Staphylococcus aureus - Antimicrobial Agents and Chemotherapy

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Nov. 1994, P. 2590-2598 0066-4804/94/$04.00+0 Copyright © 1994, American Society for Microbiology

Vol. 38, No. 11

Reassessment of the Number of Auxiliary Genes Essential for Expression of High-Level Methicillin Resistance in

Staphylococcus aureus HERMINIA

DE

LENCASTRE1'2 AND ALEXANDER TOMASZl*

Laboratory of Microbiology, The Rockefeller University, New York, New York 10021,1 and Instituto de Tecnologia Quimica e Biol6gica, Universidade Nova de Lisboa, Oeiras, Portugal2 Received 21 March 1994/Accepted 16 August 1994

A new transposon library constructed in the background of the highly and homogeneously methicillinresistant Staphylococcus aureus strain COL yielded 70 independent insertional mutants with reduced levels of antibiotic resistance. Restriction analysis with HindlIl, EcoRV, EcoRI, and PstI and then Southern hybridization with probes for the transposon and for thefemA-femB gene demonstrated that 41 of the 70 Tn551 mutants carried distinct and novel, as yet undescribed insertion sites, all of which were outside of the mecA gene and were also outside the already-characterized auxiliary genes femA, femB, femC, and femD. All previously described Tn551 mutations of this type were in genes located either on SmaI fragment A or SmaI fragment I. In contrast, inserts of the new library were located in 7 of the 16 SmaI chromosomal fragments, fragments A, B, C, D, E, F, and I. In all of the mutants, expression of methicillin resistance became heterogeneous, and the MIC for the majority of cells was reduced (1.5 to 200 ,ug ml-') from the homogeneous methicillin MIC (1,600 ,ug ml-) of the parental cells. Although identification of the exact number of genes inactivated through the new set of transposon inserts will require cloning and sequencing, a rough estimate of this number from mapping data suggests a minimum of at least 10 to 12 new genetic determinants, all of which are needed together with femA,femB,femC, and femD for the optimal expression of methicillin resistance. Methicillin-resistant Staphylococcus aureus (MRSA) strains the mecA gene, which is the central genetic element of this beta-lactam antibiotic resistance mechanism, and yet for individual MRSA isolates there is a tremendous range in resistance levels, from barely above that for susceptible staphylococci to MICs of several thousand micrograms per milliliter. This variation in level of resistance is not always related to the expression of mecA, since bacteria for which MICs were as diverse as 1.5 pug ml-' to 1.5 mg ml-' were shown to contain comparable amounts of the mecA gene product, penicillinbinding protein 2A (PBP 2A) (13). This surprising fact has led to the suggestion that additional factors (factor X) besides the mecA gene product are also essential for the optimal expression of methicillin resistance (13). A transposon mutant (Q2003, femzA) located outside mecA and causing a reduced methicillin resistance phenotype (and behaving as the postulated factor X) was described in 1983 (1, 2). Two additional genes (femB and femC) were subsequently identified on the large SmaI-A fragment, and a fourth gene (femD) was identified on the SmaI fragment I of the chromosome of S. aureus COL (4, 9). In the experiments that led to the identification of the fem genes, Tn551 mutants with severely reduced methicillin resistance were selected. In order to identify additional genetic determinants that may also be involved in the expression of resistance, we increased the resolution of the screen so that mutants with relatively lower reductions in resistance levels were also detected. The results of these studies suggest that the number of such auxiliary genes (23) capable of profoundly influencing the methicillin resistance phenotype is even larger than was thought until now. carry

Corresponding author. Mailing address: Laboratory of MicrobiolThe Rockefeller University, 1230 York Avenue, New York, NY 10021. Phone: (212) 327-8277. Fax: (212) 327-8688. *

ogy,

2590

MATERIALS AND METHODS Strains and growth conditions. The S. aureus reference strains used in the study are listed in Table 1 and were grown as described before (19). Transposon mutants and their backcrosses were also grown as described previously (19). Selection of Tn551 mutants. The transposition experiment and the selection of mutants were carried out by a modification of a previously described method (20). The parent strain COL harboring the thermosensitive plasmid pRN3208 carrying TnS51 with the erythromycin resistance determinant [referred to as COL(pRN3208)] was grown overnight at 30°C and was then diluted and plated at different cell concentrations on tryptic soy agar (TSA; Difco) containing erythromycin (20 ,ug ml-'). The plates were incubated at 30 and 43°C for 48 h. The approximate frequency of erythromycin-resistant colonies that grew at the nonpermissive temperature (number of colonies at 43°C/number of colonies at 30°C) was 2.1 x 10-5. The erythromycin-resistant colonies grown at 43°C were checked for cadmium resistance (0.25 mM CdNO3). Since a large proportion of these colonies was cadmium resistant (indicating the retention of the whole plasmid), the colonies that were identified as being affected by methicillin resistance in the two first screens described below were grown on TSA at 43°C for 16 h to eliminate the plasmid and were tested again for cadmium resistance and methicillin resistance. Only colonies that were erythromycin resistant and cadmium susceptible were kept for further study. Screening of mutants with decreased methicillin resistance. The erythromycin-resistant (Eryr) colonies were tested sequentially by three screens. In the first screen, all Eryr colonies were streaked onto TSA plates with erythromycin (20 ,ug ml-1) and different concentrations of methicillin (0, 25, 50, and 400 ,ug ml-1); the same Eryr colonies were also streaked onto TSA plates with cadmium (0.25 mM CdNO3). We identified as putative mutants all

NEW Tn551 MUTANTS OF MRSA

VOL. 38, 1994

2591

TABLE 1. Reference strains used in the study Strain

Relevant genotype

Relevant phenotype

PAP expression class

Parental COL COL(pRN3208) Mioo

Homogeneous Mcr Mcr Emr Cdr COL with pRN3208 [Rep(Ts)] Laboratory step mutant of strain 27s Homogeneous

TnS5l mutants of COL Mapped in SmaI-A BB403 RUSAIII-8 RUSA10 RUSAIII-3 RUSAlH1 RUSA2OF RUSAIII-2 RUSAII-1 RUSA208b RUSA330 (1H)C RUSA101

COL f12003 (femA::TnSSl) COL f1560 (femA::Tn551) COL Q1552 (femB::TnS5l) COL Q1553 (femB::TnS5l) COL Q1554 (femB::TnS5l) COL Q1555 (femB::Tn5Sl) COL Q1556 (femB::TnS5l) COL Q557 (femB::Tn551) COL Q1561 (femC::TnS5l) COL Q12005 (femC::TnSSl) COL Q1559 (femE::TnS5l)

Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous Emr heterogeneous

Mapped in SmaI-F, RUSA4

COL £1551 (mecA::Tn551)

Emr reduced Mcr

RUSA4 type

Mapped in SmaI-I, RUSA12F

COL £1558 (femD::TnS5l)

Emr heterogeneous Mcr

2

4 4

Mcr Mcr Mcr Mcr Mcr Mcr Mcr Mcr Mcr Mcr Mcr

2-3 2 1 1 1 1 1 1 1 1 2-3

( gmc-)

Origin or reference

1,600 800 25

RU collection 15 9, 25

25a 12 3 3 3 3 3 3 3 3 12-25 3 12

1, 2 2, 9, 15 K. Murakami; 9 J. Komblum; 9 J. Kornblum; 9 J. Kornblum; 9 2, 9, 15 2, 9, 15 18 4, 15 9 9, 16, 17 4, 15

a The MIC for COL derivatives with the mutation Q12003 is greater than the one for the original mutant in the BB270 background (1). Ten other mutants (QI562 to f1571) have the same restriction pattern as strain RUSA208 (Q561) (18). C This strain was obtained by transduction of the insert from mutant 1H (QI2005) (15) into strain COL from the RU collection.

b

colonies that failed to grow on plates with any of the methicillin concentrations. Strain COL(pRN3208) was used as the positive control. Strains RUSA10 and RUSA12F (Table 1) were also used in all steps of the work as additional internal controls for the concentration of methicillin either on the plates or in the methicillin discs. In the second screen, the colonies whose methicillin resistance was identified as being affected in the first screen were grown overnight in 5 ml of tryptic soy broth (TSB) with 10 jig of erythromycin ml-1 and were then tested on TSA plates containing a 1-mg methicillin disc. The controls referred to above were tested under the same conditions. After 24 h of incubation at 37°C, the halos of inhibition were measured. The use of discs containing 1 mg of methicillin have been extensively used as a reliable and easy test for the comparison of strains (lOa). Inhibition zones, determined after 24 h of incubation, were highly reproducible but were dependent on the cell concentration in the culture. For the following control strains used, the diameters of the zones of inhibition were as indicated at both 30 and 37°C: RUSA10, 52 mm; RUSA12F, 40 mm; COL(pRN3208), 15 mm; COL, 11 mm. Strains that gave inhibition halos larger than the ones of COL(pRN3208) and that were shown to be cadmium resistant were streaked onto TSA for single colonies and were incubated at 43°C for 24 to 48 h to eliminate the plasmid pRN3208. Four to eight individual colonies were picked and tested for erythromycin and cadmium resistance. The cadmium-susceptible colonies were reanalyzed with 1-mg methicillin discs. If the halos of inhibition for these strains, which were cured of their plasmids, remained larger than the ones of COL(pRN3208), the strains were studied by a third screen. Before this step all of the putative mutants were streaked for single colonies on TSA plates with erythromycin (20 jig ml-'), and the plates were incubated at 37°C for 24 to 48 h. A single colony was then inoculated into 5

ml of TSB with 10 ,ug of erythromycin ml-' to prepare frozen stocks. In the third screen for mutants with decreased levels of methicillin resistance, candidate colonies were analyzed by population analysis profiles (PAPs) (10) (see below). Four controls were used: RUSA10, RUSA12F, COL(pRN3208), and COL. Colonies that showed a PAP different from the one of COL(pRN3208) were kept for further study. For all 70 mutants selected in this manner the MIC of methicillin was significantly lower than the MIC of methicillin for the parent strain, and all but 2 of the 70 mutants also showed heterogeneous methicillin resistance phenotypes. The presence of plasmid pRN3208 caused a moderate reduction in the MIC of methicillin for COL, from 1,600 to 800 ,ug ml-' (calculated from the PAPs). For this reason the resistance phenotypes of all mutants were also compared in the second and third screens with that of COL(pRN3208). PAPs. PAPs were determined as described previously (10) on plates containing methicillin (0, 1.5, 3, 6, 12.5, 25, 50, 100, 200, 400, and 800 ,ug ml-l). Mutants were assigned to different expression classes according to a previous classification (24), with the addition of an intermediate class, class 2-3. The critical parameter for the assignment of a PAP to a particular class was the MIC for the majority of the cells, as follows: class 1, 1.5 to 3 jig ml-1; class 2, 6 to 12 jig ml-; class 2-3, 25 to 50 jig ml-'; class 3, 100 to 200 ,ug ml-'; class 4, -400 jig ml-l; RUSA4 type (mecA::Tn551), 3 ,ug ml-'. Strains of classes 1 to 3 are heterogeneous, whereas strains of class 4 or of the RUSA4 type are homogeneous. For a review of the PAPs of representative TnS5l mutants, see Table 1 and reference 9. Transduction crosses and analysis of transductants. Transduction crosses in which the newly isolated mutants were used as donors and the homogeneously resistant strain COL was used as the recipient were performed with phage 80 alpha as

2592

DE LENCASTRE AND TOMASZ

A 485.0 43eM

>

388.0

>

339 29O 24235

>

194.0 145J 97.0 48S

--.

>

*~ ~ ~ ~ ~ ~ ~~

> -

>

B 485.0 436.5 388.0

339.5 292.0

> > > > >

242.5

194.0

>

145.5

>

97.0

>

48.5

>

FIG. 1. Localization of TnS5S inserts in chromosomal SmaI fragby Southern hybridization. Chromosomal DNA prepared from a selected group of TnSSl mutants was treated with SmaI endonuclease, and the fragments were separated by pulsed-field gel electrophoresis (A). After transfer of the DNA, membranes were hybridized with a radiolabeled TnS5S DNA probe as described in Materials and Methods (B). Numbers to the left of the gels are in kilobases. Letters to the right of the gel in panel B identify the particular SmaI fragments. ments

described previously (19). The primary selection was for the transposon marker erythromycin (10 ,ug ml-'). A total of 50 to 100 Eryr transductants from each transduction were streaked onto TSA plates containing erythromycin (10 jig ml-') and erythromycin (10 ,ig ml-') plus methicillin (400 ,ug ml-') by using as positive and negative controls COL(pRN3208) and the particular donor strain used in the cross, respectively. All erythromycin-resistant transductants were found to show reduced levels of methicillin resistance as well. From each cross, eight transductants were further tested for decreased levels of methicillin resistance by the 1-mg methicillin disc method, and two or three transductants were also tested by PAP analysis for their antibiotic resistance phenotypes. The location of the insert was tested by comparing the HindIlI hybridization patterns of transductants and their donors. Conventional and pulsed-field gel electrophoresis. Preparation of chromosomal DNA for conventional and pulsed-field

ANTiMICROB. AGENTS CHEMOTHER.

gel electrophoresis was performed as described previously (8). Restriction digestions with SmaI, EcoRI, EcoRV, PstI, HindIII, and XbaI were carried out according to the manufacturer's recommendations. Conventional gel electrophoresis in 1% agarose was carried out in ix TAE buffer (21) for 16 h at 5 V cm For pulsed-field gel electrophoresis, the gels were prepared with 1.1% agarose (SeaKem LE; FMC Bioproducts) in 0.5x TBE buffer as described previously (8). The gels were run in an LKB 2015 Pulsaphor System (Pharmacia) or in a Chef-DR II apparatus (Bio-Rad). The running conditions for SmaI restriction digests in the two systems were as described previously (8). In most cases, when restriction with other enzymes generated chromosomal fragments larger than 15 kb (difficult to separate by conventional gel electrophoresis), the digests were run in a Chef-DR II apparatus for 23 h at 14°C in 0.5 x TBE buffer. DNA transfer. For blotting of conventional and pulsed-field gel electrophoresis gels to nitrocellulose or nylon membranes, a vacuum blotting apparatus was used (Vaccu-blot; Pharmacia/ LKB) as described in previously published protocols (8). Preparation of DNA probes and hybridization. The whole plasmid pRT1 (which contains an internal fragment of the transposon TnS51) was used as a probe. pRT1 contains a 4-kb HpaI-XbaI fragment from the transposon TnSSl cloned in the SmaI site of plasmid pGEM.1 (Promega) (16a). To ascertain the location of insertions in the femA-femB region, plasmids pBBB31 and pBBB13, containing the 2.2-kb femA fragment containing EcoRV and a 10.5-kb femA and femB fragment containing PstI, respectively, were used through the courtesy of Brigitte Berger-Bachi (2). A standard methodology was followed for 3"P labeling of the probes by nick translation, prehybridization, and hybridization (21). The hybridization was carried out at 42°C in 50% formaldehyde. Nick-translated plasmid DNA was denatured and added to the hybridization mixtures without the separation of unincorporated nucleotides. When the membranes were rehybridized, the previous probe was removed by boiling in 0.1% sodium dodecyl sulfate for 10 min. Physical characterization of the mutants with EcoRI, PstI, EcoRV, and HindHI by probing with the Tn551 probe. The enzymes EcoRI, PstI, and EcoRV have no restriction sites in TnS5, whereas the enzyme HindIII has two recognition sites (22), both of which are included in the probe used (for the physical map of TnSSl, see reference 12). The sizes of the DNA fragments generated after restriction with EcoRI, PstI, and EcoRV and hybridization with the TnSSl probe represent the sum of the chromosomal fragment size, in which the transposon is integrated with the size of TnSSI (5.2 kb). In Tables 2 to 5 the molecular sizes of the fragments are given after subtraction of 5.2 kb. Hybridization of the DNAs restricted with HindIII generates three bands; one corresponds to the internal TnS51-HindIII fragment, and there are two others (one includes the 1.0-kb HindIII-TnS51 right junction and the other includes the 3.0-kb HindIII-TnS51 left junction). As with the other enzymes the results presented in Tables 2 to 5 indicate the size of the fragments in which the insertions were located (i.e., sum of the three hybridization bands minus 5.2 kb). HindIlI was found to be the most useful enzyme for ascertaining the identity of two different insertions and was used to identify the number of different insertion sites in the new transposon library. RESULTS Isolation of the Tn5S1 mutants with reduced methicillin resistance. A total of 1,012 erythromycin-resistant colonies

2593

NEW TnSSl MUTANTS OF MRSA

VOL. 38, 1994

TABLE 2. New TnS5l mutants with mutations mapping in EcoRI fragment a Strain


Size of restriction fragment (kb)

Ql no.

PstI

EcoRV

HindIll

MIC

(.g

Backcross

ml-1 )

RUSA251 RUSA148 RUSA270 RUSA291 RUSA217a

COL f1573 COL f1574 COL f1575 COL Q1576

a

= a= a= a= a=

10.5 10.5 10.5 10.5 10.5

1.2 1.6 4.3 4.0 4.3

5.4 6.8 2.2 7.3 a

1 2-3 2-3 2-3 2-3

3 25 25 25 25

RUSA251 RUSA148 RUSA270

RUSA101 RUSA321 RUSA252 RUSA279 RUSA289 RUSA301

COL fQ559 COL Q1577 COL Q1579 COL £1580 COL £1581 COL £1582

a a a , a a

12.4 12.4 12.4 12.4 12.4 12.4

2.5 2.5 0.5 2.5 7.8 5.9

2.9 2.9 7.3 2.4 10.1 11.6

2-3 2-3 3 3 3 2-3

25 25 100 50 100 25

RUSA101 RUSA321 RUSA252 RUSA279

a

,

= = = = = =

only two HindIII fragments were visible after hybridization with the TnSSI probe; the fragment containing the TnS51

obtained by transposition of TnS51 into the chromosome of the highly and homogeneously resistant strain COL (methicillin MIC, 1600 ,ug ml-') were screened for reduced levels of methicillin resistance. In 70 colonies, a decrease in methicillin resistance was observed, in comparison with the levels of methicillin resistance in the parent strains COL or COL(pRN3208). Moreover, in all but two of the mutants the Tn551 insertion also changed the phenotypic expression of resistance from homogeneous to heterogeneous. A wide range of reductions in the methicillin MICs for the 70 mutants was found, from strains for which the MIC was as low as 1.5 jig ml-' to strains for which the MIC was as high as 200 ,ug ml-1 (see data in Tables 2 through 5). The locations of the TnS51 inserts relative to those of the already characterized femA, femB, femC, andfemD genes were determined by physical mapping in order to identify how many of the mutants had novel insertion sites and, possibly, new genes. Location of the TnSSI inserts in the staphylococcal chromosome. The TnS5l inserts of the 70 independently selected mutants mapped to 7 of the 16 fragments obtained by restriction of the S. aureus chromosome of strain COL with SmaI. The distribution of the inserts among the fragments was as follows: SmaI-A, 48; SmaI-B, 4; SmaI-C, 1; SmaI-D, 1; SmaI-E, 2; SmaI-F, 3, and SmaI-I, 11. Figures 1A and B show the locations of representative mutations on the seven SmaI fragments. Physical characterization of the Tn551 mutants: localization of inserts on the various SmaI chromosomal fragments. The DNAs of the mutants were digested with the appropriate restriction endonuclease, separated by gel electrophoresis, and hybridized with a TnS51-specific probe (pRT1). In some cases, after removing the TnS5l probe, the same DNAs were hybridized with afemA probe (pBBB31) or with afemA-femB probe

right junction is missing.

kb, which includes the fragment size (6.2 kb) of femC (Table 3). Mutations in fragment EcoRI-a. The 11 mutants whose inserts map in the 40-kb EcoRI-a fragment were further analyzed by PstI, EcoRV, and HindIII restriction as follows. PstI restriction and then hybridization with the TnSSl probe showed that in five mutants ((1573 to Q1576 and mutant RUSA217a) the hybridizing band had a size of 15.7 kb, while in six mutants (Q559, Q577, and Q1579 to Q1582) the size of this band was 17.6 kb. After removing the TnS5S probe, the same gels were hybridized with plasmid pBBB31 containing the 2.2-kb EcoRV fragment, where the femA gene is located. The five mutants with insertions in their 15.7-kb PstI fragments showed afemA-hybridizing band of the same size (15.7 kb) as the one that hybridized with the TnS51 probe, indicating (after subtracting the 5.2-kb size of TnS51) that in these mutants the

1

4

2 3

5

6 7 8

9

10 11

kb 4.3

|

~~~~~~~~~~~~~~~~4.0 2.6 2.2

(pBBB13).

Mutations in fragment SmaI-A. The previously studied

1.6

TnSSl mutations in the femA, femB, and femC genes map in SmaI fragment A (4). In order to compare the 48 new mutants with the previously characterized ones, their DNAs were restricted with EcoRI, an enzyme that has no restriction sites in TnSSl (22). We found that the 48 mutants could be divided into two groups according to their restrictions with EcoRI. The first group includes 11 mutants (Table 2) whose inserts map in the largest EcoRI fragment of approximately 40 kb (EcoRI-a), where the femA and femB genes are also located (2); the second group includes 37 mutants with inserts located in EcoRI fragments of decreasing sizes ranging from 14.3 to 1.2

1.2

FIG. 2. Southern blots of EcoRV restriction digests of mutants located in EcoRI-a probed with the 10.5-kb PstI fragment from plasmid pBBB13 covering the femA-femB region (2). Lanes: 1 and 7, bacteriophage lambda plus HindIII; 2, strain BB403 (12003; femA::TnS5l); 3, strain RUSA251 (£573); 4, strain RUSA148 (£574); 5, strain RUSA270 (£575); 6, strain RUSA217a; 8, strain RUSA291 (£576); 9, strain RUSA252 (£579); 10, strain RUSA101 (£559); 11, strain RUSA279

(£580).

2594


ANTiMICROB. AGENTS CHEMOTHER.

TABLE 3. New Tn551 mutants with mutations mapping in SmaI-A but not EcoRI-a Size of restriction fragment (kb)

Strain

RUSA208 RUSA178 RUSA247 RUSA112 RUSA114 RUSA158 RUSA176 RUSA303 RUSA219 RUSA254 RUSA296 RUSA164 RUSA305 RUSA182 RUSA237 RUSA188 RUSA233 RUSA260 RUSA190 RUSA319 RUSA152 RUSA239 RUSA130 RUSA172 RUSA317 RUSA264 RUSA256

EcoRV

PstI


2.2 2.2 2.2 8.5 8.5 8.5 8.5 1.3 2.2 5.5 5.5 NDb 4.0 2.3 17 17 2.9 2.6 16 4.7 2.8 1.5 1.4 6.1 2.5 10.0

12.4 12.6 12.6 11.7 11.7 11.7 11.7 12.6 11.5 12.2 12.6 12.2 13.1 5.7 9.7 15.3 20.6 6.8 14.9 6.8 7.5 6.8 7.4 6.5 7.5 10.1 ND

1 2-3 2 2-3 2-3 2 2 3 2-3 3 2-3 2 3 2-3 2 3 3 2 3 2-3 3 2-3 2 2-3 2 2-3 3

fl no.

COL fQ511 COL f1591 COL Q1596 COL Q1583 COL Q1584 COL £1585 COL Q1586 COL Q1590 COL Q1595 COL Q701 COL 1707 COL Q708 COL (1587 COL (1588 COL (1589 COL (1592 COL (1593 COL Q594 COL (1598 COL Q1599 COL 1700 COL (703 COL (704 COL (705 COL Q706

a

HindIlI

EcoRI

1.7 1.7 1.7 2.3 2.3 2.3 2.3 10.8 6.6 __a 12.8 1.0 1.9 5.7 3.0 2.8 7.0 1.7 2.6 5.8 6.2 6.0 5.2 5.6 5.8 10.35

6.2 7.1 6.5 14 14 14 14 8.3 6.4 6.1 4.2 2.4 1.2 11 11 9.0 7.1 7.1 6.4 5.0 4.5 4.2 3.7 3.7 3.7 2.9

c

4.Od

, only two HindIII fragments were visible after hybridization with the ND, not determined. c-, two insertions. d Doublet. b

c

TnSSI probe; the fragment containing the Tn5Sl

insertions must be in the 10.5-kb PstI fragment (PstI-alpha) known to contain the femA-femB genes (2). In contrast, with the other six mutants, in which the TnS51-hybridizing band had a size of 17.6 kb, pBBB31 still hybridized with a 10.5-kb fragment. Therefore, in this group of six mutants the TnS5l insert is outside of the femA-femB region in a 12.4-kb PstI

fragment (PstI-beta) (Table 2). These conclusions were confirmed by restriction of the same DNAs with EcoRV and hybridization with the Tn551 probe, which also allowed the assignment of the insertions to the EcoRV fragments, as indicated in Table 2. After removing the TnS5l probe, the gels were hybridized with another probe (pBBB13) carrying a 10.5-kb PstI fragment covering the entire femAl-femB region, including their flanking regions (2). In the six mutants in which the insertions were located in PstI-beta, the hybridization with this probe generated six bands of 4.3, 4.0, 2.5, 2.2, 1.6, and 1.2 kb (Fig. 2), as was expected from previously published results (4). In the five mutants whose inserts mapped in PstI-alpha, there was a loss of one EcoRV fragment which was replaced by another one, the molecular size of which corresponded to that of the lost fragment plus the size of TnS5l (5.2 kb). For instance, in mutant RUSA251 ((Q573) the 1.2-kb band was replaced by a 6.4-kb band, in RUSA148 (Q574) the 1.6-kb band was replaced by a 6.8-kb band, in RUSA291 (Q576) the 4.0-kb band was replaced by a 9.2-kb band, and in mutants RUSA270 (Q575) and RUSA217a the 4.3-kb band was replaced by a 9.5-kb band (Fig. 2). Next, these five mutants were further characterized by HindIII restriction. The insert in mutant RUSA251 (Q573) was mapped to the 1.2-kb EcoRV fragment known to containfemB

MIC

(g

-

3 12-25 12 12-25 12-25 12 12 100 12-25 50 25 12 50 12-25 12 100 50 12 50 25 100 25 6 25 6 25 50

Backcross

RUSA208


RUSA188

right junction is missing.

(2). HindIII restriction of this mutant showed that its insertion is located in the same distal part of femB as those for a group of six previously described mutants (RUSAII-1, RUSAIII-2, RUSAIII-3, RUSA10, RUSAiH1, and RUSA20F) (see references in Table 1, and for a review, see reference 9). On the other hand, mutants RUSA270 (omega 575) and RUSA217a, whose mutations are located in the 4.3-kb EcoRV fragment (Table 2), were shown to be distinct by HindlIl restriction (Table 2). Strain RUSA217a was not analyzed further since after HindlIl restriction and hybridization with Tn551 it produced only two (instead of the expected three) bands. The mutations in the last two of the PstI-alpha mutants were located in 1.6-kb (Q574) and 4.0-kb (Q576) EcoRV fragments, respectively, which were clearly different from the fragments in which femA (2.2 kb) or femB (1.2 kb) reside (Fig. 2). Thus, among the five new PstI-alpha mutants, at least three (Q574, (1575, and M576) represent mutants with new inserts. From the six mutants with insertions located in PstI-beta, two (RUSA101 and RUSA321) produced identical restriction fragment length polymorphisms (RFLPs) which differed from the RFLPs of the four other mutants (RUSA252, RUSA279, RUSA289, and RUSA301), each one of which was unique to the strain (Table 2). Thus, the six new PstI-beta mutants represent mutants with five new insertion sites. Mutations in fragment SmaI-A located outside EcoRI-a. The chromosomes of 37 of the 48 new mutants had insertions outside of the EcoRI fragment a in fragments of decreasing sizes, ranging from 14 to 1.2 kb; these included the EcoRI fragment of 6.2 kb in which the femC locus resides (Table 3). From these 37 mutants, 11 ((561 to (1571) were already


VOL. 38, 1994

2595

TABLE 4. New TnS51 mutants with mutations mapping in SmaI fragments B, C, D, E, and F PAP

Size of restriction fragment (kb) MIC noeeprssoncrs Restrlctlonk no. Backcross expression (g ml-') Xba EcoRV Hindll EcoRI

Restriction enzyme and strain

SmaI-B RUSA281 RUSA262 RUSA235 RUSA196

COL Q709 COL Q710 COL fQ711 COL Q712

5.0 5.2 7.3 3.0

9.6 2.1 10.8 10.8

SmaI-C, RUSA223

COL Q713

5.7

18

SmaI-D, RUSA162

COL Q714

2.0

SmaI-E RUSA206 RUSA277

COL Q715 COL Q716

SmaI-F RUSA138 RUSA311 RUSA221

COL Q717

a

ND, not determined.

b_, Only two HindlIl

COL Q719

1.12 16.6 2.96 2.96

NDa ND ND ND

1 2 3 2-3

5.1

ND

3

3.4

4.1

ND

2-3

12-25

3.8 5.8

7.1 19.2

1.0 1.8

ND ND

3 2-3

200 12-25

4.8 b 2.8

ND ND ND

18 18 7.1

4.6 7.2 8.4

RUSA4 type RUSA4 type 3

3 12 100 12-25

RUSA281 RUSA261 RUSA235

100

3 0.75 100

RUSA138

fragments were visible after hybridization with the TnS51 probe; the fragment containing the Tn551 right junction is missing.

described as the RUSA208 insertional cluster (18). Mutants in this cluster share common RFLP patterns with mutant 1H (15), the insert of which was subsequently named fQ2005 and was used to define femC (4). Reexamination of our previous data (18) indicated that the mutations in all RUSA208 insertional cluster mutants and mutant 1H (represented by strain RUSA330 in Table 1) are located in identical PstI, EcoRI, EcoRV, and HindIII fragments of 12.4, 6.2, 2.2, and 1.7 kb, respectively. Two additional new mutants (RUSA178 and RUSA247) also generated similar RFLPs. Two mutants, RUSA256 (double insert) and RUSA254 (one HindIII band was missing), were not analyzed further. Of the remaining 22 mutants (i.e., 37 minus 15; fQ583 to f1590, fQ592 to fQ595, f1598 to f1599, Q700 to Q1701, and Q703 to N708), none appeared to have a mutation in the femC locus because of the incompatibilities of restriction patterns (Table 3). Four of these mutants (RUSA112, RUSA114, RUSA158, and RUSA176) showed identical RFLPs with each one of the four restriction enzymes, while the remaining 18 mutants each had unique RFLPs, thus generating a total of 19 new and distinct insertion sites in this particular group of mutants. In summary, from the 48 new mutants with insertions in SmaI-A, a total of 27 new insertion sites can provisionally be defined as follows: 8 in EcoRI-a, outside the femA and femB genes, and 19 in fragments different from EcoRI fragment a and outside the femC locus. Mutations in fragments SmaI-B, SmaI-C, SmaI-D, and SmaI-E. Prior to the present work no TnSSJ insertions affecting methicillin resistance were assigned to SmaI fragments B, C, D, and E. There were four mutations on SmaI-B, one mutation each on SmaI-C and SmaI-D, and two mutations on SmaI-E. Each one of the mutants with these mutations showed unique restriction patterns with EcoRI, EcoRV, and HindIll, defining eight new and distinct insertion sites (fQ709 to M716)

(Table 4). Mutations in fragment SmaI-F. In a previously isolated mutant, RUSA4 (17), the insert was localized within the mecA gene (16) and was mapped in the SmaI-F fragment of strain COL (9). Among the mutants in the new TnS5I library we isolated

three new mutants with inserts in the SmaI-F fragment: RUSA138, RUSA311, and RUSA221. Whereas RUSA138 and RUSA311 have PAPs very similar to that of RUSA4 (MIC, 3 ,ug ml-'), without subpopulations of more resistant colonies, strain RUSA221 had a typical heterogeneous resistance phenotype (class 3). By restriction analysis it was possible to confirm that the insertion sites in RUSA138 and RUSA311 lie in mecA, whereas the insertion site in RUSA221 probably lies outside mecA in a new auxiliary gene that has not yet been characterized (Table 4). A more definitive localization of the insert in RUSA221 must wait until hybridization with the mecA-specific DNA probe is performed. Mutations in fragment SmaI-I. The insert in a previously isolated mutant, RUSA12F (15), was mapped to SmaI fragment I and was used to define a new locus, referred to asfemD (4). In our current study 11 new inserts (Q1720 to Q726, Q1728 to Q1730 and mutant RUSA168) were mapped to SmaI-I (Table 5). Of these 11 mutants, the inserts in 7 (Q1720 to Q1726) were located in the largest HindIlI fragment (34 kb) and in the same EcoRI (10.3 kb) and EcoRV (7.5 kb) fragments; 6 of the inserts (the RUSA315 cluster) were also on a common PstI fragment (5.8 kb) and shared identical RFLP patterns after HindIII restriction. These 6 mutants differed from the previously described mutant 12F (femD) only in their RFLP patterns with HindIII. These mutations may indeed lie in femD, but at an insertion site different from that in 12F. The mutation in the seventh mutant (RUSA266; Q1726) was located in a different PstI fragment (1.8 kb) and had a different RFLP pattern with HindIII that also differed from the pattern of the RUSA315 cluster. Of the remaining four new mutants (RUSA192, RUSA122, RUSA150, and RUSA168), RUSA192 had distinct RFLP patterns with PstI, EcoRI and EcoRV, and HindIlI. The mutations in the three other mutants were located on identical PstI and EcoRI fragments but different EcoRV fragments, and RUSA150 and RUSA168, but not RUSA122, also shared a common EcoRV fragment. RUSA168 was not assigned an omega number (it had only two Hindlll fragments). The 11 new mutants with mutations in SmaI-I define five

2596


ANTIMICROB. AGENTS CHEMOTHER.

TABLE 5. Tn551 mutants with mutations mapping in the SmaI-I fragment Size of restriction fragment (kb)

Strain

fl no.

HindlIl

RUSA12F RUSA315 RUSA243 RUSA184A RUSA299 RUSA313 RUSA18F RUSA266

COL £1558 COL Q720 COL Q721 COL Q722 COL Q723 COL Q724 COL Q725 COL Q726


COL Q728 COL Q729 COL Q730

a

,

34 34 34 34 34 34 34 34

_a 5.6 1.9 1.25


EcoRI

EcoRV

PstI

10.3 10.3 10.3 10.3 10.3 10.3 10.3 10.3

7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5

5.8 5.8 5.8 5.8 5.8 5.8 5.8 1.8

2 2 2 2 2 2 2

9.8 9.8 9.8 4.2

0.8 0.8 1.4 6.5

10.8 10.8 10.8 12.8

2-3 3 3 3

only two HindIII fragments were visible after hybridization with the

MIC

(g m-) 6 6 12 12 12 12 6

Backcross

RUSA12F RUSA315 RUSA243

RUSA18F RUSA266

12-25 50 50 50

TnSSl probe; the fragment containing the Tn5Sl right junction is missing.

new insertion sites: the 315 cluster, RUSA266, RUSA192, RUSA122, and RUSA150. In summary, of the 70 new mutants with decreased methicillin resistance, the number of novel insertion sites is 43, of which only 2 are in mecA. The 41 novel insertion sites outside mecA were distributed as follows: 27 in SmaI-A, 4 in SmaI-B, 1 in SmaI-C, 1 in SmaI-D, 2 in SmaI-E, 1 in SmaI-F, and 5 in SmaI-I. Transduction crosses and analysis of transductants. A number of the auxiliary mutations isolated were transduced back into the parental strain COL. All backcrosses analyzed both by PAP analysis and by Hindlll restriction analysis, as described in Materials and Methods, are listed in Tables 2 to 5. In all of these 21 crosses 100% cotransduction of the Eryr marker with reduced methicillin resistance was obtained. There was no evidence of reversion to resistance or transpo-

sition. Effects of Tn551 mutation on bacterial physiology. Several of the auxiliary mutants that were analyzed showed clearly reduced growth rates compared with that of the parental bacterium. This was also true when the Tn551 mutation was crossed out of the MRSA genetic background into a laboratory-derived methicillin-resistant step mutant, strain M100 (25) (data not shown). The effect of the Tn551 insertion on the phenotypic expression of antibiotic resistance traits other than resistance against beta-lactam antibiotics was tested with mutant RUSA10 by using susceptibilities to D-cycloserine, tetracycline, ofloxacin, and gentamicin. In contrast to the massive reduction in the methicillin resistance level and the appearance of a heterogeneous phenotype, the mutation in RUSA10 did not change the susceptibilities (MICs) of the parental strain COL to tetracycline (200 ,ug ml-'), ofloxacin (1.5 ,ug ml-'), gentamicin (0.75 ,ug ml-'), or D-cycloserine (50 ,ug ml-1), and the expression of resistance to these antibiotics remained homogeneous (Fig. 3). DISCUSSION While all MRSA isolates carry the central genetic element of methicillin resistance, the mecA gene, the level (MIC) and mode of expression (homogeneous versus heterogeneous) of antibiotic resistance is not controlled only through the transcription and translation of mecA but can also be profoundly influenced by a variety of environmental factors and by a surprisingly large number of chromosomal genes (auxiliary or fem genes), the primary functions of which are not understood.

Transposon mutagenesis combined with high-resolution biochemical analysis has begun to provide some fascinating insights into the steps of a complex pathway that seems to lead from the mecA gene to the antibiotic resistance phenotype in these bacteria. In the studies described here, we used a new Tn551 library constructed in the background of an MRSA strain with homogeneous and an unusually high methicillin resistance level (MIC, 1,600 ,ug/ml). We selected mutants for which the methicillin MIC was reduced by at least 10-fold. In the characterization of the new TnS5l library, we wished to establish how many of these new mutations were in already identified fem genes and how many involved new and distinct insertion sites. Using physical mapping data we also hoped to provide a reasonable, minimal estimate for the number of genetic elements represented by the new insertion sites. Among the 70 new insertional mutations described here only two were Tn551 inserts inside the mecA gene. Thus, it appears that the overwhelming majority of the new TnS51 mutations are in auxiliary (23) orfem (3) genes, i.e., in the unique genetic determinants needed for the phenotypic expression of optimal (high) levels of methicillin resistance. Previous work has described 12 Tn551 MRSA mutants of this type. The first one of these, (12003, has led to the identification of thefernA gene by Berger-Bachi and colleagues (1, 2). A second mutation (mutant III/8) isolated by John Kornblum (15) was also mapped infemnA (2). Six additional mutations were mapped in the distal part of a second locus, femB (2, 9), and more recently, a mutation was also identified in the open reading frame of femB (14). Two additional mutants, 1H and 12F (15), have led to the identification of genes femC and femD, respectively (4, 11). A single Tn551 mutation in mutant RUSA4 (17) was located inside the mecA gene (16) in the SmaI chromosome fragment F of COL (9). In contrast to these 12 previously described insertional mutations which were located on either SmaI fragment A or I, the large crop of new insertion sites described in this report were scattered over 7 of the 16 SmaI fragments, 48 sites in fragment A (where femA, femB, and femC are located), 4 in fragment B, 1 in fragment C, 1 in fragment D, 2 in fragment E, 1 in fragment F, and 11 in fragment I (where femD resides). Of the 48 inserts on SmaI-A, 27 represented new and distinct insertion sites. None of the 48 inserts were in femrA; 1 was located in the distal part of the femB locus and 11 (and probably 13) were in the femC locus. By using the same arguments used to estimate the numbers of new inserts on the


VOL. 38, 1994

joio

io8

102 10 100 O

.75 1.5 3 6

12

25

50 100200 400800 160V

Methicillin or D-cycloserine (jig/ml) FIG. 3. Effect of a TnSSl mutation on the phenotypic expression of

methicillin resistance and resistance to D-cycloserine. The methicillin resistance phenotypes of the parental MRSA strain COL (O) and its Tn551 mutant RUSA10 (*) were determined by population analysis as described in Materials and Methods. The same method was also used to compare the effect of the mutation in RUSA10 on the expression of resistance against another antibiotic, D-cycloserine, in strain COL (0) and the mutant RUSA10 (0).

SmaI fragment A, the inserts located on SmaI fragments B, C, D, E, F, and I represent an additional group of 14 new and distinct insertional sites (not counting the two inserts inside mecA). For several mutants only two bands were visible (instead of the expected 3) after HindIII digestion and hybridization with TnSSl. This result implies either that the flanking region reaching into the TnS5I insert was deleted or that the missing band was identical in size to that of one of the other two fragments. In one of these cases (RUSA168), HindIll digests were hybridized separately with probes (generated by cloning) containing either the right or the left junction of TnS5l. The results showed that the two hybridizing fragments were of the same size (26). Several sites of the chromosome appeared to represent preferential integration sites ("hot spots") for TnS5l. These are the distal part of femB (7 independent inserts), the femC locus (12 or 14 inserts), the RUSA112 cluster (4 inserts) located on SmaI-A and on a 14-kb EcoRI fragment, and the RUSA315 cluster (6 inserts) on the SmaI-I fragment. The new mutants were isolated through a screen designed to pick up mutants even with relatively modest (10-fold) decrease in resistance. Nevertheless, for many of the mutants with new inserts outside the femA, femB, femC, and femD loci, the methicillin MICs were drastically reduced (3 to 25 ,ug/ml).

2597

How many of these 41 new insertion sites represent new genetic determinants will remain unknown until cloning and sequencing are completed. Nevertheless, one may attempt to make a minimal estimate on the basis of the mapping data alone. Among the new inserts, the one in mutant RUSA270 is in the 4.3-kb EcoRV fragment located upstream of femA (2), and those in the two mutants (RUSA148 and RUSA291) are located in the EcoRV fragments 1.6 and 4.0 kb, respectively, downstream of femB (2). These three mutants must represent at least two new auxiliary genes. The five new inserts on the PstI-beta fragment should represent at least one additional determinant, i.e., a third new auxiliary gene. The 19 new inserts outside of the femC locus must represent at least one additional gene, raising the number of new auxiliary determinants to four. By similar arguments, the new inserts located on SmaaI fragments C, D, E, and F must each represent at least one additional auxiliary determinant, i.e., a total of at least four new genes. There are four new inserts located on SmaI fragment B; three of these (in RUSA281, RUSA262, and RUSA235) have distinct restriction fragmentation patterns. Cell wall analysis has shown that while RUSA281 had a normal muropeptide composition, RUSA262 and RUSA235 each exhibited distinct abnormalities in their peptidoglycans (unpublished data). Therefore, this group of insertional mutants is likely to include at least three new genes. Thus, not even counting the new inserts in SmaI fragment I, a likely and minimal estimate of the number of new auxiliary genes represented by the insertional mutants described here is 11. This added to the already identifiedfemnA,femB,femC, and femD genes raises the number of auxiliary determinants to 15. Why so many genetic determinants outside of the mecA locus are needed for optimizing the methicillin resistance phenotype is not clear at present. It is also unclear how these auxiliary determinants influence the mode of expression and the quantitative level of beta-lactam resistance. In the auxiliary mutants that have already been studied in some detail, the mecA gene was intact and was normally expressed into its gene product, PBP 2A. Thus, the auxiliary genes do not fall in the category of regulatory genes but are perhaps better referred to as antibiotic response genes. Some clues concerning the nature of at least some of the fem genes began to emerge through recent biochemical analysis of the cell wall peptidoglycan in the femA, femB, femC, and femD mutants. The femrA and femB mutants had abnormal peptidoglycan cross bridge structures (5, 6, 7); a femC mutant was shown to be blocked in the amidation of the alpha carboxyl group of the D-glutamic acid residues in the muropeptide subunits (18), and a defect in a gene controlling the rate of biosynthesis of the unsubstituted disaccharide pentapeptide precursor was postulated as a way of explaining the structural abnormalities detected in the cell wall of afemD mutant (6). A model for the mechanism of how such abnormalities of muropeptide structure may lead to reduction in the level of beta-lactam antibiotic resistance has recently been proposed (9). The consistent appearance of heterogeneity among virtually all auxiliary mutants described so far is not well understood. It is possible that in most of these mutants the Tn551 insert is either in the promoter or in the distal part of the open reading frame. This is consistent with the finding that normal (parental) muropeptide species are present (albeit in greatly reduced quantities) in the cell walls of several of the auxiliary mutants analyzed so far. It is indeed possible that insertions in the open reading frames of these genes would be lethal. The auxiliary mutants may indeed represent "methicillin-conditional" mu-

2598


tants in essential genes of staphylococcal peptidoglycan metabolism. ACKNOWLEDGMENTS The investigation described here received partial support from a grant by the U.S. Public Health Service (NIH AI 16794). The excellent technical assistance of Marilyn Chung is acknowledged. We thank Brigitte Berger-Bachi for strains and for making the femA and femA-femB DNA probes available to us. REFERENCES 1. Berger-Bachi, B. 1983. Insertional inactivation of staphylococcal methicillin resistance by Tn551. J. Bacteriol. 154:479-487. 2. Berger-Bachi, B., L. Barberis-Maino, A. Strassle, and F. H. Kayser. 1989. FemA, a host-mediated factor essential for methicillin resistance in Staphylococcus aureus: molecular cloning and characterization. Mol. Gen. Genet. 219:263-269. 3. Berger-Bachi, B., A. Strassle, L. Barberis-Maino, W. Tesch, C. Ryffel, and F. H. Kayser. 1990. FemA, a host-mediated factor essential for methicillin resistance in Staphylococcus aureus, p. 509-520. In R. P. Skurray and R. A. Novick (ed.), Molecular biology of the staphylococci. VCH Publishers, New York. 4. Berger-Bachi, B., A. Strassle, J. E. Gustafson, and F. H. Kayser. 1992. Mapping and characterization of multiple chromosomal factors involved in methicillin resistance in Staphylococcus aureus. Antimicrob. Agents Chemother. 36:1367-1373. 5. De Jonge, B. L. M., Y.-S. Chang, D. Gage, and A. Tomasz. 1992. Peptidoglycan composition of a highly methicillin-resistant Staphylococcus aureus strain. J. Biol. Chem. 267:11248-11254. 6. De Jonge, B. L. M., Y.-S. Chang, D. Gage, and A. Tomasz. 1992. Peptidoglycan composition in heterogeneous TnS51 mutants of a methicillin-resistant Staphylococcus aureus strain. J. Biol. Chem. 267:11255-11259. 7. De Jonge, B. L. M., T. Sidow, Y.-S. Chang, H. Labischinski, B. Berger-Bachi, D. A. Gage, and A. Tomasz. 1993. Altered muropeptide composition in Staphylococcus aureus strains with an inactivated femnA locus. J. Bacteriol. 175:2779-2782. 8. De Lencastre, H., I. Couto, I. Santos, J. Melo-Cristino, A. TorresPereira, and A. Tomasz. 1994. Methicillin-resistant Staphylococcus aureus disease in a Portuguese hospital: characterization of clonal types by a combination of DNA typing methods. Eur. J. Clin. Microbiol. Infect. Dis. 13:64-73. 9. De Lencastre, H., B. L. M. De Jonge, P. R. Matthews, and A. Tomasz. 1994. Molecular aspects of methicillin resistance in Staphylococcus aureus. J. Antimicrob. Chemother. 33:7-24. 10. De Lencastre, H., A. M. S. Figueiredo, C. Urban, J. Rahal, and A. Tomasz. 1991. Multiple mechanisms of methicillin resistance and improved methods for detection in clinical isolates of Staphylococcus aureus. Antimicrob. Agents Chemother. 35:632-639. 10a.De Lencastre, H., and A. Tomasz. Unpublished data. 11. Gustafson, J., A. Strassle, H. Hachler, F. H. Kayser, and B. Berger-Bachi. 1994. The femC locus of Staphylococcus aureus required for methicillin resistance includes the glutamine syn-

ANTIMICROB. AGENTS CHEMOTHER.

thetase operon. J. Bacteriol. 176:1460-1467. 12. Hachler, H., F. H. Kayser, and B. Berger-Bachi. 1987. Homology of a transferable tetracycline resistance determinant of Clostridium difficile with Streptococcus (Enterococcus) faecalis transposon Tn916. Antimicrob. Agents Chemother. 31:1033-1038. 13. Hartman, B. J., and A. Tomasz. 1986. Expression of methicillin resistance in heterogeneous strains of Staphylococcus aureus. Antimicrob. Agents Chemother. 29:85-92. 14. Henze, U., T. Sidow, J. Wecke, H. Labischinski, and B. BergerBachi. 1993. Influence of femB on methicillin resistance and peptidoglycan metabolism in Staphylococcus aureus. J. Bacteriol. 175:1612-1620. 15. Kornblum, J., B. J. Hartman, R. P. Novick, and A. Tomasz. 1986. Conversion of a homogeneously methicillin-resistant strain of Staphylococcus aureus to heterogeneous resistance by Tn551mediated insertional inactivation. Eur. J. Clin. Microbiol. 5:714718. 16. Matthews, P., and A. Tomasz. 1990. Insertional inactivation of the mec gene in a transposon mutant of a methicillin-resistant clinical isolate of Staphylococcus aureus. Antimicrob. Agents Chemother. 34:1777-1779. 16a.Matthews, P. R. Personal communication. 17. Murakami, K., and A. Tomasz. 1989. Involvement of multiple genetic determinants in high-level methicillin resistance in Staphylococcus aureus. J. Bacteriol. 171:874-879. 18. Ornelas-Soares, O., H. de Lencastre, B. L. M. de Jonge, D. Gage, Y.-S. Chang, and A. Tomasz. 1993. The peptidoglycan composition of a Staphylococcus aureus mutant selected for reduced methicillin resistance. J. Biol. Chem. 268:26268-26272. 19. Oshida, T., and A. Tomasz. 1992. Isolation and characterization of a Tn551-autolysis mutant of Staphylococcus aureus. J. Bacteriol. 174:4952-4959. 20. Pattee, P. A. 1981. Distribution of TnS51 sites responsible for auxotrophy on the Staphylococcus aureus chromosome. J. Bacteriol. 145:479-488. 21. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1990. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 22. Shaw, J. H., and D. B. Clewell. 1985. Complete nucleotide sequence of macrolide-lincosamine-streptogramin B-resistance transposon Tn917 in Streptococcus faecalis. J. Bacteriol. 164:782796. 23. Tomasz, A. 1990. Auxiliary genes assisting in the expression of methicillin resistance in Staphylococcus aureus, p. 565-583. In R. P. Skurray and R. A. Novick (ed.), Molecular biology of the staphylococci. VCH Publishers, New York. 24. Tomasz, A., S. Nachman, and H. Leaf. 1991. Stable classes of phenotypic expression in methicillin-resistant clinical isolates of staphylococci. Antimicrob. Agents Chemother. 35:124-129. 25. Tonin, E., and A. Tomasz. 1986. Beta-lactam-specific resistant mutants of Staphylococcus aureus. Antimicrob. Agents Chemother. 30:577-583. 26. Wu, S. Unpublished data.