![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Antimicrobial Agents and Chemotherapy, September 2002, p. 2901-2907, Vol. 46, No. 9
0066-4804/02/$04.00+0 DOI: 10.1128/AAC.46.9.2901-2907.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.
Alain Dubus, Bernard Joris, and Jean-Marie Frère*
Laboratoire d'Enzymologie and Centre d'Ingénierie des Protéines, Université de Liège, Institut de Chimie, B-4000 Liège, Belgium
Received 19 October 2001/ Returned for modification 7 February 2002/ Accepted 21 May 2002
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On this basis, a model which allowed a quantitative prediction of the MICs for gram-negative bacteria was developed and applied with success to Escherichia coli and Enterobacter cloacae (3, 5, 25). However, as demonstrated by Livermore and Davy (18), this model was not applicable to Pseudomonas aeruginosa, one of the most common opportunistic pathogens. This was due to efflux pumps which also significantly contribute to intrinsic ß-lactam resistance in P. aeruginosa (15, 16, 19, 22). This was unexpected, because ß-lactam targets are located outside the cytoplasmic membrane and also because in E. coli this mechanism does not produce such a high and broad level of resistance (21). Li et al., pointed out that the low outer membrane permeability of P. aeruginosa renders this efflux mechanism particularly efficient (15, 16). In fact, while the few permeability coefficients (P) published for P. aeruginosa are low, they are also highly variable. For cephaloridine and nitrocefin, Nikaido's group reported P values of 10-6 and 6 x 10-7 cm/s, respectively (33), but others have measured values equal to or less than 2 x 10-8 cm/s (9, 14). All these values were determined on the basis of the method of Zimmermann and Rosselet (35), in which the hydrolysis of ß-lactam by intact cells is compared to that obtained with a lysate. The analysis is then based on the assumption that a steady state is rapidly established in the periplasm, where the rate of penetration of the antibiotic which obeys Fick's law of diffusion is equal to that of the ß-lactamase-catalyzed hydrolysis, which follows Henri-Michaelis kinetics. If the validity of this assumption has been demonstrated experimentally (2), the method is not accurate when ß-lactamase activity is present in the medium or on the cell surface. Under these conditions, the error introduced remains negligible if the outer membrane permeability is high, as with E. coli (17), but when the permeability is low, as with P. aeruginosa, the correction factors range from 4 to 10 depending on the experimental conditions (9), and in addition it should be carefully verified that steady-state conditions really prevail. It is therefore not surprising that only a few P values have been determined for Enterobacter and Pseudomonas species.
In the present study, we have developed a new method based on the properties of a soluble, periplasmic high-affinity penicillin-binding protein (PBP) which can easily be used to measure low outer membrane permeability coefficients. This PBP is the C-terminal domain of the BlaR penicillin sensory transducer involved in ß-lactamase induction in Bacillus licheniformis (M. Swinnen, S. Lepage, A. Brans, B. Granier, J. M. Frere, and B. Joris, submitted for publication).
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Plasmids pDML309 and pDML310 were constructed by inserting the 1.2-kb HindIII-BalI fragment of plasmid pRTW8 coding for the 256 residues of the C-terminal domain of the BlaR penicillin sensory transducer of B. licheniformis (BlaR-CTD) in plasmid pIN-III-ompA (28). This vector allows periplasmic expression of proteins. It contains the strong lppp promoter of the E. coli lipoprotein controlled by the lac-UV5 promoter-operator. The E. coli ompA signal peptide is under the control of the two latter promoters and a polylinker is present just at the end of the ompA signal peptide sequence. The ampicillin resistance gene of pIN-III-ompA was replaced by the tetracycline and kanamycin resistance genes, yielding pDML309 and pDML310, respectively (Swinnen et al., submitted). The resulting plasmids were used to transform E. cloacae AD2 and E. aerogenes 008-4, respectively. Plasmid pBL2 was used to produce BlaR-CTD in the periplasm of P. aeruginosa. It was constructed by ligating the 5.8-kb PstI-HpaI fragment of plasmid pDML309 to the 5.9-kb PstI-PvuII fragment of pKT240 (1) (purchased from the Laboratorium voor Moleculaire Biologie Plasmidencollectie culture collection, University of Ghent, Ghent, Belgium).
Antibiotics. The following compounds were kind gifts of various companies: carbenicillin from Beecham (Brentford, United Kingdom), cefotaxime from Hoechst-Roussel (Romainville, France), cephaloridine from Eli-Lilly (Indianapolis, Ind.), aztreonam from Squibb (Princeton, N.J.), benzylpenicillin from Rhône-Poulenc (Paris, France), ampicillin from Bristol-Myers Squibb (Brussels, Belgium), nitrocefin from Oxoid (Basingstoke, United Kingdom), and [14C]cephacetrile (4.5 mCi/mmol) from Ciba-Geigy (Basel, Switzerland).
3H- and 14C-labeled benzylpenicillin (19 Ci and 59 mCi/mmol, respectively) were purchased from Amersham (Little Chalfont, United Kingdom), and tetracycline and kanamycin were purchased from Sigma (St. Louis, Mo.) and Merck (Darmstadt, Germany), respectively.
MIC determination. MICs were determined in Isosensitest broth (Oxoid) with serial doubling dilutions of the antibiotics. For each antibiotic concentration, an inoculum of 2 x 104 CFU taken from a growing culture in Isosensitest broth (A600 = 1), was added to 2 ml of medium. Results were recorded after a 24-h incubation at 37°C with shaking (200 rpm). The MIC was taken as the minimal antibiotic concentration inhibiting bacterial growth.
Measurement of benzylpenicillin hydrolysis. An aliquot of cells was centrifuged at 5,000 x g and 4°C for 3 min. A 10-µl aliquot of the supernatant was applied to a silica thin-layer chromatography plate (Merck), and benzylpenicillin was separated from benzylpenicilloic acid using chloroform-methanol-acetic acid (88:10:2, vol/vol/vol) as a solvent.
SDS-PAGE and fluorography. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and fluorography were performed according to standard methods (10, 13).
Cell fractionation and labeling with fluorescent antibiotic. Enterobacter cells were harvested by centrifugation when the culture had reached an A600 value of 0.8; resuspended in 30 mM Tris-HCl, pH 8, containing 27% sucrose, 5 mM EDTA, and lysozyme (0.5 mg/ml); and incubated at 4°C for 20 min. The periplasmic fraction was separated from the spheroplasts by centrifugation at 30,000 x g for 15 min. To lyse the spheroplasts, they were resuspended in 10 mM Tris-HCl, pH 8, containing 10 mM MgCl2 and DNase (20 µg/ml). The lysed spheroplasts are referred to hereafter as the cellular fraction.
For P. aeruginosa, the periplasmic fraction was isolated by three freeze-thaw cycles.
Labeling of the periplasmic and cellular fractions was performed with 5'-fluoresceyl-glycyl-6-aminopenicillanic acid (5'-Flu-Gly-6APA) as previously described (11).
Determination of ß-lactamase content of cells. Aliquots of cells growing in Isosensitest broth were withdrawn after various time intervals. A portion (50%) of each aliquot was filtered on a Millipore GVWP filter (pore size, 0.2 µm) for dry weight measurement.
The other 50% was kept in dry ice until sonication on ice (three 30-s bursts; amplitude, 6 µm). The ß-lactamase activity was determined in 10 mM HEPES buffer (pH 8.2) and at 30°C with 115 µM nitrocefin as a substrate.
Determination of outer membrane permeability. Cells were grown overnight in Luria-Bertani broth containing kanamycin or tetracycline (50 µg/ml) as plasmid selection agents. The resulting culture was diluted 100-fold into the appropriate culture medium without any isopropyl-ß-D-thiogalactopyranoside or antibiotic, and growth was resumed at 37°C up to an A600 of 1.5. Cells were harvested by centrifugation at 15 to 20°C and resuspended in the appropriate prewarmed medium. The assay was started 3 to 5 min after equilibration at the desired temperature. A sample was also taken for dry weight determination.
(i) Determination of outer membrane permeability by the method of Zimmermann and Rosselet. A portion of cell suspension (cell density [dry weight] between 0.05 and 0.15 mg/ml) was sonicated on ice (three 30-s bursts; amplitude, 6 µm) and used for determination of maximum rate (Vmax). The remaining part of the intact cell suspension was mixed with 400 to 800 µM cephacetrile or 250 to 1,000 µM cephaloridine and transferred to a cuvette with a 1-mm light path, and the absorbance at 260 nm was recorded with the help of an Uvikon 860 device (Kontron Instruments).
(ii) Determination of outer membrane permeability by the new method. (a) Preliminary comment. The protocol described here is optimized for 3H- or 14C-labeled benzylpenicillin and [14C]cephacetrile. However, it can probably be adapted according to the properties of the radiolabeled antibiotic. Optimal results are obtained if BlaR-CTD labeling can be quenched as fast as possible and if the excess of free antibiotic can be almost completely removed.
(b) Determination of outer membrane permeability if the studied antibiotic is radiolabeled. Nine hundred microliters of cells (cell density [dry weight], 5 mg/ml) in the appropriate prewarmed medium was mixed with 100 µl of prewarmed antibiotic (10 µM [3H]benzylpenicillin [see Fig. 1]; 4.5 to 18 µM [14C]benzylpenicillin [see Fig. 3]; 1.5 µM [14C]cephacetrile). Aliquots of 100 µl were withdrawn after various intervals and mixed to 25 µl of 5 N HCl to instantaneously stop the binding to BlaR-CTD. The excess of free radiolabeled antibiotic was eliminated by three extractions with the same volume of water-saturated butanol. The proteins were then precipitated by adding 25 µl of a mixture of 50% trichloroacetic acid-3.75% tungstosilicic acid (Fluka, Buchs, Switzerland). Samples were left on ice for 10 min and then centrifuged 45 min at 15,000 x g. The supernatant was removed and the tubes were washed once with water to remove the last traces of radiolabeled antibiotic. After centrifugation, the pellet was solubilized in 10% SDS and the radioactivity was determined with a scintillation counter.
|
|
Nine hundred microliters of cells (cell density [dry weight], 20 to 30 mg/ml) in the appropriate prewarmed medium was mixed with 100 µl of prewarmed nonradioactive antibiotic. After different time intervals, 100-µl samples were removed and added with 50 µl of 45 µM [14C]cephacetrile. After 2 min of incubation at 25°C, the reaction was stopped and the samples were treated as above.
The antibiotic concentrations used for permeability measurements were as follows: 7.5, 15, and 30 µM cefotaxime; 50, 100, and 200 µM carbenicillin; 5, 10, and 20 µM ampicillin; and 75, 100, and 200 µM aztreonam. The longest incubation time was 45 min.
Model and simulation.
Using the new method, the situation that prevails during the measurement of the outer membrane permeability for a ß-lactamase producing strain is represented by the following model:
 |
The equations describing the variations of Ip, E1I*, and E2I* are
 |
 |
 |
Simulations were performed on the basis of these equations with the help of a numerical integration program based on a Runge Kutta method of the fourth order (6).
For ß-lactamase-deficient strains, the same model was used with [E2] and [E2I*] equal to zero.
Calculation of the permeability coefficient (P) by the new method.
Since kd is equal to P · A/Vp and [E1I*] is equal to nE1I*/Vp, where Vp is the periplasmic volume and nE1I* is the quantity of labeled BlaR-CTD (in picomoles)
 | (1) |
A is equal to 132 cm2 · mg (dry weight)-1, and the unit for [Ie] is nanomolar. d(nE1I*)/dt can be obtained by multiplying the slope of the observed labeling versus time (expressed in counts per minute second-1 · milligram [dry weight]-1) by the specific activity of the labeled antibiotic (0.0096 pmol · cpm-1 for [14C]benzylpenicillin and 0.123 pmol · cpm-1 for [14C]cephacetrile as estimated using [14C]hexadecane [1.1 µCi/g] as a standard for estimating the counting yield).
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
 | (2) |
Simulations were done keeping E1tot between 10 and 40 µM, Ie between 0.5 and 100 µM, and kd at <0.26 s-1. In all cases the acylation of E1 up to 70% saturation (E1I*/E1tot = 0.7) was analyzed by linear regression. The kd values calculated on the basis of equation 2 were then compared to those used in the simulation. From these analyses (data not shown), it appears that diffusion can be considered limiting if
 | (3) |
This relation gives useful information on the limitations of the method. For example, it will be difficult to study the outer membrane permeability of E. coli because most of its kd values are larger than 0.13 s-1 (25), and on the basis of equation 3, it would thus be necessary to produce a periplasmic PBP with kf values larger than 30,000 M-1 s-1 (if we assume E1tot = 40 µM). On the contrary, this method applies well to P. aeruginosa because all reported kd values are below 0.13 s-1. BlaR-CTD is therefore an ideal PBP for such studies, because the kf values for various ß-lactams are between 103 and 106 M-1 s-1 (Swinnen et al., submitted). Note that if the value of kf1 for the studied compound is not known, it can be easily measured to demonstrate that diffusion is the rate-limiting factor (equation 3).
ß-Lactamase production is another important factor, as these enzymes compete with PBPs for the free antibiotic. Consequently this method is not applicable to ß-lactamase-overproducing strains. For ß-lactamase-inducible strains, which produce low levels of enzyme in the noninduced state, the validity of the method will depend on (i) the relative concentration of the two enzymes, (ii) their relative affinity for the antibiotic, and (iii) the stability of the ß-lactam against hydrolysis.
Development of the method. Because some ß-lactams might cross the cytoplasmic membrane (15), the cellular localization of BlaR-CTD was analyzed. Periplasmic and cellular fractions were added with 5'-Flu-Gly-6APA, and the labeling was assessed under UV after SDS-PAGE. As expected from the plasmid construction, more than 95% of BlaR-CTD is found in the periplasm, and only a small amount of precursor is detected in the cellular fraction (result not shown). The influence of antibiotics on outer membrane integrity is still largely unknown, and therefore fresh cultures were grown without selective pressure. Under these conditions, 5 to 15% of cells lose their plasmid. Although these cells influence the dry weight but do not participate in the permeability measurement, no correction was introduced when calculating kd.
Incubation of E. aerogenes 008-4 intact cells with [3H]benzylpenicillin shows that they bind more ß-lactam if they express BlaR-CTD (Fig. 1). Some protein-bound radioactivity is released into the medium during the assay, but it only represents a small percentage of the cell-bound counts and is not expected to influence the slope of the line over the first 15 min. The binding appears to be a three-step process. First, a rapid step of small amplitude occurs during the mixing time (i.e., t = 0 [5 to 10 s in Fig. 1 ]), followed by a second step of larger amplitude over a 15-min time scale and finally a third step with a small amplitude on a larger time scale. The same experiment analyzed by fluorography shows that during the first 15 min, when the highest binding rate is observed, only BlaR-CTD is acylated (Fig. 2). Acylation of the membrane-bound PBPs only takes place during the third step of the experiment when BlaR is fully labeled. Because the total amount of BlaR-CTD is constant during the experiment and no significant release into the medium occurs during that time, the labeling observed over the first 15 min after mixing is diffusion dependent and can be used to measure the outer membrane permeability. The binding which occurs during the mixing time remains unexplained, since in the case of E. aerogenes 008-4 devoid of plasmid, no labeled protein is detected when samples are directly denatured and analyzed by SDS-PAGE. This labeling is probably due to nonspecific adsorption or trapping of the antibiotic because of the high cell density and acidic treatment. When pDML310 is present in the cells, in addition to this phenomenon, some BlaR-CTD might be freely accessible because it is inserted into the outer membrane with its active site directed to the medium, as already observed with ß-lactamases (9). Moreover, whatever its cause, this phenomenon does not alter the linearity of BlaR-CTD labeling during the second phase (Fig. 3). For the cells which do not produce BlaR-CTD, the slow labeling is due to the membrane bound PBPs.
|
For unlabeled antibiotics, the method is only applicable by counterlabeling with a radioactive compound. Of course, the result is only valid if the diffusion rate of the labeled antibiotic (i.e., kd · [Ie]) is at least 10 times larger than that of the unlabeled ß-lactam. Due to its high P value, cephacetrile is adequate in this case. Using the counterlabeling method with [14C]cephacetrile, the P value of E. aerogenes 008-4 for benzylpenicillin was found to be 3.1 x 10-8 cm/s. This value is close to that measured directly with radiolabeled benzylpenicillin, demonstrating that counterlabeling with cephacetrile is accurate. Therefore, the permeability coefficients of E. aerogenes 008-4 and E. cloacae AD2 for ampicillin, carbenicillin, cefotaxime and aztreonam were measured at 25°C (Fig. 4; Table 1). In all cases, the outer membrane permeability of E. aerogenes is lower than that of E. cloacae by a factor of about 2.
|
|
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
By our method, we have measured the outer membrane P values for two Enterobacter strains and six antibiotics. It should be noted that whatever the species, P values for benzylpenicillin, ampicillin, carbenicillin, and aztreonam had never been measured on whole cells. As shown in Table 1, with the exception of cephaloridine and cephacetrile, the P values do not exceed 10-7 cm/s. Also note that the permeability for aztreonam is the lowest ever measured and this antibiotic is the only one for which the half-equilibration times are larger than the generation times of the bacteria. This very low diffusion rate can also invalidate the method of Zimmermann and Rosselet, since the steady-state assumption is not verified. It also appears that the outer membrane of E. cloacae is two- to threefold more permeable than that of E. aerogenes, and as reported for other Enterobacter species, P values are lower than those of E. coli (3, 6). For cephaloridine, cephacetrile, carbenicillin, and benzylpenicillin, E. cloacae AD2 is 20-to 60-fold less permeable than E. coli SNO3. For cefotaxime, aztreonam, and ampicillin, E. cloacae AD2 is, respectively, 400-, 1,000-, and 700-fold less permeable than E. coli. In decreasing order of permeation rate, the classification for E. cloacae is cephaloridine, cephacetrile, ampicillin, benzylpenicillin, cefotaxime, carbenicillin, and aztreonam. These results suggest that the porins of these Enterobacter strains have a different selectivity compared to those of E. coli. They seem to be more restrictive and/or more affected by the drug structure. Note that the only obvious common characteristic of ampicillin, cefotaxime, and aztreonam, the compounds for which the largest difference is observed between E. coli and Enterobacter, is the presence of a free amine on the side chain (in C-6 for the penam, C-7 for the cephem, and C3 for the monobactam). Whether or not this could influence the diffusion through Enterobacter porins remains to be established or confirmed. A similar wider dispersion of the P values in Serratia marcescens when compared to E. coli has been reported by Raimondi et al. (27).
At 37°C, the P value of E. cloacae AD2 for benzylpenicillin does not significantly increase when compared to that at 25°C, whereas for E. aerogenes 008-4 and cefotaxime, benzylpenicillin, and aztreonam, these values increase by factors of 1.8, 2.4, and 3.5, respectively (data not shown). Consequently it appears that the studied ß-lactams mostly cross the outer membrane through porins.
Because the measured diffusion rates are very low, one can wonder if active efflux mechanisms impair the measurement of the P values. The two ß-lactamase-deficient strains studied here are derived from ß-lactamase overproducing strains obtained by a multistep selection process. Based on recent results, one might suppose that these four strains possess an increased efflux activity that can compete with BlaR-CTD for the periplasmic ß-lactam. Various data do not favor this hypothesis. First, E. cloacae 908S (the wild-type ß-lactamase-inducible strain) and E. cloacae 908R (the ß-lactamase-overproducing strain) show the same level of resistance to various non-ß-lactam antibiotics such as chloramphenicol, norfloxacin, nalidixic acid, or erythromycin, and these levels are the same as those observed for Salmonella wild-type strains (24, 31). Moreover, the P for benzylpenicillin, a ß-lactam sensitive to efflux systems, is the same for E. aerogenes 006, the wild-type strain, and E. aerogenes 008-4. Therefore, the efflux activity of the ß-lactamase-deficient and -overproducing strains used in this study is not significantly different from that of the wild-type strains.
To demonstrate that, in the cases we studied, the basal efflux system does not interfere with the method, we have tried to predict the MICs (25) for the ß-lactamase-overproducing strains on the basis of the following equation: MIC = Ipl + Vmax · Ipl/[P · A · (Km + Ipl)]. Because its ß-lactamase is still partially inducible (data not shown), the prediction was not done with E. aerogenes 008. In Table 3, the lethal periplasmic concentration of ß-lactam (Ipl) is the MIC for E. cloacae AD2. These values are higher than expected on the basis of the PBP sensitivity, but as discussed previously (12), except for cephaloridine, this has little impact on the MICs for E. cloacae 908R, because they are much larger than the corresponding Kms, and therefore the Vmax·Ipl/P · A · (Km + Ipl) term simplifies to Vmax/P · A. This "abnormal" resistance of E. cloacae AD2 probably results from an efflux system naturally expressed in this strain or from the selection of a mutant overproducing an efflux system during MIC measurements. How ever, while efflux certainly contributes to increase the MICs for AD2, it is unlikely to affect the ß-lactam resistance of 908R (compare the MICs for the isogenic AD2 and 908R strains). If efflux was interfering with P measurement, it would certainly decrease the periplasmic concentration of antibiotic and thus the rate of BlaR-CTD acylation. This would result in underestimated P values and overestimated predicted MICs, which is not the case except for cephaloridine, for which P was determined by the method of Zimmerman and Rosselet, and benzylpenicillin, but in the latter case the MIC might be higher as higher antibiotic concentrations have not been tested. Otherwise the predicted and observed MICs are in fair (benzylpenicillin and carbenicillin) to good (all others) agreement.
|
Thanks to the progress in genetic engineering methods, it is now easy to selectively inactivate chromosomal ß-lactamases or efflux systems in gram-negative bacteria. Therefore, we think that despite some limitations, the method described here will be of great interest to more accurately measure low outer membrane permeabilities encountered in some genera and/or resistant mutants.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Present address: Centre de Recherches en Neurobiologie Cellulaire et Moleculaire, Université de Liège, B-4020 Liège, Belgium.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Clin. Vaccine Immunol. | Clin. Microbiol. Rev. |
|---|---|
| J. Clin. Microbiol. | ALL ASM JOURNALS |
, cellular fraction from cells producing BlaR-CTD; 
, supernatant fraction from cells producing BlaR-CTD; 
, cellular fraction from nontransformed cells.
