Feb 6, 1984 - cillin (50 mCi/mmol; Amersham International, Amersham, Bucks) according ..... it has been postulated that many of the pores are closed (Angus.
J. MED. MICROBIO1.-VOL.
18 (1984),261-270
8 1984 The Pathological Society of Great Britain and Ireland
P ENICILLIN-BINDING PROTEINS, P OR IN S A N D O U T E R- MEMBRANE PERMEABILITY O F C AR B E N IC IL L IN RESISTANT A N D -SUSCEPTIBLE S T R AIN S OF P S E U D O M O N A S AERUGINOSA D. M. LIVERMORE Department of Medical Microbiology, The London Hospital Medical College, Turner Street, London E l 2AD
SUMMARY.Reduced cell permeability and target penicillin-binding protein modification were investigated as mechanisms of intrinsic resistance in strains of Pseudomonas aeruginosa resistant to carbenicillin (MIC > 128 mg/L) independently of p-lactamase production. The carbenicillin-resistant strains were also remarkably resistant to other P-lactams, quinolones, tetracyline and chloramphenicol, whereas carbenicillin-hypersusceptible strains (MIC < 2 mg/L) were very sensitive to these antimicrobial compounds. These observations suggested a non-specific mechanism of resistance involving reduced permeability of the outer layers of the bacterial cell. However, carbenicillin-resistant and carbenicillin-sensitive strains had identical porin levels and the target penicillin-binding proteins of carbenicillin-resistant (MIC 256-2048 mg/L), carbenicillin-sensitive (MIC 64 mg/L) and carbenicillin-hypersusceptible (MIC 0.015 mg/L) strains were equally sensitive to p-lactams. Thus, subtle changes in porin function or additional outer-membrane barriers regulating permeability may be involved in intrinsic resistance. INTRODUCTION Between 10% and 15% of strains of Pseudomonas aeruginosa from British hospitals are resistant to carbenicillin (MIC > 128 mg/L) (King et al., 1980; Williams et al., 1984 and in press). Various plasmid-encoded P-lactamases account for resistance in only 20-25% of the carbenicillin-resistant isolates; in the remainder, resistance is termed ‘intrinsic’ (Richmond, 1975; Williams et al., 1984). Target modification and cell impermeability are widely proposed as sources of intrinsic resistance but their relative importance remains undefined. Various modifications of the target penicillin-binding proteins (PBPs) have been shown to be associated with carbenicillin resistance in P . aeruginosa (Curtis et al., 1978; Mirelman, Nuchamowitz, and Rubinstein, 1981; Godfrey, Bryan and Rabin, 1981; Rodriguez-Tebar et al., 1982); nevertheless secondary defences including p-lactamase synthesis and impermeability were also implicated in some of the strains described by these authors. Received 6 Feb. 1984; accepted 16 Mar. 1984. 26 I
262
D . M . LIVERMORE
Because p-lactams act on targets in the cytoplasmic membrane, it is assumed that the major barrier to their penetration is the outer membrane, which is unusually impermeable to small aqueous solutes even in carbenicillin-susceptible strains of P . aeruginosa (Yoshimura and Nikaido, 1982). Beta-lactams cross the outer membrane by diffusion through pores composed of “porin” proteins (Nikaido, 198 1). Reduced porin expression has been associated with intrinsic resistance in Escherichia coli (Jaffe, Chabbert and Semonin, 1982) and Salmonella typhimurium (Nikaido et al., 1977) but such a relationship has not been reported in P . aeruginosa. In the present study, target insensitivity, outer-membrane impermeability and porin deficiency were examined as possible sources of intrinsic resistance to carbenicillin in wild-type strains of P . aeruginosa. MATERIALS AND METHODS Bacterial strains and cultural conditions. Strains of P . aeruginosa were from a previously described collection (Livermore, Williams and Williams, 1981). P. aeruginosa strain Z799/6 1 a hyperpermeable strain (carbenicillin MIC < 0.0 15 mg/L), with fully susceptible PBPs (Zimmermann, 1979and 1980)was used as a control organism. Strains were maintained at 4°C on slopes of Nutrient Agar (Southern Group Media). For PBP and porin studies, bacteria were grown in Antibiotic Medium No. 3 (Difco) in conical flasks incubated at 37°C with orbital shaking. A 3-10% v/v inoculum of overnight broth culture was added to fresh prewarmed (37°C) broth and incubated at 37°C for 5 h to obtain late exponential-phase cultures. Antibiotics and other reagents. The p-lactams carbenicillin (Beecham Research Laboratories), latamoxef (Moxalactam; Eli Lilly Inc.) and cefsulodin (Ciba-Geigy Pharmaceuticals); the quinolones nalidixic acid (Sterling Winthrop) and norfloxacin (Merck, Sharpe and Dohme); chloramphenicol (Parke-Davis), tetracycline (Lederle), polymyxin E (Pharmax Ltd) and gentamicin (Hoechst-Roussel) were obtained as powders and stored at 4°C. Solutions were prepared on the day of use. Other reagents of “Analar” or equivalent grade were obtained from Sigma Chemical Co., St Louis, MO, or BDH, Poole, Dorset. Antibiotic susceptibility tests. MICs of antibiotics were determined by the previously described agar-dilution method of Livermore et al. (198 1); the inoculum contained c. lo4 cfu. Beta-lactamaseproduction.Strains were examined for constitutive B-lactamase expression by the EDTA-nitrocefin technique of Williams et al. (1984). Beta-lactamase inducibility of the strains used in PBP, porin and permeability studies was investigated with cephaloridine or cefoxitin (500 mg/L) as inducers, as previously described (Livermore et al., 1982). In addition, cephaloridine-induced fl-lactamase extracts were clarified by ultracentrifugation ( 100 000 g for 30 min) at 4°C and subjected to isoelectric focusing (Matthew et al., 1975). Target PBP modiJication. Bacterial membranes were prepared by differential centrifugation of sonicated late exponential-phase bacteria, as described by Spratt (1977); the preparations were adjusted to contain 4-6 mg of protein/L. The affinities of the PBP components in the membranes for different fl-lactams were determined by competition assay with 14C-benzylpenicillin (50 mCi/mmol; Amersham International, Amersham, Bucks) according to the method of Spratt (1977). The fluorography period was extended from 60 to 90 days at - 70°C. Extraction and electrophoresis of F porin. Late exponential-phase bacteria were harvested and resuspended 250 g (wet weight)/L in 10 mM phosphate buffer (PH 7.0) containing sodium dodecyl sulphate (SDS) 2% w/v. Bacterial suspensions (2 ml) were briefly sonicated and held at 25°C for 30 rnin before centrifugation (100 000 g for 30 min) at 15°C. The resultant pellet was retained, washed twice, and resuspended in the same SDS-containing phosphate buffer (500 pl), then warmed to 60°C for 10 min to liberate porin from the peptidoglycan (Hancock et al., 1981). After centrifugation, as described above, the supernate was retained and prepared for electrophoresis by heating to 37°C for 10 min with an equal volume of 0.2 M Tris (hydroxymethyl-methylamine) buffer (PH 6.8) containing SDS 4% w/v, p-mercaptoethanollO% v/v, glycerol 20% v/v, and bromophenol blue 0.005% w/v. Higher temperatures and more
INTRINSIC RESISTANCE IN P . AERUGINOSA
263
prolonged heating were occasionally used (see Results). These preparations were then electrophoresed on a discontinuous slab-gel system based on the method of Laemmli and Favre (I 973). The running gel, which contained acrylamide 12% w/v, NN'methylene-bis-acrylamide 0.2% w/v, and SDS 0.1% w/v in 0-375M tris-HC1 buffer (PH 8.8), was polymerised by addition of ammonium persulphate and NNNN'-tetramethylethylenediamine(TEMED) at final concentrations (w/v) of 0.005% and O.Ol%, respectively. The stacking gel contained acrylamide 6% w/v, NN'-methylene-bis-acrylamide 0.06% w/v and SDS 0.1% w/v in 0.125 M Tris-HC1 buffer (PH 6.8); it was polymerised as described above. The running buffer (PH 84-8.5) contained 0.05 M Tris, 0.38 M glycine and SDS 0.1% w/v. Electrophoresis was performed at 15 mA/gel through the stacking gel, and 25 mA/gel through the running gel. Proteins were located by staining overnight in Coomassie Blue G 250 0.1% w/v in methanol: acetic acid:water (50: 10:40); gels were destained with multiple changes of methanol: acetic acid: water (5: 10:85). Estimation of Fporin in whole-cell extracts. Bacteria were harvested from late exponentialphase cultures (100 ml), resuspended in 2 ml of the above SDS-containing phosphate buffer (pH7-0) and heated to 95°C for 10 min to solublise proteins. Insoluble matter was removed by centrifugation (100 OOOg for 30 min) at 12°C and the supernate retained. Samples, containing 5 mg protein/L, were electrophoresed and stained as described for porin preparations. The
1024512256128-I
64-
32v
0 E
168-
Q
4-
.-0
2-
E
l -
;
+-L
s
03-
0.25
-
0.125.
ON%
c 1
1
2
M I C ( mg/L )
1
4
of
1
8
1
16
1
32
carbenicillin
1
64
(and
1
1
1
I
1
128 256 512 1024 2048
number of strains )
FIG. 1.-Relationships between MICs of carbenicillin for P.aeruginosa strains and MICs of: tetracycline, B;-. chloramphenicol, 0-0; nalidixic acid, 0-0; norfloxacin, v- - -v; latamoxef, 0-0; cefsulodin, A-A; polymyxin E, 0- - -0; and gentamicin, A-A.
Carbenicillin
Latamoxef
c-
MICs (mg/L) of:
1 1 1
0.5 0-5
0.015 64 64 128 256 128/256
8 >8 ND >8 >8 ND ND
\
3
4
0.25 2 4 0.06 2 0.125 ND 1 0.25 1 0.25 1 1 0.25 0.25 0.06 0.5 0.06 >8 0.25 0.125 > 6 4 0.125 > 8
516 >8 32 >8 >8 >8 >8 2 2 2 >8 >64 >8
* Competitor concentration (mg/L) required to inhibit 14C-benzylpenicillinbinding by 50%. Latamoxef and cefsulodin were not tested with strains R8, R28 and R39. ND = not detected.
266
D . M . LIVERMORE
FIG.2.-Binding of latamoxef by PBPs of P.aeruginosa: (a), strain Z799/61; and, (b), strain R70. Tracks show PBP-binding of 14C-benzylpenicillinafter pre-exposure of membranes to latamoxef (mg/L): (1) 0, (2) 0.008, (3) 0.015, (4)0.031, (5) 0.062, (6) 0.125, (7) 0.25, (8) 0-5, (9) 1, (10) 2, (I 1) 4,(12) 8 and (13) 0.
observed differenceswere unrelated to levels of resistance. Specimen fluorograms are presented (figs. 2a and 2b). Cell-wall permeability Similar amounts of a protein of mol. wt 37 000 were extracted from all five test strains and from the control strain Z799/61 (fig. 3). With increases in the solubilisation temperature or the time of heating this protein was converted in such a way that it had an apparent mol.wt of 43 000 (e.g., for strain R20, see fig. 3). This finding is in agreement with the description of F porin by Hancock and Carey (1979), the mol. wts of 37 000 and 43 000 corresponding to those of the F and F* forms,
INTRINSIC RESISTANCE IN P . AERUGINOSA
267
FIG. 3.-Behaviour of F porin from strains of P . aeruginosa. From left to right, tracks 1-5 show extracts from strain R20 after heating with electrophoresis buffer for: (1) 30 min at 100°C;(2) 10 min at 100°C; (3) 10 rnin at 80°C; (4) 10 rnin at 60°C; and, ( 5 ) 10 rnin at 37°C. Track 6 contains mo1.-wt markers: phosphorylase B (92 500); bovine-serum albumin (68 000); ovalbumin (45 000); carbonic anhydrase (31 000); and, soyabean-trypsin inhibitor (21 500). Tracks 7-12 contain porin extracts, solubilised for 10 rnin at 37”C, from strains: R70, R8, R39, R28, R20 and Z799/61.
respectively. These extracts were substantially contaminated with a protein of mol. wt 19 500 that was identified as H2 protein which, like F porin, requires hot SDS extraction procedures (Hancock et al., 1981). Contamination with other proteins was minimal. F porin, migrating as a mixture of its F and F* forms, was recognised by its electrophoretic mobility in whole-cell extracts. These extracts were prepared from 24 wild-type strains, including those listed in table I, representing a wide range of antibiotic susceptibilities. A relationship between quantity of porin and carbenicillin resistance was not observed. Again, major interstrain differences were not evident in the many other (c. 100) proteins resolved by this technique. DISCUSSION Carbenicillin resistance in P . aeruginosa more commonly depends on “intrinsic factors” than on carbenicillinase production (Williams et al., 1984) and we have studied the nature of these intrinsic factors. Beta-lactamase-mediated resistance in the present strains was discounted by EDTA-nitrocefin testing, and for the strains studied in detail, by enzyme extraction and electrofocusing. These strains produced only an inducible P-lactamase of alkaline PI which was identified as the cephalosporinase first described by Sabath, Jag0 and Abraham (1965). This enzyme is known to be ineffective against carbenicillin (Rosselet and Zimmerman, 1973; Furth, 1979). Antibiogram studies showed the more intrinsically carbenicillin-resistant strains
268
D . M . LIVERMORE
were also more resistant than sensitive strains to other p-lactams, quinolones, chloramphenicol and tetracycline. It should, perhaps, be mentioned that the latter agents were used solely as probes of resistance and that we do not suggest any role for them in antipseudomonal chemotherapy. Resistance to gentamicin and polymyxin E was unlinked to carbenicillin resistance. This degree of cross-resistance between unrelated antimicrobial agents implied a non-specific mechanism of resistance and argued against specific target modification. PBP-competition assays supported that conclusion; thus, PBPs from strain R70 (carbenicillin MIC 2048 mg/L) were as sensitive to P-lactams as those from strain Z799/6 1 (carbenicillin MIC 0.0 15 mg/L). The apparent absence of PBP 2 from some strains may not have been significant because (i) other workers have failed to detect PBP 2 in P.aeruginosa (Zimmermann, 1980; Mirelman et al., 1981), and (ii) PBP 2 is thermolabile and present in low concentrations in the cell (Noguchi, Matsuhashi and Mitsuhashi, 1979). Authentic PBP 2-deficient gram-negative bacteria grow as round forms (Spratt, 1980); none of the strains examined in this study exhibited this morphology. A non-specific reduction in cell permeability provides the most plausible explanation of broad-spectrum intrinsic resistance and, because P lactams act on cytoplasmicmembrane targets, the outer membrane seemed initially to be the most likely site of such a barrier. Most small aqueous solutes, except perhaps polymyxins and aminoglycosides, cross the outer membrane of P. aeruginosa by diffusion through water-filled pores composed of “porin” protein (Nikaido, 1981; Chopra and Ball, 1982). The principal porin of P. aeruginosa is the F protein (Hancock, Decad and Nikaido, 1979)which, however, was present at similar levels in all strains, regardless of their resistance (fig. 3). This result is in contrast with findings for enteric bacteria for which increased intrinsic resistance has been linked to porin deficiency (Nikaido et a1.,1977; Jaffe et al., 1982). However, there may be more subtle changes associated with F porin in the more resistant strains. In order to reconcile the large quantity of F porin present in P.aeruginosa, and the large size of its pores, with the impermeability of the P.aeruginosa cell, it has been postulated that many of the pores are closed (Angus et al., 1982). The inoperative fraction may be greater in the more resistant strains. P. aeruginosa possesses two other porins, D1 (Hancock and Carey, 1980) and P (Hancock, Poole and Benz, 1982) that seem unlikely to have modulated intrinsic resistance in our tests. Porin D1 is glucose inducible, yet little difference was found in carbenicillin MICs regardless of whether complex media or glucose- or acetate-mineral agar (the latter not inducing D 1) were used for susceptibility tests (unpublished data). Porin P is induced by conditions of phosphate starvation that were not relevant to the present study. Another possible explanation of the porin results is that afforded by Scudamore and Goldner (1982) who contended that P-lactam permeation in P. aeruginosa was regulated by an additional barrier internal to the outer membrane. Support for that hypothesis comes from preliminary studies with the “modified crypticity” technique (Zimmermann and Rosselet, 1977)which failed to detect resistance-related variation in outer-membrane permeability (D. M. Livermore, unpublished results). However, these assays are of low sensitivity (Yoshimura and Nikaido, 1982) and sufficient P-lactamase was obtained by inducing the bacteria with cephaloridine or 6 amino-penicillanic acid, a procedure which might have impaired outer-wall structure (Furth, 1979). Accordingly, these findings are still tentative.
INTRINSIC RESISTANCE IN P. AERUGINOSA
269
Function of this additional layer could explain the anomalies which are apparent between some other studies. Zimmermann (1980) found that P . aeruginosa strain Z799/61 had PBPs which were almost totally accessible to radiolabelled p-lactams, whereas the PBPs of its parent strain (1000 times more resistant to carbenicillin) were protected by a considerable permeability barrier. Nicas and Hancock (1983), however, found that the outer-membrane permeability coefficient for strain Z799/6 1 was only six-fold greater than that for its parent strain. Thus, the “intrinsic” type of carbenicillin resistance most frequently encountered in clinical strains of P . aeruginosa did not involve p-lactamase or target PBP modification. The extent to which increased carbenicillin resistance also enhanced resistance to unrelated antimicrobials suggested a non-specific mechanism involving reduced permeability of the outer layers, not necessarily the outer membrane itself. This reduced permeability did not involve reduced porin expression I am grateful to the In-patient X-Ray Department of The London Hospital and the Photographic Department of The London Hospital Medical College for their assistance in preparing illustrations. REFERENCES Angus B L, Carey A M, Caron D A, Kropinski A M B, Hancock R E W 1982 Outer membrane permeability in Pseudomonas aeruginosa: comparison of a wild-type with an antibioticsupersusceptible mutant. Antimicrobial Agents and Chemotherapy 2 1:299-309. Chopra I, Ball P 1982 Transport of antibiotics into bacteria. Advances in Microbial Physiology 23 183-240. Curtis N A C, Brown C, Boxall M, Boulton M G 1978 Modified peptidoglycan transpeptidase activity in a carbenicillin-resistant mutant of Pseudomonas aeruginosa 18s. Antimicrobial Agents and Chemotherapy 14:246-25 1. Furth A 1979 The p-lactamases of Pseudomonas aeruginosa. In: Hamilton-Miller J M T, Smith J T (eds) Beta-lactamases. Academic Press, London, pp 403-428. Godfrey A J, Bryan L.E, Rabin H R 1981 Q-lactam-resistant Pseudomonas aeruginosa with modified penicillin-binding proteins emerging during cystic fibrosis treatment. Antimicrobial Agents and Chemoiherapy 19:705-711. Hancock R E W, Carey A M 1979 Outer membrane of Pseudomonas aeruginosa: heat- and 2-mercaptoethanol-modifiable proteins. Journal of Bacteriology 140:902-9 10. Hancock R E W, Decad G M, Nikaido H 1979 Identification of the protein producing transmembrane diffusion pores in the outer membrane of Pseudomonas aeruginosa PA 01. Biochimica et Biophysica Acta 554: 323-331. Hancock R E W, Carey A M 1980 Protein D1-A glucose-inducible, pore-forming protein from the outer membrane of Pseudomonas aeruginosa. FEMS Microbiology Letters 8 : 105-109. Hancock R E W, Irvin R T, Costerton J W, Carey A M 1981 Pseudomonas aeruginosa outer membrane: peptidoglycan associated proteins. Journal of Bacteriology 145:628-631. Hancock R E W, Poole K, Benz R 1982 Outer membrane protein P of Pseudomonas aeruginosa: regulation by phosphate deficiency and formation of small anion-specific channels in lipid bilayer membranes. Journal of Bacteriology 15:730-738. Jaffe A, Chabbert Y A, Semonin 0 1982 Role of porin proteins Omp F and Omp C in the permeation of p-lactams. Antimicrobial Agents and Chemotherapy 22 :946-948. King J D, Farmer T, Reading C, Sutherland R 1980 Sensitivity to carbenicillin and ticarcillin and the Q-lactamases of Pseudomonas aeruginosa in the U K in 1978-79. Journal of Clinical Pathology 33 :297-30 1. Laemmli U K, Favre M 1973 Maturation of the head of bacteriophage T4. 1. DNA packaging events. Journal of Molecular Biology 80 :575-599. Livermore D M, Williams R J, Williams J D 1981 Comparison of the Q-lactamase stability and the in uitro activity of cefoperazone, cefotaxime, cefsulodin, ceftazidime, moxalactam and
270
D . M . LIVERMORE
ceftriaxone against Pseudomonas aeruginosa. Journal of Antimicrobial Chemotherapy 8 :323-331. Livermore D M, Williams R J, Lindridge M A, Slack R C B, Williams J D 1982 Pseudomonas aeruginosa isolates with modified beta-lactamase inducibility: effects on beta-lactam sensitivity. Lancet 1: 1466-1467. Lowry 0 H, Rosebrough N J, Farr A L, Randall R J 1951 Protein measurement with the Folin phenol reagent. Journal of Biological Chemistry 193:265-275. Matthew M, Harris A M, Marshall M J, Ross G W 1975 The use of analytical isoelectric focusing for detection and identification of 8-lactamases. Journal of General Microbiology 88: 169-178. Mirelman D, Nuchamowitz Y, Rubinstein E 1981 Insensitivity of peptidoglycan biosynthetic reactions to 8-lactam antibiotics in a clinical isolate of Pseudomonas aeruginosa. Antimicrobial Agents and Chemotherapy 19:687-695. Nicas T I, Hancock R E W 1983 Pseudomonas aeruginosa outer membrane permeability: isolation of a porin protein F-deficient mutant. Journal of Bacteriology 153:28 1-285. Nikaido H, Song S A, Shaltiel L, Nurminen M 1977 Outer membrane of Salmonella XIV. Reduced transmembrane diffusion rates in porin deficient mutants. Biochemical and Biophysical Research Communications 76 :324-330. Nikaido H 1981 Outer membrane permeability of bacteria: resistance and accessibility of targets. In: Salton M, Shockman G D (eds) P-Lactam antibiotics: mode of action, new developments and future prospects. Academic Press, New York, pp 249-260. Noguchi H, Matsuhashi M, Mitsuhashi S 1979 Comparative studies of penicillin-binding proteins in Pseudomonas aeruginosa and Escherichia coli. European Journal of Biochemistry 100:41-49. Richmond M H 1975 Antibiotic inactivation and its genetic basis. In: Brown M R W (ed) Resistance of Pseudomonas aeruginosa. John Wiley, London, pp 1-33. Rodriguez-Tebar A, Rojo F, Damaso D, Vazquez D 1982 Carbenicillin resistance of Pseudomonas aeruginosa. Antimicrobial Agents and Chemotherapy 22 :255-26 1. Rosselet A, Zimmermann W 1973 Mutants of Pseudomonas aeruginosa with impaired p-lactamase inducibility and increased sensitivity to p-lactam antibiotics. Journal of General Microbiology 76 :455-457. Sabath L D, Jag0 M, Abraham E P 1965 Cephalosporinase and penicillinase activities of a 8-lactamase from Pseudomonas pyocyanea. Biochemical Journal 96 :739-752. Scudamore R A, Goldner M 1982 Limited contribution of the outer-membrane penetration barrier towards intrinsic antibiotic resistance in Pseudomonas aeruginosa. Canadian Journal of Microbiology 28 : 169- 175. Spratt B G 1977 Properties of the penicillin-binding proteins of Escherichia coli K 12. European Journal of Biochemistry 72: 341-352. Spratt B G 1980 Biochemical and genetical approaches to the mechanism of action of penicillin. Philosophical Transactions of the Royal Society (London) 289B :273-283. Williams R J, Lindridge M A, Said A A, Livermore D M, Williams J D National survey of antibiotic resistance in Pseudomonas aeruginosa. Journal of Antimicrobial Chemotherapy, in press. Williams R J, Livermore D M, Lindridge M A, Said A A, Williams J D 1984 Mechanisms of beta-lactam resistance in British isolates of Pseudomonas aeruginosa. Journal of Medical Microbiology, 17:283-294. Yoshimura F, Nikaido H 1982 Permeability of Pseudomonas aeruginosa outer membrane to hydrophilic solutes. Journal of Bacteriology 152 :636-642. Zimmermann W 1979 Penetration through the Gram-negative cell wall: a co-determinant of the efficacy of beta-lactam antibiotics. International Journal of Clinical Pharmacology and Biopharmacy 17 : 13 1- 134. Zimmermann W 1980 Penetration of p-lactam antibiotics into their target enzymes in Pseudomonas aeruginosa: comparison of a highly sensitive mutant with its parent strain. Antimicrobial Agents and Chemotherapy 18 :94- 100. Zimmermann W, Rosselet A 1977 Function of the outer membrane of Escherichia coli as a permeability barrier to beta-lactam antibotics. Antimicrobial Agents and Chemotherapy 12:368-372.
