sensitive ribosomes andthat formation of a complex between erythro-mycin and a sensitive ribosome plays a key rolein the induction process.On the basis of the DNA sequence of the

control region for inducible resistance specifiedby planid pE194, including sequences of 11constitutive mutants, we proposed an explicittranslational attenuation model for induction ofMLS resistance (10) which we believe is alsoutilized in regulating MLS resistance in Strep-tomyces spp. The significant feature ofthe modelof induction pertinent to our present work is theexperimental fact that members belonging to thethree classes of antibiotics to which methylated

1 0 20 30 40 10 20 30 40

FRACTION NUMBER

FIG. 8. Inducible resistance in S. fradiae: methyl-ation of23S rRNA. (A) Wild type, uninduced control;(B) constitutive mutant selected with erythromycin at10 pg/ml.

ribosomes are resistant can, in principle, showinducing activity. Whereas our model helps tounderstand why any MIS antibiotic can haveinducing activity, we can only speculate why itis that only a limited number can be shown toinduce in any given organism.

Sorting out the variables that determine in-duction specificity will depend on distinguishingamong: (i) primary effects on the inductionmechanism, (ii) secondary effects such as anti-biotic transport or conversion to either active orinactive substances, and (iii) incidental varia-tions or differences in ribosome structure involv-ing primary sequence differences or secondarysequence modifications of rRNA or ribosomalproteins. A system which may provide usefulinformation involves transformation of Bacillussubtilis with plsmids from different sources andcomparison of the induction specificities of thetwo plasmids in the new host cell with the re-

J. BACTERIOL.

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDUCIBLE MLS RESISTANCE IN STREPTOMYCES 629

A S.fradiae NRRL 2702 B S.fradiae NJRRL 2702

Ery Car TylCha1 Cir Kit

I.. lr

FIG. 9. Inducible resistance in S. hygroscopicus: disk sensitivity test. (A) Induction of resistance toerythromycin by vernamycin B and ostreogrycin; (B) resistance to MLS antibiotics in the wild type and in aconstitutive mutant selected with erythromycin at 10 pg/ml. Drug abbreviations as for Fig. 1.

A B

10 20 30 40 10 20 30 40

FRACTION NUMBER

FIG. 10. Inducible resistance in S. hygroscopicus:methylation of23S rRNA. (A) Wild type, control; (B)constitutive mutant selected with erythromycin at 10pg/ml.

spective specificities found in the original host.In one such comparison we noted that pAM77from Streptococcus sanguis and pE194 from S.aureus, when introduced into B. subtilis bytransformation, gave phenotypes (disk assay)that resemble those seen in the original hosts(data not shown); both erythromycin and linco-mycin induced in the former, whereas erythro-mycin alone induced in the latter. This suggeststhat the induction specificity is inherent in theDNA and not in the host cell background. Interms of the explicit model for control of MLSresistance we have postulated (10), the aminoacid composition of the control region peptideand its variability make a major contribution tothe specificity of induction, mediated by thedifferent relative potencies of MLS antibioticsin inhibiting incorporation of specific aminoacids. For example, in experiments performedusing polyuridylate:polycytidylate-directed invitro polypeptide synthesis as a model test sys-

tem, we reported that leucine incorporation wasstrongly inhibited by lincomycin and minimallyinhibited by erythromycin (4).

In our experience with Streptomyces spp. weobserved that many false negatives, but no falsepositives, were obtained in testing for inductionby the disk method. In terms of the model forinduction that we have proposed, formation ofa complex between the ribosome and the induc-ing antibiotic comprises the critical step in in-duction; i.e., the sensitive ribosome is the mac-romolecular sensor that determines the presenceof an inducing stimulus in the environment. Ithas been shown previously that 50S subunitinhibitors generally can interact competitivelyin their binding to the ribosome (reviewed byVazquez [20]). This would apply to the MLSantibiotics, whose binding to the ribosome ap-pears to share overlapping specificities.

In our previous studies of S. aureus we havenoted that inducible cells exposed to a nonin-ducing MLS antibiotic can respond in at leastfour ways. The types of mutants obtained in-clude: (i) "true constitutives," coresistant to allMLS antibiotics tested, by the disk assay (23);(ii) 'partial constitutives," comprising a variedgroup of phenotypes involving reduced zone sizefor some test antibiotics, turbid inhibition zonesfor others, and no change for still others (23);(iii) "altered specificity of induction," as in thecase of the carbomycin- and lincomycin-induc-ible mutant (19); and (iv) "high plasmid copynumber," as in the case of pE194 mutants se-

lected for resistance to tylosin in a B. subtilisbackground (22). The first three of these cate-gories seem to have counterparts in Strepto-myces spp.

In several instances, induction tested by thedisk method produced only minimal distortionof the respective inhibition zones, which hasraised some question about the validity of our

lolS.hygroscopicus, control S.hygroscopicus, constitutive

105 _ ___Adenine

104 Aden_ine

10 _ m6Am2lmA ~~~~~~~~~6A6j

10 I I I I I I I I

VOL. 146, 1981

c0

,to.I4uCL

iCL1;x

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

630 FUJISAWA AND WEISBLUM

p cu~1,_ 1 :^ - '-. S N RR s. .X .i ...

FIG. 11. Inducible resistance in S. diastaticus: disk sensitivity test. (A) Induction of resistance to kitasa-mycin and maridomycin by cirramycin; (B) resistance toMLS antibiotics in the wild type and in a constitutivemutant selected with maridomycin at 10 pg/ml. Drug abbreviations as for Fig. 1.

Adenine

10~~~~~~~~~~~~~~Adenine

410 4z r.6

1o2 ~~6 2A

10 I J II I I

10 20 30 40 1 0 20 30 40

FRACTION NUMBER

FIG. 12. Inducible resistance in S. diastatiwus:methykation of 23S rRNA. (A) Wild type, uninducedcontrol; (B) wild type, induced by growth in mediumcontaining 10 pg of ostreogrycin B per ml.

interpretation. The ability ofoneMLS antibioticto induce resistance to another requires not onlythat the inducer have optimal specificity, butalso that the inducing member of the test pairbind more strongly to the ribosome than theantibiotic to which resistance is induced, or elsethe net effect would be dominated by noninduc-tion characteristic of the more strongly bindingantibiotic.

Naturally occurring mechanisms of resistanceto MLS antibiotics unrelated to rRNA methyl-ation might produce spurious effects in studiesof induction. Such mechanis include phos-phorylation ofclindamycin by Streptomyces coe-licolor as described by Coats (5) and hydrolyticcleavage of pristinamycin I (a streptogramintype B antibiotic) by S. aureus as described byLeGoffic et al. (14). We have not yet checked forthe extent to which these alternative mecha-nisms may operate in the producing strains, andwhether they act instead of or in addition tomethylation, if at all, is not yet known.

In the case of the macrolide producers, wehave found the presence of methylated adeninein cells grown under noninducing conditions (i.e.,in the absence of exogenously added inducer)and apparent inducibility of these strains whentested by the disk method. We attempt to rec-oncile these apparently inconsistent findings bysuggesting the possibility that such strains maysynthesize endogenous inducers. This must bethe case for S. lincolnensis, since we have shownthat the ribosomes lack methylated adenine un-til exposed to exogenously added lincomycin.We propose that under conditions used for an-tibiotic production, lincomycin or some congenersuch as celesticetin, produced initially at subin-hibitory levels, may serve to mediate the meta-bolic switch that occurs as part of the commit-ment to antibiotic production. The differencebetween S. lincolnensis and the macrolide pro-ducers may reflect differences in regulation ofthe synthesis of endogenous inducers.

Induction of adenine monomethylation in S.lincolnensis by exogenously added lincomycinemerges as one of the most significant findingsin this study. A similar situation may obtain inthe streptogramin B producer S. diastaticus.These observations suggest, more generally, apossible link between production of other, if notall, MLS antibiotics and rRNA methylation.This can be tested; if the development of resist-ance to MIS antibiotics by rRNA methylationconstitutes a limiting step in the synthesis ofthese antibiotics, we would expect to find moreefficient production of MLS antibiotics by con-stitutively resistant derivatives of inducible par-ent producing strains.

ACKNOWLEDGMEIWe are grateful to J. C. Enaign for providing us with the S.

viridochromogenes strain used in these studies and for many

J. BACTERIOL.

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDUCIBLE MLS RESISTANCE IN STREPTOMYCES 631

helpful discussions. S. Silver provided much appreciated edi-torial assistance far above and beyond the call of duty.

This work was supported by research grants from TheNational Science Foundation (PCM-7719390) as well as bysupport from the Research Foundations of The Upjohn Co.and Eli Lilly and Co.

LITERATURE CITED1. Allen, N. E. 1977. Macrolide resistance in Staphylococcus

aureus: inducers of macrolide resistance. Antimicrob.Agents Chemother. 11:669-674.

2. Baltz, R. H. 1978. Genetic recombination in Streptomycesfradiae by protoplast fusion and cell regeneration. J.Gen. Microbiol. 107:93-102.

3. Brownlee, G. G., F. Sanger, and B. G. BarreiL 1968.The sequence of 5S ribosomal ribonucleic acid. J. Mol.Biol. 34:379-412.

4. Chang, F. N., and B. Weisblum. 1967. The specificity oflincomycin binding to ribosomes. Biochemistry 6:836-843.

5. Coats, J. H. 1975. Clindamycin phosphotransferase.Methods Enzymol. 43:755-759.

6. Dixon, J., and A. Lipinski. 1972. Resistance of group Abeta-hemolytic streptococci to lincomycin and eryth-romycin. Antimicrob. Agents Chemother. 1:333-339.

7. Dixon, J., and A. Lipinski. 1974. Infections with beta-hemolytic streptococcus resistant to lincomycin anderythromycin and observations on zonal-pattern resist-ance to lincomycin. J. Infect. Dis. 130:351-356.

8. Ensign, J. C. 1978. Formation, properties, and germina-tion of actinomycete spores. Annu. Rev. Microbiol. 32:185-219.

9. Graham, M. Y., and B. Weisblum. 1978. Methylationof adenine in 23S ribosomal RNA of macrolide-produc-ing streptomycetes. J. Bacteriol. 137:1464-1467.

10. Horinouchi, S., and B. Weisblum. 1980. Posttranscrip-tional modification ofmRNA conformation: mechanismthat regulates erythromycin-induced resistance. Proc.Natl. Acad. Sci. U.S.A. 77:7079-7083.

11. Hyder, S. L, and M. MW Streitfeld. 1973. Inducible andconstitutive resistance to macrolide antibiotics and lin-comycin in clinically isolated strains of Streptococcuspyogenes. Antimicrob. Agents Chemother. 4:327-331.

12. Lai, C.-J., J. E. Dahlberg, and B. Weisblum. 1973.Structure of an inducibly methylatable nucleotide se-

quence in 23S ribosomal ribonucleic acid from eryth-romycin-resistant Staphylococcus aureus. Biochemis-try 12:457-460.

13. Lai, C.-J., and B. Weisbhum. 1971. Altered methylationof ribosomal ribonucleic acid in an erythromycin resist-ant strain of Staphylococcus aureus. Proc. Natl. Acad.Sci. U.S.A. 68:856-860.

14. LeGoffic, F., MW. L Capmau, J. Abbe, C. Cerceau, A.Dublanchet, and J. Duval. 1977. Plasmid-mediatedpristinamycin resistance: PH 1A, a prisitinamycin 1Ahydrolase. Ann. Microbiol. Inst. Pasteur 128:471-474.

15. Malke, H. 1978. Zonal-pattem resistance to lincomycin inStreptococcuspyogenes: genetic and physical studies, p.142-145. In D. Schlessinger (ed.), Microbiology-1978.American Society for Microbiology, Washington, D.C.

16. Malke, H., H. Jacob, and K. Storl. 1976. Characteriza-tion of the antibiotic resistance plasmid ERLI fromStreptococcuspyogenes. Mol. Gen. Genet. 144:333-338.

17. Pestka, S., R. Vince, R. LeMahieu, F. Weiss, L. Fern,and J. Unowsky. 1976. Induction of erythromycinresistance in Staphylococcus aureus by erythromycinderivatives. Antimicrob. Agents Chemother. 9:128-130.

18. Shivakumar, A. G., J. Hahn, G. Grandi, Y. Kozlov,and D. Dubnau. 1980. Posttranscriptional regulationof an erythromycin protein specified by plasmid pE194.Proc. Natl. Acad. Sci. U.S.A. 77:3903-3907.

19. Tanaka, T., and B. Weisblum. 1974. A mutant of Staph-ylococcus aureus with lincomycin and carbomycin in-ducible resistance to erythromycin. Antimicrob. AgentsChemother. 5:538-Mi.

20. Vazquez, D. 1979. Inhibitors of protein biosynthesis.Springer-Verlag, Berlin.

21. Weisblum, B. 1975. Altered methylation of ribosomalribonucleic acid in erythromycin-resistant Staphylococ-cus aureus, p. 199-206. In D. Schlesinger (ed.), Micro-biology-1974, American Society for Microbiology,Washington, D.C.

22. Weisblum, B., M. Graham, T. Gryczan, and D. Dub-nau. 1979. Plasmid copy number control: isolation andcharacterization of high-copy-number mutants ofpE194. J. Bacteriol. 137:635-643.

23. Weisblum, B., C. Siddhlkol, C.-J. Lai, and V. De-mohn. 1971. Erythromycin-inducible resistance inStaphylococcus aureus: requirements for induction. J.Bacteriol. 106:8356847.

VOL. 146, 1981

on January 24, 2020 by guesthttp://jb.asm

.org/D

ownloaded from